Donate
Independent, objective, nonpartisan research

R 703JZR

Authors

R 703JZR

Tagged with:

Publication PDFs

Database

This is the content currently stored in the post and postmeta tables.

View live version

object(Timber\Post)#3711 (44) { ["ImageClass"]=> string(12) "Timber\Image" ["PostClass"]=> string(11) "Timber\Post" ["TermClass"]=> string(11) "Timber\Term" ["object_type"]=> string(4) "post" ["custom"]=> array(5) { ["_wp_attached_file"]=> string(12) "R_703JZR.pdf" ["wpmf_size"]=> string(6) "793444" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(185486) "High-Tech Start-Ups and Industry Dynamics in SiliconValley ••• Junfu Zhang 2003 PUBLIC POLICY INSTITUTE OF CALIFORNIA Library of Congress Cataloging-in-Publication Data Zhang, Junfu, 1970-. High-tech start-ups and industry dynamics in Silicon Valley / Junfu Zhang. p. cm. Includes bibliographical references. ISBN: 1-58213-074-4 1. High technology industries—California—Santa Clara Valley (Santa Clara County)—Longitudinal studies. 2. Entrepreneurship—California—Santa Clara Valley (Santa Clara County)—Longitudinal studies. I. Title. HC107.C23H5394 2003 338.4'76'0979473—dc21 2003012491 Copyright © 2003 by Public Policy Institute of California All rights reserved San Francisco, CA Short sections of text, not to exceed three paragraphs, may be quoted without written permission provided that full attribution is given to the source and the above copyright notice is included. PPIC does not take or support positions on any ballot measure or state and federal legislation nor does it endorse or support any political parties or candidates for public office. Research publications reflect the views of the authors and do not necessarily reflect the views of the staff, officers, or Board of Directors of the Public Policy Institute of California. Foreword The Bay Area economy is experiencing one of its most prolonged recessions: Unemployment continues to climb, start-ups in Silicon Valley have declined from over 3,500 a year in 1998 to well under 1,000 in recent years, and, nationwide, the high-tech sector appears to be facing a future of excess capacity. These are certainly sufficient reasons for the general mood of gloom that has settled over a region that was recently the focus of international attention for its high-tech successes. Why this dramatic turnaround in the economy of Silicon Valley? What are the prospects that the region will be booming once again? High-Tech Start-Ups and Industry Dynamics in Silicon Valley by Junfu Zhang is yet another contribution by PPIC to an improved understanding of the California economy. This research project is one of a series that PPIC has launched to gain a better understanding of California’s new economies and of the dynamic processes that underlie their cycles of boom and bust. Past PPIC studies have looked at the role of immigrant entrepreneurs and their linkage to Asia, the role of U.S. tariff policy and its effect on increasing export activity, and the role of exports and foreign direct investment in building California’s economy for future decades. Zhang’s research concludes that, collectively, new firms represent a major force in the economic dynamics of Silicon Valley. For example, firms founded after 1990 created almost all of the job growth experienced by Silicon Valley between 1990 and 2001. Why, then, do we find ourselves in the midst of the current bust cycle? The theory most applicable to the current situation was developed by Joseph Schumpeter in 1911. In The Theory of Economic Development, he explained, “The economic system does not move along continually and smoothly. Countermovements, setbacks, incidents of the most various kinds occur, which obstruct the path of development; there are breakdowns in the economic value system which interrupt it.” And, he argued, these setbacks iii lead to the development of new ideas, new entrepreneurs rise to the occasion, and soon the cycle begins all over again. The cycle of firm startups, closures, and new start-ups is very much part of the economic development process, and the very entrepreneurs who are in abundant supply in Silicon Valley will make the process happen all over again. For Silicon Valley, this cycle is as much fact as theory. In the 1950s, a handful of firms supplied electronic devices to the Defense Department. In the 1960s, the region became a center of computer chipmakers. In the 1970s and 1980s, the region developed and manufactured personal computers and workstations, and in the 1990s, the region helped commercialize Internet technology. For every major firm, such as the Hewlett-Packard Company and Intel, there were thousands of entrepreneurs starting little firms with dreams of one day becoming a leader in their field. Zhang concludes that start-ups in Silicon Valley have more rapid access to venture capital than comparable firms elsewhere in the nation; that large, established firms spin off more start-ups than firms in other parts of the country; and that the high-tech sector is subject to rapid structural change where “hot spots” of growth may appear in some industries while firms in other industries are simultaneously dying out. He observes that a dynamic labor force has been, and will be, essential to successful adaptation with each new structural change. In sum, human capital, venture capital, entrepreneurial zeal, and product cycles all contribute to the health and success of the economy of Silicon Valley. Although Zhang makes no predictions about the future, the fact that the region has weathered these cycles in the past, that the basic ingredients are still there in abundance, and that new demands for high-technology products are following on a worldwide concern for secure environments suggests that the prospects are good for yet another rebirth of the valley. Zhang suggests that the dynamics of economic development favor Silicon Valley and that yet another replay of the rebirth part of the cycle lies before us. David W. Lyon President and CEO Public Policy Institute of California iv Summary After extraordinary economic success in the late 1990s, Silicon Valley entered a deep recession in 2001. Today, policymakers, academic researchers, and the general public continue to puzzle over what made Silicon Valley such an enormous success. More important, they wonder if the region will ever experience such strong growth again. This study seeks to answer those questions by examining Silicon Valley’s high-tech economy in a dynamic context. Using two unique longitudinal databases, we investigate firm formation, growth, mortality, and migration in Silicon Valley during the 1990s and explain how the region’s economy evolves and operates through such dynamic processes. This study not only helps us better understand Silicon Valley’s success in the past but also reveals insights into how Silicon Valley can ensure its future prosperity. Major Findings New firms are important for Silicon Valley. As with other hightech centers, Silicon Valley hosts a wide variety of firms. A multitude of small firms coexist with medium-sized and big firms; and each year, many new firms are founded, which collectively are a major driver of the economic dynamics in Silicon Valley. In fact, firms founded after 1990 created almost all the job growth during 1990–2001. Young start-ups in Silicon Valley consistently attract a large amount of venture capital. Successful start-ups have remade and will continue to remake Silicon Valley. Start-ups in Silicon Valley have quick access to venture capital. On average, it takes 11.6 months for Silicon Valley’s start-ups to complete their first round of venture finance, five months faster than the national average. In addition, the quicker access to capital is found in every major industry in Silicon Valley. This gives start-ups in the region a head start—an important advantage in high-tech industries that advance at a v very fast pace. This large first-mover’s advantage implies that start-ups in the valley will have better chances to survive, all else being equal. Established firms in Silicon Valley spin off more start-ups. Compared to their counterparts in the Boston area, big companies in Silicon Valley have more previous employees who start their own venture-backed businesses. Since engineers in successful firms are in the best position to grasp and commercialize cutting-edge innovations, a high rate of spin-off helps open new markets and creates new jobs. Previous research discusses Silicon Valley’s high incidence of firm-level spin-off based on anecdotal evidence and has identified cultural and legal factors to account for it. Although the causal factors remain unclear, for the first time we have confirmed with empirical data that there are indeed more firm-level spin-offs in Silicon Valley than in other high-tech centers. Firm relocation is not a serious problem. High-tech start-ups value the hotbed of innovation because that is where new ideas emerge and entrepreneurs cluster. Silicon Valley is a perfect environment for startups whose major objective is to develop innovative ideas. On the other hand, when firms become mature and enter the phase of mass production or routine services, their major concern becomes sustainability and they naturally care about operating costs. For those firms or, rather, for certain operations of those firms, Silicon Valley is unattractive. We have investigated whether firms leave Silicon Valley when they have evolved out of the start-up stage. We find that indeed more establishments move out of Silicon Valley than move in, and establishments moving out tend to be older. Establishments still tend to stay close to the valley when they move out. When firms move across state borders, Silicon Valley does see a net job loss, because more jobs are relocated to other states than are relocated to Silicon Valley from outside California. However, the data suggest that firm relocation involves a relatively small proportion of the labor force. Firm birth and death cause much more turbulence than firm relocation. In other words, once firms are established in Silicon Valley, they are very likely to remain there. Intensive entrepreneurial activities certainly compensate for the jobs lost through firm relocation. vi Successful firms in the valley are branching out. Although relocation does not occur at significant levels, established firms in Silicon Valley frequently set up branches elsewhere. For many large high-tech companies headquartered in Silicon Valley, their employment within Silicon Valley itself is only a small proportion of their total employment. Since Silicon Valley is already tightly packed with thousands of firms, fast-growing start-ups are more likely to expand outside the immediate area. As firms begin to expand, they potentially benefit the rest of California by setting up branches elsewhere in the state. The high-tech sector experiences rapid structural changes. The high-tech sector consists of several industries, which follow different dynamics. On the one hand, the fluctuation of the macro economy has distinctive effects on different high-tech industries; on the other hand, technological innovations in different industries, the drivers of growth in those industries, do not arrive simultaneously. As a consequence, different high-tech industries may follow unsynchronized business cycles. Therefore, at different points of time, the “hot spot” of growth may appear in different industries. For example, the 1990s saw a boom in the computer industry along with a decline in the defense industry. To catch upturns and avoid downturns in high-tech industries, a high-tech center such as Silicon Valley must accommodate rapid structural changes. This implies that a dynamic labor force is necessary. Previous research has emphasized the “high-velocity labor market” through which workers move frequently from one job to another within Silicon Valley. Such a labor market certainly helps the region’s economy adapt to structural changes. In addition, a set of infrastructure and institutions that enables the labor force to quickly move into and out of Silicon Valley is also crucial for structural changes in the high-tech sector. For example, employment in the software industry in Silicon Valley increased from 48,500 to 114,600 between 1990 and 2001, a phenomenal 136 percent growth rate. It is impossible to train such a large number of technical workers within such a short period of time. This kind of rapid growth in a certain industry is achievable only through massive migration of the needed labor force. vii Policy Implications Our findings lead us to offer the following recommendations to policymakers. Promote technological innovation. More than any other sector, the high-tech economy is about innovation and entrepreneurship. State and local governments should help promote innovation. Since university research has always been a major source of innovation, state government should continue its strong support to research universities. Big budget cuts for the University of California system will severely affect the prospect of the high-tech sector off campus, which must be avoided. Moreover, the California delegation in Washington, D.C., should place a high priority on securing R&D dollars for California from the federal government. As the state economy becomes more and more reliant on high-tech industries, support for R&D and innovation not only helps Silicon Valley and the rest of the Bay Area, but it also greatly benefits the Los Angeles and San Diego areas, which continue to expand their own high-tech sectors. Encourage firm founding. Our findings show that although some firms do move out of Silicon Valley, it is not a serious problem. On the one hand, they are likely to move to nearby cities and stay within the state, and on the other hand, firm formation and growth create new jobs that overwhelmingly outnumber jobs lost by firm relocation. In addition, job creation in Silicon Valley is primarily achieved by new firms. Therefore, instead of worrying about losing firms because of the high costs of doing business in Silicon Valley, state and local governments should encourage firm founding. Offering favorable tax breaks, opening industrial parks, building high-tech incubators, and providing seed capital for commercialization of research are widely used policy levers. Continuously improving the quality of life in Silicon Valley and the Bay Area as a whole is also crucial for the vitality of the high-tech economy in this area. Look beyond Silicon Valley. The high-tech sector is not a disconnected economy, nor is Silicon Valley an isolated region. Silicon Valley is well embedded in the San Francisco Bay Area economy as well as the state economy. Most of the firms leaving Silicon Valley migrate to viii nearby cities in the Bay Area. The rest of the Bay Area has undoubtedly benefited from the proximity of Silicon Valley and has a quite strong high-tech economy. State policies regarding Silicon Valley should take into account connections between Silicon Valley and the rest of the state economy. For example, many people who work in Silicon Valley live a considerable distance from it, seeking more affordable homes. Thus, housing development and transportation policies in many other Bay Area cities help directly solve Silicon Valley’s housing problems. We have also found that large firms in Silicon Valley hire only a small proportion of their total employees from the valley or even the Bay Area. This suggests that other regions in the state have chances to benefit from the spillover from Silicon Valley by hosting branches of its firms. State government could provide incentives for large firms to set up their manufacturing or distribution arms within the state. It is also helpful to improve transportation networks between the Bay Area and the Central Valley that facilitate Silicon Valley’s branching out in other areas of the state. In addition, local governments in the rest of the Bay Area and the Central Valley should be more proactive in accommodating businesses branching out from Silicon Valley. Maintain a dynamic labor pool. Two conflicting factors characterize the high-tech labor force. On the one hand, the high-tech sector primarily hires technical workers whose skills are highly specialized and take time to acquire; on the other hand, the high-tech sector is dynamic, with its core technologies evolving quickly. This implies that the skills acquired in school three years ago may be obsolete today. Moreover, certain high-tech industries often experience explosive growth, such as the software industry did in the 1990s, which creates a high demand for certain types of technical workers within a short period. Whether Silicon Valley can evolve rapidly hinges upon whether its labor force can quickly upgrade its skills or meet completely new demands. State government should continue to rely on local universities and community colleges as a vehicle to help retool the labor force continuously. Employers in Silicon Valley need to recruit new talent not only through local universities but also by hiring qualified immigrants, who have played an important role in Silicon Valley’s growth. The immigrant pool has proved to be a major source of innovators and ix entrepreneurs. Immigrants also provide a large reserve of high-quality engineers and scientists ready to satisfy sudden surges of demand in certain industries. State government in cooperation with federal authorities should keep the door open to international talent, both at local universities and in the high-tech industries. This has emerged as a particularly crucial issue because immigration policies have now entered the equation of homeland security. x Contents Foreword ..................................... Summary..................................... Figures ...................................... Tables ....................................... Acknowledgments ............................... iii v xiii xv xvii 1. INTRODUCTION AND OVERVIEW OF THE STUDY ................................... Change in Silicon Valley ........................ A Demographic Perspective of the Silicon Valley Habitat ... Purpose of This Study .......................... Data ..................................... 1 3 6 8 9 2. START-UP, GROWTH, AND MORTALITY OF FIRMS IN SILICON VALLEY ......................... Firm Formation.............................. Rate of Firm Formation ....................... Structural Changes .......................... Firm Growth ............................... Firm Mortality .............................. Rate of Mortality ........................... Merger and Acquisition ....................... Job Creation by Start-Ups ....................... Conclusion ................................. 3. VENTURE-BACKED START-UPS IN SILICON VALLEY .................................. Venture Capital in Silicon Valley ................... Firm Formation.............................. Ownership Status and Profitability.................. Spinoffs ................................... Conclusion ................................. 11 11 11 16 19 23 24 25 28 30 31 31 35 41 47 52 4. FIRM RELOCATION IN SILICON VALLEY ......... 53 High-Tech and Nontech Relocation ................. 54 xi Trans-State Relocation ......................... Mobility vs. Vitality ........................... Relocating Out vs. Branching Out .................. Conclusion ................................. 5. CONCLUSION ............................. Major Findings .............................. Policy Implications............................ 60 65 69 71 73 73 75 Appendix A. Geographic and Industrial Definitions ............... B. The Data .................................. C. A Snapshot of the Silicon Valley Economy............. 81 85 95 Bibliography .................................. 99 About the Author ............................... 103 Related PPIC Publications .......................... 105 xii Figures 1.1. A Map of Silicon Valley ...................... 1.2. Industry Dynamics in Silicon Valley .............. 2.1. High-Tech Firm Formation in Silicon Valley, 1990– 2000 .................................. 2.2. Firm Formation in High-Tech Clusters, 1990–2000.... 2.3. High-Tech Start-Ups That Ever Hired Five or More Employees by 2001 ......................... 2.4. Employment in High-Tech Industries in Silicon Valley, 1990–2001 .............................. 2.5. Employment of High-Tech Start-Ups in Nonservice Industries, 2001 ........................... 2.6. Employment of High-Tech Start-Ups in Service Industries, 2001 ........................... 2.7. Survival Rates of High-Tech Firms in Silicon Valley .... 2.8. Comparison of Survival Rates .................. 2.9. Percentage of Firms Acquired by 2001 ............ 2.10. Employment of High-Tech Start-Ups in Silicon Valley .. 2.11. Employment of High-Tech Start-Ups Younger Than Age Five as a Percentage of Total High-Tech Employment ............................. 3.1. Total Venture Capital Investment, 1992–2001 ....... 3.2. Total Venture Capital Investment, by Region, 1992– 2001 .................................. 3.3. Venture-Backed Start-Ups, 1990–2001 ............ 3.4. Venture-Backed Start-Ups, by Region, 1990–2001 .... 3.5. Average Amount of Venture Capital Raised per Deal, 1992–2001 .............................. 3.6. Average Start-Up Age at First-Round Financing ...... 3.7. Average Start-Up Age at First-Round Financing, by Industry ................................ 3.8. Ownership Status of Venture-Backed Start-Ups in Silicon Valley, 2001 ........................ 2 9 12 13 14 17 22 22 25 26 27 29 29 32 33 36 36 37 38 39 42 xiii 3.9. Ownership Status of Venture-Backed Start-Ups in the United States, 2001 ........................ 3.10. Differences in Ownership Status in Each Cohort of Venture-Backed Start-Ups: Silicon Valley Compared to the United States .......................... 3.11. Business Status of Venture-Backed Start-Ups in Silicon Valley, 2001 ............................. 3.12. Business Status of Venture-Backed Start-Ups in the United States, 2001 ........................ 4.1. Percentage of Moving Establishments Founded Before 1990 .................................. 4.2. Average Age of Establishments Moving Between Silicon Valley and Other States ...................... 4.3. Job Movement Between Silicon Valley and Other States, 1991–2000 .............................. 4.4. Dynamics in Silicon Valley’s High-Tech Labor Market, 1991–2000 .............................. 4.5. Dynamics in Silicon Valley’s Labor Market, 1991– 2000 .................................. 43 44 46 47 63 63 64 68 68 xiv Tables 1.1. Forty Largest Technology Companies in Silicon Valley, 1982 and 2002 ........................... 2.1. High-Tech Start-Ups, by Industry, 1990–2000 ....... 2.2. Employment in High-Tech Industries in Silicon Valley, 1990–2001 .............................. 2.3. Growth of Silicon Valley’s High-Tech Firms in Nonservice Industries ....................... 2.4. Growth of Silicon Valley’s High-Tech Firms in Service Industries ............................... 2.5. Death of High-Tech Establishments in Silicon Valley, 1990–2000 .............................. 2.6. Top Headquarter States of Firms Acquired During 1990–2001 .............................. 3.1. Real Venture Capital Investment, by Industry in Silicon Valley, 1992–2001 ......................... 3.2. Number of Spinoffs from Leading Institutions in Silicon Valley and the Boston Area .................... 4.1. Relocation of Establishments in Silicon Valley, 1990– 2001 .................................. 4.2. Top Ten Destination States for Establishments Relocating Out of Silicon Valley, 1990–2001 ........ 4.3. Top Ten Destination Cities for Establishments Relocating Out of Silicon Valley, 1990–2001 ........ 4.4. Top Ten Origin States for Establishments Relocating Into Silicon Valley, 1990–2001 ................. 4.5. Top Ten Origin Cities for Establishments Relocating Into Silicon Valley, 1990–2001 ................. 4.6. High-Tech Establishments Relocating Into and Out of Silicon Valley, by Industry, 1990–2001 ............ 4.7. All Establishments Relocating Into and Out of Silicon Valley, by Industry Group, 1990–2001 ............ 5 15 18 20 21 24 28 34 50 55 56 56 58 58 59 60 xv 4.8. High-Tech Establishments Moving Between Silicon Valley and Outside California, by Industry, 1990– 2001 .................................. 4.9. All Establishments Moving Between Silicon Valley and Outside California, by Industry Group, 1990–2001 .... 4.10. Trans-State Relocation as a Percentage of Total Employment That Moved Into or Out of Silicon Valley, 1990–2001 .............................. 4.11. Employment in the High-Tech Sector of Silicon Valley, 1991–2000 .............................. 4.12. Employment in Silicon Valley, 1991–2000 ......... 4.13. Intel Operating Locations in the United States ....... B.1. Business Size Distribution in NETS and EDD Data, 2001 .................................. B.2. Employment Series in NETS and EDD Data, 1990– 2001 .................................. B.3. Real Venture Capital Investment in the United States, by Industry, 1992–2001........................ B.4. Venture Capital Investment by MoneyTree Survey and VentureOne Data .......................... C.1. Total Number of Establishments and Employees in Silicon Valley, 2001 ........................ C.2. High-Tech Establishment Category in Silicon Valley, 2001 .................................. C.3. Establishment Size Distribution in Silicon Valley, 2001 .. C.4. Establishment Age Distribution in Silicon Valley, 2001 .. C.5. Total Establishments in Silicon Valley, by Industry Group, 2001 ............................. C.6. Total High-Tech Establishments in Silicon Valley, by Industry, 2001 ............................ 61 61 62 66 67 70 88 89 92 93 95 95 95 96 96 97 xvi Acknowledgments I would like to thank Michael Teitz for his suggestions, guidance, and encouragement at every stage of this research project. I am grateful to AnnaLee Saxenian, who provided guidance during the development of the research proposal and offered invaluable comments and suggestions for finalizing the report. Thanks also go to Doug Henton, Martin Kenney, Joyce Peterson, and Karthick Ramakrishnan for their thoughtful comments on a preliminary draft of the report. Nikesh Patel did a superb job helping with data analysis. Donald Walls offered kind help in extracting the NETS data from his database. Also, I want to thank Gary Bjork and Patricia Bedrosian for their editorial assistance. The author is solely responsible for any errors of fact or interpretation. xvii 1. Introduction and Overview of the Study It took merely half a century for Santa Clara Valley, the region that curls around the southern tip of the San Francisco Bay, to become the most famous high-tech industrial cluster in the world. Silicon Valley, as it has been known since the early 1970s, is today a main driver of the California state economy (see Figure 1.1 and Appendix A for our geographic definition of Silicon Valley). It is home to more than 22,000 high-tech companies, including household names such as HewlettPackard, Intel, Apple, and eBay. Silicon Valley’s celebrity skyrocketed over the past decade as it became the center of “the largest legal creation of wealth in history.” At its peak, the Internet boom produced scores of new millionaires in Silicon Valley every day. The region had become a land of enchantment for ambitious entrepreneurs whose success stories appeared in the media all over the world, and thousands of well-paid jobs made Silicon Valley a magnet for talented people. Given the enormous success of this regional economy, policymakers around the world wondered how they could “clone Silicon Valley” in their own regions (Rosenberg, 2002). But it seems that what goes up must come down. Since 2001, the region has entered a deep recession. In Santa Clara County, the heart of Silicon Valley, the unemployment rate climbed from 1.7 percent in January 2001 to 8.9 percent in October 2002, then declined a little to 8.3 percent in December 2002.1 In 2002, Silicon Valley posted an annual unemployment rate higher than the state average for the first time in two decades. According to Joint Venture’s 2003 Index of Silicon Valley, the region lost 127,000 jobs (about 9 percent of its total employment) ____________ 1According to the California Employment Development Department, available at http://www.calmis.cahwnet.gov/htmlfile/subject/lftable.htm. 1 SOURCE: Reprinted by permission from Joint Venture: Silicon Valley Network, with adaptations. Figure 1.1—A Map of Silicon Valley 2 between the first quarter of 2001 and the second quarter of 2002. More than half of the job gains registered during 1998–2000 evaporated. At the same time, venture capital investment plummeted and personal income declined. Policymakers, academic researchers, and the general public continue to puzzle over what made Silicon Valley such a huge success. More important, they wonder if the region will ever experience such strong growth again. This study seeks to answer those questions by examining Silicon Valley’s high-tech economy in a dynamic context. Using two unique longitudinal databases, we investigate firm formation, growth, mortality, and migration in Silicon Valley during the 1990s and examine how the region’s economy evolved and operated through such dynamic processes. This study not only helps us better understand Silicon Valley’s success in the past, but it also reveals insights into how Silicon Valley can ensure its future prosperity. Change in Silicon Valley Silicon Valley has experienced both highs and lows many times. If asked to use a single word to characterize the Silicon Valley economy, many people would choose “dynamic.” Indeed, change is the only unchanging norm in Silicon Valley, as new technologies and new firms constantly emerge. Yet, as the famous economist Joseph Schumpeter observed almost a century ago, innovations are not evenly distributed over time but occur in periodic clusters (Schumpeter, 1934). This is particularly true in Silicon Valley, which has remade itself over and over again during its short history (“Silicon Valley: How It Really Works,” 1997; Henton, 2000). Until the 1950s, only a handful of high-tech firms existed in the area, most notably Hewlett-Packard and Varian. The area was a major supplier of electronic devices to the Defense Department. In the 1960s, as Fairchild spun off many semiconductor producers such as Intel and AMD, the area became a center of computer chipmakers, which later led to the name “Silicon Valley.” 3 The late 1970s and 1980s were the computer years. By then the valley was known as a developer and manufacturer of personal computers and workstations, represented by such companies as Apple, Silicon Graphics, and Sun Microsystems. In the 1990s, Silicon Valley remade itself again. This time, it helped commercialize Internet technology. The leaders of this movement included Cisco, Netscape, eBay, and Yahoo. Silicon Valley has developed through waves of innovation, with a handful of innovative start-ups initiating each wave. In fact, the continuous success of Silicon Valley must be understood as the constant emergence of successful start-ups. As Lee et al. (2000) point out, “The Silicon Valley story is predominantly one of the development of technology and its market applications by firms—especially by start-ups. The result: new companies focused on new technologies for new wealth creation.” For many decades, social scientists have noticed the important role of start-ups in carrying out radical innovations. Schumpeter (1934, p. 66) observed that innovations are, as a rule, embodied in “new firms which generally do not arise out of the old ones but start producing beside them.” Recent work has provided a rationale for this observation by emphasizing the characteristics of innovations. Foster (1986) argued that technological progress often exhibits discontinuities. That is, radical changes happen frequently. Reflected in the dynamics of high-tech industries, these discontinuities give new firms a so-called “attacker’s advantage.” When newcomers gain competitive superiority over successful incumbent firms, “leaders become losers.” More recently, Christensen (1997) further developed this idea and called it the “innovator’s dilemma.” When Schumpeter talked about “the incessant gales of creative destruction” many decades ago, he could not have imagined that the industry dynamics in Silicon Valley would provide such a vivid illustration of his notion. Silicon Valley is constantly creating the new while destroying the old. Table 1.1 lists the top 40 high-tech firms in Silicon Valley in 1982 and 2002. An overwhelming majority of the names on the 1982 list have become faded memories among the locals. To outsiders, most of the 1982 top firms are unrecognizable, because half 4 Table 1.1 Forty Largest Technology Companies in Silicon Valley, 1982 and 2002 1. Hewlett-Packard 2. National Semiconductor 3. Intel 4. Memorex 5. Varian 6. Environtecha 7. Ampex 8. Raychema 9. Amdahla 10. Tymsharea 11. AMD 12. Rolma 13. Four-Phase Systemsa 14. Cooper Laba 15. Intersil 16. SRI International 17. Spectra-Physics 18. American Microsystemsa 19. Watkins-Johnsona 20. Qumea 1. Hewlett-Packard 2. Intel 3. Ciscob 4. Sunb 5. Solectron 6. Oracle 7. Agilentb 8. Applied Materials 9. Apple 10. Seagate Technology 11. AMD 12. Sanmina-SCI 13. JDS Uniphase 14. 3Com 15. LSI Logic 16. Maxtorb 17. National Semiconductor 18. KLA Tencor 19. Atmelb 20. SGI 1982 21. Measurexa 22. Tandema 23. Plantronics 24. Monolithic 25. URS 26. Tab Products 27. Siliconix 28. Dysana 29. Racal-Vadica 20. Triad Systemsa 31. Xidexa 32. Avanteka 33. Silteca 34. Quadrexa 35. Coherent 36. Verbatim 37. Anderson-Jacobsona 38. Stanford Applied Engineering 39. Acurexa 40. Finnigan 2002 21. Bell Microproductsb 22. Siebelb 23. Xilinxb 24. Maxim Integratedb 25. Palmb 26. Lam Research 27. Quantum 28. Alterab 29. Electronic Artsb 30. Cypress Semiconductorb 31. Cadence Designb 32. Adobe Systemsb 33. Intuitb 34. Veritas Softwareb 35. Novellus Systemsb 36. Yahoob 37. Network Applianceb 38. Integrated Device 35. Linear Technology 40. Symantecb NOTES: This table was compiled using 1982 and 2002 Dun & Bradstreet (D&B) Business Rankings data. Companies are ranked by sales. aNo longer existed by 2002. bDid not exist before 1982. 5 of them no longer exist. Only four firms on the 2002 list are survivors from the 1982 list. In fact, more than half of the 2002 top firms were not even founded before 1982. In only two decades, the high-tech economy in Silicon Valley changed almost completely. The San Jose Mercury News has compiled a list of the top 150 firms in Silicon Valley each year since 1994. On average, each year’s list includes 23 new firms, reflecting the fast pace of Silicon Valley. A study of these “changes” is not only the key to understanding Silicon Valley’s past success but also the key to promoting its future success. Silicon Valley’s greatest asset is its ability to reinvent itself as soon as its leading technologies or products become standardized. Thus, the secrets of the region’s success lie in its institutions that enable the changes. To ensure a bright future, we must identify, understand, and promote those institutions, and to understand the unique features of Silicon Valley and its institutions, we must observe its dynamic context. A Demographic Perspective of the Silicon Valley Habitat Silicon Valley is often described as a “habitat” (Lee et al., 2000) or an ecosystem (Bahrami and Evans, 2000). As in a natural habitat, Silicon Valley provides a host of resources that high-tech firms require to survive and grow. This habitat includes not only people, firms, universities and research institutions, and government agencies but also networks among those players and the modes by which they interact. Previous studies have examined different constituents of the habitat (see, for example, Saxenian, 1994; Kenney and Florida, 2000; and Lee et al., 2000). These studies have provided insights into the role played by entrepreneurs, universities, social networks, and supporting players such as venture capitalists, bankers, lawyers, consultants, and so on. However, the central figure in the Silicon Valley habitat is undoubtedly high-tech firms. After all, the success of Silicon Valley is measured by the large population of high-tech firms that offer many well-paid jobs. Much like a biologist who studies animals in their natural habitats, we shall take a demographic approach to study firms in Silicon Valley. 6 The demographic approach is well developed in organizational sociology (Carroll and Hannan, 2000). In contrast to the bulk of literature in industrial economics that focuses on firm-level behavior, the demographic perspective shifts attention from individual firms to the range and diversity of firms in an industry or region. It seeks to discover insights into how industries evolve over time through processes of firm formation, growth, transformation, migration, and mortality. The demographic approach is not concerned with individual firms but, rather, focuses on properties at the population level, such as a population’s age distribution and growth rates. The demographic approach is particularly appropriate for studying the Silicon Valley economy. The high-tech sector in Silicon Valley consists of a wide range of firms. On one extreme are large companies offering thousands of local jobs, such as Hewlett-Packard and Intel; on the other are thousands of small firms that hire only a few people. Firms such as Hewlett-Packard and Varian have been around for more than six decades, whereas other high-profile firms such as eBay and Yahoo did not even exist ten years ago. Companies such as Cisco and Sun Microsystems have expanded at a stunning pace, whereas thousands of others hardly grow or disappear soon after inception. And most important, products or services are differentiated along many dimensions; rarely do any two firms provide exactly the same product or service. As Carroll and Hannan have argued, the vibrancy of the Silicon Valley economy to some extent reflects its demographic characteristics. In particular, “the high rates of turnover of constituent organizations continually reshuffle the human workforce. The great diversity of organizational forms and technological strategies means that job-changers find themselves in new and different social contexts. Ideas flow with people, get recombined, and new technical and organizational innovations result. Analysis of a putatively representative firm would not only miss the point, it would also obscure community-level dynamics” (Carroll and Hannan, 2000). Yet basic demographic facts about the Silicon Valley economy remain unknown, partly because of a lack of demographic data on industries. This means that the formulation of regional social and 7 economic policies usually ignores the implication of the full diversity of firms. Thus, a demographic study can yield very useful information for policymakers. For example, discussion of firm relocation usually draws upon anecdotal evidence from the media and often raises concerns about job loss. However, the relocating firms receiving media coverage are neither representative nor exhaustive. A statistical portrait of the whole population of moving firms would reveal the real effect of firm relocation. Purpose of This Study The purpose of this study is twofold. First, it will document the intensity of entrepreneurial activities in Silicon Valley and provide information helpful to understanding the dynamics of change in the region. Specifically, it will • Measure the rates of firm formation, growth, and mortality in Silicon Valley and compare those rates to those in other hightech centers. • Measure the proportion of start-ups in the Silicon Valley economy and their effects on job creation and dissolution. These effects will be discussed in light of the Birch (1987) debate over whether small firms create more jobs. The second purpose of this research is to track the stock and flow of high-tech firms in Silicon Valley. The study will • Determine whether most firms move to the area or are started locally. • Identify the characteristics of firms moving into or out of Silicon Valley. • Examine whether net firm relocation enhances the cluster in Silicon Valley or causes the region to lose businesses. Figure 1.2 summarizes industry dynamics in Silicon Valley’s high-tech sector. We will investigate all of the types of dynamics illustrated, except for “moving inside” Silicon Valley, which is not a major concern of our study. 8 Death Moving in Merger and acquisition Moving inside Growth Moving out Birth Figure 1.2—Industry Dynamics in Silicon Valley Data Our empirical analysis will rely on two longitudinal databases: The National Establishment Time-Series (NETS) dataset that seeks to include every firm in Silicon Valley and the nationwide VentureOne dataset that focuses on venture-backed firms. The two datasets contain an enormous amount of information that helps us better understand firm formation, growth, and industry dynamics in Silicon Valley. The abundance of data allows us to shed light on many important issues through simple descriptive analysis. For a detailed discussion of the data, see Appendix B. 9 2. Start-Up, Growth, and Mortality of Firms in Silicon Valley The high-tech sector accounts for about 11 percent of the total goods and services in the United States (DeVol, 1999). As the most concentrated high-tech center, Silicon Valley has a much larger proportion of high-tech economy than does the rest of the nation. In 2001, there were 25,787 high-tech establishments in Silicon Valley— 25 percent of the total establishments in the region. Since many hightech firms are big employers, that one-quarter of all establishments offered 42.7 percent (or 673,000) of the total jobs in Silicon Valley. (See Appendix C for a more detailed profile of the Silicon Valley economy.) This chapter documents firm formation, growth, and mortality in Silicon Valley’s high-tech sector from 1990 to 2001, using the NETS dataset. Remember, the basic observation unit in the NETS data is the “establishment,” and a big firm may have several establishments. When we study firm founding and mortality, we exclude establishments created by existing firms; and when we study firm growth, we aggregate all the establishments of a firm into a single unit. Firm Formation Rate of Firm Formation Figure 2.1 traces the trend of entrepreneurial activities in Silicon Valley’s high-tech sector. During the decade from 1990 to 2000, 29,000 high-tech firms were created in Silicon Valley. An upward trend started in the early 1990s and continued until 1998, before declining sharply in 1999 and 2000. It is interesting to note that only one-fourth of the new firms had ever hired five or more employees. Most of the new firms will always remain in the 0–4 size category. Some of the founders might be 11 Number of start-ups 4,000 3,500 3,000 2,500 Silicon Valley total Ever hired 5 or more employees Venture-backed start-ups 2,000 1,500 1,000 500 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Figure 2.1—High-Tech Firm Formation in Silicon Valley, 1990–2000 more precisely described as self-employed rather than entrepreneurs. Firms that ever employed five or more people follow a much less dramatic trend in the 1990s. That is, although many more firms were created in the hype years of Internet technology, many of them started small and never grew.1 The trend for venture-backed start-ups is also depicted in Figure 2.1. Although the high-tech sector in Silicon Valley is mostly renowned for its legendary start-ups financed by venture capital, venture-capital-backed new firms actually form only the tip of a huge iceberg. A vast majority of ____________ 1D&B, the source of raw data, did ask each establishment to report its start year. However, not all of them did so. As a consequence, the start year is missing for many establishments, especially small ones. Walls & Associates created a variable “FirstYear,” whose value is determined by the first time an establishment’s data are available at D&B. If a firm reported to D&B in 1993 for the first time, 1992 is assigned to it as its first year. For those firms that have reported their start year, the first year variable is almost always identical to the start year. But overall, the trends in the two variables are quite different, mainly because many firms that were not in the D&B database originally later chose to be included in it for common reasons, such as needing a Data Universal Numbering System (DUNS) number. With the assumption that firms that reported their start year form a representative sample of the whole population, Figures 2.1–2.3 estimate the trend of entrepreneurial activities using the number of start-ups whose start year is self-reported. For example, if x out of y start-ups reported their start year in the whole sample and z of them reported 1995 as their start year, the number of firms started in 1995 is estimated to be z*y/x. By doing so, we smooth out the noise in the trend created mainly by small firms. 