Donate
Independent, objective, nonpartisan research

R 1299JPR

Authors

R 1299JPR

Tagged with:

Publication PDFs

Database

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

View live version

object(Timber\Post)#3742 (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(13) "R_1299JPR.pdf" ["wpmf_size"]=> string(6) "158203" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(83320) "Fiscal Rules and State Borrowing Costs: Evidence from California and Other States ¥¥¥ James M. Poterba Kim S. Rueben 1999 PUBLIC POLICY INSTITUTE OF CALIFORNIA Library of Congress Cataloging-in-Publication Data Poterba, James M. Fiscal rules and state borrowing costs : evidence from California and other states / James M. Poterba, Kim S. Rueben. p. cm. Includes bibliographical references. ISBN: 1-58213-019-1 1. State bondsÑCalifornia. 2. State bondsÑUnited StatesÑ States. 3. Debts, PublicÑCalifornia. 4. Debts, PublicÑUnited StatesÑStates. 5. Interest ratesÑCalifornia. 6. Interest ratesÑ United StatesÑStates. I. Rueben, Kim S. II. Title HG4948.C2 P67 1999 336.3'43273Ñdc21 99-051905 CIP 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. Copyright © 1999 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. iii Foreword A generation of California voters and legislators has fashioned a unique system for financing government services. In doing so, they have also spawned scores of unintended consequences. Proposition 13, for example, both altered the face of public finance and generated many unforeseen outcomes. Instead of echoing conventional wisdom, several PPIC reports have described those consequences objectively. But whether policy debate trades in clichŽs, calculations, or some mixture of the two, one point is irrefutableÑCalifornia has made extensive use of the initiative process to finance and constrain state government, and we are still gauging the effects of that development. In Fiscal Rules and State Borrowing Costs, James Poterba and Kim Rueben show how some of these unintended consequences can be measured. Marshalling evidence from municipal bond markets around the nation, the authors maintain that fiscal constraints on state government affect borrowing costs in ways that are predictable but rarely acknowledged. In particular, they show that borrowing costs are iv sensitive not only to overall measures of economic health, such as unemployment rates, but also to a stateÕs fiscal rules and budget forecasts. Although some of the findings may seem counterintuitive at first, the economic reasoning is sound. Revenue limits and supermajority requirements for new taxesÑthe latter a legacy of Proposition 13Ñtend to increase borrowing costs because they hamper the stateÕs perceived ability to pay its long-term debt. In contrast, expenditure limits such as Proposition 4 lower borrowing costs because they make it easier for the state to service its debt. The authors also find that states with unexpected deficits pay more for long-term financing, and they conclude the report by measuring the costs of inaccurate state budget forecasts. These findings have immediate implications for the fiscal limits movement in California. Although this movement rarely distinguishes between tax and expenditure limits, the bond market has scaled this distinction precisely. The authors calculate that a tax limit costs the state approximately $2 million more per $1 billion of debt than a spending limit. This is not a negligible cost, even in a state as rich as California, and it indicates once again that reinventing public finance brings with it more than a few unintended consequences. In this study, however, those consequences are estimated carefully rather than asserted tendentiously. David W. Lyon President and CEO Public Policy Institute of California v Summary The vast majority of CaliforniaÕs spending on capital infrastructure, such as highways, aqueducts, prisons, and related facilities, is financed with long-term debt. Recent projections of such spending over the next decade suggest that the state will need to borrow $25 billion to $50 billion during this period. The interest rate on CaliforniaÕs debt will be an important factor in determining the overall fiscal burden of this infrastructure spending. That rate, which differs over time and from state to state, is determined by a number of factors, some of which are matters of public policy. This report investigates interstate differences in borrowing rates on general obligation bonds. It begins by presenting new evidence on how state economic conditions affect borrowing costs. By analyzing yields on tax-exempt bonds issued by different states between 1973 and 1997, the study reveals a clear relationship between a stateÕs general economic health, as measured by unemployment rates, and state borrowing costs. A 1 percent increase in a stateÕs unemployment rate is associated with an vi increase of about 0.05 percent, or five basis points, in that stateÕs bond yields. (A basis point is one one-hundredth of 1 percent.) In California, for example, the stateÕs borrowing rate rose when the state economy was weak in the early 1990s and has declined for the last several years as the economy has improved. More important from a public policy perspective, perhaps, is the finding that statesÕ fiscal rules also play an important role in determining statesÕ borrowing costs. We present three major findings in this regard. First, states with strict fiscal rules on government spending or deficits have faced lower borrowing costs during the last two decades than those with looser fiscal rules. We calculate that a state with a strict anti-deficit fiscal constitution pays about nine basis points less to issue new debt interest. Second, the bond market reacts in different ways to revenue restrictions and expenditure limits. States with expenditure limits typically borrow at lower rates than other states, but those that restrict tax increases or require supermajorities to increase taxes face higher borrowing costs. States with binding revenue limitation laws are likely to face borrowing rates more than seventeen basis points higher than those in other states. This ratio translates into an extra $1,750 in interest payments per million dollars of debt issued. Limits on the ability of local governments to increase taxes or issue debt also seem to raise state borrowing costs. The empirical findings in this study suggest that CaliforniaÕs current fiscal rules raise the interest rate on state and possibly local debt. Since the passage of Proposition 13 in 1978, a legislative supermajorityÑin this case, a two-thirds vote in both housesÑhas been required to enact new taxes. Although the Gann amendment (Proposition 4) of 1979 does vii not restrict state revenues, it limits state and local expenditure growth. More recently, Proposition 218, which was adopted in 1996, provides new mechanisms for voters to restrict local revenues. Although states with supermajority provisions face higher borrowing costs than states without such requirements, explicit limits on state revenues and stronger anti-tax provisions than those in California have an even larger effect on borrowing costs. Thus, CaliforniaÕs tax limitation strategy has probably raised interest costs to the state but by less than other tax limitation strategies might have. Our figures indicate that if California had enacted a tax limitation law for state revenues rather than an expenditure limit in 1979, borrowing costs would have been about 20 basis points higher on average during the subsequent two decades. This difference amounts to about $2,000 extra in interest payments per million dollars of debt issued. This increase represents a small but not trivial fraction of the state budget. In 1997, for example, California had outstanding state debt of $43.5 billion. If the state had enacted a restrictive revenue limit rather than an expenditure limit, the extra interest on this debt would have been approximately $90 million. Our third major finding concerns unexpected state budget surpluses or deficits and how they affect borrowing costs. Not surprisingly, unexpected state budget deficits are correlated with upward revisions in state bond yields. It is more expensive for a state to issue new debt when it is experiencing budget difficulties, but bond markets are especially responsive to the fiscal health of states with large amounts of outstanding debt. This effect is particularly pronounced in California, where an unexpected $100 per capita increase in the stateÕs deficit has historically been associated with an increase of 14 basis points in borrowing costs. Thus, if California had undertaken new infrastructure projects in 1992, viii $1 billion of new debt would have cost the state $1.4 million more than it would today. Part of this responsiveness can be explained by differences in fiscal rules; bond yields rise less during periods of financial stress for states with tight anti-deficit rules or restrictive spending rules. In California, state fiscal rules can explain about two-thirds of this effect. This pattern indicates that accurate tax and expenditure forecasts are more important for California than for most other states, which municipal bond analysts follow less closely. As California begins a new round of infrastructure borrowing and spending, it is especially important to understand how the stateÕs fiscal rules affect the overall costs of repairs and construction. According to our findings, these fiscal rules affect the costs of state tax-exempt debt by the same amount as a significant shift in the unemployment rate. Millions of dollars 0 2.5 2.0 1.5 1.0 0.5If 3.5% rise in unemploymentIf state had revenue limit instead of Gann (expenditure) limitIf state had unexpected deficit of $100 per capita Figure S.1ÑCaliforniaÕs Predicted Additional Borrowing Cost per $1 Billion of New Debt ix CaliforniaÕs interest costs are also very sensitive to unexpected deficits. Our findings indicate that voters and legislators would do well to recognize the long-term effects on borrowing costs when considering tax limits, expenditure limits, or changes to the stateÕs deficit financing rules. In addition, the state should take into account the additional costs of issuing debt during times when economic conditions or other circumstances lead to unexpected deficits. xi Contents Foreword................ iii Summary........... v Figures........... xii Tables............ xv Acknowledgments........... xvi 1. INTRODUCTION......... 1 2. CURRENT FISCAL RULES ACROSS DIFFERENT STATES............ 5 3. STATE BORROWING COSTS IN THE LAST TWO DECADES............ 15 4. BORROWING COSTS, ECONOMIC CONDITIONS, AND FISCAL RULES............. 21 How Economic Conditions and Fiscal Rules Affect the Level of Borrowing Costs............ 22 Fiscal Shocks, Fiscal Rules, and the Bond Market Reaction to Fiscal News.............. 25 Interstate Differences in Bond Market Reactions to Fiscal News................. 32 5. CONCLUSIONS AND IMPLICATIONS..... 37 xii Appendix A. State Fiscal Institutions............. 41 B. Detailed Description of Regression Results.... 43 References............... 51 About the Authors............ 55 xiii Figures S.1. CaliforniaÕs Predicted Additional Borrowing Cost per $1 Billion of New Debt............. viii 1. Balanced Budget Requirements....... 7 2. States with Debt Restriction Provisions...... 8 3. Year Local Tax Limit Passed....... 13 4. Relative Bond Yield Required on California State Debt.. 17 5. Relative Bond Yield Required for Various StatesÕ 20 Year General Obligation Bonds......... 18 6. Unexpected State Deficits as a Percentage of General Revenue, 1988Ð1997............ 27 7. Unexpected State Deficits as a Percentage of General Revenue, Selected States, 1988Ð1997....... 29 xv Tables 1. History of State Spending Limit Adoptions....... 10 2. History of State Revenue Limit Adoptions...... 10 3. Supermajority Requirements and Other Constitutional Restrictions on Legislative Tax Power....... 12 4. Effect of Fiscal Institutions on State Borrowing Costs.. 24 5. Average State Deficit Shock, Selected States, 1988Ð 1997................... 28 6. Effect of State Fiscal Rules on the Bond MarketÕs Reaction to Unexpected Budget Deficits and Budget Surpluses................. 30 7. Interstate Variability in Effect of Fiscal News on Borrowing Costs.............. 33 B.1. Regression Estimates of How Fiscal Institutions Affect State Borrowing Costs, 1973Ð1995........ 47 B.2. Regression Estimates of How Fiscal Institutions Affect Bond Market Reaction to State Fiscal News, 1988 Ð 1997.................... 49 xvii Acknowledgments We would like to thank a number of individuals who have helped us with this research project. First, we would like to thank the Chubb Corporation and especially Thomas Swartz III for access to the Relative Value Study data on which this report is based. Second, we would like to thank those who helped us with the analysis. Elizabeth Berko provided early help with data assembly, and Daniel Frakes produced the maps in this report. Third, we are grateful to James Alt, Mark Baldassare, Michael Dardia, Robert Iman, Peter Schaafsma, and Michael Teitz for their thoughtful reviews of this study. JŸergen von Hagen, Anne Case, and the other participants in a 1997 conference at the University of BonnÕs Zentrum fŸr EuropŠische Integrationsforschung provided insightful comments on an earlier paper that is closely related to this study. Finally, we wish to thank Peter Richardson of PPIC, who provided invaluable editorial assistance in completing the monograph. The authors retain responsibility for any errors of fact or interpretation. 1 1. Introduction In his proposed 1999Ð2000 budget, Governor Gray Davis established the Commission on Building for the 21st Century and called attention to the stateÕs low level of infrastructure spending over the past decade. His remarks to the commission (State of California, 1999) underscored the growing importance of capital spending as a public policy issue in California: As California moves into the 21st Century we face the dual problem of preserving the schools, highways, bridges, water systems, and housing of today, while also planning and building new facilities for a growing population. There is no choice: we must maintain our current capital investmentsÑ ÒinfrastructureÓÑand make new investments. Governor Davis charged the commission with determining how much the state should spend on infrastructure repair in the coming years and how this spending should be financed. The California Business Roundtable has also called for substantial increases in state infrastructure spending in the near future. In a special edition of CaliforniaÕs Debt Affordability Report (California State Treasurer, 1999), the state treasurer has also called for a comprehensive evaluation of infrastructure 2 and investment needs. The prospect of significant increases in future state borrowing suggests that debt policy is an increasingly important issue. The Department of Finance has identified $82.2 billion of state and state-funded local infrastructure needs over the next decade. About half of this total can be paid for out of existing bond funds and anticipated federal funds, but more than $40 billion of infrastructure spending will need to be financed with new debt. If the state maintains its current ratio of debt service to general fund revenues, the stateÕs debt capacity over the next ten years will be $32.5 billion, according to Treasurer report estimates. If spending on debt service increases from a little over 4 percent to 5 percent, the amount of debt issued could increase by $10 billion and will cover the currently identified needs of the state. CaliforniaÕs interest burden is currently lower than that of the typical state. The Census of Governments reports that although the stateÕs general obligation debt at the end of fiscal year 1997 was $45.3 billion, the second largest in the nation, per capita indebtedness was $1,405, or slightly below the national average. The Advisory Council on Intergovernmental Relations (ACIR) (1998) reports that in fiscal year 1995, CaliforniaÕs state interest burden was $76 per capita, compared with a national average of $82. These differences are not due to a smaller government sector in California than in other states. In fiscal year 1997, for example, per capita state spending in California was $3,632, compared with a national average of $3,345. Borrowing costs on state debt are determined by a variety of factors, the most dramatic of which are repayment problems or other fiscal crises. At the height of the Massachusetts fiscal crisis in 1990, for example, yields on its general obligation bonds were more than 1 percent higher 3 than CaliforniaÕs. Similarly, when New York City appeared to be on the verge of bankruptcy in 1976, yields on New York state bonds were nearly 2 percent higher than CaliforniaÕs. Although such crises account for significant differences in interstate borrowing costs, less dramatic factors contribute to these disparities as well. One such factor is the set of fiscal rules that governs a stateÕs ability to raise and spend revenue. In some states, for example, the state constitution requires a prompt tax increase or spending cut to counter a budget deficit. In other states, governors and legislatures may use short-term debt financing to cover revenue shortfalls. Disparities in these fiscal rules affect a stateÕs perceived credit risk and therefore its borrowing costs. This study explores how these fiscal rules, and the way they interact with state economic conditions, affect the cost of issuing general obligation debt. The results are interpreted with particular reference to California. Chapter 2 describes the fiscal institutions in place in California and the rest of the country that are the primary focus of this study. Chapter 3 reports summary information on state borrowing costs, in California and other states, for the three decades that provide the analytical basis for this study. Chapter 4 summarizes the empirical evidence that links state economic conditions, fiscal rules, and fiscal health to state borrowing costs. More detailed information on the fiscal institutions and the statistical analysis is presented in appendices. Finally, a brief conclusion highlights the findings and their implications for public policy. 