12 high-tech firms created in Silicon Valley are not financed by venture capital, either because they are not capital-intensive enterprises or because they do not possess a growth potential that justifies venture capital support. However, the number of venture-backed new firms grew faster proportionately than the overall trend of firm formation in the high-tech sector. In 1999, the peak year of venture capital finance, 375 start-ups were backed by venture capital—more than five times the number in 1990—whereas the total number of new firms founded in the high-tech sector did not even double from 1990 to its peak year in 1998. This reflects the fact that venture capital became much more easily available in the late 1990s. It also suggests that firm founders became more innovative as the Internet revolution created many new opportunities. We study venture-backed firms exclusively in the next chapter. Figure 2.2 compares the trend of firm formation in Silicon Valley to the trends in Boston and Washington, D.C.2 From 1990 to 1996, the 4,000 3,500 3,000 Silicon Valley Boston Washington, D.C. Number of start-ups 2,500 2,000 1,500 1,000 500 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Figure 2.2—Firm Formation in High-Tech Clusters, 1990–2000 ____________ 2By high-tech employment, Silicon Valley, Boston, and Washington, D.C., are the top three, far ahead of any other high-tech center in the United States (Cortright and Mayer, 2001). This is the primary reason why we choose Boston and Washington for comparison. 13 three areas followed almost the same upward trend. Boston lost its momentum in 1996, but Silicon Valley and Washington, D.C., continued their upward trend in firm formation until 1998. The Internet boom in the late 1990s stimulated more entrepreneurial activities in Silicon Valley and Washington than in Boston. Figure 2.3 traces the founding year of those new firms that had ever hired five or more employees in the three high-tech clusters. Silicon Valley has more firms in the 5+ category. Whereas the total number of new firms founded in Silicon Valley follows a similar trend as in the other two high-tech regions, the former consistently has more young firms hiring five or more employees. This may suggest that new firms in Silicon Valley are more growth-oriented than those in the other two areas. As mentioned above, 29,000 high-tech firms were created in Silicon Valley during the decade from 1990 to 2000. Washington, D.C., had a similar total, and Boston had about 5,000 fewer new firms. Table 2.1 presents the distribution of new firms across major high-tech industries (see Appendix A for exact definitions of those industries). In all three 900 800 700 600 500 400 Silicon Valley 300 Boston 200 Washington, D.C. 100 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Founding year Figure 2.3—High-Tech Start-Ups That Ever Hired Five or More Employees by 2001 14 Number of start-ups Table 2.1 High-Tech Start-Ups, by Industry, 1990–2000 Industry Bioscience Computers/communications Defense/aerospace Environmental Semiconductor Software Professional services Innovation services Total Silicon Valley Firms % 586 2.0 934 3.2 52 0.2 242 0.8 513 1.8 5,967 20.4 14,009 47.9 6,944 23.7 29,247 100 Boston Firms % 335 1.4 221 0.9 27 0.1 299 1.2 52 0.2 3,323 13.5 16,784 68.2 3,565 14.5 24,606 100 NOTE: Percentages may not sum to 100 because of rounding. Washington, D.C. Firms % 211 0.7 172 0.6 35 0.1 174 0.6 28 0.1 4,137 14.0 19,703 66.9 4,985 16.9 29,445 100 regions, the service industries were the most active. About 70 percent of Silicon Valley new firms were engaged in professional or innovation services. The percentage is even higher in the other two areas: for each, more than 80 percent of new firms were established in service industries. Except in the environmental industry, Silicon Valley outperformed the other two areas in every nonservice industry. Silicon Valley created more firms in the biotech, computers/communications, defense/aerospace, semiconductor, and software industries. Silicon Valley strongly led the semiconductor industry, from which it acquired its name, with 513 semiconductor start-ups during the decade, compared to 80 in Boston and Washington together. Although Boston has a long history in the defense industry and hosts Raytheon as the area’s largest employer, fewer defense/aerospace firms were founded in Boston than in the other two areas. It is also very impressive that Washington outperformed Boston (supposedly the number two high-tech cluster) in the software industry. Boston is also well known for its biotech industry. However, even in biotech, it was outnumbered by Silicon Valley. Remember, the biotech industry in the Bay Area is mainly clustered around South San Francisco and Berkeley–Emeryville, which is outside Silicon Valley. Taking that into account, the whole Bay Area did much better in biotech than reflected in the number for Silicon Valley alone. 15 Structural Changes In the high-tech sector, different industries serve different markets and employ workers with different skills. The labor forces in different industries are not entirely interchangeable. Thus, a high-tech center tends to retain a stable economic structure over time. Yet innovations do not arrive at the same rate across all industries and the macro economic climate may also have different effects on different industries. A vibrant high-tech center needs to be flexible and able to shift its emphasis when some industries slow down and others become more dynamic. Otherwise, it will not take full advantage of new areas of growth and will be hard hit when a major industry shrinks. Given the size of its hightech sector, Silicon Valley appears to be exceptionally adaptable in accommodating structural changes. Figure 2.4 presents the evolution of employment in high-tech industries in Silicon Valley. Two developments in the 1990s redefined the high-tech sector: the reduction of defense spending by the federal government after the end of the Cold War and the Internet revolution. Both have left clear marks on the structure of Silicon Valley’s high-tech economy. During 1990–2001, Silicon Valley’s defense/aerospace industry lost 60 percent of its jobs; in contrast, the software industry grew by 136 percent and the computers/communications industry by 32 percent. In 1990, total high-tech employment in Silicon Valley was 90 percent larger than in Washington, D.C., and 26 percent larger than in Boston, yet it was nimble enough to substantially change the structure of its high-tech economy over the next decade. The 136 percent growth of the software industry in Silicon Valley outpaced every high-tech industry in the other two regions. At the same time, Silicon Valley’s defense/ aerospace industry was the most heavily hit and shrank the most. For each industry, we decompose the employment growth during 1990– 2001 into the growth of firms that existed in 1990 and the jobs added by firms founded after 1990. In 2001, the high-tech economy in Silicon Valley had 672,825 employees—26 percent more than its total employment in 1990. Software, computers/communications, professional services, and semiconductor industries had each created more than 20,000 jobs. If we look only at those firms that already 16 Employment by industry 160,000 140,000 Computers/communications Innovation services Software Professional services Semiconductor Defense/aerospace Bioscience Environmental 120,000 100,000 80,000 60,000 40,000 20,000 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Figure 2.4—Employment in High-Tech Industries in Silicon Valley, 1990–2001 existed in 1990, they together lost 120,559 jobs. Old firms hired more people only in the semiconductor and environmental industries, but both increases were modest. It is interesting to note that firms founded before 1990 lost jobs during 1990–2001 in software and computers/ communications—the two industries that gained the most jobs in Silicon Valley during the 1990s (Table 2.2). On the other hand, firms founded after 1990 added a total of 258,796 jobs to the economy during the 1990s. The 136 percent growth of the software industry was all attributable to new firms, which added 72,684 jobs to the industry. In 1990, the software industry was number six by employment in Silicon Valley, after the computers/ communications, innovation services, semiconductor, professional 17 Table 2.2 Employment in High-Tech Industries in Silicon Valley, 1990–2001 Employment in 1990 Industry (1) Bioscience 46,815 Computers/communications 114,617 Defense/aerospace 68,527 Environmental 2,851 Semiconductor 79,630 Software 48,529 Innovation services 100,217 Professional services 73,402 Total 534,588 2001 Employment of Firms Existing in 1990 (2) 36,243 104,956 22,251 3,246 83,701 41,955 65,389 56,288 414,029 Total Employment in 2001 (3) 51,854 150,974 27,567 8,342 103,443 114,639 112,150 103,856 672,825 Overall Employment Growth 1990–2001 (3) – (1) 5,039 36,357 –40,960 5,491 23,813 66,110 11,933 30,454 138,237 Employment Growth of Firms Existing in 1990 (2) – (1) –10,572 –9,661 –46,276 395 4,071 –6,574 –34,828 –17,114 –120,559 Employment Growth of New Firms (3) – (2) 15,611 46,018 5,316 5,096 19,742 72,684 46,761 47,568 258,796 18 services, and defense/aerospace industries. By 2001, only the computers/ communications industry had more employees. Old firms lost jobs because not all of them survived after ten years. Also, other old firms might still be growing, but the growth occurred outside Silicon Valley. Table 2.2 provides a clear indication that Silicon Valley shifts development paths and remakes itself through the formation and growth of new firms. Firm Growth Because of the lack of sales data, firm growth is measured by employment growth. Tables 2.3 and 2.4 present the average employment of high-tech firms that are still alive. Firm sizes in service and other industries are calculated separately. On average, a high-tech start-up in nonservice industries hires 7–22 persons in the first year, depending on the cohort. As the start-up becomes older, its average employment is larger. In contrast to our general impression, the average growth of start-ups is far from explosive. It generally takes 5–6 years for an average start-up to double its employment. Firms in service industries are generally smaller and experience much slower growth. Before 1997, new firms in service industries always had an average employment below five in the first year. It takes more than nine years for service firms to double their average employment. A majority of them hardly grow at all. The growth is underestimated because the employment at a firm’s branches outside Silicon Valley is not captured here because of data limitations. Yet the number is meaningful because it measures the growth of start-ups within Silicon Valley. The growth is not accelerating as the data might have suggested. The faster growth at older ages results because many firms were defunct by those ages and only the fast-growing firms survived and were counted. Tables 2.3 and 2.4 suggest that the kind of explosive growth achieved by such stars as eBay and Yahoo is phenomenal, even by Silicon Valley’s standard. Figures 2.5 and 2.6 compare the size of high-tech firms in Silicon Valley with those in the Boston and Washington, D.C., areas. Nonservice high-tech firms seem to grow faster in Silicon Valley. Each 19 Table 2.3 Growth of Silicon Valley’s High-Tech Firms in Nonservice Industries 20 Cohort 1990 1991 1991 12.48 (58.83) 1992 1993 1994 1995 1996 1997 1998 1999 2000 1992 12.58 (58.93) 8.22 (11.76) 1993 12.59 (59.03) 8.41 (12.09) 7.34 (10.73) 1994 14.96 (63.73) 9.70 (14.17) 8.05 (12.09) 8.53 (14.28) NOTE: Standard deviations are in parentheses. Average Employment 1995 1996 1997 15.77 21.51 26.50 (65.15) (96.69) (114.3) 11.52 12.40 16.95 (19.54) (20.07) (36.36) 9.23 10.46 12.95 (14.23) (17.84) (23.35) 9.09 9.85 12.40 (14.82) (15.01) (18.59) 9.25 10.06 11.07 (13.27) (15.34) (16.39) 9.58 10.09 (16.14) (16.15) 10.24 (26.60) 1998 30.27 (127.6) 18.78 (42.69) 15.47 (30.32) 15.78 (24.28) 13.61 (20.35) 13.68 (22.67) 11.66 (28.02) 9.01 (16.83) 1999 40.54 (234.5) 21.86 (53.11) 17.99 (35.73) 18.43 (30.52) 16.55 (26.58) 16.68 (34.65) 15.00 (33.43) 9.65 (18.56) 8.41 (13.53) 2000 48.18 (281.1) 27.16 (64.50) 21.74 (58.63) 24.57 (81.62) 20.60 (34.46) 20.29 (41.17) 21.00 (50.53) 12.66 (24.43) 10.85 (23.92) 20.78 (141.8) 2001 55.74 (314.3) 35.15 (98.79) 22.90 (60.93) 35.58 (119.1) 28.19 (68.31) 28.61 (58.32) 29.12 (66.00) 20.97 (67.43) 17.23 (49.50) 20.29 (117.1) 22.51 (149.2) 21 Table 2.4 Growth of Silicon Valley’s High-Tech Firms in Service Industries Cohort 1990 1991 1991 4.88 (13.66) 1992 1993 1994 1995 1996 1997 1998 1999 2000 1992 4.90 (13.65) 4.30 (8.07) 1993 4.93 (13.70) 4.26 (7.98) 4.28 (8.82) 1994 4.98 (14.22) 4.55 (9.43) 4.46 (9.04) 4.65 (9.34) Average Employment 1995 1996 1997 5.60 5.44 6.17 (22.26) (15.96) (20.73) 4.69 5.06 5.69 (10.0) (10.28) (13.53) 4.58 4.87 5.85 (9.37) (10.30) (15.49) 4.80 5.14 5.56 (9.77) (10.77) (11.70) 4.74 4.68 4.83 (8.60) (8.06) (8.31) 4.70 4.91 (9.38) (9.70) 4.45 (10.32) 1998 6.70 (22.11) 6.52 (17.84) 5.89 (16.07) 5.88 (15.32) 5.17 (9.06) 5.53 (11.42) 4.72 (11.26) 6.61 (61.29) 1999 7.52 (26.35) 6.71 (18.42) 6.26 (17.08) 5.90 (12.71) 5.88 (13.54) 5.98 (13.08) 5.08 (10.35) 6.80 (61.79) 3.98 (7.75) NOTE: Standard deviations are in parentheses. 2000 7.90 (29.12) 6.18 (13.78) 6.62 (17.52) 5.93 (12.36) 5.84 (11.64) 6.12 (13.12) 5.66 (12.31) 7.35 (62.98) 4.18 (7.96) 6.77 (17.90) 2001 12.47 (83.00) 6.75 (15.17) 6.82 (19.00) 6.29 (14.13) 5.77 (11.22) 6.97 (18.79) 6.28 (14.43) 8.77 (66.70) 5.13 (12.80) 7.68 (19.92) 12.93 (126.5) 60 Silicon Valley 50 Boston Washington, D.C. 40 Average employment by 2001 30 20 10 Average employment by 2001 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Founding year Figure 2.5—Employment of High-Tech Start-Ups in Nonservice Industries, 2001 14 Silicon Valley 12 Boston Washington, D.C. 10 8 6 4 2 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Founding year Figure 2.6—Employment of High-Tech Start-Ups in Service Industries, 2001 22 cohort of nonservice firms founded during 1990–1999 has higher average employment in Silicon Valley. The 2000 cohort, only one year old in our data, is the only group of Silicon Valley nonservice firms that does not dominate its counterparts in the other two regions by employment size. This may suggest that Silicon Valley’s nonservice high-tech firms start with a smaller employment size but grow faster. In service industries, Silicon Valley high-tech firms are not consistently larger than those in other areas. In three cohorts, Silicon Valley firms have the smallest average employment; yet in five others, Silicon Valley firms have the largest. Silicon Valley service firms founded in the late 1990s seemed to do particularly well, which may be attributable to the Internet boom that especially benefited Silicon Valley. Figure 2.5 also shows that service firms are quite similar in size across different cohorts, which implies that they grow slowly over time. This section has demonstrated that start-ups in nonservice industries grow faster than those in service industries, and the previous section has shown that a higher proportion of start-ups in Silicon Valley occurs in nonservice industries. These together provide another reason why more firms in Silicon Valley than in Boston or Washington, D.C., had hired five or more employees by 2001 (Figure 2.3). Firm Mortality In the general practice of corporate demography literature (Carroll and Hannan, 2000), the mortality of a firm refers to any event by which a firm loses its identity. For example, a firm may disband, exit to another industry, or be merged or acquired. In this study, we are particularly interested in the disbanding of firms, since it has implications for the job market. We consider a firm dead if it drops out of the D&B dataset, since most probably it disbanded. A firm that has shifted to a different industry will simply have a new standard industrial classification (SIC) number. Those that go through merger and acquisition will simply have a different “headquarter DUNS number.” Neither will drop out of the D&B database. Firms do change their businesses sometimes. Among high-tech startups founded in Silicon Valley since 1990, 4.65 percent had changed their eight-digit SIC numbers at least once by 2001. For Boston and 23 Washington, D.C., the number was 2.59 percent and 2.54 percent, respectively. Although a high percentage of changing SIC numbers may imply a fast-changing local economy, we have little information to tell why firms exit to other industries. Rate of Mortality Table 2.5 describes the death of high-tech establishments by size between 1990 and 2000. Between 30 and 50 percent of establishments died during those 11 years. Establishments that hire fewer than 20 people have a higher chance of failing and hence provide less job security to their employees. Those with over 5,000 employees are also more likely than midsized establishments to fail, although the small sample size of establishments in that category suggests caution in the comparison. Although small establishments are more likely to disappear, the death of large establishments has a much greater effect on the labor market. Whereas the death of 21,967 establishments under size 20 left 84,453 people jobless, the death of 18 establishments with more than 2,500 employees eliminated 102,518 jobs. Figure 2.7 plots the survival rates of high-tech start-ups in Silicon Valley during 1990–2000. Nonservice start-ups have higher survival Table 2.5 Death of High-Tech Establishments in Silicon Valley, 1990–2000 Establishment Size 0–4 5–9 10–19 20–50 51–100 101–250 251–500 501–1,000 1,001–2,500 2,501–5,000 5,000+ Establishments in Sample 33,277 6,722 4,386 3,867 1,423 948 368 138 107 30 13 Establishments Dead by 2001 16,933 3,142 1,892 1,521 557 331 151 42 42 12 6 % Died 50.9 46.7 43.1 39.3 39.1 34.9 41.0 30.4 39.3 40.0 46.2 Job Loss by Death 40,530 19,805 24,118 47,149 42,572 54,505 54,248 32,400 72,234 46,800 55,718 24 Survival rate 1.2 Nonservice firms 1.0 Service firms 0.8 0.6 0.4 0.2 0 0 1 2 3 4 5 6 7 8 9 10 11 Age Figure 2.7—Survival Rates of High-Tech Firms in Silicon Valley rates than service firms in the long run. About 76 percent of the nonservice start-ups and 72 percent of service start-ups are still alive at age five. Only 46 percent of nonservice firms and 42 percent of service firms are still in business at age ten. The third year seems to be the most dangerous age. About 15 percent of Silicon Valley’s high-tech start-ups in service industries and 9 percent of those in nonservice industries died at that age. Figure 2.8 compares the survival rates of high-tech firms in Silicon Valley, Boston, and Washington, D.C. In nonservice industries, the survival rates are almost identical in the three areas. In service industries, firms in Silicon Valley have a better chance to survive than those in the other two regions. The relative size of the service industries is larger in Boston and Washington (Table 2.1), which may imply that service firms in those areas are less efficient or face harsher competition and hence have lower survival rates. Merger and Acquisition Acquisition is the generic term used to describe a transfer of ownership. A corporate acquisition occurs when a buyer purchases the stock or assets of a corporation. A merger has a strict legal meaning that 25 Survival rate 0.9 Silicon Valley 0.8 Boston Washington, D.C. 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 At age 5 At age 10 Nonservice At age 5 At age 10 Service Figure 2.8—Comparison of Survival Rates refers to the process in which one corporation is combined with and disappears into another. All mergers occur as specific transactions in accordance with the laws of the states where the firms are incorporated. Merger is a narrow technical term for a particular legal procedure that may or may not happen after an acquisition. The post-deal manner of operating or controlling a firm has no bearing on whether a merger has occurred. With regard to the NETS dataset, we consider a merger or acquisition to have happened if a firm is not a “branch” or “subsidiary” at its starting year but becomes a “branch” or “subsidiary” at the ending year.3 Figure 2.9 shows the percentage of high-tech firms acquired in each region by 2001. Note that the cohort year refers to the founding date of the firms that were acquired. The acquisition did not necessarily happen that year. In most cases, the acquisition happened a few years later. Overall, firms in Silicon Valley are most likely to change ownership. ____________ 3Alternatively, we could say a firm has changed ownership through M&A if it now has a “headquarter DUNS number” different from its own DUNS number. This gives almost identical results. 26 Percentage 4.5 4.0 Silicon Valley Boston 3.5 Washington, D.C. 3.0 2.5 2.0 1.5 1.0 0.5 0 1990 1991 1992 1993 1994 1995 1996 Founding year 1997 1998 1999 Figure 2.9—Percentage of Firms Acquired by 2001 Those in Washington, D.C., are least likely to be acquired. Among each cohort in each region, less than 4 percent of firms founded in the 1990s had been acquired by 2001. This is a relatively small number compared to how many had gone out of business. As we see in the next chapter, venture-backed firms are much more likely to be bought. Table 2.6 lists the top headquarter states whose firms tend to acquire high-tech start-ups in the three high-tech regions. Not surprisingly, a large proportion of the start-ups were acquired by local firms: California firms top the acquisition list in Silicon Valley, Massachusetts firms bought more high-tech start-ups in the Boston area, and firms in Virginia and Maryland acquired more high-tech start-ups in the Washington, D.C., area. Whereas California firms bought 56 percent of the start-ups acquired in Silicon Valley, Massachusetts firms acquired only 36 percent of those in the Boston area. In Washington, D.C., firms in Maryland, Virginia, and the city Washington bought 45 percent of the firms. Firms in California, New York, New Jersey, and Massachusetts have a strong showing in all three high-tech centers, which probably reflects the fact that those states have more established hightech companies than other states. 27 Table 2.6 Top Headquarter States of Firms Acquired During 1990–2001 Silicon Valley (Total: 1,376) State Cases 1 California 769 2 New York 97 3 Massachusetts 69 4 New Jersey 45 5 Texas 45 6 Pennsylvania 36 7 Florida 26 8 Illinois 26 9 Minnesota 24 10 Virginia 22 Boston (Total: 965) State Cases Massachusetts 350 California 134 New York 115 New Jersey 32 Texas 32 Illinois 31 Connecticut 28 Pennsylvania 27 Florida 20 Maryland 17 Washington, D.C. (Total: 814) State Cases Virginia 211 Maryland 124 California 81 New York 72 New Jersey 37 Massachusetts 36 Texas 33 Washington, D.C. 29 Florida 23 Pennsylvania 23 Job Creation by Start-Ups In this study, we refer to firms that are five years old or younger as start-ups. When new firms are founded, they create jobs. Yet many start-ups fail long before they become mature, thereby eliminating jobs. To pick up the net effect, we track the total employment of high-tech start-ups younger than certain ages, which is presented in Figure 2.10. Since 1995, high-tech start-ups in Silicon Valley have offered everincreasing numbers of jobs. In 1995, 47,200 employees worked for high-tech start-ups younger than two years old. By 2001, that number increased to 69,200. In 1998, start-ups younger than age five offered 132,500 high-tech jobs; the number had risen to 159,300 by 2001. To assess the relative importance of start-ups as job creators, we calculate the employment of start-ups younger than age five as the percentage of total high-tech employment in Silicon Valley and compare it with the same measure for the Boston area and Washington, D.C. (Figure 2.11). During 1998–2001, start-ups younger than age five consistently accounted for more than 20 percent of the high-tech employment in Silicon Valley. The percentage increased from 21.9 percent in 1998 to 23.7 percent in 2001. This means that jobs offered by start-ups grew faster than the total high-tech sector in the valley. The measure for Boston is a little higher and more stable—about 24 percent 28 Employment 180,000 160,000 140,000 Employment of start-ups younger than age 5 Employment of start-ups younger than age 4 Employment of start-ups younger than age 3 Employment of start-ups younger than age 2 120,000 100,000 80,000 60,000 40,000 20,000 0 1995 1996 1997 1998 1999 2000 2001 Percentage Figure 2.10—Employment of High-Tech Start-Ups in Silicon Valley 35 Silicon Valley 30 Boston Washington, D.C. 25 20 15 10 5 0 1998 1999 2000 2001 Figure 2.11—Employment of High-Tech Start-Ups Younger Than Age Five as a Percentage of Total High-Tech Employment 29 during the four years. The measure in the Washington, D.C., area is significantly higher than those in the other two regions. In 2001, startups offered 128,200 jobs in Washington, D.C., which amounted to 28.6 percent of the total employment in high-tech industries. In 1998, the percentage was even higher, when one out of every three employees in the high-tech sector worked for a start-up that was younger than five years old. Conclusion The whole picture of entrepreneurial activities, as presented here, differs somewhat from the public’s general impression. The media tend to direct attention to a small group of venture-backed firms. In fact, thousands of new firms are founded each year in Silicon Valley; venturebacked start-ups represent only a small proportion of the total. The public is too familiar with stories about the explosive growth of Silicon Valley start-ups but, in fact, a large proportion of every cohort of new firms founded in the valley will never hire more than five people. Hightech start-ups in service industries grow slower than other high-tech startups. Start-ups have been major job creators in Silicon Valley during the past decade; firms founded after 1990 created almost all the new jobs added to the region’s high-tech sector during 1990–2001. However, even during the decade characterized by the Internet boom, firm mortality rate was quite high in Silicon Valley. More than half of the firms started during the decade went out of business by age ten. 30 3. Venture-Backed Start-Ups in Silicon Valley This chapter examines venture-capital-backed start-ups, which are more innovative and growth-oriented than other high-tech start-ups. Venture capital refers to money managed by professionals who invest in young, rapidly growing companies that have the potential to develop into significant economic contributors. Venture capital is an important source of equity for start-up companies, particularly in the high-tech sector. In the San Francisco Bay Area, there has been a long tradition of wealthy people financing new technology firms. Yet, professional venture capital activity started later in the Bay Area than in the Boston area (Bygrave and Timmons, 1992; Kenney and Florida, 2000). In 1957, when Robert Noyce and seven fellow engineers left Shockley Semiconductor Laboratories to start their own business, they had to go to the East Coast to look for capital. The first West Coast venture capital firm—Draper, Gaither & Anderson—was not founded until 1958. The venture capital industry grew hand in hand with the high-tech industries in Silicon Valley. Since the 1960s, venture capitalists have been involved in every major successful company. Today, venture capital has become an intrinsic part of any story about Silicon Valley. Sand Hill Road in Menlo Park, the cluster of Silicon Valley’s venture capital firms, is virtually synonymous with venture investing. Venture Capital in Silicon Valley Figure 3.1 traces the nominal amount of venture capital invested in the United States and Silicon Valley over the ten years from 1992 to 2001. The trend is characterized by two big jumps and one severe crash. Between 1992 and 1994, venture capital investment first increased from $9.2 billion to $10 billion and then dropped to $8 billion. Compared to 31 Investment ($ billions) Silicon Valley as a % of U.S. total 120 30 100 25 80 20 60 U.S. total Silicon Valley 40 Silicon Valley as a % of U.S. total 15 10 20 5 00 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.1—Total Venture Capital Investment, 1992–2001 what happened later, this 10 percent increase and 20 percent decline seem to be negligible changes. Between 1994 and 1996, venture capital investment first experienced a 66 percent increase over the year, followed by another 59 percent growth. During these two years, venture capital investment jumped from $8 billion to $21.3 billion, stimulated by the promising Internet revolution. The year 1997 was relatively quiet, with venture capital investment dropping slightly to $20.4 billion. The next three years can only be described as mania: Venture capital investment increased first by 20 percent, then by 173 percent, and finally by 66 percent, ending with a total of $112.2 billion in 2000. In nominal dollars, venture capital investment in 2000 was 14 times as much as it was in 1994. This mirrors the Internet bubble seen in the NASDAQ index. The burst of the bubble is also reflected in venture capital investment. In 2001, the total crashed down to $32.5 billion, a 71 percent decline. Yet, in spite of this big falloff, the year 2001 still represents the third most heavily invested year in venture capital history. Venture capital invested in Silicon Valley followed a similar trend over the ten years. At its peak in 2000, Silicon Valley attracted nearly 32 $28 billion of venture capital investment. The decline in investment in 2001 also appeared in Silicon Valley. Still, the $7.7 billion invested in that year is the third-largest number the region has ever witnessed, second only to the venture investments in 1999 and 2000. In terms of the proportion of the U.S. total, Silicon Valley’s share has increased over the decade. In 1992, 18.7 percent of the total investment took place in Silicon Valley; in 1993, the number dropped slightly to 17.6 percent. Yet at its peak in 2000, Silicon Valley accounted for 24.8 percent of the U.S. total. Figure 3.2 compares Silicon Valley with the Bay Area as a whole, the Boston area, and Washington, D.C. Boston and Washington also experienced a large increase in venture capital investment during the late 1990s, following the national trend. However, the increases in Boston and Washington are not nearly as sharp as those in Silicon Valley and the Bay Area. It is particularly worth noting that the trend in the Bay Area shot up higher than that in Silicon Valley in the peak year 2000. That year, Bay Area firms outside Silicon Valley took in more than $10 billion 45 40 Bay Area Silicon Valley 35 Boston 30 Washington, D.C. 25 20 15 10 5 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.2—Total Venture Capital Investment, by Region, 1992–2001 33 Investment ($ billions) of venture capital. This may represent a big spillover from Silicon Valley. One possibility is that too much money was chasing too few entrepreneurs in Silicon Valley and venture capitalists had to look for opportunities nearby; or, more likely, Silicon Valley simply became too crowded and expensive, making adjacent metro areas such as San Francisco and Oakland more attractive. Table 3.1 summarizes the venture capital raised by each industry in Silicon Valley during 1992–2001. Software, communication, consumer/business services, semiconductor, electronics, information services, medical devices, and biopharmaceutical industries account for more than 96 percent of the total investment in Silicon Valley. The top three industries alone—software, communication, and consumer/business services—absorbed 63 percent of the total investment. These are also the top three industries in the nation as a whole, accounting for 59 percent of the total investment, although it is Table 3.1 Real Venture Capital Investment, by Industry in Silicon Valley, 1992–2001 Industry Software Communication Consumer/business services Semiconductor Electronics Information services Medical devices Biopharmaceutical Retailing Medical information services Advance/special material and chemical Other Healthcare Consumer/business products Energy Agriculture Total aIn 1996 dollars. Venture Capital Raised ($ millions)a 18,738.19 16,668.09 9,364.75 7,038.37 4,740.20 4,310.70 4,201.78 3,431.67 1,314.42 693.58 321.41 108.44 66.46 57.20 18.63 — 71,073.89 % of Total 26.36 23.45 13.18 9.90 6.67 6.07 5.91 4.83 1.85 0.98 0.45 0.15 0.09 0.08 0.03 — 100 No. of Deals 2,027 1,075 757 632 467 419 489 275 74 75 29 14 7 23 5 1 6,369 34 the communication industry that tops the U.S. list. The top three industries are all very much Internet-related, clearly indicating that the 1990s were the “Internet decade” for the venture capital world. Ranked eighth in the United States, the semiconductor industry is ranked fourth in Silicon Valley. Thus, the industry for which Silicon Valley was named is still relatively well-invested. Although the biopharmaceutical industry ranks fifth in the United States, it holds only the eighth position in Silicon Valley. This is partly because the biotech industry in the Bay Area is most heavily concentrated in South San Francisco, which is outside Silicon Valley by our definition. Firm Formation Figure 3.3 traces the trend of venture-backed start-ups by their founding year. The number of such start-ups steadily increased during the 1990s, peaking in 1999 and then declining sharply in 2000 and 2001. The decline reflects the burst of the Internet bubble and an economy heading toward a recession. Since it is possible that some startups founded in 2000 and 2001 will not complete their first round of financing until after 2001 and hence are not included in our data, the actual decline could be less serious than reflected in our data. The trend in Silicon Valley (where, on average, 22 percent of venture-backed startups are located) roughly parallels the national trend. Figure 3.4 depicts the trend of start-up formation for different hightech regions. Silicon Valley substantially outperformed the Boston and Washington, D.C., areas, although the three regions follow quite similar trends. In Silicon Valley, 84 start-ups founded in 1994 were financed by venture capital; the number steeply increased to 375 in 1999. During the same period, the number increased from 55 to 147 in the Boston area and from 12 to 77 in the Washington area. Percentagewise, the Washington area experienced a larger increase than Silicon Valley. The San Francisco Bay Area as a whole experienced intensive entrepreneurial activities in the late 1990s. During 1998–1999, the peak years of the Internet boom, more venture-backed start-ups were founded in the Bay Area than in the Boston area, even when excluding Silicon Valley from the Bay Area. 35 Silicon Valley as a % of U.S. total Number 2,000 25 1,800 1,600 20 1,400 1,200 1,000 800 600 U.S. total Silicon Valley Silicon Valley as a % of U.S. total 15 10 400 5 200 00 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Founding year SOURCE: Author’s calculations from the VentureOne database. Figure 3.3—Venture-Backed Start-Ups, 1990–2001 700 Bay Area 600 Silicon Valley Boston 500 Washington, D.C. 400 300 200 100 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Founding year SOURCE: Author’s calculations from the VentureOne database. Figure 3.4—Venture-Backed Start-Ups, by Region, 1990–2001 Number 36 Although so many venture-backed start-ups were founded in the late 1990s, entrepreneurs faced less-stringent capital constraints. The bull market in stocks and the enormous successes of early Internet-related start-ups attracted a large amount of money into the venture capital industry. As Figure 3.5 shows, start-ups founded in the late 1990s were much more generously financed than previous cohorts. In Silicon Valley, the average amount of venture capital per deal in 1992 was $6.33 million. By 1998, the average amount had climbed to $8.64 million. In 1999 and 2000, abundant venture capital showered on Silicon Valley: The average amount per deal jumped to $16.24 million in 1999 and further shot up to $22.34 million in 2000. Even in late 2001, when Silicon Valley had entered a deep recession, the venture capital industry still found itself in a situation of “too much money chasing too few ideas.” In the end, entrepreneurial ideas were exhausted, not the venture capital. In 2002, many venture capital funds had to downsize and return committed cash to investors because of lack of good opportunities (“The VCs Don’t Want Your Money Anymore,” July 29, 2002). Average venture capital per deal follows a similar trend in the Boston and San Francisco Bay areas. In the Boston area, the average amount dramatically increased from $6.25 million in 1998 to $18.41 million in 25 United States Silicon Valley 20 Boston Washington, D.C. 15 10 5 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.5—Average Amount of Venture Capital Raised per Deal, 1992–2001 37 Average raised ($ millions) 2000. The trend in Washington, D.C., is different. In 1999 and 2000, while venture capital deals were getting fat in Silicon Valley and Boston, they were shrinking in Washington, D.C. Also interesting to note is that the average amount of venture capital per deal was considerably higher in Washington, D.C., than in other areas during 1993–1996 and 1998. For example, in 1993, 1994, and 1998, the average in Washington, D.C., is at least 100 percent higher than in Silicon Valley or Boston. A closer look at the data for the Washington, D.C., area reveals that during 1993–1996 and 1998 a disproportionately large share of venture capital was invested in the communication and healthcare industries, with both tending to acquire extremely big deals. How quickly start-ups are able to obtain venture capital is also an indicator of the availability of capital in a region. We calculated the average span of time between the founding date of a start-up and the closing date of its first round of financing. Figure 3.6 shows that Silicon Valley firms are financed more quickly than firms elsewhere. In Silicon Valley, start-ups on average have raised their first round of venture capital at the age of 11.59 months. For the whole Bay Area, the average age is 11.86 months, only slightly higher than in Silicon Valley. The number is 16.58 for Boston and 16.62 for the nation as a whole. In Washington, D.C., it takes 22.54 months to close the first round of financing. 25 20 Average age (months) 15 10 5 0 United States Silicon Valley Bay Area Boston SOURCE: Author’s calculations from the VentureOne database. Washington, D.C. Figure 3.6—Average Start-Up Age at First-Round Financing 38 One naturally wonders whether Silicon Valley’s time-efficiency is due to its specific industry composition, since the time needed for venture capital financing may be inherently different from one industry to another. Figure 3.7 compares average firm age at the first round of venture capital financing in Silicon Valley with the national average and the Boston area average within each industry. These five industries are the top five in Silicon Valley, accounting for 80 percent of its venture capital investment. Clearly, in each industry, firms in Silicon Valley are financed more quickly. For example, Silicon Valley start-ups in the software industry can have their first rounds of venture capital in place six months sooner than Boston start-ups in the same industry. In the electronics industry, the time advantage is 7.6 months. In consumer or business services, firms in Boston are on average 19.4 months old when their first round of venture financing is completed; those in Silicon Valley are only 10.3 months old. In fact, in 14 out of 16 industry segments, Silicon Valley firms take shorter time to get venture capital 25 U.S. total Boston Silicon Valley 20 Average age (months) 15 10 5 0 Computers/ communications Consumer/ business services Electronics Semiconductor SOURCE: Author’s calculations from the VentureOne database. Software Figure 3.7—Average Start-Up Age at First-Round Financing, by Industry 39 than both the national average and the average time in Boston. The two exceptions are the healthcare industry and “other,” in which only six start-ups in Silicon Valley got financed during 1992–2001. Several possible reasons may explain the promptness of the venture capital financing in Silicon Valley: (1) The well-developed venture capital industry in the region allows start-ups to find financing locally and hence speeds up the process; (2) the well-connected business networks in Silicon Valley enable entrepreneurs to find venture capitalists (or the other way around) more quickly; or simply (3) venture capitalists in Silicon Valley work differently from their counterparts elsewhere. Start-ups in Silicon Valley naturally enjoy some advantages because of the abundance of local capital. It is well known that venture capital firms tend to finance local start-ups, so that they can closely monitor their performance and provide management guidance or assistance if needed. Silicon Valley has the world’s largest venture capital cluster. Thus, firm founders in Silicon Valley have relatively easy access to capital. However, a large venture capital industry does not seem to fully explain quick venture capital finance. For example, the Seattle area has a much smaller venture capital industry than the Boston area. In fact, Boston is undoubtedly the number two venture capital cluster in the world. According to VentureOne’s Venture Capital Sourcebook (2001), Massachusetts has 94 venture capital firms and Washington state has only 26. During 1992–2001, venture capital investment in the Boston area amounted to $31.1 billion; that number is only $10.1 billion for Seattle. However, in spite of the significant size differences in these venture capital industries, start-ups in Seattle received faster venture capital financing than those in the Boston area (16.2 months compared to 16.6 months). Thus, the proximity to considerable capital does not guarantee quick access. Silicon Valley’s risk-tolerating culture might be the real reason for the quick venture capital financing in the region. After interviewing individuals who had worked in both the Boston area and Silicon Valley, Saxenian (1994) observed that “East Coast venture capitalists were more formal and conservative in their investment strategies.” The interviewees’ experiences in the two regions help us understand the cultural difference. An entrepreneur in Silicon Valley told Saxenian, 40 “When I started Convergent [Technologies], I got commitments for $2.5 million in 20 minutes from three people over lunch who saw me write the business plan on the back of a napkin. They believed in me. In Boston, you can’t do that. It’s much more formal.” Another businessman says, “There is no real venture capital in Massachusetts. The venture capital community is a bunch of very conservative bankers. They are radically different from the venture capitalists in Silicon Valley, who have all been operational people in companies. Unless you’ve proven yourself a hundred times over, you’ll never get any money” (Saxenian, 1994). Although those comments were referring to the situation in the 1980s, it is likely that Silicon Valley preserved such cultural advantage in the 1990s. Whatever the reasons, quick financing probably gave entrepreneurs in Silicon Valley a head start over those in other regions. In the fastmoving high-tech sector, a year means a lifetime. And therefore, facilitated by local venture capital firms, innovative start-ups in Silicon Valley may enjoy some first-mover’s advantages. Ownership Status and Profitability The VentureOne data have specific variables indicating a firm’s business status and ownership status. In particular, we know with certainty whether a firm went out of business or whether it merged with another firm; we do not have to infer such events from other variables, as with the NETS data. This section focuses on ownership changes and the economic performance of venture-backed start-ups. A general impression about venture capital investment is that it is very risky. However, this is not reflected in the disbanding rate of venture-backed firms. Figure 3.8 depicts the ownership status of such firms in Silicon Valley as of the fourth quarter of 2001. In each cohort of Silicon Valley start-ups, those that have gone out of business never amount to more than 16 percent of the total. It seems that if a start-up can survive the first two years, it is very likely to succeed. As time goes by, more and more start-ups are acquired by or merged with other firms. About one-third of start-ups change ownership through merger and acquisition (M&A) before they are ten years old. This is a much higher percentage than we found in the NETS data for all the start-ups in the 41 Percentage Publicly held Private Acquired/merged Out of business 100 80 60 40 20 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Founding year SOURCE: Author’s calculations from the VentureOne database. Figure 3.8—Ownership Status of Venture-Backed Start-Ups in Silicon Valley, 2001 preceding chapter, which suggests that start-ups that are not venturebacked are much less likely to be acquired. Many start-ups will become publicly held through IPOs (Initial Public Offerings). In Silicon Valley, nearly 30 percent of venture-backed start-ups founded before 1995 had gone public by the end of 2001. IPOs and M&As provide channels for venture capitalists to exit and pay back their investors. Some start-ups remain privately held, but that proportion is declining over time. Figure 3.9 presents the ownership status of all venture-backed startups in the United States. The overall picture is similar to what we see in Silicon Valley: The disbanding rate stabilizes after two years, more and more start-ups go through M&A or IPO over time, and the number of private firms declines with time. It is also interesting to compare the ownership outcomes of Silicon Valley start-ups with the U.S. average. In each panel in Figure 3.10, a positive value means the examined proportion in Silicon Valley is higher 42 Percentage 100 90 80 70 60 50 40 30 20 10 0 1992 1993 1994 Publicly held Private Acquired/merged Out of business 1995 1996 1997 1998 Founding year 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.9—Ownership Status of Venture-Backed Start-Ups in the United States, 2001 than the national average. As the figures show, venture-backed start-ups in Silicon Valley are more likely to be acquired by or merged with other firms. The proportion of M&A is consistently higher than the national average for every cohort of start-ups. For example, M&A activities among start-ups founded in Silicon Valley in each year from 1993 to 1996 are at least 6 percent higher than the U.S. average. In the 1994 cohort, 36 percent of Silicon Valley start-ups went through M&A, compared to only 23 percent of the U.S. total. Generally speaking, M&A activities are more common in high-tech industries than in other sectors. According to Mergerstat’s industry report, 49 industries completed 7,518 M&A deals in 2002. The software industry alone accounted for 1,347 deals, 18 percent of the total. If we include computer hardware, communications, electronics, drugs, health services, and aerospace and defense, the seven high-tech industries account for nearly one-third of the total M&A deals. This is a 43 Percentage 15 10 5 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 M&A 15 10 5 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 IPO 5 0 –5 –10 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Out of business 10 0 –10 –20 –30 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Private SOURCE: Author’s calculations from the VentureOne database. Figure 3.10—Differences in Ownership Status in Each Cohort of VentureBacked Start-Ups: Silicon Valley Compared to the United States very high fraction given that even a much more broadly defined hightech sector produces less than 11 percent of the U.S. gross domestic product (GDP) (DeVol, 1999). M&A is more active in the high-tech sector for several reasons. On the one hand, for many start-up founders, being bought out is a major avenue to the aspired financial success; at the same time, M&A provides an exit channel by which venture capitalists collect the return on their investments. On the other hand, many 44 established companies in the high-tech sector have incentives to acquire start-ups. The most renowned example is Cisco, which has bought 76 high-tech start-ups since 1993. In fact, Cisco’s practice is so successful that it coined a new term: acquisition and development (A&D). An established company’s typical motivations for buying start-ups include acquiring a technology faster than through internal development, buying market share and presence, and buying talented people (Paulson, 2001). Two possible reasons may explain the fact that a higher proportion of Silicon Valley start-ups exit by M&A. First, Silicon Valley is no doubt the largest cluster of successful high-tech companies in the nation, all of which are potential buyers of young start-up firms. Being close to giants raises the possibility of being acquired. Second, Silicon Valley has probably the best developed networks that service the high-tech sector, including investment banks, venture capital firms, law firms, accountants, and consultants. They are all matchmakers that help form M&A deals. Compared to the U.S. average, Silicon Valley venture-backed startups are also more likely to go public. This is true for every cohort. The difference is especially striking for start-ups founded before 1996. Among the 2,058 start-ups founded in the United States during 1992– 1995, 403, or 20 percent of the total, had gone public by the end of 2001. In Silicon Valley, however, 118 out of 412 start-ups, or 29 percent of the total, founded in the same period were traded on the stock market by late 2001. For those founded in 1994 and 1995, the IPO rate is 12 percent higher in Silicon Valley. These two years inaugurated the era of the Internet revolution. Venture-backed start-ups in these cohorts include high-profile pioneers such as eBay, Netscape, and Yahoo. Figures 3.11 and 3.12 depict the business status of venture-backed start-ups in Silicon Valley and the nation as a whole, given that they were not disbanded. In general, only a small proportion of venture-backed start-ups founded from 1992 to 2001 were showing a profit in the fourth quarter of 2001. The older a start-up, the more likely it is profitable. However, even among the earliest cohort, those founded in 1992, less than 14 percent in Silicon Valley and less than 20 percent in the nation were making a profit in 2001. A majority of the start-ups less than two years old were still developing or testing products. 45 Percentage 100 90 80 70 60 50 40 30 20 10 0 1992 1993 1994 Profitable Shipping product Clinic/beta trial Product development Starting up or restarting 1995 1996 1997 1998 Founding year 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.11—Business Status of Venture-Backed Start-Ups in Silicon Valley, 2001 The proportion of profitable start-ups in Silicon Valley is lower than the national average in every cohort. For cohorts founded before 1995, the proportion of profitable start-ups in the United States is always at least 5 percent higher than the proportion in the valley. Moreover, startups in Silicon Valley are more likely to be in the product development or testing stages than the U.S. average. Among every cohort founded after 1995, Silicon Valley has a higher percentage of such firms. Given that Silicon Valley houses about one-fifth of the venture-backed start-ups in the nation, the difference between Silicon Valley and the rest of nation should be much more significant. This difference and the fact that start-ups in Silicon Valley have quick access to venture capital seem to suggest that the venture capital investments in the valley bear more risks than those in the rest of the United States. On the one hand, venture capitalists in Silicon Valley may have bet on many “bad” ideas and will never profit from them; on 46 Percentage 100 90 80 70 60 50 40 30 20 10 0 1992 1993 1994 Profitable Shipping product Clinic/beta trial Product development Starting up or restarting 1995 1996 1997 1998 Founding year 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.12—Business Status of Venture-Backed Start-Ups in the United States, 2001 the other hand, they may have put money into very long term deals that will take many years to reach profitability. Neither of these is necessarily a bad strategy. In venture capital investment, the returns from a few superstar deals are usually more than enough to cover the money lost in many bad deals. The venture capitalists in Silicon Valley may have acted quickly to take the first mover’s advantage in producing superstars such as Yahoo and Netscape at the cost of investing in many bad deals. In the end, although fewer of the start-ups make profits, the venture capitalists’ returns from Silicon Valley investments may not be below the national average. Spinoffs Who founds high-tech start-ups? One group of entrepreneurs who have attracted researchers’ attention is previous employees of incumbent firms. In the high-tech sector, employees sometimes leave their 47 employers and start their own firms in the same industry, which are called spinoffs. For example, many firms in the semiconductor industry trace their origin to a successful firm in the early years of the industry: Fairchild Semiconductor. The long list of “Fairchildren” includes such prestigious trade names as Intel, National Semiconductor, Advanced Micro Devices, Cypress, Linear Technology, and Xilinix, among others. Spinoffs could emerge for various reasons (Klepper, 2001). For example, employees may wish to capitalize on important discoveries they make while working for a particular firm. However, it is often impossible for them to contract with their employers to commercialize the discoveries, so they start their own firms. In other cases, successful incumbent firms may have difficulties evaluating and implementing certain types of innovations, which gives individual employees opportunities to pursue such innovations by themselves. A classic example is Apple Computers’ cofounder Stephen Wozniak, who used to work for Hewlett-Packard. His design of a personal computer gained no appreciation from senior engineers, so he eventually gave up on his employer and teamed up with Steve Jobs, and the two built their own computers out of a start-up in a garage (Freiberger and Swaine, 2000). The effects of spinoffs are debatable. If employees profit from innovations they make at their previous employers, incumbent firms will have less incentive to spend on R&D. As a consequence, the whole industry may lose out to international competition, such as that faced by the U.S. semiconductor industry. On the other hand, one may argue that spinoffs provide a vehicle for knowledge transfer and hence accelerate innovation. In her renowned study of Silicon Valley and Route 128 in the Boston area, Saxenian (1994) contends that Silicon Valley enjoys a “regional advantage” partly because its culture and institutions encourage employees to move from one firm to another. In particular, employees in Silicon Valley feel free to transfer from established firms to start-ups. This so-called “high-velocity” labor market enables knowledge gained from one firm to quickly spill over to other firms. Since knowledge circulates among a collective learning network instead of traveling in one direction, all firms benefit from this phenomenon. Saxenian argues that established firms in the Boston area, on the contrary, tend to endorse a 48 more inward-looking culture, which encourages loyalty to employers rather than job mobility. In the Boston area, climbing the promotion ladder within a firm is more socially acceptable than taking the risk of starting one’s own business. According to Saxenian, these differential approaches provide an explanation of why Silicon Valley has overtaken the Boston area as the leading high-tech center in the country. Gilson (1999), a law professor at Stanford, further proposed an account of the differential business cultures between Silicon Valley and Route 128. He argued that it is the different legal infrastructures in California and Massachusetts—particularly the enforceability of postemployment covenants not to compete—that make the difference. The so-called “covenants not to compete” are contractual agreements between employees and employers in which the employee promises not to compete with the employer for a certain period of time within a specific geographic area in case the employment relationship terminates. Generally, Massachusetts’ courts have enforced such covenants to protect trade secrets, confidential data, or the employer’s good will. Under California law, such covenants are not enforceable. According to section 16600 of California Business and Professions Code, “every contract by which anyone is restrained from engaging in a lawful profession, trade, or business of any kind is to that extent void.” California courts have consistently referred to this stipulation to prohibit covenants not to compete. Employees in Silicon Valley know that they can leave current employers and found competing start-ups or join other firms in the same business; employers know that they cannot prevent such things from happening. As a result, employers in Silicon Valley adopt an approach that emphasizes both cooperation and competition. Both Saxenian and Gilson discuss Silicon Valley’s high-velocity employment using anecdotal evidence, because of a lack of empirical data. It has yet to be verified with empirical data whether established firms in Silicon Valley indeed have more spinoffs. To resolve this issue, we matched VentureOne’s founder information with firm-level data, so that we could identify where an entrepreneur founded his or her firm. We extracted two groups of venture-backed entrepreneurs by firm location: Silicon Valley and Boston. Using the biographic information of firm founders, we were able to identify which 49 companies a person had ever worked for. If an entrepreneur ever worked for a company or a university, we counted him or her as an “employee founder” from that company or university and the start-up as a “spinoff start-up” from that company or university. The number of employee founders does not necessarily agree with the number of spinoff start-ups. On the one hand, some employee founders turned into “serial entrepreneurs,” founding two or more start-ups; on the other hand, two or more employees may cofound a single start-up. Table 3.2 compares spinoffs from leading firms and universities in Silicon Valley and the Boston area.1 Indeed, leading firms in Silicon Valley significantly outperformed their counterparts in the Boston area in terms of producing entrepreneurs. Raytheon and DEC are probably the two most prestigious names in the Boston area’s high-tech history. DEC Table 3.2 Number of Spinoffs from Leading Institutions in Silicon Valley and the Boston Area Silicon Valleya Boston Areab Employee Spinoff Founders Start-Ups Employee Spinoff Founders Start-Ups Leading Companies Apple Cisco HP Intel Oracle SGI Sun IBM 94 71 Data General 13 41 35 DEC 52 117 99 EMC 9 76 68 Lotus 29 73 57 Prime 5 50 37 Raytheon 7 101 79 Wang 11 82 77 IBM 23 13 41 6 26 5 7 11 23 Leading Universities Stanford UC Berkeley 71 20 64 MIT 20 Harvard 74 63 32 31 aFounder sample size: 2,492. bFounder sample size: 1,157. ____________ 1The VentureOne data cover only start-up founders who have ever been funded by venture capital since 1992. Therefore, no number in Table 3.2 should be interpreted as the total number of spinoffs in the firm’s (or university’s) history. 50 scored the highest in the Boston area with 52 employee founders; Raytheon, the largest employer in Boston’s high-tech sector, with about 15,000 employees locally, produced only seven entrepreneurs according to the VentureOne data.2 Together, DEC and Raytheon spun off 48 venture-backed start-ups, only about half of the 99 spinoffs from Hewlett-Packard. Sun Microsystems, a 20-year-old company in Silicon Valley, has seen more than 100 previous employees become venturebacked entrepreneurs; yet EMC, another big name in the Boston area, founded three years earlier than Sun, had only nine employees who founded start-ups. Apple Computers and Lotus Development Corporation are no doubt two of the most successful pioneers in the early years of the personal computer era. Apple in Silicon Valley has spun off 71 venture-backed start-ups, whereas Lotus in the Boston area lags far behind with only 26 spinoffs. Boston’s Data General, Prime Computer, and Wang Laboratory all once were giants in the minicomputer market created by DEC, but they are all dwarfs in terms of spinoffs, compared to leading firms in Silicon Valley such as Cisco, Oracle, or SGI. Even IBM, the New York–based high-tech conglomerate that has a presence in both areas, has many more spinoffs in Silicon Valley. However, a comparison of the leading universities in the two areas tells a different story. In this case, Boston is doing as well as Silicon Valley, if not better. Among the 1,157 venture-backed entrepreneurs in Boston, 74 have worked at MIT; yet only 71 out of the 2,492 entrepreneurs in Silicon Valley have ever been Stanford employees. Harvard also comes in better than Berkeley, 32 to 20. Notice, we consider here only entrepreneurs who have ever worked at those universities. The number of firm founders who graduated from those universities is a natural alternative measure. Unfortunately, the VentureOne data do not provide such information. Table 3.2 suggests that leading firms in Silicon Valley have spun off more entrepreneurs than those in the Boston area; however, leading universities in Boston have more employees who commercialize their innovations by founding ____________ 2Raytheon’s role as a defense contractor may have placed some restrictions on potential employee founders. 51 start-ups. Whether it is the culture/institutions or the legal infrastructure that enables Silicon Valley to surpass Boston in terms of employee founders and spinoffs, the big difference seems to be in the business world as opposed to academia. Conclusion Since 1994, Silicon Valley has consistently accounted for more than 20 percent of the total venture capital investment in the United States. Venture capital investment in Silicon Valley surged in the late 1990s. On the one hand, the increased venture capital gave a big push to entrepreneurial activities during the Internet boom; on the other hand, the size of venture capital deals during the late 1990s became much bigger. Software, communications, consumer/business services, semiconductor, and electronics industries were the most heavily invested high-tech industries in the region. Silicon Valley start-ups have quicker access to venture capital in almost every industry. Although the exact reasons for this quick access remain unclear, it certainly helps firms in Silicon Valley to get a head start. A preliminary examination of venturebacked firm founders in Silicon Valley and the Boston area confirms that successful firms in Silicon Valley tend to have more spinoffs than their counterparts in Boston. The difference in university spinoffs is not significant between the two regions. 52 4. Firm Relocation in Silicon Valley Silicon Valley is the most renowned example of an industrial cluster. Classical economic theory teaches that a cluster is attractive to firms for multiple reasons, including a specialized labor pool, specialized inputs, proximity to customers, knowledge spillovers, and so on. For high-tech start-ups, the benefits from a cluster may also include easy access to capital and the psychological support that an entrepreneur receives from his peers and the community. It is worth noting that high-tech start-ups and mature firms may have different locational concerns. Start-ups are more dependent on outside resources at the developing stage, and, as newcomers, they suffer from lack of credibility. For these reasons, the entrepreneur’s local connections and his familiarity with local institutions have a big effect on the formation and growth of a start-up. Therefore, we observe two empirical regularities: (1) high-tech firm founders are usually engineers who have working experiences in industrial clusters, where local culture and institutions are favorable to new firms, and (2) high-tech firm founders rarely move outside the immediate area when they decide to start new firms (Cooper and Folta, 2000). Thus, start-ups are more likely to emerge in clusters such as Silicon Valley than in other places. However, mature firms are more likely to follow routinized operations and are more concerned about the costs of doing business. An industrial cluster, once established, tends to face high demand for labor and land and also the overuse of infrastructure. This drives up operating costs for firms in the cluster. At the same time, the overloaded infrastructure may lower the quality of life in the cluster. For these reasons, mature firms may choose to set up branches elsewhere rather than in the cluster. When the costs of doing business are high enough in a cluster, mature firms themselves may consider relocating. 53 Silicon Valley and the San Francisco Bay Area in general have long been notorious for the high costs of living and doing business. This has usually been recognized as a threat to economic growth in the area.1 High costs may deter firm formation in Silicon Valley and may push mature firms away so that the region will not reap the fruit it grows. Our analysis in the previous chapters has shown that entrepreneurial activities were intensive in Silicon Valley during the past decade. Entrepreneurs were probably driving up costs even higher, rather than being scared away by them. In this chapter, we investigate whether mature firms tend to leave Silicon Valley and, if they do, how large the effect is. Although start-ups differ from mature firms, high-tech firms and nontech firms may weigh factors differently when they consider their locations. High-tech firms are in knowledge-intensive businesses. Therefore, they care more about the availability of a well-educated labor pool and knowledge spillovers from their competitors and partners. In contrast, nontech firms might be more responsive to land price, transportation cost, tax burden, and so on. Silicon Valley experienced an Internet revolution in the high-tech sector during the 1990s. The intensive entrepreneurial activities in that period raise the question of whether the booming high-tech sector crowded out nontech firms. In this chapter, we shed some light on this issue. High-Tech and Nontech Relocation Throughout this chapter, firm relocation is measured by locational change at the establishment level. Remember, an establishment is a business or industrial unit at a single physical location. If a firm has multiple establishments, we track each single establishment rather than the firm as a whole. An establishment has moved if its reported address has changed. We do not consider acquisitions that result in ownership changes but not physical movements of establishments. Table 4.1 reveals some interesting facts. First, establishments in Silicon Valley do move. During 1990–2001, 25,485 out of 217,169 establishments changed addresses at least once. Some establishments ____________ 1One could also argue that the high cost of doing business is not a threat but a sign of Silicon Valley’s economic health. 54 Table 4.1 Relocation of Establishments in Silicon Valley, 1990–2001 Never moved Relocated out Relocated in Relocated within Silicon Valley Total High-Tech Sector No. of % in Establishments Total 42,354 82.44 1,490 2.90 894 1.74 6,637 12.92 51,375 100 Nontech Sector No. of % in Establishments Total 149,330 90.07 3,111 1.88 1,834 1.10 11,519 6.95 165,794 100 moved into the area, some moved out, and still others moved around within Silicon Valley. For the purpose of our study, we care more about those establishments moving in and out. Second, high-tech establishments are more likely to move than nontech establishments. Nearly 18 percent of high-tech establishments moved whereas only about 10 percent of nontech establishments relocated. Two possible reasons explain this difference. On the one hand, high-tech firms by their very nature are more mobile. Many hightech firms, especially those in software and research, use portable equipment and occupy little land space. Nontech sectors include establishments in agriculture, forestry, mining, utilities, and government branches and agencies, which are all somewhat attached to well-defined territories. Establishments in nontech manufacturing and services, although generally not fixed to their locations, may have bulky equipment or need large land space, and thus face high moving costs. On the other hand, high-tech establishments tend to develop fast and quickly outgrow their office space. So moving into new office buildings could be more common among them. Third, establishments are more likely to relocate within Silicon Valley. Among all establishments that moved, 79.8 percent remained within Silicon Valley. Fourth, establishments relocating out of Silicon Valley outnumber those moving in. This is true in both the high-tech and nontech sectors. It seems consistent with our intuition that the high costs of doing business may push businesses away from Silicon Valley. 55 Tables 4.2 and 4.3 list the top destination states and cities for establishments that moved out of Silicon Valley. It seems that distance is a very important factor in business relocation. A majority of the establishments moving out of Silicon Valley remained in California— 75.6 percent of high-tech establishments and 84.6 percent of nontech establishments. Those that moved to other states tended to choose Table 4.2 Top Ten Destination States for Establishments Relocating Out of Silicon Valley, 1990–2001 High-Tech Sector Destination No. of State Establishments 1 California 2 Texas 3 Nevada 4 Oregon 5 Colorado 6 Washington 7 Massachusetts 8 Arizona 9 Florida 10 New York 1,126 34 32 30 21 21 20 20 19 18 No. of Employees 12,700 1,570 354 355 1,404 187 932 208 1,944 1,272 Nontech Sector Destination No. of State Establishments California Oregon Arizona Nevada Washington Texas Colorado Florida Illinois Utah 2,631 56 47 40 39 36 33 24 20 18 No. of Employees 27,750 275 348 547 146 1,941 2,075 303 612 229 Table 4.3 Top Ten Destination Cities for Establishments Relocating Out of Silicon Valley, 1990–2001 High-Tech Sector Destination No. of City Establishments 1 San Francisco 2 Hayward 3 Burlingame 4 Pleasanton 5 Santa Cruz 6 San Ramon 7 Oakland 8 South San Francisco 9 Livermore 10 San Diego 148 88 84 75 34 28 27 26 21 16 No. of Employees 1,744 1,211 871 1,097 207 129 411 650 209 542 Nontech Sector Destination No. of City Establishments Hayward Burlingame San Francisco Pleasanton Livermore Santa Cruz San Leandro Oakland South San Francisco Sacramento 286 219 205 131 78 73 62 55 49 43 No. of Employees 3,414 2,968 2,312 2,998 1,360 354 498 490 828 544 56 states in the west. Arizona, Colorado, Nevada, Oregon, Texas, and Washington are among the top ten destination states for both high-tech and nontech establishments. It is interesting to note that Florida appears on both lists, probably for its California-like warm weather. East Coast states Massachusetts and New York are among the top ten destination states for high-tech establishments, possibly because both states have strong high-tech sectors; neither appears on the list for nontech establishments. The importance of distance is also reflected in the lists of top destination cities. San Francisco Bay Area cities occupy the top nine spots for both high-tech and nontech establishments. San Francisco, Hayward, Burlingame, and Pleasanton are the top four on both lists. Among those that leave the Bay Area, high-tech establishments tend to favor San Diego whereas nontech establishments are likely to go to Sacramento. Tables 4.4 and 4.5 list the top origin states and cities for establishments moving into Silicon Valley. A majority of the establishments moving into Silicon Valley—76.6 percent of high-tech establishments and 89.7 percent of nontech establishments—are from other places in California. Among high-tech establishments moving in from outside California, distance does not seem to be the only major determining factor. East Coast states Massachusetts, New York, and New Jersey each had more establishments that moved to Silicon Valley than California’s neighbors such as Arizona, Nevada, and Oregon. It is not surprising that Massachusetts, New York, and Texas follow California on the high-tech list because all of them have strong high-tech economies. Although Silicon Valley is the most concentrated high-tech industrial center in the country, its nontech sectors together have more establishments and hire more employees. Thus, in Table 4.2, it is quite natural to see more nontech than high-tech establishments leaving Silicon Valley. Yet, in Table 4.4, high-tech establishments moving in from outside California outnumbered nontech establishments. It is consistent with our general impression that Silicon Valley is more attractive to high-tech than nontech firms. 57 Table 4.4 Top Ten Origin States for Establishments Relocating Into Silicon Valley, 1990–2001 High-Tech Sector No. of Origin State Establishments 1 California 2 Massachusetts 3 New York 4 Texas 5 Illinois 6 Colorado 7 New Jersey 8 Oregon 9 Pennsylvania 10 Nevada 685 33 29 21 13 12 12 8 6 6 No. of Employees 13,453 1,168 727 182 148 440 222 74 106 69 Nontech Sector No. of Origin State Establishments California New York Texas Washington Nevada New Jersey Oregon Arizona Florida Illinois 1,645 19 15 14 13 13 13 12 8 8 No. of Employees 16,420 920 126 78 95 56 53 80 242 158 Table 4.5 Top Ten Origin Cities for Establishments Relocating Into Silicon Valley, 1990–2001 High-Tech Sector Nontech Sector No. of No. of No. of No. of Origin City Establishments Employees Origin City Establishments Employees 1 San Francisco 2 Burlingame 3 Hayward 4 South San Francisco 5 Santa Cruz 6 Pleasanton 7 Oakland 8 San Diego 9 Los Angeles 10 San Bruno 130 66 48 36 33 24 17 11 11 10 1,225 915 1,306 353 1,575 274 902 195 99 150 San Francisco Burlingame Hayward South San Francisco Oakland San Leandro Santa Cruz San Bruno Pleasanton Los Angeles 267 205 199 74 54 42 38 34 30 26 2,145 2,222 2,958 1,424 349 562 190 523 314 235 Again, top origin cities are mostly in the San Francisco Bay Area, with San Diego and Los Angeles the two exceptions. San Francisco, Burlingame, Hayward, and South San Francisco are the top four on both the high-tech and nontech lists. Comparing Table 4.5 with Table 4.3, we see that more establishments move out of Silicon Valley to adjacent cities than move in from those cities. This is particularly true for nontech sectors, in which we see that not only more establishments but 58 also more employees move out of Silicon Valley to nearby cities. This seems to suggest that Silicon Valley was expanding and its economic activities spilling over into other Bay Area cities. Table 4.6 breaks out relocating high-tech establishments by industry. During 1990–2001, 1,490 establishments moved out of Silicon Valley, which together represented 26,684 jobs; 894 establishments moved into Silicon Valley, with a total of 20,999 employees. Measured by either net establishments or net employees, the Silicon Valley economy is spilling out. In every industry, establishments moving out outnumbered those moving in. Out-moving establishments also had more employees except in two industries: In computers/communications, the net flow of employees is close to zero; in the semiconductor industry, there was a net employment inflow, although more establishments moved out. The two service industries had more moving establishments than the other hightech industries. Yet because service establishments are generally smaller, the relocation in the computer and software industries involved more employees. Table 4.7 summarizes firm relocation by industry group, including both high-tech and nontech industries. During 1990–2001, every sector in Silicon Valley registered a net loss because of firm relocation, measured either by total number of establishments or by total Table 4.6 High-Tech Establishments Relocating Into and Out of Silicon Valley, by Industry, 1990–2001 Bioscience Computers/ communications Defense/aerospace Environmental Semiconductor Software Professional services Innovation services Total Moving Out No. of No. of Establishments Employees 82 2,153 Moving In No. of No. of Establishments Employees 51 1,510 117 15 17 39 281 527 412 1,490 5,737 577 178 1,356 7,023 5,343 4,317 26,684 86 5,740 1 39 12 125 35 2,918 186 5,278 282 2,389 241 3,000 894 20,999 59 Table 4.7 All Establishments Relocating Into and Out of Silicon Valley, by Industry Group, 1990–2001 Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation, communication, and utilities Wholesale trade Retail trade Finance, insurance, and real estate Services Total Moving Out Moving In No. of No. of No. of No. of Establishments Employees Establishments Employees 48 306 5 42 386 3,501 796 17,769 29 237 3 31 184 1,103 356 13,996 177 703 503 346 1,635 4,599 2,596 7,153 4,082 5,984 21,255 62,688 117 444 391 207 995 2,726 1,898 4,495 2,919 1,881 13,452 40,012 employment. Altogether, establishments relocating out of Silicon Valley outnumbered those relocating in by 1,873; those moving out offered 22,676 more jobs than those moving in. The service sector, the largest sector in the Silicon Valley economy, lost the most establishments (640) and jobs (7,803). The finance, manufacturing, construction, and wholesale sectors each lost more than 2,000 jobs. Trans-State Relocation From the state of California’s point of view, the spillover of economic activities from Silicon Valley to other Bay Area cities could be a welcome trend. However, establishments relocating out of the state may be a cause for concern. In this section, we more closely examine Silicon Valley establishments that relocated to or from other states. Tables 4.8 and 4.9 replicate Tables 4.6 and 4.7 for establishments moving between Silicon Valley and outside California. We see similar patterns on a smaller scale in both the high-tech sector and the overall economy: Silicon Valley saw more businesses relocating out than relocating in. This is true even if we separate the high-tech sector into 60 Table 4.8 High-Tech Establishments Moving Between Silicon Valley and Outside California, by Industry, 1990–2001 Bioscience Computers/communications Defense/aerospace Environmental Semiconductor Software Professional services Innovation services Total Moving Out No. of No. of Establishments Employees 27 844 48 4,458 3 337 14 16 880 88 3,925 83 2,208 98 1,328 364 13,984 Moving In No. of No. of Establishments Employees 12 486 31 3,385 00 00 11 750 53 1,614 39 264 63 1,047 209 7,546 Table 4.9 All Establishments Moving Between Silicon Valley and Outside California, by Industry Group, 1990–2001 Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation, communication, and utilities Wholesale trade Retail trade Finance, insurance, and real estate Services Total Moving Out Moving In No. of No. of No. of No. of Establishments Employees Establishments Employees 49 12 43 153 209 8,999 2 0 18 88 5 0 57 6,355 32 680 113 1,530 76 476 45 2,919 321 7,766 844 22,534 18 71 29 18 154 398 189 758 163 156 2,758 10,441 different industries or separate the overall economy into industry groups. It is really striking that the pattern is always the same in every subindustry (or industry group), whether we measure net flow of establishments or of employment. There is a clear pattern that Silicon Valley has been losing enterprises to other states over the past decade. 61 In Table 4.10, we calculate the percentage of employees moving across the state border as they relocated into or out of Silicon Valley. Of all the employees that moved, 32.1 percent did not relocate within California. The high-tech sector saw a higher percentage of interstate movement (45.2 percent) than the overall economy (32.1 percent). Except for the environmental industry, a minor industry in Silicon Valley, every high-tech industry experienced at least 32 percent employee movement one way or the other across the state border. The computers/communications industry tops the list with nearly 70 percent interstate relocation. Defense/aerospace and software also stand out with 54.7 percent and 45 percent, respectively. If we look at the overall economy with both high-tech and nontech sectors, the manufacturing sector has the highest percentage of movement between Silicon Valley and outside California (48.3 percent). At 39.1 percent, the finance sector is the only other sector with above-average trans-state movement. We also calculate the average age of establishments moving between Silicon Valley and other states. Figure 4.1 shows that in both high-tech and nontech sectors, a higher proportion of establishments that moved out were founded before 1990 compared to those that moved into Silicon Valley from other states. Figure 4.2 compares the average age of Table 4.10 Trans-State Relocation as a Percentage of Total Employment That Moved Into or Out of Silicon Valley, 1990–2001 Industry % Industry Group Computers/communications 68.3 Manufacturing Defense/aerospace 54.7 Finance, insurance, and real estate Software 45.0 Services Semiconductor 38.1 Wholesale trade Bioscience 36.3 Transportation, communication, Innovation services 32.5 and utilities Professional services 32.0 Retail trade Environmental 1.3 Construction Overall 45.2 Mining Agriculture, forestry, and fishing Overall % 48.3 39.1 30.3 19.6 19.3 9.1 4.6 2.7 2.6 32.1 62 Percentage Moving out to other states 80 Moving in from other states 73.5 70 67.1 60 58.7 50 41.0 40 30 20 10 0 High-tech Nontech Figure 4.1—Percentage of Moving Establishments Founded Before 1990 5 Moving out to other states Moving in from other states 4.03 4 3.77 3.61 4 3.08 3 3 2 2 1 1 0 High-tech Nontech Figure 4.2—Average Age of Establishments Moving Between Silicon Valley and Other States Average age (years) 63 moving establishments founded during 1990–2000. The out-moving establishments tend to be older in both high-tech and nontech sectors, although the differences are not large. Figure 4.3 tracks the number of jobs eliminated by establishments moving out of California and the number created by those moving to Silicon Valley from other states from 1991 to 2000. The moving activities in both directions seem to have accelerated since 1996, probably because of the Internet boom and the resulting “digital rush” during the late 1990s. The high-tech sector saw more jobs move into Silicon Valley than moved out only in 1996. Nontech sectors had a net inflow of jobs only in 1994. Overall, only the year 1996 saw a net inflow of total jobs. We have seen a clear pattern in firm relocation between Silicon Valley and outside California. High-tech establishments, if they move, are more likely than nontech establishments to move to or from other states. In both high-tech and nontech sectors, more establishments moved out than moved in, whether measured by total number of establishments or total employment. Out-moving establishments are older. All these facts are consistent with our intuition that Silicon Valley Number of jobs 4,500 4,000 3,500 3,000 High-tech jobs moving out High-tech jobs moving in Total jobs moving out Total jobs moving in 2,500 2,000 1,500 1,000 500 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Figure 4.3—Job Movement Between Silicon Valley and Other States, 1991–2000 64 is attractive primarily to high-tech firms, and especially high-tech startups, and is losing mature businesses. A further question is whether the trend we observe is serious enough to worry policymakers and the business community. Mobility vs. Vitality Each year, new firms are created, offering many jobs; at the same time, some existing firms are closed and their employees are laid off. Some firms leave Silicon Valley, taking jobs away; others move into the region, bringing new employment opportunities with them. Tables 4.11 and 4.12 describe these dynamics in Silicon Valley’s job market, in the high-tech sector and in the whole economy, respectively. The tables show that firm relocation has a much smaller effect on the labor market than firm birth and death. We use two indexes to measure the dynamics in the labor market in Silicon Valley: Rate of vitality = (jobs created by new establishments + jobs lost by dead establishments)/total employment; Rate of mobility = (jobs offered by in-moving establishments + jobs taken away by out-moving establishments)/total employment. Figures 4.4 and 4.5 present the rate of vitality, the rate of mobility, and the rate of interstate mobility in the Silicon Valley labor market. On average, the rate of vitality is 14.2 percent in the high-tech sector and 13.3 percent in the whole economy. The rate of mobility is only 0.8 percent in the high-tech sector and 0.7 percent in the overall economy. Compared to firm birth and death, establishment relocation has an almost negligible effect on the labor market. On average, new establishments offer 6.4 percent of Silicon Valley’s high-tech jobs, and dead establishments eliminate 7.8 percent of them. The growth of existing establishments could make up the difference. At the same time, establishments that relocate out of Silicon Valley take away only 0.43 percent of its high-tech jobs, and establishments that moved into the valley offer 0.35 percent of the total high-tech jobs. If we consider the 65 Table 4.11 Employment in the High-Tech Sector of Silicon Valley, 1991–2000 66 Year 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Total Employment 523,496 512,536 545,575 524,722 539,885 565,560 588,031 605,607 606,731 650,331 Total No. of Employees in New Firms 16,487 28,257 60,895 22,154 34,910 54,311 40,672 32,147 25,584 48,941 Total No. of Employees in Dead Firms 36,997 32,597 47,182 20,984 37,649 51,585 38,113 56,006 46,559 72,816 Total No. of Employees Moving Out of Silicon Valley 1,796 1,375 1,607 1,055 1,532 1,411 1,992 4,573 3,059 5,706 Total No. of Employees Moving Into Silicon Valley 639 3,970 566 2,363 856 1,963 2,440 1,691 2,251 2,987 Total No. of Employees Moving Out of California 764 572 882 516 774 296 790 2,844 1,607 3,253 Total No. of Employees Moving Into California 117 143 120 160 217 1,400 254 808 1,524 1,088 Table 4.12 Employment in Silicon Valley, 1991–2000 Year 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Total Employment 1,346,351 1,312,856 1,407,188 1,380,399 1,374,146 1,395,816 1,435,747 1,455,318 1,445,922 1,511,192 Total No. of Employees in New Firms 35,156 54,468 177,262 54,990 75,860 102,531 106,764 75,548 68,350 107,549 Total No. of Employees in Dead Firms 86,673 72,909 87,457 90,645 85,555 119,410 98,418 122,619 106,316 140,620 Total No. of Employees Moving Out of Silicon Valley 5,745 3,656 4,346 2,837 5,277 3,335 5,142 8,711 6,763 9,400 Total No. of Employees Moving Into Silicon Valley 2,710 5,814 2,840 4,063 2,445 3,164 4,889 4,032 3,915 5,373 Total No. of Employees Moving Out of California 1,191 1,075 1,564 933 2,489 764 1,197 3,586 2,123 4,155 Total No. of Employees Moving Into California 163 242 192 782 341 1,482 371 1,109 1,761 2,265 67 Rate (%) 25 Rate of vitality Rate of mobility Rate of interstate mobility 20 15 10 5 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Figure 4.4—Dynamics in Silicon Valley’s High-Tech Labor Market, 1991–2000 20 15 10 Rate of vitality Rate of mobility 5 Rate of interstate mobility 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Figure 4.5—Dynamics in Silicon Valley’s Labor Market, 1991–2000 nontech sectors as well, the overall Silicon Valley economy sees a little less birth, death, and relocating activities, but the relative effect of vitality and mobility is similar. Note that in calculating the rate of vitality, we included all new establishments (i.e., some are new firms and others are new offices or 68 Rate (%) plants set up by existing firms). If we focus only on jobs created by new firms, they account for 2.6 percent of the high-tech employment and 2.1 percent of the overall employment. Again, jobs created by new firms alone are much more than enough to cover job losses by firm relocation. We should thus conclude that although Silicon Valley is losing jobs as a result of firm relocation, the magnitude of the loss is relatively small. In other words, once firms are founded in Silicon Valley, they are very likely to stay there. A vibrant economy in Silicon Valley with intensive entrepreneurial activities and rapid growth will generate job opportunities that are more than enough to compensate for any leakage of jobs. Relocating Out vs. Branching Out In the preceding sections, we have examined whether firms choose to relocate out of Silicon Valley and where they go if they do move out. A related question that also concerns us is whether start-ups in Silicon Valley tend to set up branches elsewhere as they mature. Although very few establishments move to other states, many firms may choose to headquarter in Silicon Valley but to locate production or distribution capacities in other states. In such cases, California does not fully benefit from the economic growth enabled by Silicon Valley. We must recognize that some companies have an absolute need to reach out to their customers and hence have to open offices nationally or even globally. For example, software companies usually earn a large proportion of their revenue from on-site services to their customers, and thus they need to operate at locations near clusters of customers. Consider Oracle, the world’s largest enterprise software company based in Silicon Valley. It has employees all over the world; in the United States alone, it has 78 offices in 76 cities in over 35 states. Another company based in Silicon Valley, Siebel Systems, which specializes in e-business application software, has employees in 31 foreign countries. At the young age of ten, Siebel has already opened 57 offices in 28 states. It makes sense for such firms to expand beyond California. However, not all high-tech firms need to be physically close to customers. An example may shed some light on this issue. Table 4.13 shows that Intel, probably one of the most famous companies in Silicon Valley, in fact hires many more people outside California. Although it has 7,500 69 Table 4.13 Intel Operating Locations in the United States City Santa Clara (headquarters) Hillsboro Chandler Folsom Rio Rancho Hudson Dupont Colorado Springs Parsippany Riverton Austin San Diego Shrewsbury Thousand Oaks Los Angeles Columbia San Luis Obispo Chantilly Irvine San Jose Raleigh State California Oregon Arizona California New Mexico Massachusetts Washington Colorado New Jersey Utah Texas California Massachusetts California California South Carolina California Virginia California California North Carolina No. of Employees 7,500 15,000 10,000 7,300 5,500 2,700 1,500 1,000 900 625 550 400 400 300 250 150 145 140 130 100 70 SOURCE: http://www.intel.com/jobs/usa/sites/index.htm. employees in its Santa Clara headquarters, Intel’s campus in Hillsboro, Oregon, is twice as large. With 15,000 employees in its Hillsboro branch, Intel is the largest private employer in Oregon. Its second-largest location is in Chandler, Arizona, which offers 10,000 jobs. Intel’s campus in Rio Rancho, New Mexico, has 5,500 employees, which makes it the largest private industrial employer in the Albuquerque metropolitan area. A complete NETS dataset would allow us to measure more precisely how many establishments Silicon Valley firms manage outside California. Unfortunately, a dataset for the whole nation is not ready yet. What we can do is to choose some large firms in Silicon Valley, calculate their employment in the Bay Area, and compare those numbers with total 70 employment available for the Duns Business Rankings. According to our calculation, the top 40 firms (by sales) in Silicon Valley together have 32 percent of their total employment in the Bay Area. This ranges from Novellus Systems’ nearly 100 percent to 3Com’s mere 4 percent. As a successful start-up becomes a mature company, it will develop new needs that require different cost-benefit considerations, or it may be that the company simply cannot conduct its business successfully unless it branches out to other locations. Although its headquarters is likely to remain in Silicon Valley, it will look elsewhere to accommodate its growth. A famous example is the hard disk drive industry that was born in Silicon Valley but later had to move manufacturing operations to Southeast Asia to maintain its competitiveness (McKendrick, Doner, and Haggard, 2000). A serious question for the state of California is how, if possible, to keep spillovers from Silicon Valley within California. This is a particularly relevant question for the fast-growing biotech industry in Silicon Valley and the San Francisco Bay Area as a whole. Conclusion Silicon Valley firms do move. In general, more establishments leave the area than move into the area. High-tech establishments are more likely than nontech establishments to move, both into and out of the valley. Establishments moving to Silicon Valley tend to be younger than those moving out. All these findings are consistent with our intuition. Although more establishments relocating out implies that Silicon Valley is losing businesses and job opportunities, it is not a serious problem. On the one hand, establishments moving out tend to go to adjacent cities within the state; on the other hand, new firms created each year overwhelmingly outnumber those moving away, which is more than enough to compensate for the net loss resulting from firm relocation. Thus, instead of worrying about what we might do to keep the businesses, we should focus our attention on how to create new businesses and facilitate their growth. Although most firms founded in Silicon Valley will remain in the region, the most successful ones among them will almost surely set up 71 branches elsewhere for operations such as manufacturing that do not benefit much from the Silicon Valley environment. This creates the possibility for the rest of California to accommodate the branching-out of Silicon Valley’s successful firms. 