5 2. Current Fiscal Rules Across Different States Although state fiscal constitutions vary along many dimensions, our analysis of fiscal rules focuses on seven characteristics. 1 The first is the extent to which a state must have a balanced budget at different points in the legislative process. Only Vermont does not have a formal balanced budget requirement. The balanced budget requirements of the other 49 states can be divided into four groups. In 44 states, including California, the governor must submit a balanced budget. This is the weakest of the balanced budget requirements, as it does not actually require that the state enact a budget that matches expenditures and revenues. In 37 states, including California, the legislature must enact a balanced budget. Even in these states, however, actual revenues and expenditures may diverge if there are unexpected fiscal shocks after the budget is signed into law. When an unexpected deficit develops during _________ 1Detailed information on state budget rules may be found in the annual reports of the ACIR (for example, 1988 and 1998), or in Briffault (1996). 6 the fiscal year, six states require that the governor and legislature correct the deficit in the next budget cycle. Because budget cycles in some states are biennial, this requirement permits substantial periods of budget deficits. Twenty-four of the 37 states with balanced budget requirements prohibit the government from carrying deficits into the next budget cycle. This provision represents the strictest anti-deficit rule, as it requires the legislature either to cut spending, or raise taxes in the fiscal year when the deficit emerges, or to float short-term debt that will be retired in the next fiscal year. Previous studies, including Alt and Lowry (1994), Bohn and Inman (1995), and Poterba (1994), find that such anti-deficit rules are generally correlated with lower average deficits and more rapid adjustments to budget shortfalls. In 1987, the ACIR constructed an index that characterizes fiscal discipline among state governments. This index, which is presented in detail in Appendix A and summarized in Figure 1, ranges from 0 (lax) to 10 (stringent). California receives a score of 6 on this scale, in part because the state constitution permits the use of short-term borrowing to cover a deficit. Because it is difficult to interpret a one-unit change in this index, we have adapted it for the purposes of this study. Our statistical analysis contrasts states for which the index value is 5 or below with those for which the value is 6 or above. States with scores of 5 or below cannot have any stronger anti-deficit rules than a requirement that the governor present a balanced budget. Only eight states receive ACIR scores of 5 or below, whereas 26 receive a score of 10. States in the Northeast and the upper Midwest are less likely to have stringent anti- deficit requirements than states in other regions. Outside of those regions, many states with relatively weak budget rules have other fiscal constraints on state revenue or expenditures; for example, California, 7 0: No balanced budget rules 1–5: Proposed budget balances 6–8: Can borrow to balance budget 9–10: Ending budget must balance Figure 1ÑBalanced Budget Requirements Nevada, and Louisiana have passed limits on how quickly expenditures can grow over time. These revenue and expenditure limits are discussed in more detail below. The second fiscal characteristic we consider is the ease with which states issue long-term general obligation debt. Twelve states do not restrict debt issuance. Thirty-eight states have constitutional restrictions on debt issue, and two have legislative limits. The most common type of restriction limits the amount of debt outstanding. Fourteen states, including California, permit voters to override constitutional restrictions on debt levels to issue additional debt. We do not attempt to distinguish 8 among the various limits on debt issue; rather, we consider whether a state has any restriction, including requiring voter approval, on issuing general obligation debt. We use this definition to study the effects of borrowing limits on borrowing costs. Figure 2 (and Column 2 in Appendix A) shows the continental states that have borrowing limits. As with anti-deficit rules, the previous literature suggests that limits on long- term debt issues affect fiscal outcomes. Bunche (1991) finds that states with restrictions on general obligation debt are more likely to rely on revenue bonds to finance a range of government projects. Kiewiet and Szakaly (1996) also find that these rules affect state borrowing behavior. No provision Partial restrictiona Full restriction aDebt limit can be exceeded with voter approval. Figure 2ÑStates with Debt Restriction Provisions 9 However, previous studies have not considered the link between these institutions and yields on state general obligation debt. The third and fourth fiscal characteristics are tax and expenditure limits at the state level. The most common type of limit restricts the growth rate of general fund expenditures or revenues to the growth rate of personal income or to the growth rate of population and inflation. Some of these laws place binding constraints on governors and legislatures. Rueben (1996) provides information on override provisions and indicates which limits cannot be overridden by a simple legislative majority. Tax and expenditure limits have changed since the late 1970s, when most current limits were enacted. Several states enacted new limits or strengthened earlier limits in the early 1990s. In 1979, for example, Louisiana passed a law mandating that the ratio of tax revenue to personal income in any year could not exceed the 1979 ratio; however, a simple legislative majority could override the measure. In 1993, voters passed a referendum that limited the growth in appropriations. Unlike the earlier law, this one required a two-thirds vote to override the limit. Similarly, ColoradoÕs expenditure limit in 1978 was supplemented with Amendment 1, or the Taxpayers Bill of Rights, in 1992. The latter initiative limits increases in state and local spending to the change in population and increases in the Consumer Price Index. In addition, it requires voter approval for the implementation of new taxes. In 1980, California voters approved Proposition 4, the Gann amendment, which limits the growth in both state and local expenditures to changes in the Consumer Price Index and population. As Tables 1 and 2 indicate, 24 states have adopted some form of state tax or expenditure limitation since 1976. Twenty-one states currently have an expenditure limit in place, and 19 states require at least a supermajority of both houses of the 10 Table 1 History of State Spending Limit Adoptions 1976Ð1979 1980Ð1983 1984Ð1988 1989Ð1992 1993Ð1997 UtahAlaskaOklahomaColorado Nevada NevadaMontana Rhode Island Washington Oregon South Carolina Connecticut Louisiana California Missouri North Carolina TexasIdaho New Jersey TennesseeUtah Hawaii Arizona Colorado Rhode Island New Jersey NOTES: Binding limits are in bold. Other limits are advisory or can be overridden by a legislative majority. New JerseyÕs original expenditure limit expired in 1976. Table 2 History of State Revenue Limit Adoptions 1976Ð1979 1980Ð1983 1984Ð1988 1989Ð1992 1993Ð1997 Louisiana MissouriMassachusetts Colorado Missouri Washington Louisiana Florida Michigan NOTE: Binding limits are in bold. Other limits are advisory or can be overridden by a legislative majority. legislature to override spending limits. Seven states have passed a limit on revenues, five of which were enacted after 1986. Because tax and expenditure limits restrain the growth of government spending, they are often classified together. However, the tax-exempt bond market makes a strict distinction between them. Tax limits increase the risk that revenues will not cover future interest payments; they therefore increase the risk of state general obligation debt. In contrast, expenditure limits increase the likelihood that legislatures 11 will make interest payments. The empirical results presented below support this dichotomy. The fifth fiscal characteristic we consider is the supermajority requirement to raise specific taxes. Unlike revenue limits, supermajority requirements do not cap existing taxes, such as sales and income taxes, which often rise during economic upswings. Fifteen states, including California, currently have a supermajority requirement for new taxes. Historically, supermajority requirements have been concentrated in the South. More recently, however, other states have passed supermajority requirements as part of initiatives to limit government revenues. California passed a two-thirds majority requirement in 1978 as part of Proposition 13. Montana and Missouri have even stricter provisions concerning new taxes. In a budgetary emergency, the legislature in each state can pass new temporary taxes with a supermajority. However, voter approval is required to make these emergency taxes permanent. Table 3 presents information on states with supermajority requirements. Because supermajority requirements and revenue limits restrict the governmentÕs ability to raise revenue, we expect these constraints to affect bond yields in similar ways. The last two fiscal characteristics considered here are limits on local governments, specifically, the extent to which states restrict the ability of these governments to raise revenues and issue new debt. Although local restrictions do not directly affect a state governmentÕs fiscal health, they can have indirect effects insofar as they lead state governments to take over traditional local government functions. Rueben (1996) shows that states with local limits raise a higher percentage of revenues centrally than states without such limits. For example, Proposition 1A, which was enacted in California in November 1998, permits the state to borrow 12 Table 3 Supermajority Requirements and Other Constitutional Restrictions on Legislative Tax Power Ê Ê Referendum Legislative Ê Ê or Voter Majority Ê State Adopted Initiative Required Applies To Arizona 1992 Initiative 2/3 All taxes Arkansas 1934 Referendum 3/4 All taxes except sales and alcohol California 1978 Initiative 2/3 All taxes Colorado 1992 Initiative 2/3 All taxes Delaware 1980 Referendum 3/5 All taxes Florida 1971 Referendum 3/5 Corporate income tax Florida 1996 Initiative 2/3 New taxes Louisiana 1966 Referendum 2/3 All taxes Mississippi 1970 Referendum 3/5 All taxes Missouri 1996 Referendum 2/3 Emergency taxes Montana 1998 Initiative 3/4 Emergency taxes Nevada 1996 Initiative 2/3 All taxes Oklahoma 1992 Initiative 3/4 All taxes Oregon 1996 Referendum 3/5 All taxes South Dakota 1996 Referendum 2/3 All taxes Washington 1993 Initiative 2/3 All taxes NOTES: Missouri and Montana can pass temporary new taxes in an emergency. However, if either state wants to make emergency tax increases permanent, it must get voter approval. $9.2 billion for local education facilities, providing a means of financing school construction in districts that are unable to gain the supermajority required in local elections. Figure 3 shows which states have local revenue limits. Many states restrict local tax rates or assessments. CaliforniaÕs Proposition 13 is perhaps the best known of such restrictions. Our analysis considers both state-imposed limits on local revenue levels and state restrictions on both the assessed valuation of local property and the tax rate that can be 13       No limit Before 1976 1976 to 1981 1982 to 1987 Since 1987    Figure 3ÑYear Local Tax Limit Passed imposed. In addition to Proposition 13 (and Proposition 4, which limits expenditures), California has two other restrictions on local government finance: Proposition 62, which limits general taxes at the local level, and Proposition 218, which confirms the limit on general taxes and also limits local fees and assessments. Virtually all states limit localitiesÕ ability to issue municipal debt. According to a 1993 ACIR report, only Florida and Tennessee do not place any debt limits on cities, but these states require voter approval for debt over a certain amount. The most typical requirement calls for voter approval before municipalities can issue new debt. Forty states also limit counties in this regard as well. Because all states restrict the issuance of 14 local bonds, we focus on the eight states that require a supermajority to authorize new local debt. Appendix A presents information on the seven state fiscal institutions described above. It reports the actual ACIR index for each stateÕs anti - deficit program, indicates which states limit the amount of outstanding debt, and shows whether this limit can be overridden with consent of the voters. It also gives the year of passage for the more recent fiscal constraints passed and information on the kind of supermajority required. California has a unique combination of fiscal rules. The actual state budget need not be balanced, but there are limits on the amount of state debt that can be issued without voter approval. California also has a state expenditure limit and requires supermajorities to introduce new taxes or to increase existing ones. Finally, CaliforniaÕs local governments are more fiscally constrained than those in other states. Using the initiative system, CaliforniaÕs voters have enacted a number of restrictions on local governments; indeed, only OregonÕs voters have used the initiative system more. Property taxes are limited and local governments are required to receive voter approval for most other revenue sources. Finally, California is one of only eight states that require a supermajority of voters to issue local general obligation debt. The supermajority requirement for local debt is the only limit that was not passed recently using the initiative system. 15 3. State Borrowing Costs in the Last Two Decades The market value of outstanding tax-exempt bonds is roughly 40 percent as large as that of corporate bonds and roughly one-fifth the value of U.S. Treasury bonds. Although the bond market is well organized, there are three practical obstacles to obtaining information on the borrowing costs that different states pay at a given time or on those paid by a given state over time. First, there is limited trading in most tax-exempt bond issues. Second, tax-exempt bonds differ widely in their call provisions and in other detailed provisions. Finally, many tax - exempt bonds are sold in bundles, making it difficult to identify the price and yield to maturity of any single bond. For these reasons, it is not generally possible to obtain price quotes for a large and comparable sample of state general obligation bonds over any substantial period of time. In the absence of market data on state government borrowing costs, we have turned to the Chubb Insurance Company ÒRelative Value 16 Survey.Ó This survey, which has been carried out every six months since 1973, asks approximately 25 traders of tax-exempt bonds to estimate the current yields on general obligation bonds from 40 states. (Ten states with little borrowing activity are excluded.) Chubb survey respondents are asked to estimate the current yield on ÒhypotheticalÓ 20-year general obligation (GO) bonds. By asking about hypothetical bonds, the survey avoids problems related to call protection and other specific details associated with individual bonds traded in the marketplace. The Chubb survey collects information on the relative yields on the GO bonds issued by various states and by New Jersey, which serves as the comparison state of the study. The Chubb survey data show differences in borrowing costs in two ways: across states at a given point in time and over time for a given pair of states. Figure 4 presents summary information on the differences in borrowing costs across states, as well as on relative borrowing costs in California and other states. The top line represents the highest yield and the bottom line represents the lowest yield at each date. A state with a higher required yield pays more in interest payments to issue a given amount of debt. The spread between the maximum and minimum yield was less than 30 basis points in 1997; as recently as 1988, however, the yield spread between the highest yield and lowest yield borrowers was more than 120 basis points. The bold line in Figure 4 represents the relative yield on CaliforniaÕs general obligation bonds in comparison to similar bonds issued by New Jersey. The last line represents the average required yield for all other states (compared to New Jersey) in the Chubb survey. CaliforniaÕs borrowing costs follow the stateÕs economic fortunes; these costs were below the national average in the late 1980s, when the California 17 Interest required above New Jersey bond (basis points) –100150 50 0 –5073 75 77 89 9593 91 87 85 83 81 79 97 Year 100 California Other state average Highest yield required Lowest yield required Figure 4ÑRelative Bond Yield Required on California State Debt economy boomed, and rose above the national average during the 1990s. In fact, CaliforniaÕs borrowing costs were among the highest in the nation in the mid 1990s. Figure 5 shows the relative yields on the general obligation bonds of California, Massachusetts, New York, and Texas between 1974 and 1997. The figure shows the historical volatility in each stateÕs bond yield. During the New York City fiscal crisis in 1976, Chubb survey respondents estimated that bonds issued by New York or Massachusetts would yield roughly 150 basis points (one and one-half percentage points) more than bonds issued by California or Texas. This yield spread declined throughout the 1980s and widened once again around 1990, another period of fiscal stress for Northeastern states. Recent economic 18 strength in all of the states shown in Figure 5 contributes to the relatively small yield disparities at the end of the data sample. To help illustrate the relative differences in borrowing costs, consider the relative costs of issuing $1 million of debt in California, New Jersey, and New York. In 1976, California would have paid $4,700 dollars less than New Jersey and $17,000 less than New York in interest costs for every $1 million of debt. These differences reflect the effects of the New York City fiscal crisis of 1976. In July 1995, following the Orange County bankruptcy in December 1994, California would have paid $2,570 more than New Jersey and only $870 more than New York to issue the same amount of debt. Interest required above New Jersey bond (basis points) –100150 50 0 –5073 75 77 89 9593 91 87 85 83 81 79 97 Year 100 CaliforniaMassachusetts Texas New York Figure 5ÑRelative Bond Yield Required for Various StatesÕ 20 Year General Obligation Bonds 19 To evaluate the accuracy of the estimates in the Chubb survey, we compared the reported relative yields from the survey with those of bonds for states that make up the Bond Buyer General Obligation Bond Index. The results are encouraging. The index measures the yields on GO bonds maturing in 20 years, and it is supposed to have a rating roughly equivalent to MoodyÕs A1. Bonds issued by California and New York are among those included in the 20 bond index. From December 1, 1988, through August 13, 1992, CaliforniaÕs average bond yield was 38 basis points lower than New YorkÕs. The average yield differential in the Chubb survey was 36.5 basis points. 21 4. Borrowing Costs, Economic Conditions, and Fiscal Rules The substantial variation in tax-exempt bond yields across states and over time indicates that bond prices are sensitive to repayment prospects. This pattern suggests the viability of using data like those in the Chubb survey to determine how borrowing costs are affected by differences in state economic conditions as well as state fiscal rules. This chapter reports three sets of results along these lines. It begins with a discussion of how economic conditions and fiscal rules are correlated with the level of borrowing costs. It then examines how unexpected budget deficits and surpluses affect changes in borrowing costs and how the effects of budget surprises are amplified or cushioned by the presence or absence of various fiscal rules. The chapter closes with a discussion of the different ways bond markets react to fiscal news. 22 How Economic Conditions and Fiscal Rules Affect the Level of Borrowing Costs Our first empirical strategy tests whether borrowing costs are higher, on average, in states with weaker economies or with less-stringent fiscal rules. Because bond yields are influenced by state economic circumstances and other factors that affect the perceived risk of state debt, we use multivariate regression techniques. Our measured effect of fiscal rules controls for several economic variables: outstanding debt level as a percentage of personal income, state unemployment rates, real state per capita income, state general fund revenues as a fraction of per capita income, and the top state marginal tax rate on interest income. 1 As a proxy for changes in political preferences among voters for state spending and state deficits, we also include an index of the liberalness of the congressional delegation of each state as calculated by Americans for Democratic Action. These controls are discussed at more length in Poterba and Rueben (1999). Before analyzing the effect of fiscal rules on borrowing costs, we consider the effects of state economic conditions on these costs. The regression equation shown in Appendix Table B.1 indicates that a 1 percent increase in a stateÕs unemployment rate raises the stateÕs borrowing costs by five and one-half basis points. Three percent differences in unemployment rates across states are not uncommon, and the statistical results indicate that these economic differences are clearly an important factor in state borrowing costs. Because our regression equation also includes the level of state per capita income as an _________ 1Goldstein and Woglom (1992) investigate the role that economic variables play in setting the price of new municipal bonds. 23 explanatory variable, and because higher state per capita income is associated with lower borrowing costs, this analysis may understate the importance of economic conditions in affecting bond yields. In 1993, for example, the California unemployment rate was 9.4 percent. By 1998, the unemployment rate had declined to 5.9 percent. Our estimates suggest that the stateÕs borrowing cost has declined by 20 basis points as a result of this improvement in economic conditions. We now turn to the effects of our seven fiscal variables on state borrowing costs. Unlike economic conditions, fiscal rules are determined by legislators and voters, who may wish to consider the potential effects of new limits. Our econometric results provide clear support for an influence of fiscal rules on borrowing costs. Table 4 summarizes the key findings. The indicator variables for expenditure and revenue limits affect borrowing costs in different, but substantively important, ways. States with binding revenue limits pay, on average, 17.5 basis points more on their general obligation debt than states without such limits. By comparison, states with binding expenditure limits pay four basis points less than states without such limits. Although states with supermajority requirements also face higher costs, the estimate is too imprecise to permit strong conclusions. A limit on local revenues increases borrowing costs by about five basis points. The findings on the link between borrowing costs, anti-deficit rules, and limits on a legislatureÕs power to issue long-term debt are less conclusive. States with scores of 5 or lower on the ACIR index pay nine basis points more than states with higher index values, but we cannot reject the null hypothesis that anti-deficit rules do not affect borrowing costs. The coefficient on the indicator for limits on issuing debt is negative, indicating that debt limits on average reduce borrowing costs, 24 Table 4 Effect of Fiscal Institutions on State Borrowing Costs Fiscal InstitutionEstimated CoefficientCurrent California PolicyEffect of Change in Fiscal Variable on Per Capita Interest Costs for California Weak anti-deficit rules 8.99 (6.58)ÒStrongÓ rules$1.26 Limit on legislatureÕs power to issue long-term debtÐ4.72 (5.51)Limit Ð$0.66 Binding expenditure limit Ð3.65 (2.12)Limit since 1979Ð$0.51 Binding revenue limit 17.45 (5.33)No limit $2.45 Supermajority required to pass new taxes 3.15 (3.94)Limit $0.44 Limit on local revenues (Proposition-13-type limits)4.56 (2.54)Limit since 1978$0.64 Supermajority required to pass local bondsÐ4.04 (3.00)2/3 major- ity requiredÐ$0.57 NOTES: Values in parentheses are standard errors. Estimates are from regression results presented in Appendix Table B.1. California had $45.3 billion or $1,405 per capita of outstanding state debt at the end of fiscal year 1997 according to the Census of Governments (1999). The entries in the last column assume that the borrowing cost of this entire debt stock changes by the number of basis points indicated in the second column. but again, the figures do not permit precise conclusions. This result partly reflects the difficulty of making distinct estimates of the effect of many different fiscal rules, many of which are found together in various states. Poterba and Rueben (1999) estimate the effect of each fiscal rule separately and find significant effects of weak anti-deficit rules, expenditure limits, revenue limits, and supermajority taxes. If we 25 estimate separate effects for the last decade, we also find that supermajority rules are becoming increasingly important. Table 4 also describes the implications of our results for a state with an outstanding debt level comparable to CaliforniaÕs. The third column in Table 4 reports the change in per capita interest costs that would be associated with a change in fiscal rules. The two changes with the largest effects on per capita borrowing costs are the adoption of a tax limitation law and the lack of strong balanced budget rules. We estimate that if California had adopted a revenue limit at the state level, the stateÕs borrowing rate would have increased by more than 17 basis points. If we also assume that this higher borrowing cost would apply to all state debt, then this difference would translate to an increase in state borrowing costs of $2.45 per person, or an additional $79 million. Fiscal Shocks, Fiscal Rules, and the Bond Market Reaction to Fiscal News Our second empirical strategy considers how a stateÕs borrowing costs change in reaction to fiscal news. If fiscal rules are an important determinant of market interest rates, and if some rules are thought to reduce risk for bondholders, then such rules will have a larger effect on borrowing costs in some circumstances than in others. In particular, the economic effect of tight fiscal rules may be greatest when states are experiencing fiscal stress. The last decade provides a valuable opportunity to study bond market reactions to fiscal stress. Although statesÕ revenues have exceeded expectations recently, and some states have built up Òrainy day fundsÓ or enacted tax cuts, many states experienced sharp declines in their revenues during the recession of the early 1990s. This recession was especially 26 long lasting and deep in California. Poterba (1994) shows that in states with stringent anti-deficit rules, these fiscal events triggered corrective tax increases or expenditure reductions. In states with weaker anti-deficit rules, adjustments occurred more slowly. In California, the recession resulted in unexpected budget deficits in certain years. CaliforniaÕs state government reacted to the recession and the resulting budget shortfall in various ways. These reactions included increasing fees and assessments (in higher education, for example) and reallocating property tax revenues from cities and counties to school districts. The latter action allowed the state government to lower the amount of general fund revenues spent on school finance. We examine how unexpected deficits affected the bond market yields of different states. In particular, we are interested in how the tax-exempt bond market reacts to news concerning state surpluses and deficits, and how these reactions are affected by a stateÕs fiscal rules. We study the change in the Chubb surveyÕs yield spread between bonds issued by a given state and bonds issued by New Jersey over a 12 month period. After relating this change to the unexpected budget deficit or budget surplus in the intervening fiscal year, we allow the change in borrowing costs for a given fiscal shock to depend on the stateÕs fiscal rules. Thus, we are estimating how much interest rates have changed over a given fiscal year as a function of the fiscal health of the state during that time period. In addition, we are allowing the different fiscal rules to affect this relationship. Appendix B presents the details of this regression specification. Our analysis is based on annual measures of state fiscal shocks that are constructed from information collected by the National Association of State Budget Officers, which collects information on budgeted and 27 actual revenues and expenditures for each state. Each yearÕs unexpected budget deficit is defined as the deficit that would have developed, given actual economic conditions and other factors, with the tax and expenditure system that was in effect at the beginning of the fiscal year, less the forecast deficit at the beginning of the fiscal year. Poterba (1994) describes this measure in greater detail. Figure 6 graphs the unexpected deficit experienced by California relative to the average deficit experienced by other states. It also shows the largest deficit and the largest surplus of any state over the last decade. Unexpected deficits and surpluses are represented as a percentage of each stateÕs general fund revenues. Throughout this period, there were always states with either surpluses or deficits. CaliforniaÕs deficits have followed its economic health, with deficits during the recession and surpluses Percentage of general fund revenues –3040 20 0 –10 –2088 89 90 9695 94 93 92 91 97 Year 10 30 CaliforniaMaximum Other states Minimum Figure 6ÑUnexpected State Deficits as a Percentage of General Revenue, 1988Ð1997 28 more recently. Its fiscal health, however, has lagged behind that of other states. While other states experienced deficits in 1990, CaliforniaÕs budget was virtually balanced. In 1991 and 1992, however, unexpected deficits for California averaged nearly 10 percent of revenues, or approximately $90 per person. In more recent years, state finances have grown stronger, with unexpected budget surpluses on average. Table 5 presents information on the average unexpected deficit in selected states. It also presents data on the first and second halves of our data sample. In many states, the first half of the sample was a time of fiscal stress, with most states recovering during the second half. For California, the unexpected deficit averages 2.5 percent of revenue. This figure is somewhat larger than that in other large states. For New York, Table 5 Average State Deficit Shock, Selected States, 1988Ð1997 StateAverage Deficit Shock, 1988Ð1997Average Deficit Shock, 1988Ð1992Average Deficit Shock, 1993Ð1997 Alaska 3.67% 5.81 1.53 Arizona 0.63 Ð3.31 4.57 California Ð2.52 Ð4.29 Ð0.76 Colorado 1.21 0.37 2.06 Illinois Ð0.80 Ð2.42 0.82 Massachusetts Ð4.68 Ð11.06 1.70 New Jersey 0.30 Ð1.16 1.76 New York Ð1.84 Ð3.35 Ð0.34 Oregon 3.12 2.84 3.46 Pennsylvania Ð0.19 Ð1.50 1.13 Texas Ð1.23 Ð7.66 5.20 Utah 2.30 3.00 1.59 Washington 1.21 1.76 0.66 Average for all states Ð0.13 Ð0.89 1.16 NOTES: AuthorsÕ tabulations are based on information reported by the National Association of State Budget Officers. Deficit shocks are reported as a share of state revenues. A negative number indicates a deficit and a positive number is a surplus. The last row is an equal weighted average for all states. 