72 5. Conclusion Silicon Valley’s high-tech sector consists of the most dynamic industries in the economy. These industries have unique features and call for careful analysis. The high-tech economy is driven by innovation, and radical changes usually originate from innovative entrepreneurs starting new firms. For these reasons, we have studied high-tech startups and industry dynamics in Silicon Valley with the intention of discovering how Silicon Valley changed in the past and the lessons we should learn for the future. Major Findings New firms are important for Silicon Valley. As with other hightech centers, Silicon Valley hosts a wide variety of firms. A multitude of small firms coexist with medium and large firms. Each year, many new firms are founded, which collectively are a major driver of the economic dynamics in Silicon Valley. In fact, firms founded after 1990 created almost all of the job growth during 1990–2001. Young start-ups in Silicon Valley consistently attract a large amount of venture capital, which indicates that these firms are very innovative and growth-oriented. Successful start-ups have remade and will continue to remake Silicon Valley. Start-ups in Silicon Valley have quick access to venture capital. On average, it takes 11.6 months for Silicon Valley’s start-ups to complete their first round of venture finance—five months faster than the national average. The quicker access to capital is found in every major industry in Silicon Valley. This gives start-ups in the region a head start, an important advantage in high-tech industries that advance at a very rapid pace. This large first-mover’s advantage implies that start-ups in Silicon Valley will have a better chance to survive, all else equal. Established firms in Silicon Valley spin off more start-ups. Compared to their counterparts in the Boston area, big companies in 73 Silicon Valley have more previous employees who start their own venture-backed businesses. Since engineers in successful firms are in the best position to grasp and commercialize cutting-edge innovations, a high rate of spin-offs helps open new markets and creates new jobs. Previous research discusses Silicon Valley’s high incidence of firm-level spin-offs based on anecdotal evidence and has identified cultural and legal factors to account for it. Although it remains unclear which theory is closer to the truth, for the first time we have confirmed with empirical data that there are indeed more firm-level spin-offs in Silicon Valley than in other high-tech centers. Firm relocation is not a serious problem. High-tech start-ups value the hotbed of innovation because that is where new ideas emerge and entrepreneurs cluster. Silicon Valley is a perfect environment for startups whose major objective is to develop innovative ideas. On the other hand, when firms become mature and enter the phase of mass production or routine services, their major concern becomes sustainability and they naturally care about operating costs. For those firms or, rather, for certain operations of those firms, Silicon Valley is unattractive. We have investigated whether firms leave Silicon Valley when they have evolved out of the start-up stage. We find that indeed more establishments move out of Silicon Valley than move in, and establishments moving out tend to be older. Establishments tend to stay close to Silicon Valley when they move out. In terms of those moving across state borders, Silicon Valley does see a net job loss, because more jobs are relocated to other states than are relocated to Silicon Valley from outside California. However, the data suggest that firm relocation involves a relatively small proportion of the labor force. Firm birth and death cause much more turbulence than firm relocation. In other words, once firms are established in Silicon Valley, they are very likely to remain there, and intensive entrepreneurial activities certainly compensate for the jobs lost through firm relocation. Successful firms in the valley are branching out. Although relocation does not occur at significant levels, established firms in Silicon Valley frequently set up branches elsewhere. For many large high-tech companies headquartered in Silicon Valley, their employment within Silicon Valley itself is only a small proportion of their total employment. 74 Since Silicon Valley is already tightly packed with thousands of firms, fast-growing start-ups are more likely to expand outside the immediate area. As firms expand, they could benefit the rest of California by setting up branches elsewhere in the state. The high-tech sector experiences rapid structural changes. The high-tech sector consists of a number of diverse industries, which follow different dynamics. On the one hand, the fluctuation of the macro economy has distinctive effects on different high-tech industries; on the other hand, technological innovations in different industries—the drivers of growth in those industries—do not arrive simultaneously. As a result, different high-tech industries may follow unsynchronized business cycles. And thus, at different points in time, the “hot spot” of growth may appear in different industries. For example, the 1990s saw a boom in the computer industry, along with a decline in the defense industry. To catch upturns and avoid downturns in high-tech industries, a high-tech center such as Silicon Valley must accommodate rapid structural changes. This implies that a dynamic labor force is necessary. Previous research has emphasized the “high-velocity labor market” through which workers move frequently from one job to another within Silicon Valley. Such a labor market certainly helps the region’s economy adapt to structural changes. In addition, we believe, a set of infrastructure and institutions that enable the labor force to move quickly into and out of Silicon Valley is also crucial for structural changes in the high-tech sector. For example, employment in the software industry in Silicon Valley increased from 48,500 to 114,600 between 1990 and 2001, a phenomenal 136 percent rate of growth. It is impossible to train such a large number of technical workers within such a short period of time. This kind of rapid growth in a certain industry is achievable only through massive migration of the needed labor force. Policy Implications State and local governments played only a minor role in the early years of Silicon Valley. The history of Silicon Valley evolved from a tradition of innovative thinking in the region and industry-university networks such as that between the business world and Stanford University. Government’s largest effect on Silicon Valley’s high-tech 75 sector was probably the purchase of defense products by the federal government during the Cold War era. State and local governments were not actively involved in the region. Yet outside Silicon Valley, the recent trend shows that state and local governments can lend an effective helping hand to a regional high-tech economy. From Seattle and Portland to Austin and Denver, state and local governments all have supportive policies for the local high-tech sector. Governments play even bigger roles in the Silicon Valley clones in the rest of the world, such as Cambridge, England; Helsinki, Finland; Tel Aviv, Israel; Bangalore, India; and Hsinchu, Taiwan (Rosenberg, 2002). To maintain Silicon Valley’s success is by no means an easier task than building a Silicon Valley clone. Silicon Valley today faces more competition than ever from high-tech regional economies both domestically and internationally. Supportive policies have been implemented in metro areas all over the country that aim at grabbing a bigger piece of the high-tech economy. In addition, Silicon Valley’s success today could become its burden tomorrow when innovations again call for changes. How to keep Silicon Valley growing is a big challenge for California’s policymakers. This is especially true today, with the valley struggling through a deep recession. Policies directly related to Silicon Valley include the federal government’s spending on R&D and military goods and its immigration policies, state government’s R&D spending and education policy, and local governments’ land use policies, and so on. In addition, in any other areas where the private sector has no incentive or capability to solve the problems, government must step in. Examples include building infrastructure, training labor, and preventing further energy crises. Several policy implications have emerged from our examination of high-tech start-ups and industry dynamics in Silicon Valley. Promote technological innovation. More than any other sector, the high-tech economy is about innovation and entrepreneurship. Waves of innovation cause business cycles. Silicon Valley has experienced highs and lows many times, and right now the region is struggling in a deep trough. Previous experience proves that Silicon Valley always gets out of a recession on two legs: One is strong demand for high-tech products 76 from the whole economy, and the other is new demand created by innovations that add a new dimension to Silicon Valley’s economy. Although state and local governments can do little to improve the macroeconomic environment of the national economy, they could help promote innovation. University research has always been a major source of innovation, and state government should continue its strong support to research universities. Big budget cuts for the University of California system will severely affect the prospect of the high-tech sector off campus. Moreover, the California delegation in Washington, D.C., should place a high priority on securing R&D dollars for California from the federal government. As the state economy becomes more and more reliant on high-tech industries, support for R&D and innovation not only helps Silicon Valley and the rest of the Bay Area, but it also greatly benefits the Los Angeles and San Diego areas, which are continuing to expand their own high-tech sectors. Encourage firm founding. Our findings show that although some firms do move out of Silicon Valley, it is not a serious problem. On the one hand, they are likely to move to nearby cities and stay within the state; and on the other hand, firm formation and growth create new jobs that overwhelmingly outnumber jobs lost through firm relocation. Job creation in Silicon Valley is primarily achieved by new firms. Thus, instead of worrying about losing businesses because of the high cost of living and doing business in Silicon Valley, state and local governments should encourage firm founding. Offering favorable tax breaks, opening industrial parks, building high-tech incubators, and providing seed capital for commercialization of research are widely used policy levers. Previous research has shown that a primary factor determining a hightech start-up’s location is where its founder would like to live (Cooper and Folta, 2000). Thus, continuously improving the quality of life in Silicon Valley and the Bay Area as a whole is crucial for the vitality of the high-tech economy in this area. Look beyond Silicon Valley. The high-tech sector is not a disconnected economy, nor is Silicon Valley an isolated region. Silicon Valley is well embedded in the San Francisco Bay Area and well connected to the rest of the state economy. Most of the firms relocating out of Silicon Valley migrate to nearby cities in the Bay Area. The rest of 77 the Bay Area has undoubtedly benefited from the proximity of Silicon Valley and has quite a strong high-tech economy. Our data show that entrepreneurial activities in the 1990s were intensive in the whole Bay Area, both inside and outside Silicon Valley. Venture capital investment is also abundant for the rest of the Bay Area. State policies regarding Silicon Valley should take into account Silicon Valley’s connection with the rest of the state economy. For example, many people who work in Silicon Valley live a considerable distance from it, seeking more affordable homes. Thus, housing development and transportation policies in many other Bay Area cities help to solve Silicon Valley’s housing problems. We have also found that large firms in Silicon Valley often hire only a small proportion of their total employees from Silicon Valley or even the Bay Area. This suggests that other regions in the state have the opportunity to benefit from spillover from Silicon Valley by hosting branches of its firms. State government should try to understand not only new firm formation but also the concerns of mature firms in Silicon Valley. In particular, state government could provide incentives for large firms to set up their manufacturing or distribution arms within the state. State government could also improve transportation networks between the Bay Area and the Central Valley that facilitate Silicon Valley’s branching out to the latter area. In addition, local governments in the rest of the Bay Area and in the Central Valley should be more proactive in accommodating businesses branching out from Silicon Valley. Maintain a dynamic labor pool. Two conflicting factors characterize the high-tech labor force. On the one hand, the high-tech sector primarily hires technical workers whose skills are highly specialized and take time to acquire; on the other hand, the high-tech sector is dynamic, with its core technologies evolving quickly. This implies that the skills acquired in school three years ago may be obsolete today. Moreover, certain high-tech industries often experience explosive growth, such as the software industry in the 1990s, which creates a high demand for certain types of technical workers within a short period. Whether Silicon Valley can evolve rapidly hinges upon whether its labor force can quickly upgrade its skills or meet completely new demands. State government should continue to support universities and colleges as 78 vehicles for continuously retooling the labor force. Employers in Silicon Valley should recruit new talent not only through local colleges and universities but also by recruiting and hiring highly qualified immigrants. Immigrants have played an important role in Silicon Valley’s growth. They are a major source of Silicon Valley’s entrepreneurs and innovation. Immigrants also provide a large reserve of high-quality engineers and scientists capable of satisfying sudden surges of demand for certain talents in some industries. State government in cooperation with federal authorities should keep the door open to international talent, both at local universities and in the high-tech industries. This has emerged as a particularly crucial issue because immigration policies have now entered the equation of homeland security. 79 Appendix A Geographic and Industrial Definitions Geographic Definition of Silicon Valley Our definition of Silicon Valley includes all of Santa Clara County and adjacent cities in Alameda, San Mateo, and Santa Cruz Counties. City Santa Clara County All Alameda County Fremont Newark Union City San Mateo County Atherton Belmont East Palo Alto Foster City Menlo Park Redwood City San Carlos San Mateo Santa Cruz County Scotts Valley Zip Code All 94536–39, 94555 94560 94587 94027 94002 94303 94404 94025 94061–65 94070 94400–03 95066–67 Definition of Industry Groups in the NETS Data Used in This Study Industries are listed by their SIC code; “n.e.c.” means not elsewhere classified. 81 Industry Group Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation, communication, electric, gas, and sanitary services Wholesale trade Retail trade Finance, insurance, and real estate Services Government SIC Code 01–09 10–14 15–17 20–39 40–49 50–51 52–59 60–67 70–89 91–97 Definition of High-Tech Industries in the NETS Data Industry Bioscience Drugs Laboratory apparatus and analytical, optical, measuring, and controlling instruments Surgical medical and dental instruments and supplies Medical laboratories Computers/communications Electronic computers Computer storage devices Computer terminals Computer peripheral equipment, n.e.c. Telephone and telegraph apparatus Radio and television broadcasting and communications equipment Communications equipment, n.e.c. Printed circuit boards SIC Code 283 3821, 3823–24, 3827, 3829 384 8071 3571 3572 3575 3577 3661 3663 3669 3672 82 Electronic components, n.e.c. Magnetic and optical recording media Defense/aerospace Small arms, ammunition Electron tubes Aircraft and parts Guided missiles and space vehicles Tanks and tank components Search, detection, navigation, guidance, aeronautical, and nautical systems instruments and equipment Environmental Industrial and commercial fans and blowers and air purification equipment Service industry machinery, n.e.c. Sanitary services Scrap and waste materials Semiconductors Special industry machinery Semiconductors and related devices Instruments for measuring and testing electricity and electrical signals Software Computer programming services Prepackaged software Computer integrated systems design Computer processing and data preparation and processing services Information retrieval services Innovation services Wholesale of computers and computer peripheral equipment and software Wholesale of electronics parts and equipment, n.e.c. Computer facilities management services Computer rental and leasing 3679 3695 348 3671 372 376 3795 381 3564 3589 495 5093 3559 3674 3825 7371 7372 7373 7374 7375 5045 5065 7376 7377 83 Computer maintenance and repair Computer-related services, n.e.c. Engineering services Research and testing services Professional services Commercial printing Manifold business forms Service industries for the printing trade Investors, n.e.c. Advertising Consumer credit reporting agencies Mailing, reproduction, commercial art and photography, and stenographic services Personal supply services Legal services Architectural services Surveying services Accounting, auditing, and bookkeeping services Management and public relations services 7378 7379 8711 873 275 276 279 6799 731 732 733 736 81 8712 8713 872 874 84 Appendix B The Data Here we give a detailed discussion of the two longitudinal databases we used. The NETS Data The NETS database was constructed by Walls & Associates, who derived the raw data from Dun & Bradstreet (D&B). D&B, which has been collecting business data for more than 160 years, offers business-tobusiness credit information on companies throughout the world. The D&B data include information on the location, industry category, ownership, and employment of almost all businesses in the United States. Although the goal of D&B is not to collect and organize data for scholarly research, it does have an incentive to ensure the accuracy of its data. Serious inaccuracies could hurt D&B’s business and might even result in lawsuits. D&B has thus established a complicated quality control system, which has resulted in a relatively accurate and reliable database. However, D&B data are by no means without limitations. The main source of bias comes from its criterion of inclusion. Only firms that seek credit ratings or whose credit ratings are demanded by business partners have an incentive to report their activities to D&B. D&B has no information about businesses that do not report to them. Early evidence suggests that D&B data tend to overrepresent the manufacturing sector and new firms may not be completely covered or not included in their early years of existence. Nonetheless, with all their shortcomings, D&B data are one of the most widely consulted sources of information for academic research, mainly because firm-level data are always hard to acquire and D&B data are conveniently available, cover nearly the whole economy, and are of reasonably good quality. Many previous studies on industry dynamics such as Birch (1987) and Audretsch (1995) have used refined D&B data. 85 Walls & Associates teamed up with D&B to convert their archival establishment data into a time series: the NETS database. Walls & Associates first used D&B’s Duns Marketing Information file, which followed more than 22 million establishments from 1990 to 2001, to determine which establishments were active in January of each year. Then they retrieved information about each establishment from other D&B files (e.g., the credit rating file) to create a time series with rich firm-level information. In the NETS database, the basic unit of observation is the “establishment.” An establishment is a business or industrial unit at a single physical location that produces or distributes goods or provides services. For example, a single store or factory is an establishment. Many companies own or control more than one establishment, and those establishments may be located in different geographic areas and may be engaged in different kinds of business. D&B assigns a unique nine-digit DUNS (Data Universal Numbering System) number to each establishment. D&B also links the DUNS numbers of parent companies, headquarters, subsidiaries, and branches to form corporate family structures. The NETS database has all such information included, so that we are able to tell whether a new establishment is a start-up company or a newly established branch of an existing company. Specifically for the purpose of our study, Walls & Associates cut a PPIC extract from their NETS database. This dataset covers all the establishments that were ever located in 15 counties during 1990–2001. The 15 target counties include: ten counties in the San Francisco Bay Area (Alameda, Contra Costa, Marin, Napa, San Francisco, San Mateo, Santa Clara, Santa Cruz, Solano, and Sonoma),1 two counties in the Boston area (Middlesex and Suffolk), and three counties in the Washington, D.C., area (Arlington, Virginia; Fairfax, Virginia; and Montgomery, Maryland). The data for the ten counties around San Francisco Bay completely cover Silicon Valley as we defined it and, in addition, allow us to look at a bigger picture beyond the valley. The data ____________ 1For most of our analysis, we do not use all the data from the ten counties in the Bay Area. As defined in Appendix A, Silicon Valley covers only Santa Clara County and some adjacent cities in Alameda, San Mateo, and Santa Cruz Counties. 86 for the other two high-tech centers, Boston and Washington, D.C., enable us to make comparisons. One of our tasks is to measure start-up growth over time. Firm growth is usually measured by employment or sales. D&B does collect data on firm sales. However, for various reasons, a very small proportion of firms choose to report their sales numbers. Firms are more likely to report their employment. For example, our dataset has self-reported employment for 84 percent of Silicon Valley’s high-tech firms active in 2001. Small firms are more likely to have missing data. For example, if we exclude the Silicon Valley high-tech firms that never hired more than five people, 97 percent of the employment data will be self-reported. Fortunately, an establishment’s employment data are usually not missing for all years. If a data point is missing for a year between two selfreported data points, then D&B or Walls & Associates will fill it in according to simple smoothing formulae. If there is a missing data point before or after a series of self-reported data points, it is filled in by extrapolation. In rare cases where the employment data are completely missing, D&B or Walls & Associates will enter their estimates based on industry average. To test the reliability of the NETS data, we compared business size distribution in NETS with that available at the Economic Development Department (EDD) of California. The EDD dataset also counts an establishment as a business unit, which makes it comparable with NETS. It defines the employment at an establishment as “insured wage earners on the payroll.” Any employer hiring one or more persons, who pays wages in excess of $100 during a calendar quarter, and who is not engaged in an exempt activity, is subject to the Unemployment Insurance provision of the California Unemployment Insurance code. Table B.1 presents business size distribution in four counties that cover Silicon Valley: Alameda, San Mateo, Santa Clara, and Santa Cruz. Although the NETS data cover self-employed people, we have excluded them from our calculation because the EDD data do not include them. The EDD dataset is a snapshot as of September 2001. We use the NETS data collected in January 2001. As Table B.1 shows, in Alameda and Santa Clara Counties, the NETS covers more establishments than the EDD data in every size 87 Table B.1 Business Size Distribution in NETS and EDD Data, 2001 0–4 Alameda EDD 28,105 NETS 38,201 San Mateo EDD 14,061 NETS 21,277 Santa Clara EDD 27,949 NETS 48,614 Santa Cruz EDD 4,561 NETS 9,487 5–9 6,229 8,507 3,412 4,559 7,794 10,053 1,347 1,879 10–19 4,598 5,298 2,413 2,784 5,569 6,344 909 975 20–49 3,579 4,003 1,918 2,035 4,612 4,968 714 596 50–99 1,511 1,614 743 775 1,934 1,969 238 227 100– 249 788 874 399 425 1,119 1,189 121 104 250– 499 198 212 110 111 278 304 25 16 500– 999 79 88 44 31 120 161 10 8 1,000+ 43 59 22 37 76 119 3 8 NOTE: The EDD data are available at http://www.calmis.cahwnet.gov/file/indsize/ 1sfcoru.htm. category. In San Mateo County, the NETS is bigger than the EDD sample except in one category. In Santa Cruz County, the EDD picks up more firms than NETS in size categories bigger than 20, except for 1,000+. In every county, the NETS sample covers many more small firms that employ fewer than 20 people. The difference in the 0–4 category is most significant. For example, in Santa Cruz County, the NETS data include more than twice as many size 0–4 firms as the EDD data. A comparison of business size distribution for some other years yields similar results: The NETS data always capture far more small firms than the EDD data; although the difference becomes smaller for larger firms, the NETS is still likely to have more of them. A more complete coverage of the small firms is particularly valuable for studying start-ups. Table B.2 compares county-level employment series from the EDD data and those from the NETS. The NETS data consistently produce a larger employment figure. This is true in every year for every county. In some cases, the difference is very large. For example, in 1993 in Santa Clara County, the NETS data documented 30 percent more employees than the EDD data. To some extent, a larger employment number 88 Table B.2 Employment Series in NETS and EDD Data, 1990–2001 89 1990 Alameda 1991 1992 1993 1994 1995 1996 1997 1998 EDD — NETS 709,496 San Mateo —— 679,448 677,404 591,300 750,168 590,600 730,025 607,000 738,186 620,800 733,484 639,100 742,097 660,500 796,386 EDD 295,600 NETS 364,531 Santa Clara 298,100 291,500 351,646 344,535 294,200 365,465 296,300 365,310 305,800 355,550 319,100 361,851 333,300 370,309 345,100 380,917 EDD 819,500 810,900 797,200 802,000 805,000 836,400 885,000 931,700 961,500 NETS 1,055,389 1,017,015 988,208 1,054,477 1,028,791 1,032,777 1,048,374 1,070,466 1,079,035 Santa Cruz EDD 94,900 96,100 94,800 95,400 96,600 97,700 99,200 101,600 103,000 NETS 105,471 101,451 101,219 110,638 113,370 109,298 109,186 112,064 114,591 NOTE: The EDD data are available at http://www.calmis.cahwnet.gov/file/indsize/1sfcoru.htm. 1999 683,600 793,967 357,900 371,438 976,600 1,077,960 103,200 116,455 2000 71,100 802,609 375,800 383,188 1,035,000 1,131,221 105,600 117,184 2001 719,600 838,795 375,400 408,199 1,021,000 1,174,771 107,200 115,478 simply reflects the fact that the NETS data cover more firms, which could be a good feature of our data. However, this good feature is not cost-free. The NETS data contain a large number of very small firms. The data for those small firms tend to be noisy, which adds more noise to the NETS data. As we have mentioned, a firm chooses to be included in the D&B raw data when it needs a DUNS number. In certain circumstances, that need may suddenly become pertinent for many firms, and hence many existing firms that are not in the D&B database will jump in simultaneously. This kind of behavior is more common for small firms, which creates more noise in the NETS data. For example, we see a big surge in employment from 1992 to 1993 in the NETS data but not in the EDD data. In the 1992–1993 period, the California economy came out of a severe recession, and therefore an increase in employment was expected. But the 6–10 percent increase in the NETS data is too dramatic to be credible. We have attempted to discover possible reasons to explain the surge in 1993. As part of President Clinton’s mandate to streamline the procurement process through the use of electronic commerce, the federal government adopted the D&B DUNS number as a principal contractor identification code in 1993. This means that suppliers doing business with government agencies via Electronic Data Interchange would be required to submit their DUNS number as part of the registration and transaction processes. This might have pushed many existing firms into the D&B database. We see a nationwide surge in the number of business units in the 1993 edition of D&B’s business census. This is also reflected in our NETS data. The problem could be partly solved if every establishment reported its starting date as required, but a large proportion of small firms failed to do so. For this reason, we should use caution when interpreting economic trends in the NETS data. We have compared some of the NETS data and the EDD data at the county level. Our general conclusion is that the NETS data provide a more complete coverage of business enterprises and particularly of small firms. The drawback that comes with the more complete coverage is that it is subject to noise created by small firms. The above comparison reveals only some of the properties of the NETS data at the aggregate 90 level. At the firm level, the NETS data offer a very rich pool of information such as firm location, ownership, industry, employment, and the changes in such variables over time. This wealth of information is unparalleled by any other database. The VentureOne Data The second dataset is provided by VentureOne, a leading venture capital research company. VentureOne claims that it has “the most comprehensive database on venture-backed companies.” Our data cover venture capital deals completed from the first quarter of 1992 through the fourth quarter of 2001. They include 29,277 rounds of financing involving 11,029 firms. Among those firms, 83.53 percent were founded in or after 1990. The VentureOne data provide detailed information about all the venture-backed start-ups. Interesting firm-level variables include the start year, address, industry, employment, current business status, current ownership status, closing date of each round of financing, the amount of capital raised in each round, and so on. VentureOne categorizes venture-backed firms into 16 different “industry segments.” Table B.3 shows the amount of venture capital invested and the number of deals completed in each industry. An overwhelming majority of venture-backed start-ups should be classified as high-tech. Even in the retailing industry, most venture-backed firms qualify as high-tech because they are Internet-related. Only a tiny proportion of firms in our dataset do not fall into our definition of hightech, such as restaurants in the retailing industry. Since VentureOne does not use the SIC codes, we have no consistent way to exclude nontech firms from our analysis other than relying on subjective judgment. Thus, we decided to use the entire dataset. VentureOne also provided a separate dataset containing information about start-up founders. An “EntityID” variable allows us to match the firm data with the founder data. The biographical information of founders is available, including the previous working experiences of the founder. This enables us to do some elementary studies of entrepreneurs, such as what kind of people tend to found venture-backed start-ups. To do a preliminary reliability test of the VentureOne data, we compared them with the only alternative comprehensive venture capital 91 Table B.3 Real Venture Capital Investment in the United States, by Industry, 1992–2001 Industry Communications Software Consumer/business services Information services Biopharmaceutical Retailing Medical devices Semiconductor Electronics Healthcare Medical information services Consumer/business products Advance/special material and chemical Other Energy Agriculture Total aIn 1996 dollars. Venture Capital Raised ($ billions)a 72.926 57.058 52.830 26.436 21.845 14.617 13.579 11.627 11.343 7.902 7.347 5.554 1.395 1.337 1.116 0.516 307.426 % of Total Venture Capital 23.72 18.56 17.18 8.60 7.11 4.75 4.42 3.78 3.69 2.57 2.39 1.81 0.45 0.43 0.36 0.17 100 No. of Deals 3,893 7,142 5,025 2,522 2,140 1,062 1,885 1,154 1,476 932 915 579 200 199 76 77 29,277 database, the PricewaterhouseCoopers/Venture Economics/National Venture Capital Association MoneyTree Survey. The data from the MoneyTree Survey do have one advantage in that they cover a longer time period. However, our main purpose is to study industry dynamics through firm formation, growth, and mortality but not the trend of venture capital investment. So we need detailed information about venture-backed firms. By this criterion, the VentureOne data are more suitable for us. Table B.4 compares some aggregate statistics from the MoneyTree Survey and the VentureOne data. The VentureOne data show a higher sum of venture capital investment for every year except 2001. We acquired our data from VentureOne in late December 2001, when the fourth-quarter data were not completed yet. That may explain the deficit of the VentureOne data in 2001. In terms of companies covered, the 92 Table B.4 Venture Capital Investment by MoneyTree Survey and VentureOne Data MoneyTree Surveya VentureOne Year 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Sum Invested No. of Average per Sum Invested No. of Average per ($ millions) Companies Company ($ millions) Companies Company 3,827.56 1,054 3.63 9,230.75 1,126 8.20 4,565.53 945 4.83 10,220.66 1,162 8.80 3,792.89 954 3.98 8,043.74 1,230 6.54 5,693.46 1,265 4.50 13,389.43 1,536 8.72 11,386.77 1,809 6.29 21,313.05 2,105 10.12 14,823.33 2,385 6.22 20,474.79 2,329 8.79 19,843.17 2,821 7.03 24,752.63 2,568 9.64 54,499.93 4,202 12.97 67,480.78 4,027 16.76 102,308.33 5,608 18.24 112,214.10 5,483 20.47 37,672.50 3,224 11.69 32,524.21 2,933 11.09 aInformation is current as of February 20, 2002, and is available at http://www.nvca. org/. VentureOne data report more venture-backed companies from 1992 to 1996. Since then, the MoneyTree Survey has covered more companies. The discrepancies are quite small, although we have to recognize that a larger set of companies in one dataset does not necessarily encompass the smaller number of companies in the other dataset. The most significant disagreement between the two datasets is the average amount of money raised by each company. Except for 2001, the VentureOne data always produce a higher average. And the trend is the earlier the data, the bigger the difference. In 1992, the average venture capital per company in the VentureOne data is more than twice as much as in the MoneyTree Survey. Many possible reasons can explain the differences. For example, the definition of venture capital may not be identical. We notice that VentureOne actively tracks only venture capital firms that manage more than $20 million. This may bias the VentureOne data toward larger venture capital deals. Because most deals became very large in the late 1990s, this bias could have become smaller. Overall, it seems that there is not enough evidence to conclude that one dataset is better than the other. 93 Appendix C A Snapshot of the Silicon Valley Economy Using an extract from the NETS database, we assemble a collection of statistics here to describe the Silicon Valley economy in 2001. Table C.1 Total Number of Establishments and Employees in Silicon Valley, 2001 Total establishments Total employees High-Tech 25,787 672,825 Nontech 77,334 903,332 Total 103,121 1,576,157 Table C.2 High-Tech Establishment Category in Silicon Valley, 2001 Establishment Category Alive in 2001 % of total Headquarters 1,682 6.52 Branches 2,621 10.16 Stand-Alone 21,484 83.31 Total 25,787 100 Table C.3 Establishment Size Distribution in Silicon Valley, 2001 No. of Employees 0–4 5–9 10–19 20–50 51–100 101–250 251–500 501–1,000 1,001–2,500 2,500+ Total High-Tech 15,993 3,405 2,372 2,227 823 579 207 93 63 25 25,787 Nontech 51,924 10,800 6,556 5,402 1,598 739 184 83 34 14 77,334 95 Table C.4 Establishment Age Distribution in Silicon Valley, 2001 Establishment Yeara High-Tech Nontech Total 1989 or before 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Total 7,570 653 743 1,437 1,034 1,222 1,441 1,821 1,877 2,317 2,099 3,573 25,787 30,777 1,681 1,756 5,318 2,283 2,941 3,475 5,202 3,729 5,019 4,913 10,240 77,334 38,347 2,334 2,499 6,755 3,317 4,163 4,916 7,023 5,606 7,336 7,012 13,813 103,121 aThis refers to the variable “FirstYear,” which is a firm’s start year or, in case the start year is missing, the year when its data first entered the D&B database. Table C.5 Total Establishments in Silicon Valley, by Industry Group, 2001 Industry Group Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation, communication, and utilities Wholesale trade Retail trade Finance, insurance, and real estate Services Government Total No. of Establishments 1,758 35 6,886 8,163 3,402 6,907 17,291 9,237 49,039 403 103,121 No. of Employees 12,496 315 55,795 459,388 71,326 85,153 181,026 85,048 580,742 44,868 1,576,157 96 Table C.6 Total High-Tech Establishments in Silicon Valley, by Industry, 2001 Industry Semiconductors Computers/communications Bioscience Defense/aerospace Environmental Software Innovation services Professional services Total No. of Establishments 816 1,127 847 94 244 4,505 6,257 11,897 25,787 No. of Employees 103,443 150,974 51,854 27,567 8,342 114,639 112,150 103,856 672,825 97 Bibliography Arthur, W. Brian, “‘Silicon Valley’ Locational Clusters: When Do Increasing Returns Imply Monopoly?” Mathematical Social Sciences, Vol. 19, 1990, pp. 235–251. Audretsch, David B., Innovation and Industry Evolution, The MIT Press, Cambridge, Massachusetts, 1995. Bahrami, Homa, and Stuart Evans, “Flexible Recycling and HighTechnology Entrepreneurship,” in Martin Kenney, ed., Understanding Silicon Valley: The Anatomy of an Entrepreneurial Region, Stanford University Press, Stanford, California, 2000. Birch, David L., Job Creation in America: How Our Smallest Companies Put the Most People to Work, Free Press, New York, 1987. Bower, Joseph L., and Clayton M. Christensen, “Disruptive Technologies: Catching the Wave,” Harvard Business Review, January–February 1995, pp. 43–53. Bygrave, William D., and Jeffry A. Timmons, Venture Capital at the Crossroads, Harvard Business School Press, Boston, Massachusetts, 1992. Carroll, Glenn R., and Michael T. Hannan, The Demography of Corporations and Industries, Princeton University Press, Princeton, New Jersey, 2000. Christensen, Clayton M., The Innovator’s Dilemma, Harvard Business School Press, Boston, Massachusetts, 1997. Cooper, A., and T. Folta, “Entrepreneurship and High-Technology Clusters,” in D. L. Sexton and H. Landstrom, eds., The Blackwell Handbook of Entrepreneurship, Blackwell Business, Malden, Massachusetts, 2000. 99 Cortright, Joseph, and Heike Mayer, “High Tech Specialization: A Comparison of High Technology Centers,” working paper, Center on Urban and Metropolitan Policy, the Brookings Institution, Washington, D.C., 2001. DeVol, Ross C., “America’s High-Tech Economy,” Milken Institute, Santa Monica, California, 1999. Foster, Richard N., Innovation: The Attacker’s Advantage, Summit Books, New York, 1986. Freiberger, Paul, and Michael Swaine, Fire in the Valley, 2nd edition, McGraw-Hill, New York, 2000. Gilson, Ronald G., “The Legal Infrastructure of High Technology Industrial Districts: Silicon Valley, Route 128, and Covenants Not to Compete,” New York University Law Review, Vol. 74, 1999, pp. 575– 629. Henton, Doug, “A Profile of the Valley’s Evolving Structure,” in ChongMoon Lee, William F. Miller, Marguerite Gong Hancock, and Henry S. Rowen, eds., The Silicon Valley Edge: A Habitat for Innovation and Entrepreneurship, Stanford University Press, Stanford, California, 2000. Henton, Doug, Kim Walesh, Liz Brown, and Chi Nguyen, Joint Venture’s 2003 Index of Silicon Valley, Joint Venture: Silicon Valley Network, San Jose, California, 2003. Hyde, Alan, “Silicon Valley’s High-Velocity Labor Market,” Journal of Applied Corporate Finance, Vol. 11, 1998, pp. 28–37. Kenney, Martin, and Richard Florida, “Venture Capital in Silicon Valley: Fueling New Firm Formation,” in Martin Kenney, ed., Understanding Silicon Valley: The Anatomy of an Entrepreneurial Region, Stanford University Press, Stanford, California, 2000. Klepper, Steven, “Employee Startups in High-Tech Industries,” Industrial and Corporate Change, Vol. 10, 2001, pp. 639–674. Krugman, Paul, Geography and Trade, The MIT Press, Cambridge, Massachusetts, 1991. 100 Lee, Chong-Moon, William F. Miller, Marguerite Gong Hancock, and Henry S. Rowen, eds., The Silicon Valley Edge: A Habitat for Innovation and Entrepreneurship, Stanford University Press, Stanford, California, 2000. McKendrick, David G., Richard E. Doner, and Stephan Haggard, From Silicon Valley to Singapore: Location and Competitive Advantage in the Hard Disk Drive Industry, Stanford University Press, Stanford, California, 2000. Paulson, Ed, Inside Cisco: The Real Story of Sustained M&A Growth, John Wiley & Sons, New York, 2001. Rosenberg, David, Cloning Silicon Valley, Pearson Education, New York, 2002. Saxenian, Annalee, Regional Advantage: Culture and Competition in Silicon Valley and Route 128, Harvard University Press, Cambridge, Massachusetts, 1994. Schumpeter, Joseph A., The Theory of Economic Development, Harvard University Press, Cambridge, Massachusetts, 1934. “Silicon Valley: How It Really Works,” BusinessWeek, August 18–25, 1997, pp. 64–147. “The VCs Don’t Want Your Money Anymore,” BusinessWeek, July 29, 2002, pp. 81–82. VentureOne Corporation, The Venture Capital Sourcebook, San Francisco, California, 2001. 101 About the Author JUNFU ZHANG Junfu Zhang specializes in evolutionary economics and agent-based computational economics. His research interests include racial segregation in housing and schools, entrepreneurship, and innovations in the high-tech industry. He has held the Graduate Fellowship at Johns Hopkins University and the Leo Model Research Fellowship at The Brookings Institution. He received a B.A. from Renmin University of China and an M.A. and Ph.D. in economics from Johns Hopkins University. 103 Related PPIC Publications Rethinking the California Business Climate Michael Dardia and Sherman Luk California’s Vested Interest in U.S. Trade Liberalization Initiatives Jon D. Haveman The Evolution of California Manufacturing Paul W. Rhode Local and Global Networks of Immigrant Professionals in Silicon Valley AnnaLee Saxenian Silicon Valley’s New Immigrant Entrepreneurs AnnaLee Saxenian Business Without Borders? The Globalization of the California Economy Howard J. Shatz PPIC publications may be ordered by phone or from our website (800) 232-5343 [mainland U.S.] (415) 291-4400 [Canada, Hawaii, overseas] www.ppic.org 105" } ["___content":protected]=> string(102) "

R 703JZR