29 this average is 1.8 percent; for Texas, 1.2 percent; and for Illinois, 0.8 percent. Unlike California, other Pacific states averaged fiscal windfalls over this period. In Oregon, the average surplus was 3.2 percent of revenue. In Washington, the analogous surplus was 1.2 percent of revenue. Alaska had the largest fiscal windfalls, averaging 3.7 percent of revenue. Figure 7 presents deficit patterns for selected states. Notice that both New York and Texas experienced much more volatility in their budgets over this period. Table 6 presents the main findings from our regression analysis of how bond yields react to unexpected deficits and how state fiscal institutions affect these reactions. It presents the change in interest rate required to issue new debt if a state has a given fiscal rule in place and experiences an unexpected $100 per capita deficit. The results further Percentage of general fund revenues –3040 20 0 –10 –2088 89 90 9695 94 93 92 91 97 Year 10 30 New Jersey CaliforniaMassachusetts Texas New York Figure 7ÑUnexpected State Deficits as a Percentage of General Revenue, Selected States, 1988Ð1997 30 Table 6 Effect of State Fiscal Rules on the Bond MarketÕs Reaction to Unexpected Budget Deficits and Budget Surpluses Fiscal InstitutionCurrent California PolicyEffect of Fiscal Institution on Bond Market Reaction to a $100 Per Capita Deficit Shock (Basis Points) Weak anti-deficit rules ÒStrongÓ rules 6.1 Limit on legislatureÕs power to issue long-term debtLimit 3.0 Binding expenditure limit Limit since 1979Ð0.0 Binding revenue limit No limit 5.2 Supermajority required to pass new taxesLimit since 19785.6 Limit on local revenues (Proposition-13-type limits)Limit since 1978Ð1.0 Supermajority required to pass local bonds2/3 majority required1.6 Predicted effect on bond costs of deficit given current California rules9.2 NOTES: Estimates are based on regression coefficients reported in Table B.2. See the text for further details. The last row represents the additional cost of issuing new debt while experiencing a $100 unexpected per capita deficit and having CaliforniaÕs rules in place. support our conclusion that fiscal institutions have real effects on state borrowing costs. In states with weak anti-deficit rules, required bond yields increased more in reaction to a given deficit than in states with stricter rules. A $100 per capita increase in a stateÕs budget deficit, about the size of the California unexpected deficits in 1991 and 1992, raises 31 borrowing costs by 6.1 basis points more in states with weak anti-deficit rules than in states with stricter rules. The next entry in Table 6 shows the estimated effect of having a limit on issuing long-term debt and experiencing a deficit. Restricting a stateÕs ability to issue debt appears to make a stateÕs bond yield more responsive to changes in the stateÕs fiscal health, although once again it is difficult to draw precise conclusions. This result partly reflects the difficulty of making distinct estimates of the effect of many different fiscal rules, many of which are found together in various states. The next two rows report the effect of the interactions between expenditure limits, tax limits, and unexpected deficits. The results suggest that an unexpected increase in the state deficit has a much larger effect on bond yields in states with binding revenue limits than in those with binding expenditure limits. Although the effect of an expenditure limit is virtually zero, there is a positive and statistically significant effect of a revenue limit on the bond market reaction to an unexpected deficit. A $100 per capita deficit shock raises borrowing costs by 5.2 basis points more in states with tax limitation laws than in states without such limits. Thus, CaliforniaÕs debt costs would have been higher in the recession of the early 1990s if revenue limits had been in place instead of the Gann amendment. This would translate into an additional $520,000 in interest costs per billion dollars of debt issued. Supermajority rules that limit a stateÕs ability to raise taxes also increase borrowing costs. A $10 per capita deficit shock increases borrowing costs by 5.6 basis points. This result indicates that borrowing costs during the California recession of the early 1990s would have been higher than now; specifically, an additional billion dollars of debt would have cost an extra $560,000. 32 The next two rows in Table 6 present our findings on the interaction between local fiscal limits and the effect of fiscal news on state borrowing costs. The effects associated with these variables are small. So, although there was a direct effect of local tax limits on the total cost of debt (see Table 4), it appears that local fiscal limits do not affect the relationship between fiscal news and bond costs. Finally, the last row of Table 6 presents the cumulative results of CaliforniaÕs fiscal rules. Our estimates suggest that if California had issued new debt in 1992, when the state was running a deficit, California would have had to pay an additional 9.2 basis points to issue new debt. Thus, the interaction of CaliforniaÕs deficit and its fiscal rules would have meant an additional $920,000 in interest costs per billion dollars of debt. The results in Tables 6 and B.2 provide clear support for our earlier claim that bond market participants view revenue limits and expenditure limits quite differently. They also set the stage for our analysis in the next section of the sensitivity of state borrowing costs to news about state fiscal conditions. Interstate Differences in Bond Market Reactions to Fiscal News The analysis underlying our results in Table 6 assumes that different states exhibit different bond yield sensitivities to deficit shocks only because they have different fiscal rules. This is likely to be a substantial oversimplification. To study this issue, we related the change in bond yields to the unexpected state deficit for all states and then included an indicator for specific states. We discovered substantial differences across states in the effect of fiscal news on bond yields. Table 7 reports the effect on bond yields of changes in fiscal health, as measured by unexpected deficit shocks, in specific states. We have 33 Table 7 Interstate Variability in Effect of Fiscal News on Borrowing Costs (in Basis Points) StateEffect of $100 Per Capita Deficit Shock on Bond Yield for Large States California 14.1 (3.6) Texas Ð2.2 (1.9) New York 14.1 (4.7) Florida Ð4.2 (7.8) Pennsylvania 18.3 (6.9) Illinois 2.1 (6.5) Ohio Ð3.0 (5.6) Michigan 0.7 (5.7) Georgia 3.6 (5.1) Massachusetts 8.5 (1.2) Connecticut 3.0 (1.3) NOTES: Estimates are based on regressions relating the change in bond yields from one year to the next to the unexpected fiscal deficit reported within the year. Larger deficits are reported as positive values. States are included in the sample if they are large in population terms or have a large outstanding stock of debt. Standard errors are reported in parentheses 34 chosen the largest states, as well as those states with the largest amount of outstanding debt, because these states are most responsive to such changes. California is the most populous state and also had the second largest amount of outstanding debt in 1995. CaliforniaÕs borrowing costs are more responsive to unexpected budget deficits than are the borrowing costs of most other large states. Given the size of the stateÕs municipal bond market and the Orange County bankruptcy at the end of 1994, it is not surprising that California is more responsive to fiscal shocks. A $100 per capita unexpected increase in CaliforniaÕs budget deficit is predicted to raise borrowing costs by 14 basis points. The only other large states with comparable estimates are New York (14 basis points) and Pennsylvania (18 basis points). Our earlier analysis in Table 6 suggests that if municipal bond markets were only responding to fiscal rules in place, California should have to pay nine basis points more during periods of fiscal stress. We investigated whether the apparently large ÒCalifornia effectÓ could be explained by differences in fiscal rules and found that these rules seem to explain a little less than two-thirds of the California effect. We modified the regression equation that generated the estimates in Table 6 to allow for a separate California effect. The resulting estimates, which are shown in the second column of Table B.2, confirm that the California effect is not simply the result of the stateÕs fiscal rules. The effects of the fiscal variable interactions are not substantively affected by allowing for a California interaction. These results indicate that the bond market is particularly sensitive to deficit news about California and other large states. One possible explanation is that bond market participants pay closer attention to fiscal developments in states with relatively large quantities of outstanding tax- 35 exempt debt than to similar developments in states with limited debt markets. This explanation implies that deficit news has a larger effect in these states because it is more visible and it affects more investors. A second possibility is that the political circumstances in California and several other large states make it more difficult for the states to return to fiscal balance in the years immediately following an unexpected deficit. Unexpected deficits would therefore have a larger effect in these states because they provide more information about the future of state fiscal conditions. Both of these explanations are likely to contribute to our empirical findings. 37 5. Conclusions and Implications This study finds that there is a clear link between state economic circumstances and the interest rate at which a state borrows when it sells general obligation debt. Because lenders expect states with more robust economies to have less difficulty servicing their debt, states with lower unemployment rates face lower interest rates in the capital market. Changes in state economic conditions explain some of the variation over time in the relative borrowing costs of different states. State fiscal institutions, such as CaliforniaÕs Gann amendment or Proposition 218, also affect borrowing costs. Whether and how the stateÕs constitution limits the discretion of policymakers to run fiscal deficits, and whether the state has a tax limit or an expenditure limit in place, are strongly correlated with borrowing costs. These findings suggest that policymakers need to recognize that changes in fiscal policies, and in the budget process more generally, affect these costs. In California, policymakers and citizen groups have historically paid little 38 attention to the effects new fiscal restrictions can have on borrowing costs. Nor has the topic received much research attention. Rosenblum (1999) remarks that one effect of Proposition 218 and its predecessors is that CaliforniaÕs bond rating has not returned to AAA, the highest possible level. Davey (1995) describes how Ohio legislators and others who opposed a revenue limit argued that adopting it might raise the stateÕs borrowing cost. Beyond the finding that fiscal rules matter for borrowing costs, two more specific conclusions emerge from our study. First, the bond market reacts very differently to revenue limits and expenditure limits. Expenditure limits have small effects and probably reduce borrowing costs, but states with tax limitation laws face substantially higher borrowing costs than other states. States that require legislative supermajorities to pass new taxes also face higher borrowing costs. Our estimates suggest that if California had enacted a tax limitation law for state revenues instead of an expenditure limit, the stateÕs borrowing costs would be about 20 basis points higher than they currently are. That difference translates into approximately $3 of additional annual state expenditures per capita, or $95.6 million, if we assume that the level of state indebtedness was not affected by such policies. Second, there are substantial differences across states in the sensitivity of general obligation bond yields to fiscal news. The borrowing costs of states with relatively lax fiscal rules are more sensitive to budget news than are the borrowing costs of states with tighter fiscal rules. Although its fiscal rules are relatively tight, California is among the group of states with the highest responsiveness. It may be that municipal bond analysts and other credit market participants pay more attention to fiscal news in large states with many outstanding bonds than to 39 comparable news in smaller states. Our results suggest that accurate budget projections are more important in California than in many other states, because unfavorable budget news has a substantial and adverse effect on the stateÕs borrowing cost. The need for accurate budgeting is especially important over the next few years as California increasingly turns to the municipal bond market to pay for new infrastructure spending. 41 Appendix A State Fiscal Institutions State State Supermajority TaxLocal Local StateBalanced BudgetDebt RestrictSpending LimitRevenue LimitYear PassedVote RequiredTax LimitDebt Vote Alabama 10 Yes Maj Alaska 6 Yes 1982 1972 Maj Arizona 10 Yes 1978 1992 2/3 1980 Maj Arkansas 9 Yes* 1934 3/4 1981 Maj California 6 Yes* 1979 1978 2/3 1978 2/3 Colorado 10 No 1992 a1992 1992 2/3 1992 Maj Connecticut 5 No 1991 Delaware 10 No 1980 3/5 Florida 10 Yes* 1994 1971 3/5 1995 Maj Georgia 10 Yes Maj Hawaii 10 Yes 1978 Idaho 10 Yes 1980 1992 3/5 Illinois 4 No 1991 Maj Indiana 10 Yes 1973 Iowa 10 Yes 1978 3/5 Kansas 10 Yes* 1970 Maj Kentucky 10 Yes* 1979 Maj Louisiana 4 No 1993 1991 b1966 2/3 1978 Maj Maine 9 Yes* Maj Maryland 6 No Massachusetts 3 No 1986 1980 42 State State Supermajority TaxLocal Local StateBalanced BudgetDebt RestrictSpending LimitRevenue LimitYear PassedVote RequiredTax LimitDebt Vote Michigan 6 Yes* 1978 1978 Maj Minnesota 8 Yes Maj Mississippi 9 Yes 1970 3/5 Maj Missouri 10 Yes* 1980 1996 c1996 2/3 1980 2/3 Montana 10 No 1981 1998 3/4 1987 Maj Nebraska 10 Yes 1990 Maj Nevada 4 Yes 1994 d1996 2/3 1983 Maj New Hamp. 2 No Maj New Jersey 10 Yes* 1990 e New Mexico 10 Yes 1979 Maj New York 3 Yes* Maj North Carolina 10 No 1991 Maj North Dakota 8 Yes 1981 2/3 Ohio 10 Yes 1976 Maj Oklahoma 10 No 1985 1992 3/4 3/5 Oregon 8 Yes* 1979 1996 3/5 1991 Maj Pennsylvania 6 Yes* Maj Rhode Island 10 Yes* 1992 f1985 South Carolina 10 Yes* 1980 Maj South Dakota 10 Yes 1996 2/3 Maj Tennessee 10 No 1978 g Maj Texas 8 Yes 1978g1982 Maj Utah 10 Yes 1989 Maj Vermont 0 Yes Virginia 8 Yes Maj Washington 8 Yes 1993 1979 1993 2/3 1971 3/5 West Virginia 10 Yes 1990 3/5 Wisconsin 6 Yes Maj Wyoming 8 Yes Maj NOTES: Data on budget stringency rules and state and local debt restrictions are from ACIR (1987a) and Rafool (1997). Data on revenue and expenditure limits are from Rueben (1996). Data on local limits are from ACIR (1997). * denotes states that require a popular vote to approve debt issue. In states with nonbinding limits, limits can be overridden by simple legislative majorities. aColorado passed a nonbinding spending limit in 1977.bLouisiana adopted a nonbinding revenue limit in 1979 and a binding one in 1991.cMissouri adopted a nonbinding revenue limit in 1980.dNevada passed a nonbinding spending limit in 1979.eNew Jersey also passed a spending limit in 1976 that expired in 1983.fRhode Island adopted a nonbinding limit in 1977, but it was replaced with a binding limit in 1992. gTennessee and Texas have limits that are not binding. 43 Appendix B Detailed Description of Regression Results The empirical findings summarized in the text are based on two sets of regression equations. The first analyzes the relationship between the level of general obligation bond yields and the structure of state fiscal rules; the second relates the change in the bond yield for a state to the unexpected budget deficit, or surplus, in a given year. Poterba and Rueben (1999, 1998) investigate each of these statistical relationships and present a range of different regression models that are broadly supportive of the general findings reported here. In this appendix, we describe the regression equations that underlie our core findings. 44 Fiscal Rules and the Level of General Obligation Yields The basic regression model that we estimate to test for a relationship between the level of borrowing costs, and the structure of state fiscal institutions, is RR XX ZZ it jt it jt it jt t i it jt-= - + - +++-()*()* ()abqkee (1) where R it denotes the nominal interest rate on bonds issued by state i at time t, X it denotes the set of state-specific economic conditions that may affect borrowing costs, and Z it represents the set of state fiscal institutions that we are interested in studying. We estimate models for the cross-state differences in yields because the Chubb survey data are collected in this way. To make the explanatory variables consistent with the dependent variables, they must be computed as differences relative to the corresponding values for New Jersey. The variables that we include as control variables for borrowing costs are state outstanding general obligation debt per capita, the state unemployment rate, the level of real per capita income in the state, state general fund revenues as a fraction of per capita income, and the top state marginal tax rate on interest income. The first three variables are drawn from publications of the Bureau of Labor Statistics, the National Income and Product Accounts, and the Census of Governments. The marginal tax rate variable was computed using the State Tax Module of the National Bureau of Economic Research TAXSIM program and augmented through downloaded information on state income tax forms. We also include variables that proxy for the political climate in the state, on the grounds that such variables may provide information on the future evolution of state deficits and therefore on the creditworthiness of 45 the state. Our principal variable of this type is the Americans for Democratic Action (ADA) score for the stateÕs senate delegation; this should provide a general indication of the political ideology of the state. We include seven variables ( Z it) to measure the fiscal institutions in each state. The first is an indicator variable for states with weak anti- deficit rules, defined as a score of 5 or less on the ACIR index of fiscal stringency. The second is an indicator variable for legislative or constitutional limits on the legislatureÕs power to issue long-term debt. The third and fourth variables are indicator variables for binding tax limits, and binding expenditure limits, respectively. The fifth variable indicates whether a supermajority is required to enact a new revenue source. The sixth and seventh variables measure restrictions on local governmentÕs ability to raise revenues or to issue debt. Previous studies have related bond yields as reported in the Chubb survey to various state fiscal rules, although none of the previous studies have considered as broad a range of fiscal institutions, or as long a time span of bond yields, as the analysis reported here. Eichengreen (1992) and Goldstein and Woglom (1992) focus on single cross-sections of bond yields. Even with much smaller datasets than the panel dataset that underlies our study, they present evidence that borrowing costs are higher for states with weak anti-deficit fiscal rules. Bayoumi, Goldstein, and Woglom (1995) is most similar to our analysis. It is based on panel data for a period that ends before the state fiscal crises of the early 1990s, and it develops a careful model of how the outstanding stock of state debt affects borrowing costs. That study is also limited to anti-deficit rules, and it does not consider the effect of tax or expenditure limits. Our results are consistent with theirs. 46 We estimate regression equation (1) for the 40 states covered in the Chubb survey over the 1973Ð1995 period. The dependent variable is measured in basis points, which permits easy interpretation of the coefficients on the indicator variables for different fiscal rules. The coefficient estimates are reported in Table B.1. Unexpected Fiscal News and the Reaction of Bond Market Yields In our second specification, we compute the fiscal ÒsurpriseÓ that bond market participants learned about between two Chubb surveys 12 months apart. We label this variable DEFSHOCK, for deficit shock, but it can take positive or negative values depending on whether the fiscal news is unfavorable (higher deficit) or favorable (lower deficit). DEFSHOCK is measured in dollars, but we scale this variable to per capita terms and deflate to constant dollars for our regression analysis. We then compute the difference between the reported yields on a given stateÕs bonds in the two surveys. Given the way the survey data are collected, this is actually a Òdifference in differences.Ó Since each survey reports the difference between the yield on a stateÕs bonds and the yield on New JerseyÕs bonds, the difference between responses in two surveys is the change in the yield on a stateÕs bonds, minus the change in the yield on New JerseyÕs bonds. The dependent variable is therefore D()RR it jt-. DEFSHOCK is differenced from New JerseyÕs unexpected per capita deficit. In addition to DEFSHOCK, we include the change in the stateÕs unemployment rate between the two Chubb surveys relative to the change in New JerseyÕs unemployment rate, and the change in each stateÕs per capita outstanding debt relative to the change in New JerseyÕs 47 Table B.1 Regression Estimates of How Fiscal Institutions Affect State Borrowing Costs, 1973Ð1995 Explanatory Variable1973Ð 1995 Sample Outstanding debt to personal income 1.93 (1.20) Unemployment rate 5.60 (0.57) Per capita income 1.21 (0.52) State revenue/personal income Ð0.23 (0.12) Top marginal tax rate on interest income 0.47 (0.38) ADA score 4.41 (4.96) Weak anti-deficit rules 8.99 (6.58) Limit on legislatureÕs power to issue long-term debt Ð4.72 (5.51) Binding expenditure limit Ð3.65 (2.12) Binding revenue limit 17.45 (5.33) Supermajority required to pass new taxes 3.15 (3.94) Limit on local revenues (Proposition-13-type limits) 4.56 (2.54) Supermajority required to pass local bonds Ð4.04 (3.00) Adjusted R2 0.42 NOTE: Values in parentheses are standard errors. 48 debt level. These variables should capture the fact that state economic conditions are an important determinant of general obligation bond yields. Our principal hypothesis is that the effect of unexpected fiscal news on bond yields is mediated by the structure of the stateÕs fiscal institutions. In addition to the actual DEFSHOCK variable, we therefore include interactions between DEFSHOCK and the seven fiscal institutions described above. The coefficients on the interaction terms measure the effect of a stateÕs fiscal institution on the way its borrowing costs respond to fiscal news, normalized by how similar fiscal rules in New Jersey affect the response of that stateÕs bond yield to similar news. The actual estimating equation is therefore: DD D DD DD ()( )* ()*( )* ( * *)* R R DEBTSTOCK DEBTSTOCK UNEMP UNEMP DEFSHOCK DEFSHOCK DEFSHOCK Z DEFSHOCK Z it jt it jt it jt it jt it it jt jt it jt-= - +- + -+ -++-a a a aa e e1 2 3 45 (2) In this case, a 4 is a vector consisting of seven distinct coefficients on the seven fiscal institutions. Conventional F-tests allow us to reject the null hypothesis that these coefficients are jointly zero at very high confidence levels. Equation (2) is estimated on annual data for 1988Ð1997. Although we have nearly 15 years of prior data on the dependent variable, we cannot construct the DEFSHOCK variable before 1988, because that is when the National Association of State Budget Officers began collecting the full range of fiscal data that our DEFSHOCK calculations require. Table B.2 reports our estimates of equation (2). 49 Table B.2 Regression Estimates of How Fiscal Institutions Affect Bond Market Reaction to State Fiscal News, 1988Ð1997 Explanatory Variable Estimated Coefficients Change in outstanding debt 0.0024 (0.0014)0.0022 (0.0014) Change in unemployment rate 1.4870 (0.3057)1.4754 (0.3148) DEFSHOCK Ð0.0210 (0.0178)Ð0.0096 (0.0168) DEFSHOCK*(weak anti-deficit rules) 0.0608 (0.0214)0.0521 (0.0224) DEFSHOCK*(limit on legislatureÕs power to issue long-term debt)0.0299 (0.0186)0.0215 (0.0176) DEFSHOCK*(binding expenditure limit) Ð0.0006 (0.0062)Ð0.0033 (0.0062) DEFSHOCK*(binding revenue limit) 0.0518 (0.0222)0.0501 (0.0226) DEFSHOCK* (supermajority required to pass new taxes)0.0557 (0.0220) 0.0357 (0.0224) DEFSHOCK* (limit on local revenues) (Proposition-13-type limits)Ð0.0101 (0.0076)Ð0.0095 (0.0079) DEFSHOCK* (supermajority required to pass local bonds )0.0161 (0.0193)Ð0.0052 (0.0123) DEFSHOCK* (California) Ñ 0.1165 (0.0276) Adjusted R2 0.2963 0.3089 NOTES: Values in parentheses are standard errors. All equations are estimated for 1988Ð1997 using data for the 40 states included in the Chubb survey. Estimates are given for a state response to a per capita deficit. Unexpected deficits are expressed as positive values in the regression results. 50 Unlike equation (1), which resembles the estimating equations in several previous studies, only one study has explored how bond yields change in reaction to fiscal developments. Lowry and Alt (1997) find that actual deficits have a larger positive effect on borrowing costs when states have weaker anti-deficit rules. Our findings based on fiscal ÒnewsÓ rather that the level of fiscal variables are consistent with this result. 51 References Advisory Council on Intergovernmental Relations, Fiscal Discipline in the Federal System: National Reform and the Experience of the States, Washington, D.C.: Advisory Council on Intergovernmental Relations, 1987a. Advisory Council on Intergovernmental Relations, Significant Features of Fiscal Federalism: 1988. Volume 1, Budget Processes and Tax Systems, Washington, D.C.: Advisory Council on Intergovernmental Relations, 1987b. Advisory Council on Intergovernmental Relations, State Laws Governing Local Government Structure and Administration, Washington, D.C.: Advisory Council on Intergovernmental Relations, 1993. Advisory Council on Intergovernmental Relations, A Profile of Tax and Expenditure Limits in the Fifty States, Washington, D.C.: Advisory Council on Intergovernmental Relations, 1997. Advisory Council on Intergovernmental Relations, Significant Features of Fiscal Federalism: 1995. Volume 1, Budget Processes and Tax Systems and Volume 2: Revenues and Expenditures, Washington, D.C.: Advisory Council on Intergovernmental Relations, 1998. 52 Alt, James E., and Robert C. Lowry, ÒDivided Government, Fiscal Institutions, and Budget Deficits: Evidence from the States,Ó American Political Science Review, Vol. 88, 1994, pp. 811Ð828. Bayoumi, Tamim, Morris Goldstein, and Geoffrey Woglom, ÒDo Credit Markets Discipline Sovereign Borrowers: Evidence from U.S. States,Ó Journal of Money Credit and Banking, Vol. 27, 1995, pp. 1046Ð1059. Bohn, Henning, and Robert P. Inman, ÒConstitutional Limits and Public Deficits: Evidence from the U.S. States,Ó Carnegie - Rochester Conference Series on Public Policy, December 1995. Briffault, Richard, Balancing Acts: The Reality Behind State Balanced Budget Requirements, New York: Twentieth Century Fund Press, 1996. Bunche, Beverly S., ÒThe Effect of Constitutional Debt Limits on State GovernmentsÕ Use of Public Authorities,Ó Public Choice, Vol. 68, 1991, pp. 57Ð69. Bureau of Labor Statistics, ÒLocal Area Unemployment StatisticsÓ Washington, D.C., 1999, http://stats.bls.gov/cgi-bin/surveymost. California Business Roundtable, Building a Legacy for the Next Generation, Sacramento, California, 1999. California State Treasurer, Smart Investments: Special Edition of CaliforniaÕs Debt Affordability Report, Sacramento, California, 1999. Davey, C., ÒPlan to Make Tax Increase Tougher Gets Cool Reception in Senate,Ó Cincinnati Enquirer, September 24, 1995, p. C4. Eichengreen, Barry, Should the Maastricht Treaty Be Saved? Princeton Studies in International Finance No. 74, Princeton, New Jersey: Princeton University, 1992. Goldstein, Morris, and Geoffrey Woglom, ÒMarket-Based Fiscal Discipline in Monetary Unions: Evidence from the U.S. Municipal Bond Market,Ó in M. B. Canzoneri, V. Grilli, and P. R. Masson, eds., Establishing a Central Bank: Issues in Europe and Lessons from the United States, Cambridge: Cambridge University Press, 1992. 53 Inman, Robert, ÒDo Balanced Budget Rules Work? U.S. Experience and Possible Lessons for the EMU,Ó NBER Working Paper 5838, Cambridge, Massachusetts, 1996. Kiewiet, D. Roderick, and Kristin Szakaly, ÒThe Efficacy of Constitutional Restrictions on Borrowing, Taxing, and Spending: An Analysis of State Bonded Indebtedness, 1961Ð90,Ó Journal of Law, Economics, and Organization, Vol. 12, 1996, pp. 62Ð97. Knight, Brian, ÒSupermajority Voting Requirements for Tax Increases: Evidence from the States,Ó mimeo, Madison, Wisconsin: University of Wisconsin Department of Economics, 1999. Legislative AnalystÕs Office, CaliforniaÕs Fiscal Outlook: The LAOÕs Economic and Budget Projections, Sacramento, California, 1998. Lowry, Robert C., and James E. Alt, ÒA Visible Hand? Bond Markets, Political Parties, Balanced Budget Laws, and State Government Debt,Ó mimeo, Cambridge, Massachusetts: Harvard University Department of Government, 1997. National Association of State Budget Officers, Fiscal Survey of the States, Washington, D.C., 1986Ð1997. Poterba, James M., ÒState Responses to Fiscal Crises: The Effects of Budgetary Institutions and Politics,Ó Journal of Political Economy, Vol. 102, 1994, pp. 799Ð821. Poterba, James M., ÒDo Budget Rules Work?,Ó in A. Auerbach, ed., Fiscal Policy: Lessons from Economic Research, Cambridge, Massachusetts: MIT Press, 1997, pp. 53Ð86. Poterba, James, and Kim Rueben, ÒFiscal News, State Budget Rules, and Tax-Exempt Bond Yields,Ó mimeo, Cambridge, Massachusetts: MIT Department of Economics, 1998. Poterba, James, and Kim Rueben, ÒState Fiscal Institutions and the Municipal Bond Market,Ó in J. Poterba and J. von Hagen, eds., Fiscal Institutions and Fiscal Performance, Chicago, Illinois: University of Chicago Press, 1999. 54 Rafool, Mandy, ÒState Tax and Expenditure Limits,Ó State Tax Notes, Washington, D.C., January 13, 1997, pp. 115Ð132. Rosenblum, Joseph, The State of California: Update March 1999, New York: Sanford Bernstein Municipal Bond Research, 1999. Rueben, Kim S., ÒTax Limitations and Government Growth: The Effect of State Tax and Expenditure Limits on State and Local Government,Ó mimeo, Public Policy Institute of California, 1996. State of California, GovernorÕs Proposed Budget Highlights 1999Ð2000: Sacramento, California, 1999. U.S. Census Bureau, ÒState Government FinancesÓ Washington, D.C., 1999, http://www.census.gov/govs/www/state.html. 55 About the Authors JAMES M. POTERBA James M. Poterba is a professor of economics at the Massachusetts Institute of Technology who specializes in the analysis of taxation and government expenditure programs. His past research has examined the way state governments respond to unexpected fiscal windfalls and fiscal crises, and it has also considered the factors that determine interest rates in the tax-exempt bond market. He holds a B.A. from Harvard College and a D.Phil. from Oxford University. KIM S. RUEBEN Kim S. Rueben is an economist who is interested in the relationship between fiscal institutions and state and local government budgetary decisions and the relationship between institutional structure and educational resources. She has examined how state tax and expenditure limits have affected government finances, including the effects of budgetary institutions on municipal bonds and public employee salaries. Currently, she is investigating how limitation laws affect student outcomes and is also examining which local governments are successful in passing new taxes when faced with the need for voter approval. She holds a Ph.D. in economics from the Massachusetts Institute of Technology." } ["___content":protected]=> string(104) "

R 1299JPR

" ["_permalink":protected]=> string(124) "https://www.ppic.org/publication/fiscal-rules-and-state-borrowing-costs-evidence-from-california-and-other-states/r_1299jpr/" ["_next":protected]=> array(0) { } ["_prev":protected]=> array(0) { } ["_css_class":protected]=> NULL ["id"]=> int(8079) ["ID"]=> int(8079) ["post_author"]=> string(1) "1" ["post_content"]=> string(0) "" ["post_date"]=> string(19) "2017-05-20 02:34:43" ["post_excerpt"]=> string(0) "" ["post_parent"]=> int(3164) ["post_status"]=> string(7) "inherit" ["post_title"]=> string(9) "R 1299JPR" ["post_type"]=> string(10) "attachment" ["slug"]=> string(9) "r_1299jpr" ["__type":protected]=> NULL ["_wp_attached_file"]=> string(13) "R_1299JPR.pdf" ["wpmf_size"]=> string(6) "158203" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(83320) "Fiscal Rules and State Borrowing Costs: Evidence from California and Other States ¥¥¥ James M. Poterba Kim S. Rueben 1999 PUBLIC POLICY INSTITUTE OF CALIFORNIA Library of Congress Cataloging-in-Publication Data Poterba, James M. Fiscal rules and state borrowing costs : evidence from California and other states / James M. Poterba, Kim S. Rueben. p. cm. Includes bibliographical references. ISBN: 1-58213-019-1 1. State bondsÑCalifornia. 2. State bondsÑUnited StatesÑ States. 3. Debts, PublicÑCalifornia. 4. Debts, PublicÑUnited StatesÑStates. 5. Interest ratesÑCalifornia. 