" ["_permalink":protected]=> string(102) "https://www.ppic.org/publication/high-tech-start-ups-and-industry-dynamics-in-silicon-valley/r_703jzr/" ["_next":protected]=> array(0) { } ["_prev":protected]=> array(0) { } ["_css_class":protected]=> NULL ["id"]=> int(8294) ["ID"]=> int(8294) ["post_author"]=> string(1) "1" ["post_content"]=> string(0) "" ["post_date"]=> string(19) "2017-05-20 02:36:20" ["post_excerpt"]=> string(0) "" ["post_parent"]=> int(3461) ["post_status"]=> string(7) "inherit" ["post_title"]=> string(8) "R 703JZR" ["post_type"]=> string(10) "attachment" ["slug"]=> string(8) "r_703jzr" ["__type":protected]=> NULL ["_wp_attached_file"]=> string(12) "R_703JZR.pdf" ["wpmf_size"]=> string(6) "793444" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(185486) "High-Tech Start-Ups and Industry Dynamics in SiliconValley ••• Junfu Zhang 2003 PUBLIC POLICY INSTITUTE OF CALIFORNIA Library of Congress Cataloging-in-Publication Data Zhang, Junfu, 1970-. High-tech start-ups and industry dynamics in Silicon Valley / Junfu Zhang. p. cm. Includes bibliographical references. ISBN: 1-58213-074-4 1. High technology industries—California—Santa Clara Valley (Santa Clara County)—Longitudinal studies. 2. Entrepreneurship—California—Santa Clara Valley (Santa Clara County)—Longitudinal studies. I. Title. HC107.C23H5394 2003 338.4'76'0979473—dc21 2003012491 Copyright © 2003 by Public Policy Institute of California All rights reserved San Francisco, CA Short sections of text, not to exceed three paragraphs, may be quoted without written permission provided that full attribution is given to the source and the above copyright notice is included. PPIC does not take or support positions on any ballot measure or state and federal legislation nor does it endorse or support any political parties or candidates for public office. Research publications reflect the views of the authors and do not necessarily reflect the views of the staff, officers, or Board of Directors of the Public Policy Institute of California. Foreword The Bay Area economy is experiencing one of its most prolonged recessions: Unemployment continues to climb, start-ups in Silicon Valley have declined from over 3,500 a year in 1998 to well under 1,000 in recent years, and, nationwide, the high-tech sector appears to be facing a future of excess capacity. These are certainly sufficient reasons for the general mood of gloom that has settled over a region that was recently the focus of international attention for its high-tech successes. Why this dramatic turnaround in the economy of Silicon Valley? What are the prospects that the region will be booming once again? High-Tech Start-Ups and Industry Dynamics in Silicon Valley by Junfu Zhang is yet another contribution by PPIC to an improved understanding of the California economy. This research project is one of a series that PPIC has launched to gain a better understanding of California’s new economies and of the dynamic processes that underlie their cycles of boom and bust. Past PPIC studies have looked at the role of immigrant entrepreneurs and their linkage to Asia, the role of U.S. tariff policy and its effect on increasing export activity, and the role of exports and foreign direct investment in building California’s economy for future decades. Zhang’s research concludes that, collectively, new firms represent a major force in the economic dynamics of Silicon Valley. For example, firms founded after 1990 created almost all of the job growth experienced by Silicon Valley between 1990 and 2001. Why, then, do we find ourselves in the midst of the current bust cycle? The theory most applicable to the current situation was developed by Joseph Schumpeter in 1911. In The Theory of Economic Development, he explained, “The economic system does not move along continually and smoothly. Countermovements, setbacks, incidents of the most various kinds occur, which obstruct the path of development; there are breakdowns in the economic value system which interrupt it.” And, he argued, these setbacks iii lead to the development of new ideas, new entrepreneurs rise to the occasion, and soon the cycle begins all over again. The cycle of firm startups, closures, and new start-ups is very much part of the economic development process, and the very entrepreneurs who are in abundant supply in Silicon Valley will make the process happen all over again. For Silicon Valley, this cycle is as much fact as theory. In the 1950s, a handful of firms supplied electronic devices to the Defense Department. In the 1960s, the region became a center of computer chipmakers. In the 1970s and 1980s, the region developed and manufactured personal computers and workstations, and in the 1990s, the region helped commercialize Internet technology. For every major firm, such as the Hewlett-Packard Company and Intel, there were thousands of entrepreneurs starting little firms with dreams of one day becoming a leader in their field. Zhang concludes that start-ups in Silicon Valley have more rapid access to venture capital than comparable firms elsewhere in the nation; that large, established firms spin off more start-ups than firms in other parts of the country; and that the high-tech sector is subject to rapid structural change where “hot spots” of growth may appear in some industries while firms in other industries are simultaneously dying out. He observes that a dynamic labor force has been, and will be, essential to successful adaptation with each new structural change. In sum, human capital, venture capital, entrepreneurial zeal, and product cycles all contribute to the health and success of the economy of Silicon Valley. Although Zhang makes no predictions about the future, the fact that the region has weathered these cycles in the past, that the basic ingredients are still there in abundance, and that new demands for high-technology products are following on a worldwide concern for secure environments suggests that the prospects are good for yet another rebirth of the valley. Zhang suggests that the dynamics of economic development favor Silicon Valley and that yet another replay of the rebirth part of the cycle lies before us. David W. Lyon President and CEO Public Policy Institute of California iv Summary After extraordinary economic success in the late 1990s, Silicon Valley entered a deep recession in 2001. Today, policymakers, academic researchers, and the general public continue to puzzle over what made Silicon Valley such an enormous success. More important, they wonder if the region will ever experience such strong growth again. This study seeks to answer those questions by examining Silicon Valley’s high-tech economy in a dynamic context. Using two unique longitudinal databases, we investigate firm formation, growth, mortality, and migration in Silicon Valley during the 1990s and explain how the region’s economy evolves and operates through such dynamic processes. This study not only helps us better understand Silicon Valley’s success in the past but also reveals insights into how Silicon Valley can ensure its future prosperity. Major Findings New firms are important for Silicon Valley. As with other hightech centers, Silicon Valley hosts a wide variety of firms. A multitude of small firms coexist with medium-sized and big firms; and each year, many new firms are founded, which collectively are a major driver of the economic dynamics in Silicon Valley. In fact, firms founded after 1990 created almost all the job growth during 1990–2001. Young start-ups in Silicon Valley consistently attract a large amount of venture capital. Successful start-ups have remade and will continue to remake Silicon Valley. Start-ups in Silicon Valley have quick access to venture capital. On average, it takes 11.6 months for Silicon Valley’s start-ups to complete their first round of venture finance, five months faster than the national average. In addition, the quicker access to capital is found in every major industry in Silicon Valley. This gives start-ups in the region a head start—an important advantage in high-tech industries that advance at a v very fast pace. This large first-mover’s advantage implies that start-ups in the valley will have better chances to survive, all else being equal. Established firms in Silicon Valley spin off more start-ups. Compared to their counterparts in the Boston area, big companies in Silicon Valley have more previous employees who start their own venture-backed businesses. Since engineers in successful firms are in the best position to grasp and commercialize cutting-edge innovations, a high rate of spin-off helps open new markets and creates new jobs. Previous research discusses Silicon Valley’s high incidence of firm-level spin-off based on anecdotal evidence and has identified cultural and legal factors to account for it. Although the causal factors remain unclear, for the first time we have confirmed with empirical data that there are indeed more firm-level spin-offs in Silicon Valley than in other high-tech centers. Firm relocation is not a serious problem. High-tech start-ups value the hotbed of innovation because that is where new ideas emerge and entrepreneurs cluster. Silicon Valley is a perfect environment for startups whose major objective is to develop innovative ideas. On the other hand, when firms become mature and enter the phase of mass production or routine services, their major concern becomes sustainability and they naturally care about operating costs. For those firms or, rather, for certain operations of those firms, Silicon Valley is unattractive. We have investigated whether firms leave Silicon Valley when they have evolved out of the start-up stage. We find that indeed more establishments move out of Silicon Valley than move in, and establishments moving out tend to be older. Establishments still tend to stay close to the valley when they move out. When firms move across state borders, Silicon Valley does see a net job loss, because more jobs are relocated to other states than are relocated to Silicon Valley from outside California. However, the data suggest that firm relocation involves a relatively small proportion of the labor force. Firm birth and death cause much more turbulence than firm relocation. In other words, once firms are established in Silicon Valley, they are very likely to remain there. Intensive entrepreneurial activities certainly compensate for the jobs lost through firm relocation. vi Successful firms in the valley are branching out. Although relocation does not occur at significant levels, established firms in Silicon Valley frequently set up branches elsewhere. For many large high-tech companies headquartered in Silicon Valley, their employment within Silicon Valley itself is only a small proportion of their total employment. Since Silicon Valley is already tightly packed with thousands of firms, fast-growing start-ups are more likely to expand outside the immediate area. As firms begin to expand, they potentially benefit the rest of California by setting up branches elsewhere in the state. The high-tech sector experiences rapid structural changes. The high-tech sector consists of several industries, which follow different dynamics. On the one hand, the fluctuation of the macro economy has distinctive effects on different high-tech industries; on the other hand, technological innovations in different industries, the drivers of growth in those industries, do not arrive simultaneously. As a consequence, different high-tech industries may follow unsynchronized business cycles. Therefore, at different points of time, the “hot spot” of growth may appear in different industries. For example, the 1990s saw a boom in the computer industry along with a decline in the defense industry. To catch upturns and avoid downturns in high-tech industries, a high-tech center such as Silicon Valley must accommodate rapid structural changes. This implies that a dynamic labor force is necessary. Previous research has emphasized the “high-velocity labor market” through which workers move frequently from one job to another within Silicon Valley. Such a labor market certainly helps the region’s economy adapt to structural changes. In addition, a set of infrastructure and institutions that enables the labor force to quickly move into and out of Silicon Valley is also crucial for structural changes in the high-tech sector. For example, employment in the software industry in Silicon Valley increased from 48,500 to 114,600 between 1990 and 2001, a phenomenal 136 percent growth rate. It is impossible to train such a large number of technical workers within such a short period of time. This kind of rapid growth in a certain industry is achievable only through massive migration of the needed labor force. vii Policy Implications Our findings lead us to offer the following recommendations to policymakers. Promote technological innovation. More than any other sector, the high-tech economy is about innovation and entrepreneurship. State and local governments should help promote innovation. Since university research has always been a major source of innovation, state government should continue its strong support to research universities. Big budget cuts for the University of California system will severely affect the prospect of the high-tech sector off campus, which must be avoided. Moreover, the California delegation in Washington, D.C., should place a high priority on securing R&D dollars for California from the federal government. As the state economy becomes more and more reliant on high-tech industries, support for R&D and innovation not only helps Silicon Valley and the rest of the Bay Area, but it also greatly benefits the Los Angeles and San Diego areas, which continue to expand their own high-tech sectors. Encourage firm founding. Our findings show that although some firms do move out of Silicon Valley, it is not a serious problem. On the one hand, they are likely to move to nearby cities and stay within the state, and on the other hand, firm formation and growth create new jobs that overwhelmingly outnumber jobs lost by firm relocation. In addition, job creation in Silicon Valley is primarily achieved by new firms. Therefore, instead of worrying about losing firms because of the high costs of doing business in Silicon Valley, state and local governments should encourage firm founding. Offering favorable tax breaks, opening industrial parks, building high-tech incubators, and providing seed capital for commercialization of research are widely used policy levers. Continuously improving the quality of life in Silicon Valley and the Bay Area as a whole is also crucial for the vitality of the high-tech economy in this area. Look beyond Silicon Valley. The high-tech sector is not a disconnected economy, nor is Silicon Valley an isolated region. Silicon Valley is well embedded in the San Francisco Bay Area economy as well as the state economy. Most of the firms leaving Silicon Valley migrate to viii nearby cities in the Bay Area. The rest of the Bay Area has undoubtedly benefited from the proximity of Silicon Valley and has a quite strong high-tech economy. State policies regarding Silicon Valley should take into account connections between Silicon Valley and the rest of the state economy. For example, many people who work in Silicon Valley live a considerable distance from it, seeking more affordable homes. Thus, housing development and transportation policies in many other Bay Area cities help directly solve Silicon Valley’s housing problems. We have also found that large firms in Silicon Valley hire only a small proportion of their total employees from the valley or even the Bay Area. This suggests that other regions in the state have chances to benefit from the spillover from Silicon Valley by hosting branches of its firms. State government could provide incentives for large firms to set up their manufacturing or distribution arms within the state. It is also helpful to improve transportation networks between the Bay Area and the Central Valley that facilitate Silicon Valley’s branching out in other areas of the state. In addition, local governments in the rest of the Bay Area and the Central Valley should be more proactive in accommodating businesses branching out from Silicon Valley. Maintain a dynamic labor pool. Two conflicting factors characterize the high-tech labor force. On the one hand, the high-tech sector primarily hires technical workers whose skills are highly specialized and take time to acquire; on the other hand, the high-tech sector is dynamic, with its core technologies evolving quickly. This implies that the skills acquired in school three years ago may be obsolete today. Moreover, certain high-tech industries often experience explosive growth, such as the software industry did in the 1990s, which creates a high demand for certain types of technical workers within a short period. Whether Silicon Valley can evolve rapidly hinges upon whether its labor force can quickly upgrade its skills or meet completely new demands. State government should continue to rely on local universities and community colleges as a vehicle to help retool the labor force continuously. Employers in Silicon Valley need to recruit new talent not only through local universities but also by hiring qualified immigrants, who have played an important role in Silicon Valley’s growth. The immigrant pool has proved to be a major source of innovators and ix entrepreneurs. Immigrants also provide a large reserve of high-quality engineers and scientists ready to satisfy sudden surges of demand in certain industries. State government in cooperation with federal authorities should keep the door open to international talent, both at local universities and in the high-tech industries. This has emerged as a particularly crucial issue because immigration policies have now entered the equation of homeland security. x Contents Foreword ..................................... Summary..................................... Figures ...................................... Tables ....................................... Acknowledgments ............................... iii v xiii xv xvii 1. INTRODUCTION AND OVERVIEW OF THE STUDY ................................... Change in Silicon Valley ........................ A Demographic Perspective of the Silicon Valley Habitat ... Purpose of This Study .......................... Data ..................................... 1 3 6 8 9 2. START-UP, GROWTH, AND MORTALITY OF FIRMS IN SILICON VALLEY ......................... Firm Formation.............................. Rate of Firm Formation ....................... Structural Changes .......................... Firm Growth ............................... Firm Mortality .............................. Rate of Mortality ........................... Merger and Acquisition ....................... Job Creation by Start-Ups ....................... Conclusion ................................. 3. VENTURE-BACKED START-UPS IN SILICON VALLEY .................................. Venture Capital in Silicon Valley ................... Firm Formation.............................. Ownership Status and Profitability.................. Spinoffs ................................... Conclusion ................................. 11 11 11 16 19 23 24 25 28 30 31 31 35 41 47 52 4. FIRM RELOCATION IN SILICON VALLEY ......... 53 High-Tech and Nontech Relocation ................. 54 xi Trans-State Relocation ......................... Mobility vs. Vitality ........................... Relocating Out vs. Branching Out .................. Conclusion ................................. 5. CONCLUSION ............................. Major Findings .............................. Policy Implications............................ 60 65 69 71 73 73 75 Appendix A. Geographic and Industrial Definitions ............... B. The Data .................................. C. A Snapshot of the Silicon Valley Economy............. 81 85 95 Bibliography .................................. 99 About the Author ............................... 103 Related PPIC Publications .......................... 105 xii Figures 1.1. A Map of Silicon Valley ...................... 1.2. Industry Dynamics in Silicon Valley .............. 2.1. High-Tech Firm Formation in Silicon Valley, 1990– 2000 .................................. 2.2. Firm Formation in High-Tech Clusters, 1990–2000.... 2.3. High-Tech Start-Ups That Ever Hired Five or More Employees by 2001 ......................... 2.4. Employment in High-Tech Industries in Silicon Valley, 1990–2001 .............................. 2.5. Employment of High-Tech Start-Ups in Nonservice Industries, 2001 ........................... 2.6. Employment of High-Tech Start-Ups in Service Industries, 2001 ........................... 2.7. Survival Rates of High-Tech Firms in Silicon Valley .... 2.8. Comparison of Survival Rates .................. 2.9. Percentage of Firms Acquired by 2001 ............ 2.10. Employment of High-Tech Start-Ups in Silicon Valley .. 2.11. Employment of High-Tech Start-Ups Younger Than Age Five as a Percentage of Total High-Tech Employment ............................. 3.1. Total Venture Capital Investment, 1992–2001 ....... 3.2. Total Venture Capital Investment, by Region, 1992– 2001 .................................. 3.3. Venture-Backed Start-Ups, 1990–2001 ............ 3.4. Venture-Backed Start-Ups, by Region, 1990–2001 .... 3.5. Average Amount of Venture Capital Raised per Deal, 1992–2001 .............................. 3.6. Average Start-Up Age at First-Round Financing ...... 3.7. Average Start-Up Age at First-Round Financing, by Industry ................................ 3.8. Ownership Status of Venture-Backed Start-Ups in Silicon Valley, 2001 ........................ 2 9 12 13 14 17 22 22 25 26 27 29 29 32 33 36 36 37 38 39 42 xiii 3.9. Ownership Status of Venture-Backed Start-Ups in the United States, 2001 ........................ 3.10. Differences in Ownership Status in Each Cohort of Venture-Backed Start-Ups: Silicon Valley Compared to the United States .......................... 3.11. Business Status of Venture-Backed Start-Ups in Silicon Valley, 2001 ............................. 3.12. Business Status of Venture-Backed Start-Ups in the United States, 2001 ........................ 4.1. Percentage of Moving Establishments Founded Before 1990 .................................. 4.2. Average Age of Establishments Moving Between Silicon Valley and Other States ...................... 4.3. Job Movement Between Silicon Valley and Other States, 1991–2000 .............................. 4.4. Dynamics in Silicon Valley’s High-Tech Labor Market, 1991–2000 .............................. 4.5. Dynamics in Silicon Valley’s Labor Market, 1991– 2000 .................................. 43 44 46 47 63 63 64 68 68 xiv Tables 1.1. Forty Largest Technology Companies in Silicon Valley, 1982 and 2002 ........................... 2.1. High-Tech Start-Ups, by Industry, 1990–2000 ....... 2.2. Employment in High-Tech Industries in Silicon Valley, 1990–2001 .............................. 2.3. Growth of Silicon Valley’s High-Tech Firms in Nonservice Industries ....................... 2.4. Growth of Silicon Valley’s High-Tech Firms in Service Industries ............................... 2.5. Death of High-Tech Establishments in Silicon Valley, 1990–2000 .............................. 2.6. Top Headquarter States of Firms Acquired During 1990–2001 .............................. 3.1. Real Venture Capital Investment, by Industry in Silicon Valley, 1992–2001 ......................... 3.2. Number of Spinoffs from Leading Institutions in Silicon Valley and the Boston Area .................... 4.1. Relocation of Establishments in Silicon Valley, 1990– 2001 .................................. 4.2. Top Ten Destination States for Establishments Relocating Out of Silicon Valley, 1990–2001 ........ 4.3. Top Ten Destination Cities for Establishments Relocating Out of Silicon Valley, 1990–2001 ........ 4.4. Top Ten Origin States for Establishments Relocating Into Silicon Valley, 1990–2001 ................. 4.5. Top Ten Origin Cities for Establishments Relocating Into Silicon Valley, 1990–2001 ................. 4.6. High-Tech Establishments Relocating Into and Out of Silicon Valley, by Industry, 1990–2001 ............ 4.7. All Establishments Relocating Into and Out of Silicon Valley, by Industry Group, 1990–2001 ............ 5 15 18 20 21 24 28 34 50 55 56 56 58 58 59 60 xv 4.8. High-Tech Establishments Moving Between Silicon Valley and Outside California, by Industry, 1990– 2001 .................................. 4.9. All Establishments Moving Between Silicon Valley and Outside California, by Industry Group, 1990–2001 .... 4.10. Trans-State Relocation as a Percentage of Total Employment That Moved Into or Out of Silicon Valley, 1990–2001 .............................. 4.11. Employment in the High-Tech Sector of Silicon Valley, 1991–2000 .............................. 4.12. Employment in Silicon Valley, 1991–2000 ......... 4.13. Intel Operating Locations in the United States ....... B.1. Business Size Distribution in NETS and EDD Data, 2001 .................................. B.2. Employment Series in NETS and EDD Data, 1990– 2001 .................................. B.3. Real Venture Capital Investment in the United States, by Industry, 1992–2001........................ B.4. Venture Capital Investment by MoneyTree Survey and VentureOne Data .......................... C.1. Total Number of Establishments and Employees in Silicon Valley, 2001 ........................ C.2. High-Tech Establishment Category in Silicon Valley, 2001 .................................. C.3. Establishment Size Distribution in Silicon Valley, 2001 .. C.4. Establishment Age Distribution in Silicon Valley, 2001 .. C.5. Total Establishments in Silicon Valley, by Industry Group, 2001 ............................. C.6. Total High-Tech Establishments in Silicon Valley, by Industry, 2001 ............................ 61 61 62 66 67 70 88 89 92 93 95 95 95 96 96 97 xvi Acknowledgments I would like to thank Michael Teitz for his suggestions, guidance, and encouragement at every stage of this research project. I am grateful to AnnaLee Saxenian, who provided guidance during the development of the research proposal and offered invaluable comments and suggestions for finalizing the report. Thanks also go to Doug Henton, Martin Kenney, Joyce Peterson, and Karthick Ramakrishnan for their thoughtful comments on a preliminary draft of the report. Nikesh Patel did a superb job helping with data analysis. Donald Walls offered kind help in extracting the NETS data from his database. Also, I want to thank Gary Bjork and Patricia Bedrosian for their editorial assistance. The author is solely responsible for any errors of fact or interpretation. xvii 1. Introduction and Overview of the Study It took merely half a century for Santa Clara Valley, the region that curls around the southern tip of the San Francisco Bay, to become the most famous high-tech industrial cluster in the world. Silicon Valley, as it has been known since the early 1970s, is today a main driver of the California state economy (see Figure 1.1 and Appendix A for our geographic definition of Silicon Valley). It is home to more than 22,000 high-tech companies, including household names such as HewlettPackard, Intel, Apple, and eBay. Silicon Valley’s celebrity skyrocketed over the past decade as it became the center of “the largest legal creation of wealth in history.” At its peak, the Internet boom produced scores of new millionaires in Silicon Valley every day. The region had become a land of enchantment for ambitious entrepreneurs whose success stories appeared in the media all over the world, and thousands of well-paid jobs made Silicon Valley a magnet for talented people. Given the enormous success of this regional economy, policymakers around the world wondered how they could “clone Silicon Valley” in their own regions (Rosenberg, 2002). But it seems that what goes up must come down. Since 2001, the region has entered a deep recession. In Santa Clara County, the heart of Silicon Valley, the unemployment rate climbed from 1.7 percent in January 2001 to 8.9 percent in October 2002, then declined a little to 8.3 percent in December 2002.1 In 2002, Silicon Valley posted an annual unemployment rate higher than the state average for the first time in two decades. According to Joint Venture’s 2003 Index of Silicon Valley, the region lost 127,000 jobs (about 9 percent of its total employment) ____________ 1According to the California Employment Development Department, available at http://www.calmis.cahwnet.gov/htmlfile/subject/lftable.htm. 1 SOURCE: Reprinted by permission from Joint Venture: Silicon Valley Network, with adaptations. Figure 1.1—A Map of Silicon Valley 2 between the first quarter of 2001 and the second quarter of 2002. More than half of the job gains registered during 1998–2000 evaporated. At the same time, venture capital investment plummeted and personal income declined. Policymakers, academic researchers, and the general public continue to puzzle over what made Silicon Valley such a huge success. More important, they wonder if the region will ever experience such strong growth again. This study seeks to answer those questions by examining Silicon Valley’s high-tech economy in a dynamic context. Using two unique longitudinal databases, we investigate firm formation, growth, mortality, and migration in Silicon Valley during the 1990s and examine how the region’s economy evolved and operated through such dynamic processes. This study not only helps us better understand Silicon Valley’s success in the past, but it also reveals insights into how Silicon Valley can ensure its future prosperity. Change in Silicon Valley Silicon Valley has experienced both highs and lows many times. If asked to use a single word to characterize the Silicon Valley economy, many people would choose “dynamic.” Indeed, change is the only unchanging norm in Silicon Valley, as new technologies and new firms constantly emerge. Yet, as the famous economist Joseph Schumpeter observed almost a century ago, innovations are not evenly distributed over time but occur in periodic clusters (Schumpeter, 1934). This is particularly true in Silicon Valley, which has remade itself over and over again during its short history (“Silicon Valley: How It Really Works,” 1997; Henton, 2000). Until the 1950s, only a handful of high-tech firms existed in the area, most notably Hewlett-Packard and Varian. The area was a major supplier of electronic devices to the Defense Department. In the 1960s, as Fairchild spun off many semiconductor producers such as Intel and AMD, the area became a center of computer chipmakers, which later led to the name “Silicon Valley.” 3 The late 1970s and 1980s were the computer years. By then the valley was known as a developer and manufacturer of personal computers and workstations, represented by such companies as Apple, Silicon Graphics, and Sun Microsystems. In the 1990s, Silicon Valley remade itself again. This time, it helped commercialize Internet technology. The leaders of this movement included Cisco, Netscape, eBay, and Yahoo. Silicon Valley has developed through waves of innovation, with a handful of innovative start-ups initiating each wave. In fact, the continuous success of Silicon Valley must be understood as the constant emergence of successful start-ups. As Lee et al. (2000) point out, “The Silicon Valley story is predominantly one of the development of technology and its market applications by firms—especially by start-ups. The result: new companies focused on new technologies for new wealth creation.” For many decades, social scientists have noticed the important role of start-ups in carrying out radical innovations. Schumpeter (1934, p. 66) observed that innovations are, as a rule, embodied in “new firms which generally do not arise out of the old ones but start producing beside them.” Recent work has provided a rationale for this observation by emphasizing the characteristics of innovations. Foster (1986) argued that technological progress often exhibits discontinuities. That is, radical changes happen frequently. Reflected in the dynamics of high-tech industries, these discontinuities give new firms a so-called “attacker’s advantage.” When newcomers gain competitive superiority over successful incumbent firms, “leaders become losers.” More recently, Christensen (1997) further developed this idea and called it the “innovator’s dilemma.” When Schumpeter talked about “the incessant gales of creative destruction” many decades ago, he could not have imagined that the industry dynamics in Silicon Valley would provide such a vivid illustration of his notion. Silicon Valley is constantly creating the new while destroying the old. Table 1.1 lists the top 40 high-tech firms in Silicon Valley in 1982 and 2002. An overwhelming majority of the names on the 1982 list have become faded memories among the locals. To outsiders, most of the 1982 top firms are unrecognizable, because half 4 Table 1.1 Forty Largest Technology Companies in Silicon Valley, 1982 and 2002 1. Hewlett-Packard 2. National Semiconductor 3. Intel 4. Memorex 5. Varian 6. Environtecha 7. Ampex 8. Raychema 9. Amdahla 10. Tymsharea 11. AMD 12. Rolma 13. Four-Phase Systemsa 14. Cooper Laba 15. Intersil 16. SRI International 17. Spectra-Physics 18. American Microsystemsa 19. Watkins-Johnsona 20. Qumea 1. Hewlett-Packard 2. Intel 3. Ciscob 4. Sunb 5. Solectron 6. Oracle 7. Agilentb 8. Applied Materials 9. Apple 10. Seagate Technology 11. AMD 12. Sanmina-SCI 13. JDS Uniphase 14. 3Com 15. LSI Logic 16. Maxtorb 17. National Semiconductor 18. KLA Tencor 19. Atmelb 20. SGI 1982 21. Measurexa 22. Tandema 23. Plantronics 24. Monolithic 25. URS 26. Tab Products 27. Siliconix 28. Dysana 29. Racal-Vadica 20. Triad Systemsa 31. Xidexa 32. Avanteka 33. Silteca 34. Quadrexa 35. Coherent 36. Verbatim 37. Anderson-Jacobsona 38. Stanford Applied Engineering 39. Acurexa 40. Finnigan 2002 21. Bell Microproductsb 22. Siebelb 23. Xilinxb 24. Maxim Integratedb 25. Palmb 26. Lam Research 27. Quantum 28. Alterab 29. Electronic Artsb 30. Cypress Semiconductorb 31. Cadence Designb 32. Adobe Systemsb 33. Intuitb 34. Veritas Softwareb 35. Novellus Systemsb 36. Yahoob 37. Network Applianceb 38. Integrated Device 35. Linear Technology 40. Symantecb NOTES: This table was compiled using 1982 and 2002 Dun & Bradstreet (D&B) Business Rankings data. Companies are ranked by sales. aNo longer existed by 2002. bDid not exist before 1982. 5 of them no longer exist. Only four firms on the 2002 list are survivors from the 1982 list. In fact, more than half of the 2002 top firms were not even founded before 1982. In only two decades, the high-tech economy in Silicon Valley changed almost completely. The San Jose Mercury News has compiled a list of the top 150 firms in Silicon Valley each year since 1994. On average, each year’s list includes 23 new firms, reflecting the fast pace of Silicon Valley. A study of these “changes” is not only the key to understanding Silicon Valley’s past success but also the key to promoting its future success. Silicon Valley’s greatest asset is its ability to reinvent itself as soon as its leading technologies or products become standardized. Thus, the secrets of the region’s success lie in its institutions that enable the changes. To ensure a bright future, we must identify, understand, and promote those institutions, and to understand the unique features of Silicon Valley and its institutions, we must observe its dynamic context. A Demographic Perspective of the Silicon Valley Habitat Silicon Valley is often described as a “habitat” (Lee et al., 2000) or an ecosystem (Bahrami and Evans, 2000). As in a natural habitat, Silicon Valley provides a host of resources that high-tech firms require to survive and grow. This habitat includes not only people, firms, universities and research institutions, and government agencies but also networks among those players and the modes by which they interact. Previous studies have examined different constituents of the habitat (see, for example, Saxenian, 1994; Kenney and Florida, 2000; and Lee et al., 2000). These studies have provided insights into the role played by entrepreneurs, universities, social networks, and supporting players such as venture capitalists, bankers, lawyers, consultants, and so on. However, the central figure in the Silicon Valley habitat is undoubtedly high-tech firms. After all, the success of Silicon Valley is measured by the large population of high-tech firms that offer many well-paid jobs. Much like a biologist who studies animals in their natural habitats, we shall take a demographic approach to study firms in Silicon Valley. 6 The demographic approach is well developed in organizational sociology (Carroll and Hannan, 2000). In contrast to the bulk of literature in industrial economics that focuses on firm-level behavior, the demographic perspective shifts attention from individual firms to the range and diversity of firms in an industry or region. It seeks to discover insights into how industries evolve over time through processes of firm formation, growth, transformation, migration, and mortality. The demographic approach is not concerned with individual firms but, rather, focuses on properties at the population level, such as a population’s age distribution and growth rates. The demographic approach is particularly appropriate for studying the Silicon Valley economy. The high-tech sector in Silicon Valley consists of a wide range of firms. On one extreme are large companies offering thousands of local jobs, such as Hewlett-Packard and Intel; on the other are thousands of small firms that hire only a few people. Firms such as Hewlett-Packard and Varian have been around for more than six decades, whereas other high-profile firms such as eBay and Yahoo did not even exist ten years ago. Companies such as Cisco and Sun Microsystems have expanded at a stunning pace, whereas thousands of others hardly grow or disappear soon after inception. And most important, products or services are differentiated along many dimensions; rarely do any two firms provide exactly the same product or service. As Carroll and Hannan have argued, the vibrancy of the Silicon Valley economy to some extent reflects its demographic characteristics. In particular, “the high rates of turnover of constituent organizations continually reshuffle the human workforce. The great diversity of organizational forms and technological strategies means that job-changers find themselves in new and different social contexts. Ideas flow with people, get recombined, and new technical and organizational innovations result. Analysis of a putatively representative firm would not only miss the point, it would also obscure community-level dynamics” (Carroll and Hannan, 2000). Yet basic demographic facts about the Silicon Valley economy remain unknown, partly because of a lack of demographic data on industries. This means that the formulation of regional social and 7 economic policies usually ignores the implication of the full diversity of firms. Thus, a demographic study can yield very useful information for policymakers. For example, discussion of firm relocation usually draws upon anecdotal evidence from the media and often raises concerns about job loss. However, the relocating firms receiving media coverage are neither representative nor exhaustive. A statistical portrait of the whole population of moving firms would reveal the real effect of firm relocation. Purpose of This Study The purpose of this study is twofold. First, it will document the intensity of entrepreneurial activities in Silicon Valley and provide information helpful to understanding the dynamics of change in the region. Specifically, it will • Measure the rates of firm formation, growth, and mortality in Silicon Valley and compare those rates to those in other hightech centers. • Measure the proportion of start-ups in the Silicon Valley economy and their effects on job creation and dissolution. These effects will be discussed in light of the Birch (1987) debate over whether small firms create more jobs. The second purpose of this research is to track the stock and flow of high-tech firms in Silicon Valley. The study will • Determine whether most firms move to the area or are started locally. • Identify the characteristics of firms moving into or out of Silicon Valley. • Examine whether net firm relocation enhances the cluster in Silicon Valley or causes the region to lose businesses. Figure 1.2 summarizes industry dynamics in Silicon Valley’s high-tech sector. We will investigate all of the types of dynamics illustrated, except for “moving inside” Silicon Valley, which is not a major concern of our study. 8 Death Moving in Merger and acquisition Moving inside Growth Moving out Birth Figure 1.2—Industry Dynamics in Silicon Valley Data Our empirical analysis will rely on two longitudinal databases: The National Establishment Time-Series (NETS) dataset that seeks to include every firm in Silicon Valley and the nationwide VentureOne dataset that focuses on venture-backed firms. The two datasets contain an enormous amount of information that helps us better understand firm formation, growth, and industry dynamics in Silicon Valley. The abundance of data allows us to shed light on many important issues through simple descriptive analysis. For a detailed discussion of the data, see Appendix B. 9 2. Start-Up, Growth, and Mortality of Firms in Silicon Valley The high-tech sector accounts for about 11 percent of the total goods and services in the United States (DeVol, 1999). As the most concentrated high-tech center, Silicon Valley has a much larger proportion of high-tech economy than does the rest of the nation. In 2001, there were 25,787 high-tech establishments in Silicon Valley— 25 percent of the total establishments in the region. Since many hightech firms are big employers, that one-quarter of all establishments offered 42.7 percent (or 673,000) of the total jobs in Silicon Valley. (See Appendix C for a more detailed profile of the Silicon Valley economy.) This chapter documents firm formation, growth, and mortality in Silicon Valley’s high-tech sector from 1990 to 2001, using the NETS dataset. Remember, the basic observation unit in the NETS data is the “establishment,” and a big firm may have several establishments. When we study firm founding and mortality, we exclude establishments created by existing firms; and when we study firm growth, we aggregate all the establishments of a firm into a single unit. Firm Formation Rate of Firm Formation Figure 2.1 traces the trend of entrepreneurial activities in Silicon Valley’s high-tech sector. During the decade from 1990 to 2000, 29,000 high-tech firms were created in Silicon Valley. An upward trend started in the early 1990s and continued until 1998, before declining sharply in 1999 and 2000. It is interesting to note that only one-fourth of the new firms had ever hired five or more employees. Most of the new firms will always remain in the 0–4 size category. Some of the founders might be 11 Number of start-ups 4,000 3,500 3,000 2,500 Silicon Valley total Ever hired 5 or more employees Venture-backed start-ups 2,000 1,500 1,000 500 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Figure 2.1—High-Tech Firm Formation in Silicon Valley, 1990–2000 more precisely described as self-employed rather than entrepreneurs. Firms that ever employed five or more people follow a much less dramatic trend in the 1990s. That is, although many more firms were created in the hype years of Internet technology, many of them started small and never grew.1 The trend for venture-backed start-ups is also depicted in Figure 2.1. Although the high-tech sector in Silicon Valley is mostly renowned for its legendary start-ups financed by venture capital, venture-capital-backed new firms actually form only the tip of a huge iceberg. A vast majority of ____________ 1D&B, the source of raw data, did ask each establishment to report its start year. However, not all of them did so. As a consequence, the start year is missing for many establishments, especially small ones. Walls & Associates created a variable “FirstYear,” whose value is determined by the first time an establishment’s data are available at D&B. If a firm reported to D&B in 1993 for the first time, 1992 is assigned to it as its first year. For those firms that have reported their start year, the first year variable is almost always identical to the start year. But overall, the trends in the two variables are quite different, mainly because many firms that were not in the D&B database originally later chose to be included in it for common reasons, such as needing a Data Universal Numbering System (DUNS) number. With the assumption that firms that reported their start year form a representative sample of the whole population, Figures 2.1–2.3 estimate the trend of entrepreneurial activities using the number of start-ups whose start year is self-reported. For example, if x out of y start-ups reported their start year in the whole sample and z of them reported 1995 as their start year, the number of firms started in 1995 is estimated to be z*y/x. By doing so, we smooth out the noise in the trend created mainly by small firms. 12 high-tech firms created in Silicon Valley are not financed by venture capital, either because they are not capital-intensive enterprises or because they do not possess a growth potential that justifies venture capital support. However, the number of venture-backed new firms grew faster proportionately than the overall trend of firm formation in the high-tech sector. In 1999, the peak year of venture capital finance, 375 start-ups were backed by venture capital—more than five times the number in 1990—whereas the total number of new firms founded in the high-tech sector did not even double from 1990 to its peak year in 1998. This reflects the fact that venture capital became much more easily available in the late 1990s. It also suggests that firm founders became more innovative as the Internet revolution created many new opportunities. We study venture-backed firms exclusively in the next chapter. Figure 2.2 compares the trend of firm formation in Silicon Valley to the trends in Boston and Washington, D.C.2 From 1990 to 1996, the 4,000 3,500 3,000 Silicon Valley Boston Washington, D.C. Number of start-ups 2,500 2,000 1,500 1,000 500 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Figure 2.2—Firm Formation in High-Tech Clusters, 1990–2000 ____________ 2By high-tech employment, Silicon Valley, Boston, and Washington, D.C., are the top three, far ahead of any other high-tech center in the United States (Cortright and Mayer, 2001). This is the primary reason why we choose Boston and Washington for comparison. 