6. Interest ratesÑ United StatesÑStates. I. Rueben, Kim S. II. Title HG4948.C2 P67 1999 336.3'43273Ñdc21 99-051905 CIP 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. Copyright © 1999 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. iii Foreword A generation of California voters and legislators has fashioned a unique system for financing government services. In doing so, they have also spawned scores of unintended consequences. Proposition 13, for example, both altered the face of public finance and generated many unforeseen outcomes. Instead of echoing conventional wisdom, several PPIC reports have described those consequences objectively. But whether policy debate trades in clichŽs, calculations, or some mixture of the two, one point is irrefutableÑCalifornia has made extensive use of the initiative process to finance and constrain state government, and we are still gauging the effects of that development. In Fiscal Rules and State Borrowing Costs, James Poterba and Kim Rueben show how some of these unintended consequences can be measured. Marshalling evidence from municipal bond markets around the nation, the authors maintain that fiscal constraints on state government affect borrowing costs in ways that are predictable but rarely acknowledged. In particular, they show that borrowing costs are iv sensitive not only to overall measures of economic health, such as unemployment rates, but also to a stateÕs fiscal rules and budget forecasts. Although some of the findings may seem counterintuitive at first, the economic reasoning is sound. Revenue limits and supermajority requirements for new taxesÑthe latter a legacy of Proposition 13Ñtend to increase borrowing costs because they hamper the stateÕs perceived ability to pay its long-term debt. In contrast, expenditure limits such as Proposition 4 lower borrowing costs because they make it easier for the state to service its debt. The authors also find that states with unexpected deficits pay more for long-term financing, and they conclude the report by measuring the costs of inaccurate state budget forecasts. These findings have immediate implications for the fiscal limits movement in California. Although this movement rarely distinguishes between tax and expenditure limits, the bond market has scaled this distinction precisely. The authors calculate that a tax limit costs the state approximately $2 million more per $1 billion of debt than a spending limit. This is not a negligible cost, even in a state as rich as California, and it indicates once again that reinventing public finance brings with it more than a few unintended consequences. In this study, however, those consequences are estimated carefully rather than asserted tendentiously. David W. Lyon President and CEO Public Policy Institute of California v Summary The vast majority of CaliforniaÕs spending on capital infrastructure, such as highways, aqueducts, prisons, and related facilities, is financed with long-term debt. Recent projections of such spending over the next decade suggest that the state will need to borrow $25 billion to $50 billion during this period. The interest rate on CaliforniaÕs debt will be an important factor in determining the overall fiscal burden of this infrastructure spending. That rate, which differs over time and from state to state, is determined by a number of factors, some of which are matters of public policy. This report investigates interstate differences in borrowing rates on general obligation bonds. It begins by presenting new evidence on how state economic conditions affect borrowing costs. By analyzing yields on tax-exempt bonds issued by different states between 1973 and 1997, the study reveals a clear relationship between a stateÕs general economic health, as measured by unemployment rates, and state borrowing costs. A 1 percent increase in a stateÕs unemployment rate is associated with an vi increase of about 0.05 percent, or five basis points, in that stateÕs bond yields. (A basis point is one one-hundredth of 1 percent.) In California, for example, the stateÕs borrowing rate rose when the state economy was weak in the early 1990s and has declined for the last several years as the economy has improved. More important from a public policy perspective, perhaps, is the finding that statesÕ fiscal rules also play an important role in determining statesÕ borrowing costs. We present three major findings in this regard. First, states with strict fiscal rules on government spending or deficits have faced lower borrowing costs during the last two decades than those with looser fiscal rules. We calculate that a state with a strict anti-deficit fiscal constitution pays about nine basis points less to issue new debt interest. Second, the bond market reacts in different ways to revenue restrictions and expenditure limits. States with expenditure limits typically borrow at lower rates than other states, but those that restrict tax increases or require supermajorities to increase taxes face higher borrowing costs. States with binding revenue limitation laws are likely to face borrowing rates more than seventeen basis points higher than those in other states. This ratio translates into an extra $1,750 in interest payments per million dollars of debt issued. Limits on the ability of local governments to increase taxes or issue debt also seem to raise state borrowing costs. The empirical findings in this study suggest that CaliforniaÕs current fiscal rules raise the interest rate on state and possibly local debt. Since the passage of Proposition 13 in 1978, a legislative supermajorityÑin this case, a two-thirds vote in both housesÑhas been required to enact new taxes. Although the Gann amendment (Proposition 4) of 1979 does vii not restrict state revenues, it limits state and local expenditure growth. More recently, Proposition 218, which was adopted in 1996, provides new mechanisms for voters to restrict local revenues. Although states with supermajority provisions face higher borrowing costs than states without such requirements, explicit limits on state revenues and stronger anti-tax provisions than those in California have an even larger effect on borrowing costs. Thus, CaliforniaÕs tax limitation strategy has probably raised interest costs to the state but by less than other tax limitation strategies might have. Our figures indicate that if California had enacted a tax limitation law for state revenues rather than an expenditure limit in 1979, borrowing costs would have been about 20 basis points higher on average during the subsequent two decades. This difference amounts to about $2,000 extra in interest payments per million dollars of debt issued. This increase represents a small but not trivial fraction of the state budget. In 1997, for example, California had outstanding state debt of $43.5 billion. If the state had enacted a restrictive revenue limit rather than an expenditure limit, the extra interest on this debt would have been approximately $90 million. Our third major finding concerns unexpected state budget surpluses or deficits and how they affect borrowing costs. Not surprisingly, unexpected state budget deficits are correlated with upward revisions in state bond yields. It is more expensive for a state to issue new debt when it is experiencing budget difficulties, but bond markets are especially responsive to the fiscal health of states with large amounts of outstanding debt. This effect is particularly pronounced in California, where an unexpected $100 per capita increase in the stateÕs deficit has historically been associated with an increase of 14 basis points in borrowing costs. Thus, if California had undertaken new infrastructure projects in 1992, viii $1 billion of new debt would have cost the state $1.4 million more than it would today. Part of this responsiveness can be explained by differences in fiscal rules; bond yields rise less during periods of financial stress for states with tight anti-deficit rules or restrictive spending rules. In California, state fiscal rules can explain about two-thirds of this effect. This pattern indicates that accurate tax and expenditure forecasts are more important for California than for most other states, which municipal bond analysts follow less closely. As California begins a new round of infrastructure borrowing and spending, it is especially important to understand how the stateÕs fiscal rules affect the overall costs of repairs and construction. According to our findings, these fiscal rules affect the costs of state tax-exempt debt by the same amount as a significant shift in the unemployment rate. Millions of dollars 0 2.5 2.0 1.5 1.0 0.5If 3.5% rise in unemploymentIf state had revenue limit instead of Gann (expenditure) limitIf state had unexpected deficit of $100 per capita Figure S.1ÑCaliforniaÕs Predicted Additional Borrowing Cost per $1 Billion of New Debt ix CaliforniaÕs interest costs are also very sensitive to unexpected deficits. Our findings indicate that voters and legislators would do well to recognize the long-term effects on borrowing costs when considering tax limits, expenditure limits, or changes to the stateÕs deficit financing rules. In addition, the state should take into account the additional costs of issuing debt during times when economic conditions or other circumstances lead to unexpected deficits. xi Contents Foreword................ iii Summary........... v Figures........... xii Tables............ xv Acknowledgments........... xvi 1. INTRODUCTION......... 1 2. CURRENT FISCAL RULES ACROSS DIFFERENT STATES............ 5 3. STATE BORROWING COSTS IN THE LAST TWO DECADES............ 15 4. BORROWING COSTS, ECONOMIC CONDITIONS, AND FISCAL RULES............. 21 How Economic Conditions and Fiscal Rules Affect the Level of Borrowing Costs............ 22 Fiscal Shocks, Fiscal Rules, and the Bond Market Reaction to Fiscal News.............. 25 Interstate Differences in Bond Market Reactions to Fiscal News................. 32 5. CONCLUSIONS AND IMPLICATIONS..... 37 xii Appendix A. State Fiscal Institutions............. 41 B. Detailed Description of Regression Results.... 43 References............... 51 About the Authors............ 55 xiii Figures S.1. CaliforniaÕs Predicted Additional Borrowing Cost per $1 Billion of New Debt............. viii 1. Balanced Budget Requirements....... 7 2. States with Debt Restriction Provisions...... 8 3. Year Local Tax Limit Passed....... 13 4. Relative Bond Yield Required on California State Debt.. 17 5. Relative Bond Yield Required for Various StatesÕ 20 Year General Obligation Bonds......... 18 6. Unexpected State Deficits as a Percentage of General Revenue, 1988Ð1997............ 27 7. Unexpected State Deficits as a Percentage of General Revenue, Selected States, 1988Ð1997....... 29 xv Tables 1. History of State Spending Limit Adoptions....... 10 2. History of State Revenue Limit Adoptions...... 10 3. Supermajority Requirements and Other Constitutional Restrictions on Legislative Tax Power....... 12 4. Effect of Fiscal Institutions on State Borrowing Costs.. 24 5. Average State Deficit Shock, Selected States, 1988Ð 1997................... 28 6. Effect of State Fiscal Rules on the Bond MarketÕs Reaction to Unexpected Budget Deficits and Budget Surpluses................. 30 7. Interstate Variability in Effect of Fiscal News on Borrowing Costs.............. 33 B.1. Regression Estimates of How Fiscal Institutions Affect State Borrowing Costs, 1973Ð1995........ 47 B.2. Regression Estimates of How Fiscal Institutions Affect Bond Market Reaction to State Fiscal News, 1988 Ð 1997.................... 49 xvii Acknowledgments We would like to thank a number of individuals who have helped us with this research project. First, we would like to thank the Chubb Corporation and especially Thomas Swartz III for access to the Relative Value Study data on which this report is based. Second, we would like to thank those who helped us with the analysis. Elizabeth Berko provided early help with data assembly, and Daniel Frakes produced the maps in this report. Third, we are grateful to James Alt, Mark Baldassare, Michael Dardia, Robert Iman, Peter Schaafsma, and Michael Teitz for their thoughtful reviews of this study. JŸergen von Hagen, Anne Case, and the other participants in a 1997 conference at the University of BonnÕs Zentrum fŸr EuropŠische Integrationsforschung provided insightful comments on an earlier paper that is closely related to this study. Finally, we wish to thank Peter Richardson of PPIC, who provided invaluable editorial assistance in completing the monograph. The authors retain responsibility for any errors of fact or interpretation. 1 1. Introduction In his proposed 1999Ð2000 budget, Governor Gray Davis established the Commission on Building for the 21st Century and called attention to the stateÕs low level of infrastructure spending over the past decade. His remarks to the commission (State of California, 1999) underscored the growing importance of capital spending as a public policy issue in California: As California moves into the 21st Century we face the dual problem of preserving the schools, highways, bridges, water systems, and housing of today, while also planning and building new facilities for a growing population. There is no choice: we must maintain our current capital investmentsÑ ÒinfrastructureÓÑand make new investments. Governor Davis charged the commission with determining how much the state should spend on infrastructure repair in the coming years and how this spending should be financed. The California Business Roundtable has also called for substantial increases in state infrastructure spending in the near future. In a special edition of CaliforniaÕs Debt Affordability Report (California State Treasurer, 1999), the state treasurer has also called for a comprehensive evaluation of infrastructure 2 and investment needs. The prospect of significant increases in future state borrowing suggests that debt policy is an increasingly important issue. The Department of Finance has identified $82.2 billion of state and state-funded local infrastructure needs over the next decade. About half of this total can be paid for out of existing bond funds and anticipated federal funds, but more than $40 billion of infrastructure spending will need to be financed with new debt. If the state maintains its current ratio of debt service to general fund revenues, the stateÕs debt capacity over the next ten years will be $32.5 billion, according to Treasurer report estimates. If spending on debt service increases from a little over 4 percent to 5 percent, the amount of debt issued could increase by $10 billion and will cover the currently identified needs of the state. CaliforniaÕs interest burden is currently lower than that of the typical state. The Census of Governments reports that although the stateÕs general obligation debt at the end of fiscal year 1997 was $45.3 billion, the second largest in the nation, per capita indebtedness was $1,405, or slightly below the national average. The Advisory Council on Intergovernmental Relations (ACIR) (1998) reports that in fiscal year 1995, CaliforniaÕs state interest burden was $76 per capita, compared with a national average of $82. These differences are not due to a smaller government sector in California than in other states. In fiscal year 1997, for example, per capita state spending in California was $3,632, compared with a national average of $3,345. Borrowing costs on state debt are determined by a variety of factors, the most dramatic of which are repayment problems or other fiscal crises. At the height of the Massachusetts fiscal crisis in 1990, for example, yields on its general obligation bonds were more than 1 percent higher 3 than CaliforniaÕs. Similarly, when New York City appeared to be on the verge of bankruptcy in 1976, yields on New York state bonds were nearly 2 percent higher than CaliforniaÕs. Although such crises account for significant differences in interstate borrowing costs, less dramatic factors contribute to these disparities as well. One such factor is the set of fiscal rules that governs a stateÕs ability to raise and spend revenue. In some states, for example, the state constitution requires a prompt tax increase or spending cut to counter a budget deficit. In other states, governors and legislatures may use short-term debt financing to cover revenue shortfalls. Disparities in these fiscal rules affect a stateÕs perceived credit risk and therefore its borrowing costs. This study explores how these fiscal rules, and the way they interact with state economic conditions, affect the cost of issuing general obligation debt. The results are interpreted with particular reference to California. Chapter 2 describes the fiscal institutions in place in California and the rest of the country that are the primary focus of this study. Chapter 3 reports summary information on state borrowing costs, in California and other states, for the three decades that provide the analytical basis for this study. Chapter 4 summarizes the empirical evidence that links state economic conditions, fiscal rules, and fiscal health to state borrowing costs. More detailed information on the fiscal institutions and the statistical analysis is presented in appendices. Finally, a brief conclusion highlights the findings and their implications for public policy. 5 2. Current Fiscal Rules Across Different States Although state fiscal constitutions vary along many dimensions, our analysis of fiscal rules focuses on seven characteristics. 1 The first is the extent to which a state must have a balanced budget at different points in the legislative process. Only Vermont does not have a formal balanced budget requirement. The balanced budget requirements of the other 49 states can be divided into four groups. In 44 states, including California, the governor must submit a balanced budget. This is the weakest of the balanced budget requirements, as it does not actually require that the state enact a budget that matches expenditures and revenues. In 37 states, including California, the legislature must enact a balanced budget. Even in these states, however, actual revenues and expenditures may diverge if there are unexpected fiscal shocks after the budget is signed into law. When an unexpected deficit develops during _________ 1Detailed information on state budget rules may be found in the annual reports of the ACIR (for example, 1988 and 1998), or in Briffault (1996). 