13 three areas followed almost the same upward trend. Boston lost its momentum in 1996, but Silicon Valley and Washington, D.C., continued their upward trend in firm formation until 1998. The Internet boom in the late 1990s stimulated more entrepreneurial activities in Silicon Valley and Washington than in Boston. Figure 2.3 traces the founding year of those new firms that had ever hired five or more employees in the three high-tech clusters. Silicon Valley has more firms in the 5+ category. Whereas the total number of new firms founded in Silicon Valley follows a similar trend as in the other two high-tech regions, the former consistently has more young firms hiring five or more employees. This may suggest that new firms in Silicon Valley are more growth-oriented than those in the other two areas. As mentioned above, 29,000 high-tech firms were created in Silicon Valley during the decade from 1990 to 2000. Washington, D.C., had a similar total, and Boston had about 5,000 fewer new firms. Table 2.1 presents the distribution of new firms across major high-tech industries (see Appendix A for exact definitions of those industries). In all three 900 800 700 600 500 400 Silicon Valley 300 Boston 200 Washington, D.C. 100 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Founding year Figure 2.3—High-Tech Start-Ups That Ever Hired Five or More Employees by 2001 14 Number of start-ups Table 2.1 High-Tech Start-Ups, by Industry, 1990–2000 Industry Bioscience Computers/communications Defense/aerospace Environmental Semiconductor Software Professional services Innovation services Total Silicon Valley Firms % 586 2.0 934 3.2 52 0.2 242 0.8 513 1.8 5,967 20.4 14,009 47.9 6,944 23.7 29,247 100 Boston Firms % 335 1.4 221 0.9 27 0.1 299 1.2 52 0.2 3,323 13.5 16,784 68.2 3,565 14.5 24,606 100 NOTE: Percentages may not sum to 100 because of rounding. Washington, D.C. Firms % 211 0.7 172 0.6 35 0.1 174 0.6 28 0.1 4,137 14.0 19,703 66.9 4,985 16.9 29,445 100 regions, the service industries were the most active. About 70 percent of Silicon Valley new firms were engaged in professional or innovation services. The percentage is even higher in the other two areas: for each, more than 80 percent of new firms were established in service industries. Except in the environmental industry, Silicon Valley outperformed the other two areas in every nonservice industry. Silicon Valley created more firms in the biotech, computers/communications, defense/aerospace, semiconductor, and software industries. Silicon Valley strongly led the semiconductor industry, from which it acquired its name, with 513 semiconductor start-ups during the decade, compared to 80 in Boston and Washington together. Although Boston has a long history in the defense industry and hosts Raytheon as the area’s largest employer, fewer defense/aerospace firms were founded in Boston than in the other two areas. It is also very impressive that Washington outperformed Boston (supposedly the number two high-tech cluster) in the software industry. Boston is also well known for its biotech industry. However, even in biotech, it was outnumbered by Silicon Valley. Remember, the biotech industry in the Bay Area is mainly clustered around South San Francisco and Berkeley–Emeryville, which is outside Silicon Valley. Taking that into account, the whole Bay Area did much better in biotech than reflected in the number for Silicon Valley alone. 15 Structural Changes In the high-tech sector, different industries serve different markets and employ workers with different skills. The labor forces in different industries are not entirely interchangeable. Thus, a high-tech center tends to retain a stable economic structure over time. Yet innovations do not arrive at the same rate across all industries and the macro economic climate may also have different effects on different industries. A vibrant high-tech center needs to be flexible and able to shift its emphasis when some industries slow down and others become more dynamic. Otherwise, it will not take full advantage of new areas of growth and will be hard hit when a major industry shrinks. Given the size of its hightech sector, Silicon Valley appears to be exceptionally adaptable in accommodating structural changes. Figure 2.4 presents the evolution of employment in high-tech industries in Silicon Valley. Two developments in the 1990s redefined the high-tech sector: the reduction of defense spending by the federal government after the end of the Cold War and the Internet revolution. Both have left clear marks on the structure of Silicon Valley’s high-tech economy. During 1990–2001, Silicon Valley’s defense/aerospace industry lost 60 percent of its jobs; in contrast, the software industry grew by 136 percent and the computers/communications industry by 32 percent. In 1990, total high-tech employment in Silicon Valley was 90 percent larger than in Washington, D.C., and 26 percent larger than in Boston, yet it was nimble enough to substantially change the structure of its high-tech economy over the next decade. The 136 percent growth of the software industry in Silicon Valley outpaced every high-tech industry in the other two regions. At the same time, Silicon Valley’s defense/ aerospace industry was the most heavily hit and shrank the most. For each industry, we decompose the employment growth during 1990– 2001 into the growth of firms that existed in 1990 and the jobs added by firms founded after 1990. In 2001, the high-tech economy in Silicon Valley had 672,825 employees—26 percent more than its total employment in 1990. Software, computers/communications, professional services, and semiconductor industries had each created more than 20,000 jobs. If we look only at those firms that already 16 Employment by industry 160,000 140,000 Computers/communications Innovation services Software Professional services Semiconductor Defense/aerospace Bioscience Environmental 120,000 100,000 80,000 60,000 40,000 20,000 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Figure 2.4—Employment in High-Tech Industries in Silicon Valley, 1990–2001 existed in 1990, they together lost 120,559 jobs. Old firms hired more people only in the semiconductor and environmental industries, but both increases were modest. It is interesting to note that firms founded before 1990 lost jobs during 1990–2001 in software and computers/ communications—the two industries that gained the most jobs in Silicon Valley during the 1990s (Table 2.2). On the other hand, firms founded after 1990 added a total of 258,796 jobs to the economy during the 1990s. The 136 percent growth of the software industry was all attributable to new firms, which added 72,684 jobs to the industry. In 1990, the software industry was number six by employment in Silicon Valley, after the computers/ communications, innovation services, semiconductor, professional 17 Table 2.2 Employment in High-Tech Industries in Silicon Valley, 1990–2001 Employment in 1990 Industry (1) Bioscience 46,815 Computers/communications 114,617 Defense/aerospace 68,527 Environmental 2,851 Semiconductor 79,630 Software 48,529 Innovation services 100,217 Professional services 73,402 Total 534,588 2001 Employment of Firms Existing in 1990 (2) 36,243 104,956 22,251 3,246 83,701 41,955 65,389 56,288 414,029 Total Employment in 2001 (3) 51,854 150,974 27,567 8,342 103,443 114,639 112,150 103,856 672,825 Overall Employment Growth 1990–2001 (3) – (1) 5,039 36,357 –40,960 5,491 23,813 66,110 11,933 30,454 138,237 Employment Growth of Firms Existing in 1990 (2) – (1) –10,572 –9,661 –46,276 395 4,071 –6,574 –34,828 –17,114 –120,559 Employment Growth of New Firms (3) – (2) 15,611 46,018 5,316 5,096 19,742 72,684 46,761 47,568 258,796 18 services, and defense/aerospace industries. By 2001, only the computers/ communications industry had more employees. Old firms lost jobs because not all of them survived after ten years. Also, other old firms might still be growing, but the growth occurred outside Silicon Valley. Table 2.2 provides a clear indication that Silicon Valley shifts development paths and remakes itself through the formation and growth of new firms. Firm Growth Because of the lack of sales data, firm growth is measured by employment growth. Tables 2.3 and 2.4 present the average employment of high-tech firms that are still alive. Firm sizes in service and other industries are calculated separately. On average, a high-tech start-up in nonservice industries hires 7–22 persons in the first year, depending on the cohort. As the start-up becomes older, its average employment is larger. In contrast to our general impression, the average growth of start-ups is far from explosive. It generally takes 5–6 years for an average start-up to double its employment. Firms in service industries are generally smaller and experience much slower growth. Before 1997, new firms in service industries always had an average employment below five in the first year. It takes more than nine years for service firms to double their average employment. A majority of them hardly grow at all. The growth is underestimated because the employment at a firm’s branches outside Silicon Valley is not captured here because of data limitations. Yet the number is meaningful because it measures the growth of start-ups within Silicon Valley. The growth is not accelerating as the data might have suggested. The faster growth at older ages results because many firms were defunct by those ages and only the fast-growing firms survived and were counted. Tables 2.3 and 2.4 suggest that the kind of explosive growth achieved by such stars as eBay and Yahoo is phenomenal, even by Silicon Valley’s standard. Figures 2.5 and 2.6 compare the size of high-tech firms in Silicon Valley with those in the Boston and Washington, D.C., areas. Nonservice high-tech firms seem to grow faster in Silicon Valley. Each 19 Table 2.3 Growth of Silicon Valley’s High-Tech Firms in Nonservice Industries 20 Cohort 1990 1991 1991 12.48 (58.83) 1992 1993 1994 1995 1996 1997 1998 1999 2000 1992 12.58 (58.93) 8.22 (11.76) 1993 12.59 (59.03) 8.41 (12.09) 7.34 (10.73) 1994 14.96 (63.73) 9.70 (14.17) 8.05 (12.09) 8.53 (14.28) NOTE: Standard deviations are in parentheses. Average Employment 1995 1996 1997 15.77 21.51 26.50 (65.15) (96.69) (114.3) 11.52 12.40 16.95 (19.54) (20.07) (36.36) 9.23 10.46 12.95 (14.23) (17.84) (23.35) 9.09 9.85 12.40 (14.82) (15.01) (18.59) 9.25 10.06 11.07 (13.27) (15.34) (16.39) 9.58 10.09 (16.14) (16.15) 10.24 (26.60) 1998 30.27 (127.6) 18.78 (42.69) 15.47 (30.32) 15.78 (24.28) 13.61 (20.35) 13.68 (22.67) 11.66 (28.02) 9.01 (16.83) 1999 40.54 (234.5) 21.86 (53.11) 17.99 (35.73) 18.43 (30.52) 16.55 (26.58) 16.68 (34.65) 15.00 (33.43) 9.65 (18.56) 8.41 (13.53) 2000 48.18 (281.1) 27.16 (64.50) 21.74 (58.63) 24.57 (81.62) 20.60 (34.46) 20.29 (41.17) 21.00 (50.53) 12.66 (24.43) 10.85 (23.92) 20.78 (141.8) 2001 55.74 (314.3) 35.15 (98.79) 22.90 (60.93) 35.58 (119.1) 28.19 (68.31) 28.61 (58.32) 29.12 (66.00) 20.97 (67.43) 17.23 (49.50) 20.29 (117.1) 22.51 (149.2) 21 Table 2.4 Growth of Silicon Valley’s High-Tech Firms in Service Industries Cohort 1990 1991 1991 4.88 (13.66) 1992 1993 1994 1995 1996 1997 1998 1999 2000 1992 4.90 (13.65) 4.30 (8.07) 1993 4.93 (13.70) 4.26 (7.98) 4.28 (8.82) 1994 4.98 (14.22) 4.55 (9.43) 4.46 (9.04) 4.65 (9.34) Average Employment 1995 1996 1997 5.60 5.44 6.17 (22.26) (15.96) (20.73) 4.69 5.06 5.69 (10.0) (10.28) (13.53) 4.58 4.87 5.85 (9.37) (10.30) (15.49) 4.80 5.14 5.56 (9.77) (10.77) (11.70) 4.74 4.68 4.83 (8.60) (8.06) (8.31) 4.70 4.91 (9.38) (9.70) 4.45 (10.32) 1998 6.70 (22.11) 6.52 (17.84) 5.89 (16.07) 5.88 (15.32) 5.17 (9.06) 5.53 (11.42) 4.72 (11.26) 6.61 (61.29) 1999 7.52 (26.35) 6.71 (18.42) 6.26 (17.08) 5.90 (12.71) 5.88 (13.54) 5.98 (13.08) 5.08 (10.35) 6.80 (61.79) 3.98 (7.75) NOTE: Standard deviations are in parentheses. 2000 7.90 (29.12) 6.18 (13.78) 6.62 (17.52) 5.93 (12.36) 5.84 (11.64) 6.12 (13.12) 5.66 (12.31) 7.35 (62.98) 4.18 (7.96) 6.77 (17.90) 2001 12.47 (83.00) 6.75 (15.17) 6.82 (19.00) 6.29 (14.13) 5.77 (11.22) 6.97 (18.79) 6.28 (14.43) 8.77 (66.70) 5.13 (12.80) 7.68 (19.92) 12.93 (126.5) 60 Silicon Valley 50 Boston Washington, D.C. 40 Average employment by 2001 30 20 10 Average employment by 2001 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Founding year Figure 2.5—Employment of High-Tech Start-Ups in Nonservice Industries, 2001 14 Silicon Valley 12 Boston Washington, D.C. 10 8 6 4 2 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Founding year Figure 2.6—Employment of High-Tech Start-Ups in Service Industries, 2001 22 cohort of nonservice firms founded during 1990–1999 has higher average employment in Silicon Valley. The 2000 cohort, only one year old in our data, is the only group of Silicon Valley nonservice firms that does not dominate its counterparts in the other two regions by employment size. This may suggest that Silicon Valley’s nonservice high-tech firms start with a smaller employment size but grow faster. In service industries, Silicon Valley high-tech firms are not consistently larger than those in other areas. In three cohorts, Silicon Valley firms have the smallest average employment; yet in five others, Silicon Valley firms have the largest. Silicon Valley service firms founded in the late 1990s seemed to do particularly well, which may be attributable to the Internet boom that especially benefited Silicon Valley. Figure 2.5 also shows that service firms are quite similar in size across different cohorts, which implies that they grow slowly over time. This section has demonstrated that start-ups in nonservice industries grow faster than those in service industries, and the previous section has shown that a higher proportion of start-ups in Silicon Valley occurs in nonservice industries. These together provide another reason why more firms in Silicon Valley than in Boston or Washington, D.C., had hired five or more employees by 2001 (Figure 2.3). Firm Mortality In the general practice of corporate demography literature (Carroll and Hannan, 2000), the mortality of a firm refers to any event by which a firm loses its identity. For example, a firm may disband, exit to another industry, or be merged or acquired. In this study, we are particularly interested in the disbanding of firms, since it has implications for the job market. We consider a firm dead if it drops out of the D&B dataset, since most probably it disbanded. A firm that has shifted to a different industry will simply have a new standard industrial classification (SIC) number. Those that go through merger and acquisition will simply have a different “headquarter DUNS number.” Neither will drop out of the D&B database. Firms do change their businesses sometimes. Among high-tech startups founded in Silicon Valley since 1990, 4.65 percent had changed their eight-digit SIC numbers at least once by 2001. For Boston and 23 Washington, D.C., the number was 2.59 percent and 2.54 percent, respectively. Although a high percentage of changing SIC numbers may imply a fast-changing local economy, we have little information to tell why firms exit to other industries. Rate of Mortality Table 2.5 describes the death of high-tech establishments by size between 1990 and 2000. Between 30 and 50 percent of establishments died during those 11 years. Establishments that hire fewer than 20 people have a higher chance of failing and hence provide less job security to their employees. Those with over 5,000 employees are also more likely than midsized establishments to fail, although the small sample size of establishments in that category suggests caution in the comparison. Although small establishments are more likely to disappear, the death of large establishments has a much greater effect on the labor market. Whereas the death of 21,967 establishments under size 20 left 84,453 people jobless, the death of 18 establishments with more than 2,500 employees eliminated 102,518 jobs. Figure 2.7 plots the survival rates of high-tech start-ups in Silicon Valley during 1990–2000. Nonservice start-ups have higher survival Table 2.5 Death of High-Tech Establishments in Silicon Valley, 1990–2000 Establishment Size 0–4 5–9 10–19 20–50 51–100 101–250 251–500 501–1,000 1,001–2,500 2,501–5,000 5,000+ Establishments in Sample 33,277 6,722 4,386 3,867 1,423 948 368 138 107 30 13 Establishments Dead by 2001 16,933 3,142 1,892 1,521 557 331 151 42 42 12 6 % Died 50.9 46.7 43.1 39.3 39.1 34.9 41.0 30.4 39.3 40.0 46.2 Job Loss by Death 40,530 19,805 24,118 47,149 42,572 54,505 54,248 32,400 72,234 46,800 55,718 24 Survival rate 1.2 Nonservice firms 1.0 Service firms 0.8 0.6 0.4 0.2 0 0 1 2 3 4 5 6 7 8 9 10 11 Age Figure 2.7—Survival Rates of High-Tech Firms in Silicon Valley rates than service firms in the long run. About 76 percent of the nonservice start-ups and 72 percent of service start-ups are still alive at age five. Only 46 percent of nonservice firms and 42 percent of service firms are still in business at age ten. The third year seems to be the most dangerous age. About 15 percent of Silicon Valley’s high-tech start-ups in service industries and 9 percent of those in nonservice industries died at that age. Figure 2.8 compares the survival rates of high-tech firms in Silicon Valley, Boston, and Washington, D.C. In nonservice industries, the survival rates are almost identical in the three areas. In service industries, firms in Silicon Valley have a better chance to survive than those in the other two regions. The relative size of the service industries is larger in Boston and Washington (Table 2.1), which may imply that service firms in those areas are less efficient or face harsher competition and hence have lower survival rates. Merger and Acquisition Acquisition is the generic term used to describe a transfer of ownership. A corporate acquisition occurs when a buyer purchases the stock or assets of a corporation. A merger has a strict legal meaning that 25 Survival rate 0.9 Silicon Valley 0.8 Boston Washington, D.C. 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 At age 5 At age 10 Nonservice At age 5 At age 10 Service Figure 2.8—Comparison of Survival Rates refers to the process in which one corporation is combined with and disappears into another. All mergers occur as specific transactions in accordance with the laws of the states where the firms are incorporated. Merger is a narrow technical term for a particular legal procedure that may or may not happen after an acquisition. The post-deal manner of operating or controlling a firm has no bearing on whether a merger has occurred. With regard to the NETS dataset, we consider a merger or acquisition to have happened if a firm is not a “branch” or “subsidiary” at its starting year but becomes a “branch” or “subsidiary” at the ending year.3 Figure 2.9 shows the percentage of high-tech firms acquired in each region by 2001. Note that the cohort year refers to the founding date of the firms that were acquired. The acquisition did not necessarily happen that year. In most cases, the acquisition happened a few years later. Overall, firms in Silicon Valley are most likely to change ownership. ____________ 3Alternatively, we could say a firm has changed ownership through M&A if it now has a “headquarter DUNS number” different from its own DUNS number. This gives almost identical results. 26 Percentage 4.5 4.0 Silicon Valley Boston 3.5 Washington, D.C. 3.0 2.5 2.0 1.5 1.0 0.5 0 1990 1991 1992 1993 1994 1995 1996 Founding year 1997 1998 1999 Figure 2.9—Percentage of Firms Acquired by 2001 Those in Washington, D.C., are least likely to be acquired. Among each cohort in each region, less than 4 percent of firms founded in the 1990s had been acquired by 2001. This is a relatively small number compared to how many had gone out of business. As we see in the next chapter, venture-backed firms are much more likely to be bought. Table 2.6 lists the top headquarter states whose firms tend to acquire high-tech start-ups in the three high-tech regions. Not surprisingly, a large proportion of the start-ups were acquired by local firms: California firms top the acquisition list in Silicon Valley, Massachusetts firms bought more high-tech start-ups in the Boston area, and firms in Virginia and Maryland acquired more high-tech start-ups in the Washington, D.C., area. Whereas California firms bought 56 percent of the start-ups acquired in Silicon Valley, Massachusetts firms acquired only 36 percent of those in the Boston area. In Washington, D.C., firms in Maryland, Virginia, and the city Washington bought 45 percent of the firms. Firms in California, New York, New Jersey, and Massachusetts have a strong showing in all three high-tech centers, which probably reflects the fact that those states have more established hightech companies than other states. 27 Table 2.6 Top Headquarter States of Firms Acquired During 1990–2001 Silicon Valley (Total: 1,376) State Cases 1 California 769 2 New York 97 3 Massachusetts 69 4 New Jersey 45 5 Texas 45 6 Pennsylvania 36 7 Florida 26 8 Illinois 26 9 Minnesota 24 10 Virginia 22 Boston (Total: 965) State Cases Massachusetts 350 California 134 New York 115 New Jersey 32 Texas 32 Illinois 31 Connecticut 28 Pennsylvania 27 Florida 20 Maryland 17 Washington, D.C. (Total: 814) State Cases Virginia 211 Maryland 124 California 81 New York 72 New Jersey 37 Massachusetts 36 Texas 33 Washington, D.C. 29 Florida 23 Pennsylvania 23 Job Creation by Start-Ups In this study, we refer to firms that are five years old or younger as start-ups. When new firms are founded, they create jobs. Yet many start-ups fail long before they become mature, thereby eliminating jobs. To pick up the net effect, we track the total employment of high-tech start-ups younger than certain ages, which is presented in Figure 2.10. Since 1995, high-tech start-ups in Silicon Valley have offered everincreasing numbers of jobs. In 1995, 47,200 employees worked for high-tech start-ups younger than two years old. By 2001, that number increased to 69,200. In 1998, start-ups younger than age five offered 132,500 high-tech jobs; the number had risen to 159,300 by 2001. To assess the relative importance of start-ups as job creators, we calculate the employment of start-ups younger than age five as the percentage of total high-tech employment in Silicon Valley and compare it with the same measure for the Boston area and Washington, D.C. (Figure 2.11). During 1998–2001, start-ups younger than age five consistently accounted for more than 20 percent of the high-tech employment in Silicon Valley. The percentage increased from 21.9 percent in 1998 to 23.7 percent in 2001. This means that jobs offered by start-ups grew faster than the total high-tech sector in the valley. The measure for Boston is a little higher and more stable—about 24 percent 28 Employment 180,000 160,000 140,000 Employment of start-ups younger than age 5 Employment of start-ups younger than age 4 Employment of start-ups younger than age 3 Employment of start-ups younger than age 2 120,000 100,000 80,000 60,000 40,000 20,000 0 1995 1996 1997 1998 1999 2000 2001 Percentage Figure 2.10—Employment of High-Tech Start-Ups in Silicon Valley 35 Silicon Valley 30 Boston Washington, D.C. 25 20 15 10 5 0 1998 1999 2000 2001 Figure 2.11—Employment of High-Tech Start-Ups Younger Than Age Five as a Percentage of Total High-Tech Employment 29 during the four years. The measure in the Washington, D.C., area is significantly higher than those in the other two regions. In 2001, startups offered 128,200 jobs in Washington, D.C., which amounted to 28.6 percent of the total employment in high-tech industries. In 1998, the percentage was even higher, when one out of every three employees in the high-tech sector worked for a start-up that was younger than five years old. Conclusion The whole picture of entrepreneurial activities, as presented here, differs somewhat from the public’s general impression. The media tend to direct attention to a small group of venture-backed firms. In fact, thousands of new firms are founded each year in Silicon Valley; venturebacked start-ups represent only a small proportion of the total. The public is too familiar with stories about the explosive growth of Silicon Valley start-ups but, in fact, a large proportion of every cohort of new firms founded in the valley will never hire more than five people. Hightech start-ups in service industries grow slower than other high-tech startups. Start-ups have been major job creators in Silicon Valley during the past decade; firms founded after 1990 created almost all the new jobs added to the region’s high-tech sector during 1990–2001. However, even during the decade characterized by the Internet boom, firm mortality rate was quite high in Silicon Valley. More than half of the firms started during the decade went out of business by age ten. 30 3. Venture-Backed Start-Ups in Silicon Valley This chapter examines venture-capital-backed start-ups, which are more innovative and growth-oriented than other high-tech start-ups. Venture capital refers to money managed by professionals who invest in young, rapidly growing companies that have the potential to develop into significant economic contributors. Venture capital is an important source of equity for start-up companies, particularly in the high-tech sector. In the San Francisco Bay Area, there has been a long tradition of wealthy people financing new technology firms. Yet, professional venture capital activity started later in the Bay Area than in the Boston area (Bygrave and Timmons, 1992; Kenney and Florida, 2000). In 1957, when Robert Noyce and seven fellow engineers left Shockley Semiconductor Laboratories to start their own business, they had to go to the East Coast to look for capital. The first West Coast venture capital firm—Draper, Gaither & Anderson—was not founded until 1958. The venture capital industry grew hand in hand with the high-tech industries in Silicon Valley. Since the 1960s, venture capitalists have been involved in every major successful company. Today, venture capital has become an intrinsic part of any story about Silicon Valley. Sand Hill Road in Menlo Park, the cluster of Silicon Valley’s venture capital firms, is virtually synonymous with venture investing. Venture Capital in Silicon Valley Figure 3.1 traces the nominal amount of venture capital invested in the United States and Silicon Valley over the ten years from 1992 to 2001. The trend is characterized by two big jumps and one severe crash. Between 1992 and 1994, venture capital investment first increased from $9.2 billion to $10 billion and then dropped to $8 billion. Compared to 31 Investment ($ billions) Silicon Valley as a % of U.S. total 120 30 100 25 80 20 60 U.S. total Silicon Valley 40 Silicon Valley as a % of U.S. total 15 10 20 5 00 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.1—Total Venture Capital Investment, 1992–2001 what happened later, this 10 percent increase and 20 percent decline seem to be negligible changes. Between 1994 and 1996, venture capital investment first experienced a 66 percent increase over the year, followed by another 59 percent growth. During these two years, venture capital investment jumped from $8 billion to $21.3 billion, stimulated by the promising Internet revolution. The year 1997 was relatively quiet, with venture capital investment dropping slightly to $20.4 billion. The next three years can only be described as mania: Venture capital investment increased first by 20 percent, then by 173 percent, and finally by 66 percent, ending with a total of $112.2 billion in 2000. In nominal dollars, venture capital investment in 2000 was 14 times as much as it was in 1994. This mirrors the Internet bubble seen in the NASDAQ index. The burst of the bubble is also reflected in venture capital investment. In 2001, the total crashed down to $32.5 billion, a 71 percent decline. Yet, in spite of this big falloff, the year 2001 still represents the third most heavily invested year in venture capital history. Venture capital invested in Silicon Valley followed a similar trend over the ten years. At its peak in 2000, Silicon Valley attracted nearly 32 $28 billion of venture capital investment. The decline in investment in 2001 also appeared in Silicon Valley. Still, the $7.7 billion invested in that year is the third-largest number the region has ever witnessed, second only to the venture investments in 1999 and 2000. In terms of the proportion of the U.S. total, Silicon Valley’s share has increased over the decade. In 1992, 18.7 percent of the total investment took place in Silicon Valley; in 1993, the number dropped slightly to 17.6 percent. Yet at its peak in 2000, Silicon Valley accounted for 24.8 percent of the U.S. total. Figure 3.2 compares Silicon Valley with the Bay Area as a whole, the Boston area, and Washington, D.C. Boston and Washington also experienced a large increase in venture capital investment during the late 1990s, following the national trend. However, the increases in Boston and Washington are not nearly as sharp as those in Silicon Valley and the Bay Area. It is particularly worth noting that the trend in the Bay Area shot up higher than that in Silicon Valley in the peak year 2000. That year, Bay Area firms outside Silicon Valley took in more than $10 billion 45 40 Bay Area Silicon Valley 35 Boston 30 Washington, D.C. 25 20 15 10 5 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.2—Total Venture Capital Investment, by Region, 1992–2001 33 Investment ($ billions) of venture capital. This may represent a big spillover from Silicon Valley. One possibility is that too much money was chasing too few entrepreneurs in Silicon Valley and venture capitalists had to look for opportunities nearby; or, more likely, Silicon Valley simply became too crowded and expensive, making adjacent metro areas such as San Francisco and Oakland more attractive. Table 3.1 summarizes the venture capital raised by each industry in Silicon Valley during 1992–2001. Software, communication, consumer/business services, semiconductor, electronics, information services, medical devices, and biopharmaceutical industries account for more than 96 percent of the total investment in Silicon Valley. The top three industries alone—software, communication, and consumer/business services—absorbed 63 percent of the total investment. These are also the top three industries in the nation as a whole, accounting for 59 percent of the total investment, although it is Table 3.1 Real Venture Capital Investment, by Industry in Silicon Valley, 1992–2001 Industry Software Communication Consumer/business services Semiconductor Electronics Information services Medical devices Biopharmaceutical Retailing Medical information services Advance/special material and chemical Other Healthcare Consumer/business products Energy Agriculture Total aIn 1996 dollars. Venture Capital Raised ($ millions)a 18,738.19 16,668.09 9,364.75 7,038.37 4,740.20 4,310.70 4,201.78 3,431.67 1,314.42 693.58 321.41 108.44 66.46 57.20 18.63 — 71,073.89 % of Total 26.36 23.45 13.18 9.90 6.67 6.07 5.91 4.83 1.85 0.98 0.45 0.15 0.09 0.08 0.03 — 100 No. of Deals 2,027 1,075 757 632 467 419 489 275 74 75 29 14 7 23 5 1 6,369 34 the communication industry that tops the U.S. list. The top three industries are all very much Internet-related, clearly indicating that the 1990s were the “Internet decade” for the venture capital world. Ranked eighth in the United States, the semiconductor industry is ranked fourth in Silicon Valley. Thus, the industry for which Silicon Valley was named is still relatively well-invested. Although the biopharmaceutical industry ranks fifth in the United States, it holds only the eighth position in Silicon Valley. This is partly because the biotech industry in the Bay Area is most heavily concentrated in South San Francisco, which is outside Silicon Valley by our definition. Firm Formation Figure 3.3 traces the trend of venture-backed start-ups by their founding year. The number of such start-ups steadily increased during the 1990s, peaking in 1999 and then declining sharply in 2000 and 2001. The decline reflects the burst of the Internet bubble and an economy heading toward a recession. Since it is possible that some startups founded in 2000 and 2001 will not complete their first round of financing until after 2001 and hence are not included in our data, the actual decline could be less serious than reflected in our data. The trend in Silicon Valley (where, on average, 22 percent of venture-backed startups are located) roughly parallels the national trend. Figure 3.4 depicts the trend of start-up formation for different hightech regions. Silicon Valley substantially outperformed the Boston and Washington, D.C., areas, although the three regions follow quite similar trends. In Silicon Valley, 84 start-ups founded in 1994 were financed by venture capital; the number steeply increased to 375 in 1999. During the same period, the number increased from 55 to 147 in the Boston area and from 12 to 77 in the Washington area. Percentagewise, the Washington area experienced a larger increase than Silicon Valley. The San Francisco Bay Area as a whole experienced intensive entrepreneurial activities in the late 1990s. During 1998–1999, the peak years of the Internet boom, more venture-backed start-ups were founded in the Bay Area than in the Boston area, even when excluding Silicon Valley from the Bay Area. 35 Silicon Valley as a % of U.S. total Number 2,000 25 1,800 1,600 20 1,400 1,200 1,000 800 600 U.S. total Silicon Valley Silicon Valley as a % of U.S. total 15 10 400 5 200 00 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Founding year SOURCE: Author’s calculations from the VentureOne database. Figure 3.3—Venture-Backed Start-Ups, 1990–2001 700 Bay Area 600 Silicon Valley Boston 500 Washington, D.C. 400 300 200 100 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Founding year SOURCE: Author’s calculations from the VentureOne database. Figure 3.4—Venture-Backed Start-Ups, by Region, 1990–2001 Number 36 Although so many venture-backed start-ups were founded in the late 1990s, entrepreneurs faced less-stringent capital constraints. The bull market in stocks and the enormous successes of early Internet-related start-ups attracted a large amount of money into the venture capital industry. As Figure 3.5 shows, start-ups founded in the late 1990s were much more generously financed than previous cohorts. In Silicon Valley, the average amount of venture capital per deal in 1992 was $6.33 million. By 1998, the average amount had climbed to $8.64 million. In 1999 and 2000, abundant venture capital showered on Silicon Valley: The average amount per deal jumped to $16.24 million in 1999 and further shot up to $22.34 million in 2000. Even in late 2001, when Silicon Valley had entered a deep recession, the venture capital industry still found itself in a situation of “too much money chasing too few ideas.” In the end, entrepreneurial ideas were exhausted, not the venture capital. In 2002, many venture capital funds had to downsize and return committed cash to investors because of lack of good opportunities (“The VCs Don’t Want Your Money Anymore,” July 29, 2002). Average venture capital per deal follows a similar trend in the Boston and San Francisco Bay areas. In the Boston area, the average amount dramatically increased from $6.25 million in 1998 to $18.41 million in 25 United States Silicon Valley 20 Boston Washington, D.C. 15 10 5 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.5—Average Amount of Venture Capital Raised per Deal, 1992–2001 37 Average raised ($ millions) 2000. The trend in Washington, D.C., is different. In 1999 and 2000, while venture capital deals were getting fat in Silicon Valley and Boston, they were shrinking in Washington, D.C. Also interesting to note is that the average amount of venture capital per deal was considerably higher in Washington, D.C., than in other areas during 1993–1996 and 1998. For example, in 1993, 1994, and 1998, the average in Washington, D.C., is at least 100 percent higher than in Silicon Valley or Boston. A closer look at the data for the Washington, D.C., area reveals that during 1993–1996 and 1998 a disproportionately large share of venture capital was invested in the communication and healthcare industries, with both tending to acquire extremely big deals. How quickly start-ups are able to obtain venture capital is also an indicator of the availability of capital in a region. We calculated the average span of time between the founding date of a start-up and the closing date of its first round of financing. Figure 3.6 shows that Silicon Valley firms are financed more quickly than firms elsewhere. In Silicon Valley, start-ups on average have raised their first round of venture capital at the age of 11.59 months. For the whole Bay Area, the average age is 11.86 months, only slightly higher than in Silicon Valley. The number is 16.58 for Boston and 16.62 for the nation as a whole. In Washington, D.C., it takes 22.54 months to close the first round of financing. 25 20 Average age (months) 15 10 5 0 United States Silicon Valley Bay Area Boston SOURCE: Author’s calculations from the VentureOne database. Washington, D.C. Figure 3.6—Average Start-Up Age at First-Round Financing 38 One naturally wonders whether Silicon Valley’s time-efficiency is due to its specific industry composition, since the time needed for venture capital financing may be inherently different from one industry to another. Figure 3.7 compares average firm age at the first round of venture capital financing in Silicon Valley with the national average and the Boston area average within each industry. These five industries are the top five in Silicon Valley, accounting for 80 percent of its venture capital investment. Clearly, in each industry, firms in Silicon Valley are financed more quickly. For example, Silicon Valley start-ups in the software industry can have their first rounds of venture capital in place six months sooner than Boston start-ups in the same industry. In the electronics industry, the time advantage is 7.6 months. In consumer or business services, firms in Boston are on average 19.4 months old when their first round of venture financing is completed; those in Silicon Valley are only 10.3 months old. In fact, in 14 out of 16 industry segments, Silicon Valley firms take shorter time to get venture capital 25 U.S. total Boston Silicon Valley 20 Average age (months) 15 10 5 0 Computers/ communications Consumer/ business services Electronics Semiconductor SOURCE: Author’s calculations from the VentureOne database. Software Figure 3.7—Average Start-Up Age at First-Round Financing, by Industry 39 than both the national average and the average time in Boston. The two exceptions are the healthcare industry and “other,” in which only six start-ups in Silicon Valley got financed during 1992–2001. Several possible reasons may explain the promptness of the venture capital financing in Silicon Valley: (1) The well-developed venture capital industry in the region allows start-ups to find financing locally and hence speeds up the process; (2) the well-connected business networks in Silicon Valley enable entrepreneurs to find venture capitalists (or the other way around) more quickly; or simply (3) venture capitalists in Silicon Valley work differently from their counterparts elsewhere. Start-ups in Silicon Valley naturally enjoy some advantages because of the abundance of local capital. It is well known that venture capital firms tend to finance local start-ups, so that they can closely monitor their performance and provide management guidance or assistance if needed. Silicon Valley has the world’s largest venture capital cluster. Thus, firm founders in Silicon Valley have relatively easy access to capital. However, a large venture capital industry does not seem to fully explain quick venture capital finance. For example, the Seattle area has a much smaller venture capital industry than the Boston area. In fact, Boston is undoubtedly the number two venture capital cluster in the world. According to VentureOne’s Venture Capital Sourcebook (2001), Massachusetts has 94 venture capital firms and Washington state has only 26. During 1992–2001, venture capital investment in the Boston area amounted to $31.1 billion; that number is only $10.1 billion for Seattle. However, in spite of the significant size differences in these venture capital industries, start-ups in Seattle received faster venture capital financing than those in the Boston area (16.2 months compared to 16.6 months). Thus, the proximity to considerable capital does not guarantee quick access. Silicon Valley’s risk-tolerating culture might be the real reason for the quick venture capital financing in the region. After interviewing individuals who had worked in both the Boston area and Silicon Valley, Saxenian (1994) observed that “East Coast venture capitalists were more formal and conservative in their investment strategies.” The interviewees’ experiences in the two regions help us understand the cultural difference. An entrepreneur in Silicon Valley told Saxenian, 40 “When I started Convergent [Technologies], I got commitments for $2.5 million in 20 minutes from three people over lunch who saw me write the business plan on the back of a napkin. They believed in me. In Boston, you can’t do that. It’s much more formal.” Another businessman says, “There is no real venture capital in Massachusetts. The venture capital community is a bunch of very conservative bankers. They are radically different from the venture capitalists in Silicon Valley, who have all been operational people in companies. Unless you’ve proven yourself a hundred times over, you’ll never get any money” (Saxenian, 1994). Although those comments were referring to the situation in the 1980s, it is likely that Silicon Valley preserved such cultural advantage in the 1990s. Whatever the reasons, quick financing probably gave entrepreneurs in Silicon Valley a head start over those in other regions. In the fastmoving high-tech sector, a year means a lifetime. And therefore, facilitated by local venture capital firms, innovative start-ups in Silicon Valley may enjoy some first-mover’s advantages. Ownership Status and Profitability The VentureOne data have specific variables indicating a firm’s business status and ownership status. In particular, we know with certainty whether a firm went out of business or whether it merged with another firm; we do not have to infer such events from other variables, as with the NETS data. This section focuses on ownership changes and the economic performance of venture-backed start-ups. A general impression about venture capital investment is that it is very risky. However, this is not reflected in the disbanding rate of venture-backed firms. Figure 3.8 depicts the ownership status of such firms in Silicon Valley as of the fourth quarter of 2001. In each cohort of Silicon Valley start-ups, those that have gone out of business never amount to more than 16 percent of the total. It seems that if a start-up can survive the first two years, it is very likely to succeed. As time goes by, more and more start-ups are acquired by or merged with other firms. About one-third of start-ups change ownership through merger and acquisition (M&A) before they are ten years old. This is a much higher percentage than we found in the NETS data for all the start-ups in the 41 Percentage Publicly held Private Acquired/merged Out of business 100 80 60 40 20 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Founding year SOURCE: Author’s calculations from the VentureOne database. Figure 3.8—Ownership Status of Venture-Backed Start-Ups in Silicon Valley, 2001 preceding chapter, which suggests that start-ups that are not venturebacked are much less likely to be acquired. Many start-ups will become publicly held through IPOs (Initial Public Offerings). In Silicon Valley, nearly 30 percent of venture-backed start-ups founded before 1995 had gone public by the end of 2001. IPOs and M&As provide channels for venture capitalists to exit and pay back their investors. Some start-ups remain privately held, but that proportion is declining over time. Figure 3.9 presents the ownership status of all venture-backed startups in the United States. The overall picture is similar to what we see in Silicon Valley: The disbanding rate stabilizes after two years, more and more start-ups go through M&A or IPO over time, and the number of private firms declines with time. It is also interesting to compare the ownership outcomes of Silicon Valley start-ups with the U.S. average. In each panel in Figure 3.10, a positive value means the examined proportion in Silicon Valley is higher 42 Percentage 100 90 80 70 60 50 40 30 20 10 0 1992 1993 1994 Publicly held Private Acquired/merged Out of business 1995 1996 1997 1998 Founding year 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.9—Ownership Status of Venture-Backed Start-Ups in the United States, 2001 than the national average. As the figures show, venture-backed start-ups in Silicon Valley are more likely to be acquired by or merged with other firms. The proportion of M&A is consistently higher than the national average for every cohort of start-ups. For example, M&A activities among start-ups founded in Silicon Valley in each year from 1993 to 1996 are at least 6 percent higher than the U.S. average. In the 1994 cohort, 36 percent of Silicon Valley start-ups went through M&A, compared to only 23 percent of the U.S. total. Generally speaking, M&A activities are more common in high-tech industries than in other sectors. According to Mergerstat’s industry report, 49 industries completed 7,518 M&A deals in 2002. The software industry alone accounted for 1,347 deals, 18 percent of the total. If we include computer hardware, communications, electronics, drugs, health services, and aerospace and defense, the seven high-tech industries account for nearly one-third of the total M&A deals. This is a 43 Percentage 15 10 5 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 M&A 15 10 5 0 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 IPO 5 0 –5 –10 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Out of business 10 0 –10 –20 –30 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Private SOURCE: Author’s calculations from the VentureOne database. Figure 3.10—Differences in Ownership Status in Each Cohort of VentureBacked Start-Ups: Silicon Valley Compared to the United States very high fraction given that even a much more broadly defined hightech sector produces less than 11 percent of the U.S. gross domestic product (GDP) (DeVol, 1999). M&A is more active in the high-tech sector for several reasons. On the one hand, for many start-up founders, being bought out is a major avenue to the aspired financial success; at the same time, M&A provides an exit channel by which venture capitalists collect the return on their investments. On the other hand, many 44 established companies in the high-tech sector have incentives to acquire start-ups. The most renowned example is Cisco, which has bought 76 high-tech start-ups since 1993. In fact, Cisco’s practice is so successful that it coined a new term: acquisition and development (A&D). An established company’s typical motivations for buying start-ups include acquiring a technology faster than through internal development, buying market share and presence, and buying talented people (Paulson, 2001). Two possible reasons may explain the fact that a higher proportion of Silicon Valley start-ups exit by M&A. First, Silicon Valley is no doubt the largest cluster of successful high-tech companies in the nation, all of which are potential buyers of young start-up firms. Being close to giants raises the possibility of being acquired. Second, Silicon Valley has probably the best developed networks that service the high-tech sector, including investment banks, venture capital firms, law firms, accountants, and consultants. They are all matchmakers that help form M&A deals. Compared to the U.S. average, Silicon Valley venture-backed startups are also more likely to go public. This is true for every cohort. The difference is especially striking for start-ups founded before 1996. Among the 2,058 start-ups founded in the United States during 1992– 1995, 403, or 20 percent of the total, had gone public by the end of 2001. In Silicon Valley, however, 118 out of 412 start-ups, or 29 percent of the total, founded in the same period were traded on the stock market by late 2001. For those founded in 1994 and 1995, the IPO rate is 12 percent higher in Silicon Valley. These two years inaugurated the era of the Internet revolution. Venture-backed start-ups in these cohorts include high-profile pioneers such as eBay, Netscape, and Yahoo. Figures 3.11 and 3.12 depict the business status of venture-backed start-ups in Silicon Valley and the nation as a whole, given that they were not disbanded. In general, only a small proportion of venture-backed start-ups founded from 1992 to 2001 were showing a profit in the fourth quarter of 2001. The older a start-up, the more likely it is profitable. However, even among the earliest cohort, those founded in 1992, less than 14 percent in Silicon Valley and less than 20 percent in the nation were making a profit in 2001. A majority of the start-ups less than two years old were still developing or testing products. 45 Percentage 100 90 80 70 60 50 40 30 20 10 0 1992 1993 1994 Profitable Shipping product Clinic/beta trial Product development Starting up or restarting 1995 1996 1997 1998 Founding year 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.11—Business Status of Venture-Backed Start-Ups in Silicon Valley, 2001 The proportion of profitable start-ups in Silicon Valley is lower than the national average in every cohort. For cohorts founded before 1995, the proportion of profitable start-ups in the United States is always at least 5 percent higher than the proportion in the valley. Moreover, startups in Silicon Valley are more likely to be in the product development or testing stages than the U.S. average. Among every cohort founded after 1995, Silicon Valley has a higher percentage of such firms. Given that Silicon Valley houses about one-fifth of the venture-backed start-ups in the nation, the difference between Silicon Valley and the rest of nation should be much more significant. This difference and the fact that start-ups in Silicon Valley have quick access to venture capital seem to suggest that the venture capital investments in the valley bear more risks than those in the rest of the United States. On the one hand, venture capitalists in Silicon Valley may have bet on many “bad” ideas and will never profit from them; on 46 Percentage 100 90 80 70 60 50 40 30 20 10 0 1992 1993 1994 Profitable Shipping product Clinic/beta trial Product development Starting up or restarting 1995 1996 1997 1998 Founding year 1999 2000 2001 SOURCE: Author’s calculations from the VentureOne database. Figure 3.12—Business Status of Venture-Backed Start-Ups in the United States, 2001 the other hand, they may have put money into very long term deals that will take many years to reach profitability. Neither of these is necessarily a bad strategy. In venture capital investment, the returns from a few superstar deals are usually more than enough to cover the money lost in many bad deals. The venture capitalists in Silicon Valley may have acted quickly to take the first mover’s advantage in producing superstars such as Yahoo and Netscape at the cost of investing in many bad deals. In the end, although fewer of the start-ups make profits, the venture capitalists’ returns from Silicon Valley investments may not be below the national average. Spinoffs Who founds high-tech start-ups? One group of entrepreneurs who have attracted researchers’ attention is previous employees of incumbent firms. In the high-tech sector, employees sometimes leave their 47 employers and start their own firms in the same industry, which are called spinoffs. For example, many firms in the semiconductor industry trace their origin to a successful firm in the early years of the industry: Fairchild Semiconductor. The long list of “Fairchildren” includes such prestigious trade names as Intel, National Semiconductor, Advanced Micro Devices, Cypress, Linear Technology, and Xilinix, among others. Spinoffs could emerge for various reasons (Klepper, 2001). For example, employees may wish to capitalize on important discoveries they make while working for a particular firm. However, it is often impossible for them to contract with their employers to commercialize the discoveries, so they start their own firms. In other cases, successful incumbent firms may have difficulties evaluating and implementing certain types of innovations, which gives individual employees opportunities to pursue such innovations by themselves. A classic example is Apple Computers’ cofounder Stephen Wozniak, who used to work for Hewlett-Packard. His design of a personal computer gained no appreciation from senior engineers, so he eventually gave up on his employer and teamed up with Steve Jobs, and the two built their own computers out of a start-up in a garage (Freiberger and Swaine, 2000). The effects of spinoffs are debatable. If employees profit from innovations they make at their previous employers, incumbent firms will have less incentive to spend on R&D. As a consequence, the whole industry may lose out to international competition, such as that faced by the U.S. semiconductor industry. On the other hand, one may argue that spinoffs provide a vehicle for knowledge transfer and hence accelerate innovation. In her renowned study of Silicon Valley and Route 128 in the Boston area, Saxenian (1994) contends that Silicon Valley enjoys a “regional advantage” partly because its culture and institutions encourage employees to move from one firm to another. In particular, employees in Silicon Valley feel free to transfer from established firms to start-ups. This so-called “high-velocity” labor market enables knowledge gained from one firm to quickly spill over to other firms. Since knowledge circulates among a collective learning network instead of traveling in one direction, all firms benefit from this phenomenon. Saxenian argues that established firms in the Boston area, on the contrary, tend to endorse a 48 more inward-looking culture, which encourages loyalty to employers rather than job mobility. In the Boston area, climbing the promotion ladder within a firm is more socially acceptable than taking the risk of starting one’s own business. According to Saxenian, these differential approaches provide an explanation of why Silicon Valley has overtaken the Boston area as the leading high-tech center in the country. Gilson (1999), a law professor at Stanford, further proposed an account of the differential business cultures between Silicon Valley and Route 128. He argued that it is the different legal infrastructures in California and Massachusetts—particularly the enforceability of postemployment covenants not to compete—that make the difference. The so-called “covenants not to compete” are contractual agreements between employees and employers in which the employee promises not to compete with the employer for a certain period of time within a specific geographic area in case the employment relationship terminates. Generally, Massachusetts’ courts have enforced such covenants to protect trade secrets, confidential data, or the employer’s good will. Under California law, such covenants are not enforceable. According to section 16600 of California Business and Professions Code, “every contract by which anyone is restrained from engaging in a lawful profession, trade, or business of any kind is to that extent void.” California courts have consistently referred to this stipulation to prohibit covenants not to compete. Employees in Silicon Valley know that they can leave current employers and found competing start-ups or join other firms in the same business; employers know that they cannot prevent such things from happening. As a result, employers in Silicon Valley adopt an approach that emphasizes both cooperation and competition. Both Saxenian and Gilson discuss Silicon Valley’s high-velocity employment using anecdotal evidence, because of a lack of empirical data. It has yet to be verified with empirical data whether established firms in Silicon Valley indeed have more spinoffs. To resolve this issue, we matched VentureOne’s founder information with firm-level data, so that we could identify where an entrepreneur founded his or her firm. We extracted two groups of venture-backed entrepreneurs by firm location: Silicon Valley and Boston. Using the biographic information of firm founders, we were able to identify which 49 companies a person had ever worked for. If an entrepreneur ever worked for a company or a university, we counted him or her as an “employee founder” from that company or university and the start-up as a “spinoff start-up” from that company or university. The number of employee founders does not necessarily agree with the number of spinoff start-ups. On the one hand, some employee founders turned into “serial entrepreneurs,” founding two or more start-ups; on the other hand, two or more employees may cofound a single start-up. Table 3.2 compares spinoffs from leading firms and universities in Silicon Valley and the Boston area.1 Indeed, leading firms in Silicon Valley significantly outperformed their counterparts in the Boston area in terms of producing entrepreneurs. Raytheon and DEC are probably the two most prestigious names in the Boston area’s high-tech history. DEC Table 3.2 Number of Spinoffs from Leading Institutions in Silicon Valley and the Boston Area Silicon Valleya Boston Areab Employee Spinoff Founders Start-Ups Employee Spinoff Founders Start-Ups Leading Companies Apple Cisco HP Intel Oracle SGI Sun IBM 94 71 Data General 13 41 35 DEC 52 117 99 EMC 9 76 68 Lotus 29 73 57 Prime 5 50 37 Raytheon 7 101 79 Wang 11 82 77 IBM 23 13 41 6 26 5 7 11 23 Leading Universities Stanford UC Berkeley 71 20 64 MIT 20 Harvard 74 63 32 31 aFounder sample size: 2,492. bFounder sample size: 1,157. ____________ 1The VentureOne data cover only start-up founders who have ever been funded by venture capital since 1992. Therefore, no number in Table 3.2 should be interpreted as the total number of spinoffs in the firm’s (or university’s) history. 50 scored the highest in the Boston area with 52 employee founders; Raytheon, the largest employer in Boston’s high-tech sector, with about 15,000 employees locally, produced only seven entrepreneurs according to the VentureOne data.2 Together, DEC and Raytheon spun off 48 venture-backed start-ups, only about half of the 99 spinoffs from Hewlett-Packard. Sun Microsystems, a 20-year-old company in Silicon Valley, has seen more than 100 previous employees become venturebacked entrepreneurs; yet EMC, another big name in the Boston area, founded three years earlier than Sun, had only nine employees who founded start-ups. Apple Computers and Lotus Development Corporation are no doubt two of the most successful pioneers in the early years of the personal computer era. Apple in Silicon Valley has spun off 71 venture-backed start-ups, whereas Lotus in the Boston area lags far behind with only 26 spinoffs. Boston’s Data General, Prime Computer, and Wang Laboratory all once were giants in the minicomputer market created by DEC, but they are all dwarfs in terms of spinoffs, compared to leading firms in Silicon Valley such as Cisco, Oracle, or SGI. Even IBM, the New York–based high-tech conglomerate that has a presence in both areas, has many more spinoffs in Silicon Valley. However, a comparison of the leading universities in the two areas tells a different story. In this case, Boston is doing as well as Silicon Valley, if not better. Among the 1,157 venture-backed entrepreneurs in Boston, 74 have worked at MIT; yet only 71 out of the 2,492 entrepreneurs in Silicon Valley have ever been Stanford employees. Harvard also comes in better than Berkeley, 32 to 20. Notice, we consider here only entrepreneurs who have ever worked at those universities. The number of firm founders who graduated from those universities is a natural alternative measure. Unfortunately, the VentureOne data do not provide such information. Table 3.2 suggests that leading firms in Silicon Valley have spun off more entrepreneurs than those in the Boston area; however, leading universities in Boston have more employees who commercialize their innovations by founding ____________ 2Raytheon’s role as a defense contractor may have placed some restrictions on potential employee founders. 51 start-ups. Whether it is the culture/institutions or the legal infrastructure that enables Silicon Valley to surpass Boston in terms of employee founders and spinoffs, the big difference seems to be in the business world as opposed to academia. Conclusion Since 1994, Silicon Valley has consistently accounted for more than 20 percent of the total venture capital investment in the United States. Venture capital investment in Silicon Valley surged in the late 1990s. On the one hand, the increased venture capital gave a big push to entrepreneurial activities during the Internet boom; on the other hand, the size of venture capital deals during the late 1990s became much bigger. Software, communications, consumer/business services, semiconductor, and electronics industries were the most heavily invested high-tech industries in the region. Silicon Valley start-ups have quicker access to venture capital in almost every industry. Although the exact reasons for this quick access remain unclear, it certainly helps firms in Silicon Valley to get a head start. A preliminary examination of venturebacked firm founders in Silicon Valley and the Boston area confirms that successful firms in Silicon Valley tend to have more spinoffs than their counterparts in Boston. The difference in university spinoffs is not significant between the two regions. 52 4. Firm Relocation in Silicon Valley Silicon Valley is the most renowned example of an industrial cluster. Classical economic theory teaches that a cluster is attractive to firms for multiple reasons, including a specialized labor pool, specialized inputs, proximity to customers, knowledge spillovers, and so on. For high-tech start-ups, the benefits from a cluster may also include easy access to capital and the psychological support that an entrepreneur receives from his peers and the community. It is worth noting that high-tech start-ups and mature firms may have different locational concerns. Start-ups are more dependent on outside resources at the developing stage, and, as newcomers, they suffer from lack of credibility. For these reasons, the entrepreneur’s local connections and his familiarity with local institutions have a big effect on the formation and growth of a start-up. Therefore, we observe two empirical regularities: (1) high-tech firm founders are usually engineers who have working experiences in industrial clusters, where local culture and institutions are favorable to new firms, and (2) high-tech firm founders rarely move outside the immediate area when they decide to start new firms (Cooper and Folta, 2000). Thus, start-ups are more likely to emerge in clusters such as Silicon Valley than in other places. However, mature firms are more likely to follow routinized operations and are more concerned about the costs of doing business. An industrial cluster, once established, tends to face high demand for labor and land and also the overuse of infrastructure. This drives up operating costs for firms in the cluster. At the same time, the overloaded infrastructure may lower the quality of life in the cluster. For these reasons, mature firms may choose to set up branches elsewhere rather than in the cluster. When the costs of doing business are high enough in a cluster, mature firms themselves may consider relocating. 53 Silicon Valley and the San Francisco Bay Area in general have long been notorious for the high costs of living and doing business. This has usually been recognized as a threat to economic growth in the area.1 High costs may deter firm formation in Silicon Valley and may push mature firms away so that the region will not reap the fruit it grows. Our analysis in the previous chapters has shown that entrepreneurial activities were intensive in Silicon Valley during the past decade. Entrepreneurs were probably driving up costs even higher, rather than being scared away by them. In this chapter, we investigate whether mature firms tend to leave Silicon Valley and, if they do, how large the effect is. Although start-ups differ from mature firms, high-tech firms and nontech firms may weigh factors differently when they consider their locations. High-tech firms are in knowledge-intensive businesses. Therefore, they care more about the availability of a well-educated labor pool and knowledge spillovers from their competitors and partners. In contrast, nontech firms might be more responsive to land price, transportation cost, tax burden, and so on. Silicon Valley experienced an Internet revolution in the high-tech sector during the 1990s. The intensive entrepreneurial activities in that period raise the question of whether the booming high-tech sector crowded out nontech firms. In this chapter, we shed some light on this issue. High-Tech and Nontech Relocation Throughout this chapter, firm relocation is measured by locational change at the establishment level. Remember, an establishment is a business or industrial unit at a single physical location. If a firm has multiple establishments, we track each single establishment rather than the firm as a whole. An establishment has moved if its reported address has changed. We do not consider acquisitions that result in ownership changes but not physical movements of establishments. Table 4.1 reveals some interesting facts. First, establishments in Silicon Valley do move. During 1990–2001, 25,485 out of 217,169 establishments changed addresses at least once. Some establishments ____________ 1One could also argue that the high cost of doing business is not a threat but a sign of Silicon Valley’s economic health. 54 Table 4.1 Relocation of Establishments in Silicon Valley, 1990–2001 Never moved Relocated out Relocated in Relocated within Silicon Valley Total High-Tech Sector No. of % in Establishments Total 42,354 82.44 1,490 2.90 894 1.74 6,637 12.92 51,375 100 Nontech Sector No. of % in Establishments Total 149,330 90.07 3,111 1.88 1,834 1.10 11,519 6.95 165,794 100 moved into the area, some moved out, and still others moved around within Silicon Valley. For the purpose of our study, we care more about those establishments moving in and out. Second, high-tech establishments are more likely to move than nontech establishments. Nearly 18 percent of high-tech establishments moved whereas only about 10 percent of nontech establishments relocated. Two possible reasons explain this difference. On the one hand, high-tech firms by their very nature are more mobile. Many hightech firms, especially those in software and research, use portable equipment and occupy little land space. Nontech sectors include establishments in agriculture, forestry, mining, utilities, and government branches and agencies, which are all somewhat attached to well-defined territories. Establishments in nontech manufacturing and services, although generally not fixed to their locations, may have bulky equipment or need large land space, and thus face high moving costs. On the other hand, high-tech establishments tend to develop fast and quickly outgrow their office space. So moving into new office buildings could be more common among them. Third, establishments are more likely to relocate within Silicon Valley. Among all establishments that moved, 79.8 percent remained within Silicon Valley. Fourth, establishments relocating out of Silicon Valley outnumber those moving in. This is true in both the high-tech and nontech sectors. It seems consistent with our intuition that the high costs of doing business may push businesses away from Silicon Valley. 55 Tables 4.2 and 4.3 list the top destination states and cities for establishments that moved out of Silicon Valley. It seems that distance is a very important factor in business relocation. A majority of the establishments moving out of Silicon Valley remained in California— 75.6 percent of high-tech establishments and 84.6 percent of nontech establishments. Those that moved to other states tended to choose Table 4.2 Top Ten Destination States for Establishments Relocating Out of Silicon Valley, 1990–2001 High-Tech Sector Destination No. of State Establishments 1 California 2 Texas 3 Nevada 4 Oregon 5 Colorado 6 Washington 7 Massachusetts 8 Arizona 9 Florida 10 New York 1,126 34 32 30 21 21 20 20 19 18 No. of Employees 12,700 1,570 354 355 1,404 187 932 208 1,944 1,272 Nontech Sector Destination No. of State Establishments California Oregon Arizona Nevada Washington Texas Colorado Florida Illinois Utah 2,631 56 47 40 39 36 33 24 20 18 No. of Employees 27,750 275 348 547 146 1,941 2,075 303 612 229 Table 4.3 Top Ten Destination Cities for Establishments Relocating Out of Silicon Valley, 1990–2001 High-Tech Sector Destination No. of City Establishments 1 San Francisco 2 Hayward 3 Burlingame 4 Pleasanton 5 Santa Cruz 6 San Ramon 7 Oakland 8 South San Francisco 9 Livermore 10 San Diego 148 88 84 75 34 28 27 26 21 16 No. of Employees 1,744 1,211 871 1,097 207 129 411 650 209 542 Nontech Sector Destination No. of City Establishments Hayward Burlingame San Francisco Pleasanton Livermore Santa Cruz San Leandro Oakland South San Francisco Sacramento 286 219 205 131 78 73 62 55 49 43 No. of Employees 3,414 2,968 2,312 2,998 1,360 354 498 490 828 544 56 states in the west. Arizona, Colorado, Nevada, Oregon, Texas, and Washington are among the top ten destination states for both high-tech and nontech establishments. It is interesting to note that Florida appears on both lists, probably for its California-like warm weather. East Coast states Massachusetts and New York are among the top ten destination states for high-tech establishments, possibly because both states have strong high-tech sectors; neither appears on the list for nontech establishments. The importance of distance is also reflected in the lists of top destination cities. San Francisco Bay Area cities occupy the top nine spots for both high-tech and nontech establishments. San Francisco, Hayward, Burlingame, and Pleasanton are the top four on both lists. Among those that leave the Bay Area, high-tech establishments tend to favor San Diego whereas nontech establishments are likely to go to Sacramento. Tables 4.4 and 4.5 list the top origin states and cities for establishments moving into Silicon Valley. A majority of the establishments moving into Silicon Valley—76.6 percent of high-tech establishments and 89.7 percent of nontech establishments—are from other places in California. Among high-tech establishments moving in from outside California, distance does not seem to be the only major determining factor. East Coast states Massachusetts, New York, and New Jersey each had more establishments that moved to Silicon Valley than California’s neighbors such as Arizona, Nevada, and Oregon. It is not surprising that Massachusetts, New York, and Texas follow California on the high-tech list because all of them have strong high-tech economies. Although Silicon Valley is the most concentrated high-tech industrial center in the country, its nontech sectors together have more establishments and hire more employees. Thus, in Table 4.2, it is quite natural to see more nontech than high-tech establishments leaving Silicon Valley. Yet, in Table 4.4, high-tech establishments moving in from outside California outnumbered nontech establishments. It is consistent with our general impression that Silicon Valley is more attractive to high-tech than nontech firms. 57 Table 4.4 Top Ten Origin States for Establishments Relocating Into Silicon Valley, 1990–2001 High-Tech Sector No. of Origin State Establishments 1 California 2 Massachusetts 3 New York 4 Texas 5 Illinois 6 Colorado 7 New Jersey 8 Oregon 9 Pennsylvania 10 Nevada 685 33 29 21 13 12 12 8 6 6 No. of Employees 13,453 1,168 727 182 148 440 222 74 106 69 Nontech Sector No. of Origin State Establishments California New York Texas Washington Nevada New Jersey Oregon Arizona Florida Illinois 1,645 19 15 14 13 13 13 12 8 8 No. of Employees 16,420 920 126 78 95 56 53 80 242 158 Table 4.5 Top Ten Origin Cities for Establishments Relocating Into Silicon Valley, 1990–2001 High-Tech Sector Nontech Sector No. of No. of No. of No. of Origin City Establishments Employees Origin City Establishments Employees 1 San Francisco 2 Burlingame 3 Hayward 4 South San Francisco 5 Santa Cruz 6 Pleasanton 7 Oakland 8 San Diego 9 Los Angeles 10 San Bruno 130 66 48 36 33 24 17 11 11 10 1,225 915 1,306 353 1,575 274 902 195 99 150 San Francisco Burlingame Hayward South San Francisco Oakland San Leandro Santa Cruz San Bruno Pleasanton Los Angeles 267 205 199 74 54 42 38 34 30 26 2,145 2,222 2,958 1,424 349 562 190 523 314 235 Again, top origin cities are mostly in the San Francisco Bay Area, with San Diego and Los Angeles the two exceptions. San Francisco, Burlingame, Hayward, and South San Francisco are the top four on both the high-tech and nontech lists. Comparing Table 4.5 with Table 4.3, we see that more establishments move out of Silicon Valley to adjacent cities than move in from those cities. This is particularly true for nontech sectors, in which we see that not only more establishments but 58 also more employees move out of Silicon Valley to nearby cities. This seems to suggest that Silicon Valley was expanding and its economic activities spilling over into other Bay Area cities. Table 4.6 breaks out relocating high-tech establishments by industry. During 1990–2001, 1,490 establishments moved out of Silicon Valley, which together represented 26,684 jobs; 894 establishments moved into Silicon Valley, with a total of 20,999 employees. Measured by either net establishments or net employees, the Silicon Valley economy is spilling out. In every industry, establishments moving out outnumbered those moving in. Out-moving establishments also had more employees except in two industries: In computers/communications, the net flow of employees is close to zero; in the semiconductor industry, there was a net employment inflow, although more establishments moved out. The two service industries had more moving establishments than the other hightech industries. Yet because service establishments are generally smaller, the relocation in the computer and software industries involved more employees. Table 4.7 summarizes firm relocation by industry group, including both high-tech and nontech industries. During 1990–2001, every sector in Silicon Valley registered a net loss because of firm relocation, measured either by total number of establishments or by total Table 4.6 High-Tech Establishments Relocating Into and Out of Silicon Valley, by Industry, 1990–2001 Bioscience Computers/ communications Defense/aerospace Environmental Semiconductor Software Professional services Innovation services Total Moving Out No. of No. of Establishments Employees 82 2,153 Moving In No. of No. of Establishments Employees 51 1,510 117 15 17 39 281 527 412 1,490 5,737 577 178 1,356 7,023 5,343 4,317 26,684 86 5,740 1 39 12 125 35 2,918 186 5,278 282 2,389 241 3,000 894 20,999 59 Table 4.7 All Establishments Relocating Into and Out of Silicon Valley, by Industry Group, 1990–2001 Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation, communication, and utilities Wholesale trade Retail trade Finance, insurance, and real estate Services Total Moving Out Moving In No. of No. of No. of No. of Establishments Employees Establishments Employees 48 306 5 42 386 3,501 796 17,769 29 237 3 31 184 1,103 356 13,996 177 703 503 346 1,635 4,599 2,596 7,153 4,082 5,984 21,255 62,688 117 444 391 207 995 2,726 1,898 4,495 2,919 1,881 13,452 40,012 employment. Altogether, establishments relocating out of Silicon Valley outnumbered those relocating in by 1,873; those moving out offered 22,676 more jobs than those moving in. The service sector, the largest sector in the Silicon Valley economy, lost the most establishments (640) and jobs (7,803). The finance, manufacturing, construction, and wholesale sectors each lost more than 2,000 jobs. Trans-State Relocation From the state of California’s point of view, the spillover of economic activities from Silicon Valley to other Bay Area cities could be a welcome trend. However, establishments relocating out of the state may be a cause for concern. In this section, we more closely examine Silicon Valley establishments that relocated to or from other states. Tables 4.8 and 4.9 replicate Tables 4.6 and 4.7 for establishments moving between Silicon Valley and outside California. We see similar patterns on a smaller scale in both the high-tech sector and the overall economy: Silicon Valley saw more businesses relocating out than relocating in. This is true even if we separate the high-tech sector into 60 Table 4.8 High-Tech Establishments Moving Between Silicon Valley and Outside California, by Industry, 1990–2001 Bioscience Computers/communications Defense/aerospace Environmental Semiconductor Software Professional services Innovation services Total Moving Out No. of No. of Establishments Employees 27 844 48 4,458 3 337 14 16 880 88 3,925 83 2,208 98 1,328 364 13,984 Moving In No. of No. of Establishments Employees 12 486 31 3,385 00 00 11 750 53 1,614 39 264 63 1,047 209 7,546 Table 4.9 All Establishments Moving Between Silicon Valley and Outside California, by Industry Group, 1990–2001 Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation, communication, and utilities Wholesale trade Retail trade Finance, insurance, and real estate Services Total Moving Out Moving In No. of No. of No. of No. of Establishments Employees Establishments Employees 49 12 43 153 209 8,999 2 0 18 88 5 0 57 6,355 32 680 113 1,530 76 476 45 2,919 321 7,766 844 22,534 18 71 29 18 154 398 189 758 163 156 2,758 10,441 different industries or separate the overall economy into industry groups. It is really striking that the pattern is always the same in every subindustry (or industry group), whether we measure net flow of establishments or of employment. There is a clear pattern that Silicon Valley has been losing enterprises to other states over the past decade. 61 In Table 4.10, we calculate the percentage of employees moving across the state border as they relocated into or out of Silicon Valley. Of all the employees that moved, 32.1 percent did not relocate within California. The high-tech sector saw a higher percentage of interstate movement (45.2 percent) than the overall economy (32.1 percent). Except for the environmental industry, a minor industry in Silicon Valley, every high-tech industry experienced at least 32 percent employee movement one way or the other across the state border. The computers/communications industry tops the list with nearly 70 percent interstate relocation. Defense/aerospace and software also stand out with 54.7 percent and 45 percent, respectively. If we look at the overall economy with both high-tech and nontech sectors, the manufacturing sector has the highest percentage of movement between Silicon Valley and outside California (48.3 percent). At 39.1 percent, the finance sector is the only other sector with above-average trans-state movement. We also calculate the average age of establishments moving between Silicon Valley and other states. Figure 4.1 shows that in both high-tech and nontech sectors, a higher proportion of establishments that moved out were founded before 1990 compared to those that moved into Silicon Valley from other states. Figure 4.2 compares the average age of Table 4.10 Trans-State Relocation as a Percentage of Total Employment That Moved Into or Out of Silicon Valley, 1990–2001 Industry % Industry Group Computers/communications 68.3 Manufacturing Defense/aerospace 54.7 Finance, insurance, and real estate Software 45.0 Services Semiconductor 38.1 Wholesale trade Bioscience 36.3 Transportation, communication, Innovation services 32.5 and utilities Professional services 32.0 Retail trade Environmental 1.3 Construction Overall 45.2 Mining Agriculture, forestry, and fishing Overall % 48.3 39.1 30.3 19.6 19.3 9.1 4.6 2.7 2.6 32.1 62 Percentage Moving out to other states 80 Moving in from other states 73.5 70 67.1 60 58.7 50 41.0 40 30 20 10 0 High-tech Nontech Figure 4.1—Percentage of Moving Establishments Founded Before 1990 5 Moving out to other states Moving in from other states 4.03 4 3.77 3.61 4 3.08 3 3 2 2 1 1 0 High-tech Nontech Figure 4.2—Average Age of Establishments Moving Between Silicon Valley and Other States Average age (years) 63 moving establishments founded during 1990–2000. The out-moving establishments tend to be older in both high-tech and nontech sectors, although the differences are not large. Figure 4.3 tracks the number of jobs eliminated by establishments moving out of California and the number created by those moving to Silicon Valley from other states from 1991 to 2000. The moving activities in both directions seem to have accelerated since 1996, probably because of the Internet boom and the resulting “digital rush” during the late 1990s. The high-tech sector saw more jobs move into Silicon Valley than moved out only in 1996. Nontech sectors had a net inflow of jobs only in 1994. Overall, only the year 1996 saw a net inflow of total jobs. We have seen a clear pattern in firm relocation between Silicon Valley and outside California. High-tech establishments, if they move, are more likely than nontech establishments to move to or from other states. In both high-tech and nontech sectors, more establishments moved out than moved in, whether measured by total number of establishments or total employment. Out-moving establishments are older. All these facts are consistent with our intuition that Silicon Valley Number of jobs 4,500 4,000 3,500 3,000 High-tech jobs moving out High-tech jobs moving in Total jobs moving out Total jobs moving in 2,500 2,000 1,500 1,000 500 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Figure 4.3—Job Movement Between Silicon Valley and Other States, 1991–2000 64 is attractive primarily to high-tech firms, and especially high-tech startups, and is losing mature businesses. A further question is whether the trend we observe is serious enough to worry policymakers and the business community. Mobility vs. Vitality Each year, new firms are created, offering many jobs; at the same time, some existing firms are closed and their employees are laid off. Some firms leave Silicon Valley, taking jobs away; others move into the region, bringing new employment opportunities with them. Tables 4.11 and 4.12 describe these dynamics in Silicon Valley’s job market, in the high-tech sector and in the whole economy, respectively. The tables show that firm relocation has a much smaller effect on the labor market than firm birth and death. We use two indexes to measure the dynamics in the labor market in Silicon Valley: Rate of vitality = (jobs created by new establishments + jobs lost by dead establishments)/total employment; Rate of mobility = (jobs offered by in-moving establishments + jobs taken away by out-moving establishments)/total employment. Figures 4.4 and 4.5 present the rate of vitality, the rate of mobility, and the rate of interstate mobility in the Silicon Valley labor market. On average, the rate of vitality is 14.2 percent in the high-tech sector and 13.3 percent in the whole economy. The rate of mobility is only 0.8 percent in the high-tech sector and 0.7 percent in the overall economy. Compared to firm birth and death, establishment relocation has an almost negligible effect on the labor market. On average, new establishments offer 6.4 percent of Silicon Valley’s high-tech jobs, and dead establishments eliminate 7.8 percent of them. The growth of existing establishments could make up the difference. At the same time, establishments that relocate out of Silicon Valley take away only 0.43 percent of its high-tech jobs, and establishments that moved into the valley offer 0.35 percent of the total high-tech jobs. If we consider the 65 Table 4.11 Employment in the High-Tech Sector of Silicon Valley, 1991–2000 66 Year 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Total Employment 523,496 512,536 545,575 524,722 539,885 565,560 588,031 605,607 606,731 650,331 Total No. of Employees in New Firms 16,487 28,257 60,895 22,154 34,910 54,311 40,672 32,147 25,584 48,941 Total No. of Employees in Dead Firms 36,997 32,597 47,182 20,984 37,649 51,585 38,113 56,006 46,559 72,816 Total No. of Employees Moving Out of Silicon Valley 1,796 1,375 1,607 1,055 1,532 1,411 1,992 4,573 3,059 5,706 Total No. of Employees Moving Into Silicon Valley 639 3,970 566 2,363 856 1,963 2,440 1,691 2,251 2,987 Total No. of Employees Moving Out of California 764 572 882 516 774 296 790 2,844 1,607 3,253 Total No. of Employees Moving Into California 117 143 120 160 217 1,400 254 808 1,524 1,088 Table 4.12 Employment in Silicon Valley, 1991–2000 Year 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Total Employment 1,346,351 1,312,856 1,407,188 1,380,399 1,374,146 1,395,816 1,435,747 1,455,318 1,445,922 1,511,192 Total No. of Employees in New Firms 35,156 54,468 177,262 54,990 75,860 102,531 106,764 75,548 68,350 107,549 Total No. of Employees in Dead Firms 86,673 72,909 87,457 90,645 85,555 119,410 98,418 122,619 106,316 140,620 Total No. of Employees Moving Out of Silicon Valley 5,745 3,656 4,346 2,837 5,277 3,335 5,142 8,711 6,763 9,400 Total No. of Employees Moving Into Silicon Valley 2,710 5,814 2,840 4,063 2,445 3,164 4,889 4,032 3,915 5,373 Total No. of Employees Moving Out of California 1,191 1,075 1,564 933 2,489 764 1,197 3,586 2,123 4,155 Total No. of Employees Moving Into California 163 242 192 782 341 1,482 371 1,109 1,761 2,265 67 Rate (%) 25 Rate of vitality Rate of mobility Rate of interstate mobility 20 15 10 5 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Figure 4.4—Dynamics in Silicon Valley’s High-Tech Labor Market, 1991–2000 20 15 10 Rate of vitality Rate of mobility 5 Rate of interstate mobility 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Figure 4.5—Dynamics in Silicon Valley’s Labor Market, 1991–2000 nontech sectors as well, the overall Silicon Valley economy sees a little less birth, death, and relocating activities, but the relative effect of vitality and mobility is similar. Note that in calculating the rate of vitality, we included all new establishments (i.e., some are new firms and others are new offices or 68 Rate (%) plants set up by existing firms). If we focus only on jobs created by new firms, they account for 2.6 percent of the high-tech employment and 2.1 percent of the overall employment. Again, jobs created by new firms alone are much more than enough to cover job losses by firm relocation. We should thus conclude that although Silicon Valley is losing jobs as a result of firm relocation, the magnitude of the loss is relatively small. In other words, once firms are founded in Silicon Valley, they are very likely to stay there. A vibrant economy in Silicon Valley with intensive entrepreneurial activities and rapid growth will generate job opportunities that are more than enough to compensate for any leakage of jobs. Relocating Out vs. Branching Out In the preceding sections, we have examined whether firms choose to relocate out of Silicon Valley and where they go if they do move out. A related question that also concerns us is whether start-ups in Silicon Valley tend to set up branches elsewhere as they mature. Although very few establishments move to other states, many firms may choose to headquarter in Silicon Valley but to locate production or distribution capacities in other states. In such cases, California does not fully benefit from the economic growth enabled by Silicon Valley. We must recognize that some companies have an absolute need to reach out to their customers and hence have to open offices nationally or even globally. For example, software companies usually earn a large proportion of their revenue from on-site services to their customers, and thus they need to operate at locations near clusters of customers. Consider Oracle, the world’s largest enterprise software company based in Silicon Valley. It has employees all over the world; in the United States alone, it has 78 offices in 76 cities in over 35 states. Another company based in Silicon Valley, Siebel Systems, which specializes in e-business application software, has employees in 31 foreign countries. At the young age of ten, Siebel has already opened 57 offices in 28 states. It makes sense for such firms to expand beyond California. However, not all high-tech firms need to be physically close to customers. An example may shed some light on this issue. Table 4.13 shows that Intel, probably one of the most famous companies in Silicon Valley, in fact hires many more people outside California. Although it has 7,500 69 Table 4.13 Intel Operating Locations in the United States City Santa Clara (headquarters) Hillsboro Chandler Folsom Rio Rancho Hudson Dupont Colorado Springs Parsippany Riverton Austin San Diego Shrewsbury Thousand Oaks Los Angeles Columbia San Luis Obispo Chantilly Irvine San Jose Raleigh State California Oregon Arizona California New Mexico Massachusetts Washington Colorado New Jersey Utah Texas California Massachusetts California California South Carolina California Virginia California California North Carolina No. of Employees 7,500 15,000 10,000 7,300 5,500 2,700 1,500 1,000 900 625 550 400 400 300 250 150 145 140 130 100 70 SOURCE: http://www.intel.com/jobs/usa/sites/index.htm. employees in its Santa Clara headquarters, Intel’s campus in Hillsboro, Oregon, is twice as large. With 15,000 employees in its Hillsboro branch, Intel is the largest private employer in Oregon. Its second-largest location is in Chandler, Arizona, which offers 10,000 jobs. Intel’s campus in Rio Rancho, New Mexico, has 5,500 employees, which makes it the largest private industrial employer in the Albuquerque metropolitan area. A complete NETS dataset would allow us to measure more precisely how many establishments Silicon Valley firms manage outside California. Unfortunately, a dataset for the whole nation is not ready yet. What we can do is to choose some large firms in Silicon Valley, calculate their employment in the Bay Area, and compare those numbers with total 70 employment available for the Duns Business Rankings. According to our calculation, the top 40 firms (by sales) in Silicon Valley together have 32 percent of their total employment in the Bay Area. This ranges from Novellus Systems’ nearly 100 percent to 3Com’s mere 4 percent. As a successful start-up becomes a mature company, it will develop new needs that require different cost-benefit considerations, or it may be that the company simply cannot conduct its business successfully unless it branches out to other locations. Although its headquarters is likely to remain in Silicon Valley, it will look elsewhere to accommodate its growth. A famous example is the hard disk drive industry that was born in Silicon Valley but later had to move manufacturing operations to Southeast Asia to maintain its competitiveness (McKendrick, Doner, and Haggard, 2000). A serious question for the state of California is how, if possible, to keep spillovers from Silicon Valley within California. This is a particularly relevant question for the fast-growing biotech industry in Silicon Valley and the San Francisco Bay Area as a whole. Conclusion Silicon Valley firms do move. In general, more establishments leave the area than move into the area. High-tech establishments are more likely than nontech establishments to move, both into and out of the valley. Establishments moving to Silicon Valley tend to be younger than those moving out. All these findings are consistent with our intuition. Although more establishments relocating out implies that Silicon Valley is losing businesses and job opportunities, it is not a serious problem. On the one hand, establishments moving out tend to go to adjacent cities within the state; on the other hand, new firms created each year overwhelmingly outnumber those moving away, which is more than enough to compensate for the net loss resulting from firm relocation. Thus, instead of worrying about what we might do to keep the businesses, we should focus our attention on how to create new businesses and facilitate their growth. Although most firms founded in Silicon Valley will remain in the region, the most successful ones among them will almost surely set up 71 branches elsewhere for operations such as manufacturing that do not benefit much from the Silicon Valley environment. This creates the possibility for the rest of California to accommodate the branching-out of Silicon Valley’s successful firms. 