6 the fiscal year, six states require that the governor and legislature correct the deficit in the next budget cycle. Because budget cycles in some states are biennial, this requirement permits substantial periods of budget deficits. Twenty-four of the 37 states with balanced budget requirements prohibit the government from carrying deficits into the next budget cycle. This provision represents the strictest anti-deficit rule, as it requires the legislature either to cut spending, or raise taxes in the fiscal year when the deficit emerges, or to float short-term debt that will be retired in the next fiscal year. Previous studies, including Alt and Lowry (1994), Bohn and Inman (1995), and Poterba (1994), find that such anti-deficit rules are generally correlated with lower average deficits and more rapid adjustments to budget shortfalls. In 1987, the ACIR constructed an index that characterizes fiscal discipline among state governments. This index, which is presented in detail in Appendix A and summarized in Figure 1, ranges from 0 (lax) to 10 (stringent). California receives a score of 6 on this scale, in part because the state constitution permits the use of short-term borrowing to cover a deficit. Because it is difficult to interpret a one-unit change in this index, we have adapted it for the purposes of this study. Our statistical analysis contrasts states for which the index value is 5 or below with those for which the value is 6 or above. States with scores of 5 or below cannot have any stronger anti-deficit rules than a requirement that the governor present a balanced budget. Only eight states receive ACIR scores of 5 or below, whereas 26 receive a score of 10. States in the Northeast and the upper Midwest are less likely to have stringent anti- deficit requirements than states in other regions. Outside of those regions, many states with relatively weak budget rules have other fiscal constraints on state revenue or expenditures; for example, California, 7 0: No balanced budget rules 1–5: Proposed budget balances 6–8: Can borrow to balance budget 9–10: Ending budget must balance Figure 1ÑBalanced Budget Requirements Nevada, and Louisiana have passed limits on how quickly expenditures can grow over time. These revenue and expenditure limits are discussed in more detail below. The second fiscal characteristic we consider is the ease with which states issue long-term general obligation debt. Twelve states do not restrict debt issuance. Thirty-eight states have constitutional restrictions on debt issue, and two have legislative limits. The most common type of restriction limits the amount of debt outstanding. Fourteen states, including California, permit voters to override constitutional restrictions on debt levels to issue additional debt. We do not attempt to distinguish 8 among the various limits on debt issue; rather, we consider whether a state has any restriction, including requiring voter approval, on issuing general obligation debt. We use this definition to study the effects of borrowing limits on borrowing costs. Figure 2 (and Column 2 in Appendix A) shows the continental states that have borrowing limits. As with anti-deficit rules, the previous literature suggests that limits on long- term debt issues affect fiscal outcomes. Bunche (1991) finds that states with restrictions on general obligation debt are more likely to rely on revenue bonds to finance a range of government projects. Kiewiet and Szakaly (1996) also find that these rules affect state borrowing behavior. No provision Partial restrictiona Full restriction aDebt limit can be exceeded with voter approval. Figure 2ÑStates with Debt Restriction Provisions 9 However, previous studies have not considered the link between these institutions and yields on state general obligation debt. The third and fourth fiscal characteristics are tax and expenditure limits at the state level. The most common type of limit restricts the growth rate of general fund expenditures or revenues to the growth rate of personal income or to the growth rate of population and inflation. Some of these laws place binding constraints on governors and legislatures. Rueben (1996) provides information on override provisions and indicates which limits cannot be overridden by a simple legislative majority. Tax and expenditure limits have changed since the late 1970s, when most current limits were enacted. Several states enacted new limits or strengthened earlier limits in the early 1990s. In 1979, for example, Louisiana passed a law mandating that the ratio of tax revenue to personal income in any year could not exceed the 1979 ratio; however, a simple legislative majority could override the measure. In 1993, voters passed a referendum that limited the growth in appropriations. Unlike the earlier law, this one required a two-thirds vote to override the limit. Similarly, ColoradoÕs expenditure limit in 1978 was supplemented with Amendment 1, or the Taxpayers Bill of Rights, in 1992. The latter initiative limits increases in state and local spending to the change in population and increases in the Consumer Price Index. In addition, it requires voter approval for the implementation of new taxes. In 1980, California voters approved Proposition 4, the Gann amendment, which limits the growth in both state and local expenditures to changes in the Consumer Price Index and population. As Tables 1 and 2 indicate, 24 states have adopted some form of state tax or expenditure limitation since 1976. Twenty-one states currently have an expenditure limit in place, and 19 states require at least a supermajority of both houses of the 10 Table 1 History of State Spending Limit Adoptions 1976Ð1979 1980Ð1983 1984Ð1988 1989Ð1992 1993Ð1997 UtahAlaskaOklahomaColorado Nevada NevadaMontana Rhode Island Washington Oregon South Carolina Connecticut Louisiana California Missouri North Carolina TexasIdaho New Jersey TennesseeUtah Hawaii Arizona Colorado Rhode Island New Jersey NOTES: Binding limits are in bold. Other limits are advisory or can be overridden by a legislative majority. New JerseyÕs original expenditure limit expired in 1976. Table 2 History of State Revenue Limit Adoptions 1976Ð1979 1980Ð1983 1984Ð1988 1989Ð1992 1993Ð1997 Louisiana MissouriMassachusetts Colorado Missouri Washington Louisiana Florida Michigan NOTE: Binding limits are in bold. Other limits are advisory or can be overridden by a legislative majority. legislature to override spending limits. Seven states have passed a limit on revenues, five of which were enacted after 1986. Because tax and expenditure limits restrain the growth of government spending, they are often classified together. However, the tax-exempt bond market makes a strict distinction between them. Tax limits increase the risk that revenues will not cover future interest payments; they therefore increase the risk of state general obligation debt. In contrast, expenditure limits increase the likelihood that legislatures 11 will make interest payments. The empirical results presented below support this dichotomy. The fifth fiscal characteristic we consider is the supermajority requirement to raise specific taxes. Unlike revenue limits, supermajority requirements do not cap existing taxes, such as sales and income taxes, which often rise during economic upswings. Fifteen states, including California, currently have a supermajority requirement for new taxes. Historically, supermajority requirements have been concentrated in the South. More recently, however, other states have passed supermajority requirements as part of initiatives to limit government revenues. California passed a two-thirds majority requirement in 1978 as part of Proposition 13. Montana and Missouri have even stricter provisions concerning new taxes. In a budgetary emergency, the legislature in each state can pass new temporary taxes with a supermajority. However, voter approval is required to make these emergency taxes permanent. Table 3 presents information on states with supermajority requirements. Because supermajority requirements and revenue limits restrict the governmentÕs ability to raise revenue, we expect these constraints to affect bond yields in similar ways. The last two fiscal characteristics considered here are limits on local governments, specifically, the extent to which states restrict the ability of these governments to raise revenues and issue new debt. Although local restrictions do not directly affect a state governmentÕs fiscal health, they can have indirect effects insofar as they lead state governments to take over traditional local government functions. Rueben (1996) shows that states with local limits raise a higher percentage of revenues centrally than states without such limits. For example, Proposition 1A, which was enacted in California in November 1998, permits the state to borrow 12 Table 3 Supermajority Requirements and Other Constitutional Restrictions on Legislative Tax Power Ê Ê Referendum Legislative Ê Ê or Voter Majority Ê State Adopted Initiative Required Applies To Arizona 1992 Initiative 2/3 All taxes Arkansas 1934 Referendum 3/4 All taxes except sales and alcohol California 1978 Initiative 2/3 All taxes Colorado 1992 Initiative 2/3 All taxes Delaware 1980 Referendum 3/5 All taxes Florida 1971 Referendum 3/5 Corporate income tax Florida 1996 Initiative 2/3 New taxes Louisiana 1966 Referendum 2/3 All taxes Mississippi 1970 Referendum 3/5 All taxes Missouri 1996 Referendum 2/3 Emergency taxes Montana 1998 Initiative 3/4 Emergency taxes Nevada 1996 Initiative 2/3 All taxes Oklahoma 1992 Initiative 3/4 All taxes Oregon 1996 Referendum 3/5 All taxes South Dakota 1996 Referendum 2/3 All taxes Washington 1993 Initiative 2/3 All taxes NOTES: Missouri and Montana can pass temporary new taxes in an emergency. However, if either state wants to make emergency tax increases permanent, it must get voter approval. $9.2 billion for local education facilities, providing a means of financing school construction in districts that are unable to gain the supermajority required in local elections. Figure 3 shows which states have local revenue limits. Many states restrict local tax rates or assessments. CaliforniaÕs Proposition 13 is perhaps the best known of such restrictions. Our analysis considers both state-imposed limits on local revenue levels and state restrictions on both the assessed valuation of local property and the tax rate that can be 13       No limit Before 1976 1976 to 1981 1982 to 1987 Since 1987    Figure 3ÑYear Local Tax Limit Passed imposed. In addition to Proposition 13 (and Proposition 4, which limits expenditures), California has two other restrictions on local government finance: Proposition 62, which limits general taxes at the local level, and Proposition 218, which confirms the limit on general taxes and also limits local fees and assessments. Virtually all states limit localitiesÕ ability to issue municipal debt. According to a 1993 ACIR report, only Florida and Tennessee do not place any debt limits on cities, but these states require voter approval for debt over a certain amount. The most typical requirement calls for voter approval before municipalities can issue new debt. Forty states also limit counties in this regard as well. Because all states restrict the issuance of 14 local bonds, we focus on the eight states that require a supermajority to authorize new local debt. Appendix A presents information on the seven state fiscal institutions described above. It reports the actual ACIR index for each stateÕs anti - deficit program, indicates which states limit the amount of outstanding debt, and shows whether this limit can be overridden with consent of the voters. It also gives the year of passage for the more recent fiscal constraints passed and information on the kind of supermajority required. California has a unique combination of fiscal rules. The actual state budget need not be balanced, but there are limits on the amount of state debt that can be issued without voter approval. California also has a state expenditure limit and requires supermajorities to introduce new taxes or to increase existing ones. Finally, CaliforniaÕs local governments are more fiscally constrained than those in other states. Using the initiative system, CaliforniaÕs voters have enacted a number of restrictions on local governments; indeed, only OregonÕs voters have used the initiative system more. Property taxes are limited and local governments are required to receive voter approval for most other revenue sources. Finally, California is one of only eight states that require a supermajority of voters to issue local general obligation debt. The supermajority requirement for local debt is the only limit that was not passed recently using the initiative system. 15 3. State Borrowing Costs in the Last Two Decades The market value of outstanding tax-exempt bonds is roughly 40 percent as large as that of corporate bonds and roughly one-fifth the value of U.S. Treasury bonds. Although the bond market is well organized, there are three practical obstacles to obtaining information on the borrowing costs that different states pay at a given time or on those paid by a given state over time. First, there is limited trading in most tax-exempt bond issues. Second, tax-exempt bonds differ widely in their call provisions and in other detailed provisions. Finally, many tax - exempt bonds are sold in bundles, making it difficult to identify the price and yield to maturity of any single bond. For these reasons, it is not generally possible to obtain price quotes for a large and comparable sample of state general obligation bonds over any substantial period of time. In the absence of market data on state government borrowing costs, we have turned to the Chubb Insurance Company ÒRelative Value 16 Survey.Ó This survey, which has been carried out every six months since 1973, asks approximately 25 traders of tax-exempt bonds to estimate the current yields on general obligation bonds from 40 states. (Ten states with little borrowing activity are excluded.) Chubb survey respondents are asked to estimate the current yield on ÒhypotheticalÓ 20-year general obligation (GO) bonds. By asking about hypothetical bonds, the survey avoids problems related to call protection and other specific details associated with individual bonds traded in the marketplace. The Chubb survey collects information on the relative yields on the GO bonds issued by various states and by New Jersey, which serves as the comparison state of the study. The Chubb survey data show differences in borrowing costs in two ways: across states at a given point in time and over time for a given pair of states. Figure 4 presents summary information on the differences in borrowing costs across states, as well as on relative borrowing costs in California and other states. The top line represents the highest yield and the bottom line represents the lowest yield at each date. A state with a higher required yield pays more in interest payments to issue a given amount of debt. The spread between the maximum and minimum yield was less than 30 basis points in 1997; as recently as 1988, however, the yield spread between the highest yield and lowest yield borrowers was more than 120 basis points. The bold line in Figure 4 represents the relative yield on CaliforniaÕs general obligation bonds in comparison to similar bonds issued by New Jersey. The last line represents the average required yield for all other states (compared to New Jersey) in the Chubb survey. CaliforniaÕs borrowing costs follow the stateÕs economic fortunes; these costs were below the national average in the late 1980s, when the California 17 Interest required above New Jersey bond (basis points) –100150 50 0 –5073 75 77 89 9593 91 87 85 83 81 79 97 Year 100 California Other state average Highest yield required Lowest yield required Figure 4ÑRelative Bond Yield Required on California State Debt economy boomed, and rose above the national average during the 1990s. In fact, CaliforniaÕs borrowing costs were among the highest in the nation in the mid 1990s. Figure 5 shows the relative yields on the general obligation bonds of California, Massachusetts, New York, and Texas between 1974 and 1997. The figure shows the historical volatility in each stateÕs bond yield. During the New York City fiscal crisis in 1976, Chubb survey respondents estimated that bonds issued by New York or Massachusetts would yield roughly 150 basis points (one and one-half percentage points) more than bonds issued by California or Texas. This yield spread declined throughout the 1980s and widened once again around 1990, another period of fiscal stress for Northeastern states. Recent economic 18 strength in all of the states shown in Figure 5 contributes to the relatively small yield disparities at the end of the data sample. To help illustrate the relative differences in borrowing costs, consider the relative costs of issuing $1 million of debt in California, New Jersey, and New York. In 1976, California would have paid $4,700 dollars less than New Jersey and $17,000 less than New York in interest costs for every $1 million of debt. These differences reflect the effects of the New York City fiscal crisis of 1976. In July 1995, following the Orange County bankruptcy in December 1994, California would have paid $2,570 more than New Jersey and only $870 more than New York to issue the same amount of debt. Interest required above New Jersey bond (basis points) –100150 50 0 –5073 75 77 89 9593 91 87 85 83 81 79 97 Year 100 CaliforniaMassachusetts Texas New York Figure 5ÑRelative Bond Yield Required for Various StatesÕ 20 Year General Obligation Bonds 19 To evaluate the accuracy of the estimates in the Chubb survey, we compared the reported relative yields from the survey with those of bonds for states that make up the Bond Buyer General Obligation Bond Index. The results are encouraging. The index measures the yields on GO bonds maturing in 20 years, and it is supposed to have a rating roughly equivalent to MoodyÕs A1. Bonds issued by California and New York are among those included in the 20 bond index. From December 1, 1988, through August 13, 1992, CaliforniaÕs average bond yield was 38 basis points lower than New YorkÕs. The average yield differential in the Chubb survey was 36.5 basis points. 