72 5. Conclusion Silicon Valley’s high-tech sector consists of the most dynamic industries in the economy. These industries have unique features and call for careful analysis. The high-tech economy is driven by innovation, and radical changes usually originate from innovative entrepreneurs starting new firms. For these reasons, we have studied high-tech startups and industry dynamics in Silicon Valley with the intention of discovering how Silicon Valley changed in the past and the lessons we should learn for the future. Major Findings New firms are important for Silicon Valley. As with other hightech centers, Silicon Valley hosts a wide variety of firms. A multitude of small firms coexist with medium and large firms. Each year, many new firms are founded, which collectively are a major driver of the economic dynamics in Silicon Valley. In fact, firms founded after 1990 created almost all of the job growth during 1990–2001. Young start-ups in Silicon Valley consistently attract a large amount of venture capital, which indicates that these firms are very innovative and growth-oriented. Successful start-ups have remade and will continue to remake Silicon Valley. Start-ups in Silicon Valley have quick access to venture capital. On average, it takes 11.6 months for Silicon Valley’s start-ups to complete their first round of venture finance—five months faster than the national average. The quicker access to capital is found in every major industry in Silicon Valley. This gives start-ups in the region a head start, an important advantage in high-tech industries that advance at a very rapid pace. This large first-mover’s advantage implies that start-ups in Silicon Valley will have a better chance to survive, all else equal. Established firms in Silicon Valley spin off more start-ups. Compared to their counterparts in the Boston area, big companies in 73 Silicon Valley have more previous employees who start their own venture-backed businesses. Since engineers in successful firms are in the best position to grasp and commercialize cutting-edge innovations, a high rate of spin-offs helps open new markets and creates new jobs. Previous research discusses Silicon Valley’s high incidence of firm-level spin-offs based on anecdotal evidence and has identified cultural and legal factors to account for it. Although it remains unclear which theory is closer to the truth, for the first time we have confirmed with empirical data that there are indeed more firm-level spin-offs in Silicon Valley than in other high-tech centers. Firm relocation is not a serious problem. High-tech start-ups value the hotbed of innovation because that is where new ideas emerge and entrepreneurs cluster. Silicon Valley is a perfect environment for startups whose major objective is to develop innovative ideas. On the other hand, when firms become mature and enter the phase of mass production or routine services, their major concern becomes sustainability and they naturally care about operating costs. For those firms or, rather, for certain operations of those firms, Silicon Valley is unattractive. We have investigated whether firms leave Silicon Valley when they have evolved out of the start-up stage. We find that indeed more establishments move out of Silicon Valley than move in, and establishments moving out tend to be older. Establishments tend to stay close to Silicon Valley when they move out. In terms of those moving across state borders, Silicon Valley does see a net job loss, because more jobs are relocated to other states than are relocated to Silicon Valley from outside California. However, the data suggest that firm relocation involves a relatively small proportion of the labor force. Firm birth and death cause much more turbulence than firm relocation. In other words, once firms are established in Silicon Valley, they are very likely to remain there, and intensive entrepreneurial activities certainly compensate for the jobs lost through firm relocation. Successful firms in the valley are branching out. Although relocation does not occur at significant levels, established firms in Silicon Valley frequently set up branches elsewhere. For many large high-tech companies headquartered in Silicon Valley, their employment within Silicon Valley itself is only a small proportion of their total employment. 74 Since Silicon Valley is already tightly packed with thousands of firms, fast-growing start-ups are more likely to expand outside the immediate area. As firms expand, they could benefit the rest of California by setting up branches elsewhere in the state. The high-tech sector experiences rapid structural changes. The high-tech sector consists of a number of diverse industries, which follow different dynamics. On the one hand, the fluctuation of the macro economy has distinctive effects on different high-tech industries; on the other hand, technological innovations in different industries—the drivers of growth in those industries—do not arrive simultaneously. As a result, different high-tech industries may follow unsynchronized business cycles. And thus, at different points in time, the “hot spot” of growth may appear in different industries. For example, the 1990s saw a boom in the computer industry, along with a decline in the defense industry. To catch upturns and avoid downturns in high-tech industries, a high-tech center such as Silicon Valley must accommodate rapid structural changes. This implies that a dynamic labor force is necessary. Previous research has emphasized the “high-velocity labor market” through which workers move frequently from one job to another within Silicon Valley. Such a labor market certainly helps the region’s economy adapt to structural changes. In addition, we believe, a set of infrastructure and institutions that enable the labor force to move quickly into and out of Silicon Valley is also crucial for structural changes in the high-tech sector. For example, employment in the software industry in Silicon Valley increased from 48,500 to 114,600 between 1990 and 2001, a phenomenal 136 percent rate of growth. It is impossible to train such a large number of technical workers within such a short period of time. This kind of rapid growth in a certain industry is achievable only through massive migration of the needed labor force. Policy Implications State and local governments played only a minor role in the early years of Silicon Valley. The history of Silicon Valley evolved from a tradition of innovative thinking in the region and industry-university networks such as that between the business world and Stanford University. Government’s largest effect on Silicon Valley’s high-tech 75 sector was probably the purchase of defense products by the federal government during the Cold War era. State and local governments were not actively involved in the region. Yet outside Silicon Valley, the recent trend shows that state and local governments can lend an effective helping hand to a regional high-tech economy. From Seattle and Portland to Austin and Denver, state and local governments all have supportive policies for the local high-tech sector. Governments play even bigger roles in the Silicon Valley clones in the rest of the world, such as Cambridge, England; Helsinki, Finland; Tel Aviv, Israel; Bangalore, India; and Hsinchu, Taiwan (Rosenberg, 2002). To maintain Silicon Valley’s success is by no means an easier task than building a Silicon Valley clone. Silicon Valley today faces more competition than ever from high-tech regional economies both domestically and internationally. Supportive policies have been implemented in metro areas all over the country that aim at grabbing a bigger piece of the high-tech economy. In addition, Silicon Valley’s success today could become its burden tomorrow when innovations again call for changes. How to keep Silicon Valley growing is a big challenge for California’s policymakers. This is especially true today, with the valley struggling through a deep recession. Policies directly related to Silicon Valley include the federal government’s spending on R&D and military goods and its immigration policies, state government’s R&D spending and education policy, and local governments’ land use policies, and so on. In addition, in any other areas where the private sector has no incentive or capability to solve the problems, government must step in. Examples include building infrastructure, training labor, and preventing further energy crises. Several policy implications have emerged from our examination of high-tech start-ups and industry dynamics in Silicon Valley. Promote technological innovation. More than any other sector, the high-tech economy is about innovation and entrepreneurship. Waves of innovation cause business cycles. Silicon Valley has experienced highs and lows many times, and right now the region is struggling in a deep trough. Previous experience proves that Silicon Valley always gets out of a recession on two legs: One is strong demand for high-tech products 76 from the whole economy, and the other is new demand created by innovations that add a new dimension to Silicon Valley’s economy. Although state and local governments can do little to improve the macroeconomic environment of the national economy, they could help promote innovation. University research has always been a major source of innovation, and state government should continue its strong support to research universities. Big budget cuts for the University of California system will severely affect the prospect of the high-tech sector off campus. Moreover, the California delegation in Washington, D.C., should place a high priority on securing R&D dollars for California from the federal government. As the state economy becomes more and more reliant on high-tech industries, support for R&D and innovation not only helps Silicon Valley and the rest of the Bay Area, but it also greatly benefits the Los Angeles and San Diego areas, which are continuing to expand their own high-tech sectors. Encourage firm founding. Our findings show that although some firms do move out of Silicon Valley, it is not a serious problem. On the one hand, they are likely to move to nearby cities and stay within the state; and on the other hand, firm formation and growth create new jobs that overwhelmingly outnumber jobs lost through firm relocation. Job creation in Silicon Valley is primarily achieved by new firms. Thus, instead of worrying about losing businesses because of the high cost of living and doing business in Silicon Valley, state and local governments should encourage firm founding. Offering favorable tax breaks, opening industrial parks, building high-tech incubators, and providing seed capital for commercialization of research are widely used policy levers. Previous research has shown that a primary factor determining a hightech start-up’s location is where its founder would like to live (Cooper and Folta, 2000). Thus, continuously improving the quality of life in Silicon Valley and the Bay Area as a whole is crucial for the vitality of the high-tech economy in this area. Look beyond Silicon Valley. The high-tech sector is not a disconnected economy, nor is Silicon Valley an isolated region. Silicon Valley is well embedded in the San Francisco Bay Area and well connected to the rest of the state economy. Most of the firms relocating out of Silicon Valley migrate to nearby cities in the Bay Area. The rest of 77 the Bay Area has undoubtedly benefited from the proximity of Silicon Valley and has quite a strong high-tech economy. Our data show that entrepreneurial activities in the 1990s were intensive in the whole Bay Area, both inside and outside Silicon Valley. Venture capital investment is also abundant for the rest of the Bay Area. State policies regarding Silicon Valley should take into account Silicon Valley’s connection with the rest of the state economy. For example, many people who work in Silicon Valley live a considerable distance from it, seeking more affordable homes. Thus, housing development and transportation policies in many other Bay Area cities help to solve Silicon Valley’s housing problems. We have also found that large firms in Silicon Valley often hire only a small proportion of their total employees from Silicon Valley or even the Bay Area. This suggests that other regions in the state have the opportunity to benefit from spillover from Silicon Valley by hosting branches of its firms. State government should try to understand not only new firm formation but also the concerns of mature firms in Silicon Valley. In particular, state government could provide incentives for large firms to set up their manufacturing or distribution arms within the state. State government could also improve transportation networks between the Bay Area and the Central Valley that facilitate Silicon Valley’s branching out to the latter area. In addition, local governments in the rest of the Bay Area and in the Central Valley should be more proactive in accommodating businesses branching out from Silicon Valley. Maintain a dynamic labor pool. Two conflicting factors characterize the high-tech labor force. On the one hand, the high-tech sector primarily hires technical workers whose skills are highly specialized and take time to acquire; on the other hand, the high-tech sector is dynamic, with its core technologies evolving quickly. This implies that the skills acquired in school three years ago may be obsolete today. Moreover, certain high-tech industries often experience explosive growth, such as the software industry in the 1990s, which creates a high demand for certain types of technical workers within a short period. Whether Silicon Valley can evolve rapidly hinges upon whether its labor force can quickly upgrade its skills or meet completely new demands. State government should continue to support universities and colleges as 78 vehicles for continuously retooling the labor force. Employers in Silicon Valley should recruit new talent not only through local colleges and universities but also by recruiting and hiring highly qualified immigrants. Immigrants have played an important role in Silicon Valley’s growth. They are a major source of Silicon Valley’s entrepreneurs and innovation. Immigrants also provide a large reserve of high-quality engineers and scientists capable of satisfying sudden surges of demand for certain talents in some industries. State government in cooperation with federal authorities should keep the door open to international talent, both at local universities and in the high-tech industries. This has emerged as a particularly crucial issue because immigration policies have now entered the equation of homeland security. 79 Appendix A Geographic and Industrial Definitions Geographic Definition of Silicon Valley Our definition of Silicon Valley includes all of Santa Clara County and adjacent cities in Alameda, San Mateo, and Santa Cruz Counties. City Santa Clara County All Alameda County Fremont Newark Union City San Mateo County Atherton Belmont East Palo Alto Foster City Menlo Park Redwood City San Carlos San Mateo Santa Cruz County Scotts Valley Zip Code All 94536–39, 94555 94560 94587 94027 94002 94303 94404 94025 94061–65 94070 94400–03 95066–67 Definition of Industry Groups in the NETS Data Used in This Study Industries are listed by their SIC code; “n.e.c.” means not elsewhere classified. 81 Industry Group Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation, communication, electric, gas, and sanitary services Wholesale trade Retail trade Finance, insurance, and real estate Services Government SIC Code 01–09 10–14 15–17 20–39 40–49 50–51 52–59 60–67 70–89 91–97 Definition of High-Tech Industries in the NETS Data Industry Bioscience Drugs Laboratory apparatus and analytical, optical, measuring, and controlling instruments Surgical medical and dental instruments and supplies Medical laboratories Computers/communications Electronic computers Computer storage devices Computer terminals Computer peripheral equipment, n.e.c. Telephone and telegraph apparatus Radio and television broadcasting and communications equipment Communications equipment, n.e.c. Printed circuit boards SIC Code 283 3821, 3823–24, 3827, 3829 384 8071 3571 3572 3575 3577 3661 3663 3669 3672 82 Electronic components, n.e.c. Magnetic and optical recording media Defense/aerospace Small arms, ammunition Electron tubes Aircraft and parts Guided missiles and space vehicles Tanks and tank components Search, detection, navigation, guidance, aeronautical, and nautical systems instruments and equipment Environmental Industrial and commercial fans and blowers and air purification equipment Service industry machinery, n.e.c. Sanitary services Scrap and waste materials Semiconductors Special industry machinery Semiconductors and related devices Instruments for measuring and testing electricity and electrical signals Software Computer programming services Prepackaged software Computer integrated systems design Computer processing and data preparation and processing services Information retrieval services Innovation services Wholesale of computers and computer peripheral equipment and software Wholesale of electronics parts and equipment, n.e.c. Computer facilities management services Computer rental and leasing 3679 3695 348 3671 372 376 3795 381 3564 3589 495 5093 3559 3674 3825 7371 7372 7373 7374 7375 5045 5065 7376 7377 83 Computer maintenance and repair Computer-related services, n.e.c. Engineering services Research and testing services Professional services Commercial printing Manifold business forms Service industries for the printing trade Investors, n.e.c. Advertising Consumer credit reporting agencies Mailing, reproduction, commercial art and photography, and stenographic services Personal supply services Legal services Architectural services Surveying services Accounting, auditing, and bookkeeping services Management and public relations services 7378 7379 8711 873 275 276 279 6799 731 732 733 736 81 8712 8713 872 874 84 Appendix B The Data Here we give a detailed discussion of the two longitudinal databases we used. The NETS Data The NETS database was constructed by Walls & Associates, who derived the raw data from Dun & Bradstreet (D&B). D&B, which has been collecting business data for more than 160 years, offers business-tobusiness credit information on companies throughout the world. The D&B data include information on the location, industry category, ownership, and employment of almost all businesses in the United States. Although the goal of D&B is not to collect and organize data for scholarly research, it does have an incentive to ensure the accuracy of its data. Serious inaccuracies could hurt D&B’s business and might even result in lawsuits. D&B has thus established a complicated quality control system, which has resulted in a relatively accurate and reliable database. However, D&B data are by no means without limitations. The main source of bias comes from its criterion of inclusion. Only firms that seek credit ratings or whose credit ratings are demanded by business partners have an incentive to report their activities to D&B. D&B has no information about businesses that do not report to them. Early evidence suggests that D&B data tend to overrepresent the manufacturing sector and new firms may not be completely covered or not included in their early years of existence. Nonetheless, with all their shortcomings, D&B data are one of the most widely consulted sources of information for academic research, mainly because firm-level data are always hard to acquire and D&B data are conveniently available, cover nearly the whole economy, and are of reasonably good quality. Many previous studies on industry dynamics such as Birch (1987) and Audretsch (1995) have used refined D&B data. 85 Walls & Associates teamed up with D&B to convert their archival establishment data into a time series: the NETS database. Walls & Associates first used D&B’s Duns Marketing Information file, which followed more than 22 million establishments from 1990 to 2001, to determine which establishments were active in January of each year. Then they retrieved information about each establishment from other D&B files (e.g., the credit rating file) to create a time series with rich firm-level information. In the NETS database, the basic unit of observation is the “establishment.” An establishment is a business or industrial unit at a single physical location that produces or distributes goods or provides services. For example, a single store or factory is an establishment. Many companies own or control more than one establishment, and those establishments may be located in different geographic areas and may be engaged in different kinds of business. D&B assigns a unique nine-digit DUNS (Data Universal Numbering System) number to each establishment. D&B also links the DUNS numbers of parent companies, headquarters, subsidiaries, and branches to form corporate family structures. The NETS database has all such information included, so that we are able to tell whether a new establishment is a start-up company or a newly established branch of an existing company. Specifically for the purpose of our study, Walls & Associates cut a PPIC extract from their NETS database. This dataset covers all the establishments that were ever located in 15 counties during 1990–2001. The 15 target counties include: ten counties in the San Francisco Bay Area (Alameda, Contra Costa, Marin, Napa, San Francisco, San Mateo, Santa Clara, Santa Cruz, Solano, and Sonoma),1 two counties in the Boston area (Middlesex and Suffolk), and three counties in the Washington, D.C., area (Arlington, Virginia; Fairfax, Virginia; and Montgomery, Maryland). The data for the ten counties around San Francisco Bay completely cover Silicon Valley as we defined it and, in addition, allow us to look at a bigger picture beyond the valley. The data ____________ 1For most of our analysis, we do not use all the data from the ten counties in the Bay Area. As defined in Appendix A, Silicon Valley covers only Santa Clara County and some adjacent cities in Alameda, San Mateo, and Santa Cruz Counties. 86 for the other two high-tech centers, Boston and Washington, D.C., enable us to make comparisons. One of our tasks is to measure start-up growth over time. Firm growth is usually measured by employment or sales. D&B does collect data on firm sales. However, for various reasons, a very small proportion of firms choose to report their sales numbers. Firms are more likely to report their employment. For example, our dataset has self-reported employment for 84 percent of Silicon Valley’s high-tech firms active in 2001. Small firms are more likely to have missing data. For example, if we exclude the Silicon Valley high-tech firms that never hired more than five people, 97 percent of the employment data will be self-reported. Fortunately, an establishment’s employment data are usually not missing for all years. If a data point is missing for a year between two selfreported data points, then D&B or Walls & Associates will fill it in according to simple smoothing formulae. If there is a missing data point before or after a series of self-reported data points, it is filled in by extrapolation. In rare cases where the employment data are completely missing, D&B or Walls & Associates will enter their estimates based on industry average. To test the reliability of the NETS data, we compared business size distribution in NETS with that available at the Economic Development Department (EDD) of California. The EDD dataset also counts an establishment as a business unit, which makes it comparable with NETS. It defines the employment at an establishment as “insured wage earners on the payroll.” Any employer hiring one or more persons, who pays wages in excess of $100 during a calendar quarter, and who is not engaged in an exempt activity, is subject to the Unemployment Insurance provision of the California Unemployment Insurance code. Table B.1 presents business size distribution in four counties that cover Silicon Valley: Alameda, San Mateo, Santa Clara, and Santa Cruz. Although the NETS data cover self-employed people, we have excluded them from our calculation because the EDD data do not include them. The EDD dataset is a snapshot as of September 2001. We use the NETS data collected in January 2001. As Table B.1 shows, in Alameda and Santa Clara Counties, the NETS covers more establishments than the EDD data in every size 87 Table B.1 Business Size Distribution in NETS and EDD Data, 2001 0–4 Alameda EDD 28,105 NETS 38,201 San Mateo EDD 14,061 NETS 21,277 Santa Clara EDD 27,949 NETS 48,614 Santa Cruz EDD 4,561 NETS 9,487 5–9 6,229 8,507 3,412 4,559 7,794 10,053 1,347 1,879 10–19 4,598 5,298 2,413 2,784 5,569 6,344 909 975 20–49 3,579 4,003 1,918 2,035 4,612 4,968 714 596 50–99 1,511 1,614 743 775 1,934 1,969 238 227 100– 249 788 874 399 425 1,119 1,189 121 104 250– 499 198 212 110 111 278 304 25 16 500– 999 79 88 44 31 120 161 10 8 1,000+ 43 59 22 37 76 119 3 8 NOTE: The EDD data are available at http://www.calmis.cahwnet.gov/file/indsize/ 1sfcoru.htm. category. In San Mateo County, the NETS is bigger than the EDD sample except in one category. In Santa Cruz County, the EDD picks up more firms than NETS in size categories bigger than 20, except for 1,000+. In every county, the NETS sample covers many more small firms that employ fewer than 20 people. The difference in the 0–4 category is most significant. For example, in Santa Cruz County, the NETS data include more than twice as many size 0–4 firms as the EDD data. A comparison of business size distribution for some other years yields similar results: The NETS data always capture far more small firms than the EDD data; although the difference becomes smaller for larger firms, the NETS is still likely to have more of them. A more complete coverage of the small firms is particularly valuable for studying start-ups. Table B.2 compares county-level employment series from the EDD data and those from the NETS. The NETS data consistently produce a larger employment figure. This is true in every year for every county. In some cases, the difference is very large. For example, in 1993 in Santa Clara County, the NETS data documented 30 percent more employees than the EDD data. To some extent, a larger employment number 88 Table B.2 Employment Series in NETS and EDD Data, 1990–2001 89 1990 Alameda 1991 1992 1993 1994 1995 1996 1997 1998 EDD — NETS 709,496 San Mateo —— 679,448 677,404 591,300 750,168 590,600 730,025 607,000 738,186 620,800 733,484 639,100 742,097 660,500 796,386 EDD 295,600 NETS 364,531 Santa Clara 298,100 291,500 351,646 344,535 294,200 365,465 296,300 365,310 305,800 355,550 319,100 361,851 333,300 370,309 345,100 380,917 EDD 819,500 810,900 797,200 802,000 805,000 836,400 885,000 931,700 961,500 NETS 1,055,389 1,017,015 988,208 1,054,477 1,028,791 1,032,777 1,048,374 1,070,466 1,079,035 Santa Cruz EDD 94,900 96,100 94,800 95,400 96,600 97,700 99,200 101,600 103,000 NETS 105,471 101,451 101,219 110,638 113,370 109,298 109,186 112,064 114,591 NOTE: The EDD data are available at http://www.calmis.cahwnet.gov/file/indsize/1sfcoru.htm. 1999 683,600 793,967 357,900 371,438 976,600 1,077,960 103,200 116,455 2000 71,100 802,609 375,800 383,188 1,035,000 1,131,221 105,600 117,184 2001 719,600 838,795 375,400 408,199 1,021,000 1,174,771 107,200 115,478 simply reflects the fact that the NETS data cover more firms, which could be a good feature of our data. However, this good feature is not cost-free. The NETS data contain a large number of very small firms. The data for those small firms tend to be noisy, which adds more noise to the NETS data. As we have mentioned, a firm chooses to be included in the D&B raw data when it needs a DUNS number. In certain circumstances, that need may suddenly become pertinent for many firms, and hence many existing firms that are not in the D&B database will jump in simultaneously. This kind of behavior is more common for small firms, which creates more noise in the NETS data. For example, we see a big surge in employment from 1992 to 1993 in the NETS data but not in the EDD data. In the 1992–1993 period, the California economy came out of a severe recession, and therefore an increase in employment was expected. But the 6–10 percent increase in the NETS data is too dramatic to be credible. We have attempted to discover possible reasons to explain the surge in 1993. As part of President Clinton’s mandate to streamline the procurement process through the use of electronic commerce, the federal government adopted the D&B DUNS number as a principal contractor identification code in 1993. This means that suppliers doing business with government agencies via Electronic Data Interchange would be required to submit their DUNS number as part of the registration and transaction processes. This might have pushed many existing firms into the D&B database. We see a nationwide surge in the number of business units in the 1993 edition of D&B’s business census. This is also reflected in our NETS data. The problem could be partly solved if every establishment reported its starting date as required, but a large proportion of small firms failed to do so. For this reason, we should use caution when interpreting economic trends in the NETS data. We have compared some of the NETS data and the EDD data at the county level. Our general conclusion is that the NETS data provide a more complete coverage of business enterprises and particularly of small firms. The drawback that comes with the more complete coverage is that it is subject to noise created by small firms. The above comparison reveals only some of the properties of the NETS data at the aggregate 90 level. At the firm level, the NETS data offer a very rich pool of information such as firm location, ownership, industry, employment, and the changes in such variables over time. This wealth of information is unparalleled by any other database. The VentureOne Data The second dataset is provided by VentureOne, a leading venture capital research company. VentureOne claims that it has “the most comprehensive database on venture-backed companies.” Our data cover venture capital deals completed from the first quarter of 1992 through the fourth quarter of 2001. They include 29,277 rounds of financing involving 11,029 firms. Among those firms, 83.53 percent were founded in or after 1990. The VentureOne data provide detailed information about all the venture-backed start-ups. Interesting firm-level variables include the start year, address, industry, employment, current business status, current ownership status, closing date of each round of financing, the amount of capital raised in each round, and so on. VentureOne categorizes venture-backed firms into 16 different “industry segments.” Table B.3 shows the amount of venture capital invested and the number of deals completed in each industry. An overwhelming majority of venture-backed start-ups should be classified as high-tech. Even in the retailing industry, most venture-backed firms qualify as high-tech because they are Internet-related. Only a tiny proportion of firms in our dataset do not fall into our definition of hightech, such as restaurants in the retailing industry. Since VentureOne does not use the SIC codes, we have no consistent way to exclude nontech firms from our analysis other than relying on subjective judgment. Thus, we decided to use the entire dataset. VentureOne also provided a separate dataset containing information about start-up founders. An “EntityID” variable allows us to match the firm data with the founder data. The biographical information of founders is available, including the previous working experiences of the founder. This enables us to do some elementary studies of entrepreneurs, such as what kind of people tend to found venture-backed start-ups. To do a preliminary reliability test of the VentureOne data, we compared them with the only alternative comprehensive venture capital 91 Table B.3 Real Venture Capital Investment in the United States, by Industry, 1992–2001 Industry Communications Software Consumer/business services Information services Biopharmaceutical Retailing Medical devices Semiconductor Electronics Healthcare Medical information services Consumer/business products Advance/special material and chemical Other Energy Agriculture Total aIn 1996 dollars. Venture Capital Raised ($ billions)a 72.926 57.058 52.830 26.436 21.845 14.617 13.579 11.627 11.343 7.902 7.347 5.554 1.395 1.337 1.116 0.516 307.426 % of Total Venture Capital 23.72 18.56 17.18 8.60 7.11 4.75 4.42 3.78 3.69 2.57 2.39 1.81 0.45 0.43 0.36 0.17 100 No. of Deals 3,893 7,142 5,025 2,522 2,140 1,062 1,885 1,154 1,476 932 915 579 200 199 76 77 29,277 database, the PricewaterhouseCoopers/Venture Economics/National Venture Capital Association MoneyTree Survey. The data from the MoneyTree Survey do have one advantage in that they cover a longer time period. However, our main purpose is to study industry dynamics through firm formation, growth, and mortality but not the trend of venture capital investment. So we need detailed information about venture-backed firms. By this criterion, the VentureOne data are more suitable for us. Table B.4 compares some aggregate statistics from the MoneyTree Survey and the VentureOne data. The VentureOne data show a higher sum of venture capital investment for every year except 2001. We acquired our data from VentureOne in late December 2001, when the fourth-quarter data were not completed yet. That may explain the deficit of the VentureOne data in 2001. In terms of companies covered, the 92 Table B.4 Venture Capital Investment by MoneyTree Survey and VentureOne Data MoneyTree Surveya VentureOne Year 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Sum Invested No. of Average per Sum Invested No. of Average per ($ millions) Companies Company ($ millions) Companies Company 3,827.56 1,054 3.63 9,230.75 1,126 8.20 4,565.53 945 4.83 10,220.66 1,162 8.80 3,792.89 954 3.98 8,043.74 1,230 6.54 5,693.46 1,265 4.50 13,389.43 1,536 8.72 11,386.77 1,809 6.29 21,313.05 2,105 10.12 14,823.33 2,385 6.22 20,474.79 2,329 8.79 19,843.17 2,821 7.03 24,752.63 2,568 9.64 54,499.93 4,202 12.97 67,480.78 4,027 16.76 102,308.33 5,608 18.24 112,214.10 5,483 20.47 37,672.50 3,224 11.69 32,524.21 2,933 11.09 aInformation is current as of February 20, 2002, and is available at http://www.nvca. org/. VentureOne data report more venture-backed companies from 1992 to 1996. Since then, the MoneyTree Survey has covered more companies. The discrepancies are quite small, although we have to recognize that a larger set of companies in one dataset does not necessarily encompass the smaller number of companies in the other dataset. The most significant disagreement between the two datasets is the average amount of money raised by each company. Except for 2001, the VentureOne data always produce a higher average. And the trend is the earlier the data, the bigger the difference. In 1992, the average venture capital per company in the VentureOne data is more than twice as much as in the MoneyTree Survey. Many possible reasons can explain the differences. For example, the definition of venture capital may not be identical. We notice that VentureOne actively tracks only venture capital firms that manage more than $20 million. This may bias the VentureOne data toward larger venture capital deals. Because most deals became very large in the late 1990s, this bias could have become smaller. Overall, it seems that there is not enough evidence to conclude that one dataset is better than the other. 93 Appendix C A Snapshot of the Silicon Valley Economy Using an extract from the NETS database, we assemble a collection of statistics here to describe the Silicon Valley economy in 2001. Table C.1 Total Number of Establishments and Employees in Silicon Valley, 2001 Total establishments Total employees High-Tech 25,787 672,825 Nontech 77,334 903,332 Total 103,121 1,576,157 Table C.2 High-Tech Establishment Category in Silicon Valley, 2001 Establishment Category Alive in 2001 % of total Headquarters 1,682 6.52 Branches 2,621 10.16 Stand-Alone 21,484 83.31 Total 25,787 100 Table C.3 Establishment Size Distribution in Silicon Valley, 2001 No. of Employees 0–4 5–9 10–19 20–50 51–100 101–250 251–500 501–1,000 1,001–2,500 2,500+ Total High-Tech 15,993 3,405 2,372 2,227 823 579 207 93 63 25 25,787 Nontech 51,924 10,800 6,556 5,402 1,598 739 184 83 34 14 77,334 95 Table C.4 Establishment Age Distribution in Silicon Valley, 2001 Establishment Yeara High-Tech Nontech Total 1989 or before 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Total 7,570 653 743 1,437 1,034 1,222 1,441 1,821 1,877 2,317 2,099 3,573 25,787 30,777 1,681 1,756 5,318 2,283 2,941 3,475 5,202 3,729 5,019 4,913 10,240 77,334 38,347 2,334 2,499 6,755 3,317 4,163 4,916 7,023 5,606 7,336 7,012 13,813 103,121 aThis refers to the variable “FirstYear,” which is a firm’s start year or, in case the start year is missing, the year when its data first entered the D&B database. Table C.5 Total Establishments in Silicon Valley, by Industry Group, 2001 Industry Group Agriculture, forestry, and fishing Mining Construction Manufacturing Transportation, communication, and utilities Wholesale trade Retail trade Finance, insurance, and real estate Services Government Total No. of Establishments 1,758 35 6,886 8,163 3,402 6,907 17,291 9,237 49,039 403 103,121 No. of Employees 12,496 315 55,795 459,388 71,326 85,153 181,026 85,048 580,742 44,868 1,576,157 96 Table C.6 Total High-Tech Establishments in Silicon Valley, by Industry, 2001 Industry Semiconductors Computers/communications Bioscience Defense/aerospace Environmental Software Innovation services Professional services Total No. of Establishments 816 1,127 847 94 244 4,505 6,257 11,897 25,787 No. of Employees 103,443 150,974 51,854 27,567 8,342 114,639 112,150 103,856 672,825 97 Bibliography Arthur, W. Brian, “‘Silicon Valley’ Locational Clusters: When Do Increasing Returns Imply Monopoly?” Mathematical Social Sciences, Vol. 19, 1990, pp. 235–251. Audretsch, David B., Innovation and Industry Evolution, The MIT Press, Cambridge, Massachusetts, 1995. Bahrami, Homa, and Stuart Evans, “Flexible Recycling and HighTechnology Entrepreneurship,” in Martin Kenney, ed., Understanding Silicon Valley: The Anatomy of an Entrepreneurial Region, Stanford University Press, Stanford, California, 2000. Birch, David L., Job Creation in America: How Our Smallest Companies Put the Most People to Work, Free Press, New York, 1987. Bower, Joseph L., and Clayton M. Christensen, “Disruptive Technologies: Catching the Wave,” Harvard Business Review, January–February 1995, pp. 43–53. Bygrave, William D., and Jeffry A. Timmons, Venture Capital at the Crossroads, Harvard Business School Press, Boston, Massachusetts, 1992. Carroll, Glenn R., and Michael T. Hannan, The Demography of Corporations and Industries, Princeton University Press, Princeton, New Jersey, 2000. Christensen, Clayton M., The Innovator’s Dilemma, Harvard Business School Press, Boston, Massachusetts, 1997. Cooper, A., and T. Folta, “Entrepreneurship and High-Technology Clusters,” in D. L. Sexton and H. Landstrom, eds., The Blackwell Handbook of Entrepreneurship, Blackwell Business, Malden, Massachusetts, 2000. 99 Cortright, Joseph, and Heike Mayer, “High Tech Specialization: A Comparison of High Technology Centers,” working paper, Center on Urban and Metropolitan Policy, the Brookings Institution, Washington, D.C., 2001. DeVol, Ross C., “America’s High-Tech Economy,” Milken Institute, Santa Monica, California, 1999. Foster, Richard N., Innovation: The Attacker’s Advantage, Summit Books, New York, 1986. Freiberger, Paul, and Michael Swaine, Fire in the Valley, 2nd edition, McGraw-Hill, New York, 2000. Gilson, Ronald G., “The Legal Infrastructure of High Technology Industrial Districts: Silicon Valley, Route 128, and Covenants Not to Compete,” New York University Law Review, Vol. 74, 1999, pp. 575– 629. Henton, Doug, “A Profile of the Valley’s Evolving Structure,” in ChongMoon Lee, William F. Miller, Marguerite Gong Hancock, and Henry S. Rowen, eds., The Silicon Valley Edge: A Habitat for Innovation and Entrepreneurship, Stanford University Press, Stanford, California, 2000. Henton, Doug, Kim Walesh, Liz Brown, and Chi Nguyen, Joint Venture’s 2003 Index of Silicon Valley, Joint Venture: Silicon Valley Network, San Jose, California, 2003. Hyde, Alan, “Silicon Valley’s High-Velocity Labor Market,” Journal of Applied Corporate Finance, Vol. 11, 1998, pp. 28–37. Kenney, Martin, and Richard Florida, “Venture Capital in Silicon Valley: Fueling New Firm Formation,” in Martin Kenney, ed., Understanding Silicon Valley: The Anatomy of an Entrepreneurial Region, Stanford University Press, Stanford, California, 2000. Klepper, Steven, “Employee Startups in High-Tech Industries,” Industrial and Corporate Change, Vol. 10, 2001, pp. 639–674. Krugman, Paul, Geography and Trade, The MIT Press, Cambridge, Massachusetts, 1991. 100 Lee, Chong-Moon, William F. Miller, Marguerite Gong Hancock, and Henry S. Rowen, eds., The Silicon Valley Edge: A Habitat for Innovation and Entrepreneurship, Stanford University Press, Stanford, California, 2000. McKendrick, David G., Richard E. Doner, and Stephan Haggard, From Silicon Valley to Singapore: Location and Competitive Advantage in the Hard Disk Drive Industry, Stanford University Press, Stanford, California, 2000. Paulson, Ed, Inside Cisco: The Real Story of Sustained M&A Growth, John Wiley & Sons, New York, 2001. Rosenberg, David, Cloning Silicon Valley, Pearson Education, New York, 2002. Saxenian, Annalee, Regional Advantage: Culture and Competition in Silicon Valley and Route 128, Harvard University Press, Cambridge, Massachusetts, 1994. Schumpeter, Joseph A., The Theory of Economic Development, Harvard University Press, Cambridge, Massachusetts, 1934. “Silicon Valley: How It Really Works,” BusinessWeek, August 18–25, 1997, pp. 64–147. “The VCs Don’t Want Your Money Anymore,” BusinessWeek, July 29, 2002, pp. 81–82. VentureOne Corporation, The Venture Capital Sourcebook, San Francisco, California, 2001. 101 About the Author JUNFU ZHANG Junfu Zhang specializes in evolutionary economics and agent-based computational economics. His research interests include racial segregation in housing and schools, entrepreneurship, and innovations in the high-tech industry. He has held the Graduate Fellowship at Johns Hopkins University and the Leo Model Research Fellowship at The Brookings Institution. He received a B.A. from Renmin University of China and an M.A. and Ph.D. in economics from Johns Hopkins University. 103 Related PPIC Publications Rethinking the California Business Climate Michael Dardia and Sherman Luk California’s Vested Interest in U.S. Trade Liberalization Initiatives Jon D. Haveman The Evolution of California Manufacturing Paul W. Rhode Local and Global Networks of Immigrant Professionals in Silicon Valley AnnaLee Saxenian Silicon Valley’s New Immigrant Entrepreneurs AnnaLee Saxenian Business Without Borders? The Globalization of the California Economy Howard J. Shatz PPIC publications may be ordered by phone or from our website (800) 232-5343 [mainland U.S.] (415) 291-4400 [Canada, Hawaii, overseas] www.ppic.org 105" ["post_date_gmt"]=> string(19) "2017-05-20 09:36:20" ["comment_status"]=> string(4) "open" ["ping_status"]=> string(6) "closed" ["post_password"]=> string(0) "" ["post_name"]=> string(8) "r_703jzr" ["to_ping"]=> string(0) "" ["pinged"]=> string(0) "" ["post_modified"]=> string(19) "2017-05-20 02:36:20" ["post_modified_gmt"]=> string(19) "2017-05-20 09:36:20" ["post_content_filtered"]=> string(0) "" ["guid"]=> string(50) "http://148.62.4.17/wp-content/uploads/R_703JZR.pdf" ["menu_order"]=> int(0) ["post_mime_type"]=> string(15) "application/pdf" ["comment_count"]=> string(1) "0" ["filter"]=> string(3) "raw" ["status"]=> string(7) "inherit" ["attachment_authors"]=> bool(false) }