21 4. Borrowing Costs, Economic Conditions, and Fiscal Rules The substantial variation in tax-exempt bond yields across states and over time indicates that bond prices are sensitive to repayment prospects. This pattern suggests the viability of using data like those in the Chubb survey to determine how borrowing costs are affected by differences in state economic conditions as well as state fiscal rules. This chapter reports three sets of results along these lines. It begins with a discussion of how economic conditions and fiscal rules are correlated with the level of borrowing costs. It then examines how unexpected budget deficits and surpluses affect changes in borrowing costs and how the effects of budget surprises are amplified or cushioned by the presence or absence of various fiscal rules. The chapter closes with a discussion of the different ways bond markets react to fiscal news. 22 How Economic Conditions and Fiscal Rules Affect the Level of Borrowing Costs Our first empirical strategy tests whether borrowing costs are higher, on average, in states with weaker economies or with less-stringent fiscal rules. Because bond yields are influenced by state economic circumstances and other factors that affect the perceived risk of state debt, we use multivariate regression techniques. Our measured effect of fiscal rules controls for several economic variables: outstanding debt level as a percentage of personal income, state unemployment rates, real state per capita income, state general fund revenues as a fraction of per capita income, and the top state marginal tax rate on interest income. 1 As a proxy for changes in political preferences among voters for state spending and state deficits, we also include an index of the liberalness of the congressional delegation of each state as calculated by Americans for Democratic Action. These controls are discussed at more length in Poterba and Rueben (1999). Before analyzing the effect of fiscal rules on borrowing costs, we consider the effects of state economic conditions on these costs. The regression equation shown in Appendix Table B.1 indicates that a 1 percent increase in a stateÕs unemployment rate raises the stateÕs borrowing costs by five and one-half basis points. Three percent differences in unemployment rates across states are not uncommon, and the statistical results indicate that these economic differences are clearly an important factor in state borrowing costs. Because our regression equation also includes the level of state per capita income as an _________ 1Goldstein and Woglom (1992) investigate the role that economic variables play in setting the price of new municipal bonds. 23 explanatory variable, and because higher state per capita income is associated with lower borrowing costs, this analysis may understate the importance of economic conditions in affecting bond yields. In 1993, for example, the California unemployment rate was 9.4 percent. By 1998, the unemployment rate had declined to 5.9 percent. Our estimates suggest that the stateÕs borrowing cost has declined by 20 basis points as a result of this improvement in economic conditions. We now turn to the effects of our seven fiscal variables on state borrowing costs. Unlike economic conditions, fiscal rules are determined by legislators and voters, who may wish to consider the potential effects of new limits. Our econometric results provide clear support for an influence of fiscal rules on borrowing costs. Table 4 summarizes the key findings. The indicator variables for expenditure and revenue limits affect borrowing costs in different, but substantively important, ways. States with binding revenue limits pay, on average, 17.5 basis points more on their general obligation debt than states without such limits. By comparison, states with binding expenditure limits pay four basis points less than states without such limits. Although states with supermajority requirements also face higher costs, the estimate is too imprecise to permit strong conclusions. A limit on local revenues increases borrowing costs by about five basis points. The findings on the link between borrowing costs, anti-deficit rules, and limits on a legislatureÕs power to issue long-term debt are less conclusive. States with scores of 5 or lower on the ACIR index pay nine basis points more than states with higher index values, but we cannot reject the null hypothesis that anti-deficit rules do not affect borrowing costs. The coefficient on the indicator for limits on issuing debt is negative, indicating that debt limits on average reduce borrowing costs, 24 Table 4 Effect of Fiscal Institutions on State Borrowing Costs Fiscal InstitutionEstimated CoefficientCurrent California PolicyEffect of Change in Fiscal Variable on Per Capita Interest Costs for California Weak anti-deficit rules 8.99 (6.58)ÒStrongÓ rules$1.26 Limit on legislatureÕs power to issue long-term debtÐ4.72 (5.51)Limit Ð$0.66 Binding expenditure limit Ð3.65 (2.12)Limit since 1979Ð$0.51 Binding revenue limit 17.45 (5.33)No limit $2.45 Supermajority required to pass new taxes 3.15 (3.94)Limit $0.44 Limit on local revenues (Proposition-13-type limits)4.56 (2.54)Limit since 1978$0.64 Supermajority required to pass local bondsÐ4.04 (3.00)2/3 major- ity requiredÐ$0.57 NOTES: Values in parentheses are standard errors. Estimates are from regression results presented in Appendix Table B.1. California had $45.3 billion or $1,405 per capita of outstanding state debt at the end of fiscal year 1997 according to the Census of Governments (1999). The entries in the last column assume that the borrowing cost of this entire debt stock changes by the number of basis points indicated in the second column. but again, the figures do not permit precise conclusions. This result partly reflects the difficulty of making distinct estimates of the effect of many different fiscal rules, many of which are found together in various states. Poterba and Rueben (1999) estimate the effect of each fiscal rule separately and find significant effects of weak anti-deficit rules, expenditure limits, revenue limits, and supermajority taxes. If we 25 estimate separate effects for the last decade, we also find that supermajority rules are becoming increasingly important. Table 4 also describes the implications of our results for a state with an outstanding debt level comparable to CaliforniaÕs. The third column in Table 4 reports the change in per capita interest costs that would be associated with a change in fiscal rules. The two changes with the largest effects on per capita borrowing costs are the adoption of a tax limitation law and the lack of strong balanced budget rules. We estimate that if California had adopted a revenue limit at the state level, the stateÕs borrowing rate would have increased by more than 17 basis points. If we also assume that this higher borrowing cost would apply to all state debt, then this difference would translate to an increase in state borrowing costs of $2.45 per person, or an additional $79 million. Fiscal Shocks, Fiscal Rules, and the Bond Market Reaction to Fiscal News Our second empirical strategy considers how a stateÕs borrowing costs change in reaction to fiscal news. If fiscal rules are an important determinant of market interest rates, and if some rules are thought to reduce risk for bondholders, then such rules will have a larger effect on borrowing costs in some circumstances than in others. In particular, the economic effect of tight fiscal rules may be greatest when states are experiencing fiscal stress. The last decade provides a valuable opportunity to study bond market reactions to fiscal stress. Although statesÕ revenues have exceeded expectations recently, and some states have built up Òrainy day fundsÓ or enacted tax cuts, many states experienced sharp declines in their revenues during the recession of the early 1990s. This recession was especially 26 long lasting and deep in California. Poterba (1994) shows that in states with stringent anti-deficit rules, these fiscal events triggered corrective tax increases or expenditure reductions. In states with weaker anti-deficit rules, adjustments occurred more slowly. In California, the recession resulted in unexpected budget deficits in certain years. CaliforniaÕs state government reacted to the recession and the resulting budget shortfall in various ways. These reactions included increasing fees and assessments (in higher education, for example) and reallocating property tax revenues from cities and counties to school districts. The latter action allowed the state government to lower the amount of general fund revenues spent on school finance. We examine how unexpected deficits affected the bond market yields of different states. In particular, we are interested in how the tax-exempt bond market reacts to news concerning state surpluses and deficits, and how these reactions are affected by a stateÕs fiscal rules. We study the change in the Chubb surveyÕs yield spread between bonds issued by a given state and bonds issued by New Jersey over a 12 month period. After relating this change to the unexpected budget deficit or budget surplus in the intervening fiscal year, we allow the change in borrowing costs for a given fiscal shock to depend on the stateÕs fiscal rules. Thus, we are estimating how much interest rates have changed over a given fiscal year as a function of the fiscal health of the state during that time period. In addition, we are allowing the different fiscal rules to affect this relationship. Appendix B presents the details of this regression specification. Our analysis is based on annual measures of state fiscal shocks that are constructed from information collected by the National Association of State Budget Officers, which collects information on budgeted and 27 actual revenues and expenditures for each state. Each yearÕs unexpected budget deficit is defined as the deficit that would have developed, given actual economic conditions and other factors, with the tax and expenditure system that was in effect at the beginning of the fiscal year, less the forecast deficit at the beginning of the fiscal year. Poterba (1994) describes this measure in greater detail. Figure 6 graphs the unexpected deficit experienced by California relative to the average deficit experienced by other states. It also shows the largest deficit and the largest surplus of any state over the last decade. Unexpected deficits and surpluses are represented as a percentage of each stateÕs general fund revenues. Throughout this period, there were always states with either surpluses or deficits. CaliforniaÕs deficits have followed its economic health, with deficits during the recession and surpluses Percentage of general fund revenues –3040 20 0 –10 –2088 89 90 9695 94 93 92 91 97 Year 10 30 CaliforniaMaximum Other states Minimum Figure 6ÑUnexpected State Deficits as a Percentage of General Revenue, 1988Ð1997 28 more recently. Its fiscal health, however, has lagged behind that of other states. While other states experienced deficits in 1990, CaliforniaÕs budget was virtually balanced. In 1991 and 1992, however, unexpected deficits for California averaged nearly 10 percent of revenues, or approximately $90 per person. In more recent years, state finances have grown stronger, with unexpected budget surpluses on average. Table 5 presents information on the average unexpected deficit in selected states. It also presents data on the first and second halves of our data sample. In many states, the first half of the sample was a time of fiscal stress, with most states recovering during the second half. For California, the unexpected deficit averages 2.5 percent of revenue. This figure is somewhat larger than that in other large states. For New York, Table 5 Average State Deficit Shock, Selected States, 1988Ð1997 StateAverage Deficit Shock, 1988Ð1997Average Deficit Shock, 1988Ð1992Average Deficit Shock, 1993Ð1997 Alaska 3.67% 5.81 1.53 Arizona 0.63 Ð3.31 4.57 California Ð2.52 Ð4.29 Ð0.76 Colorado 1.21 0.37 2.06 Illinois Ð0.80 Ð2.42 0.82 Massachusetts Ð4.68 Ð11.06 1.70 New Jersey 0.30 Ð1.16 1.76 New York Ð1.84 Ð3.35 Ð0.34 Oregon 3.12 2.84 3.46 Pennsylvania Ð0.19 Ð1.50 1.13 Texas Ð1.23 Ð7.66 5.20 Utah 2.30 3.00 1.59 Washington 1.21 1.76 0.66 Average for all states Ð0.13 Ð0.89 1.16 NOTES: AuthorsÕ tabulations are based on information reported by the National Association of State Budget Officers. Deficit shocks are reported as a share of state revenues. A negative number indicates a deficit and a positive number is a surplus. The last row is an equal weighted average for all states. 29 this average is 1.8 percent; for Texas, 1.2 percent; and for Illinois, 0.8 percent. Unlike California, other Pacific states averaged fiscal windfalls over this period. In Oregon, the average surplus was 3.2 percent of revenue. In Washington, the analogous surplus was 1.2 percent of revenue. Alaska had the largest fiscal windfalls, averaging 3.7 percent of revenue. Figure 7 presents deficit patterns for selected states. Notice that both New York and Texas experienced much more volatility in their budgets over this period. Table 6 presents the main findings from our regression analysis of how bond yields react to unexpected deficits and how state fiscal institutions affect these reactions. It presents the change in interest rate required to issue new debt if a state has a given fiscal rule in place and experiences an unexpected $100 per capita deficit. The results further Percentage of general fund revenues –3040 20 0 –10 –2088 89 90 9695 94 93 92 91 97 Year 10 30 New Jersey CaliforniaMassachusetts Texas New York Figure 7ÑUnexpected State Deficits as a Percentage of General Revenue, Selected States, 1988Ð1997 30 Table 6 Effect of State Fiscal Rules on the Bond MarketÕs Reaction to Unexpected Budget Deficits and Budget Surpluses Fiscal InstitutionCurrent California PolicyEffect of Fiscal Institution on Bond Market Reaction to a $100 Per Capita Deficit Shock (Basis Points) Weak anti-deficit rules ÒStrongÓ rules 6.1 Limit on legislatureÕs power to issue long-term debtLimit 3.0 Binding expenditure limit Limit since 1979Ð0.0 Binding revenue limit No limit 5.2 Supermajority required to pass new taxesLimit since 19785.6 Limit on local revenues (Proposition-13-type limits)Limit since 1978Ð1.0 Supermajority required to pass local bonds2/3 majority required1.6 Predicted effect on bond costs of deficit given current California rules9.2 NOTES: Estimates are based on regression coefficients reported in Table B.2. See the text for further details. The last row represents the additional cost of issuing new debt while experiencing a $100 unexpected per capita deficit and having CaliforniaÕs rules in place. support our conclusion that fiscal institutions have real effects on state borrowing costs. In states with weak anti-deficit rules, required bond yields increased more in reaction to a given deficit than in states with stricter rules. A $100 per capita increase in a stateÕs budget deficit, about the size of the California unexpected deficits in 1991 and 1992, raises 31 borrowing costs by 6.1 basis points more in states with weak anti-deficit rules than in states with stricter rules. The next entry in Table 6 shows the estimated effect of having a limit on issuing long-term debt and experiencing a deficit. Restricting a stateÕs ability to issue debt appears to make a stateÕs bond yield more responsive to changes in the stateÕs fiscal health, although once again it is difficult to draw precise conclusions. This result partly reflects the difficulty of making distinct estimates of the effect of many different fiscal rules, many of which are found together in various states. The next two rows report the effect of the interactions between expenditure limits, tax limits, and unexpected deficits. The results suggest that an unexpected increase in the state deficit has a much larger effect on bond yields in states with binding revenue limits than in those with binding expenditure limits. Although the effect of an expenditure limit is virtually zero, there is a positive and statistically significant effect of a revenue limit on the bond market reaction to an unexpected deficit. A $100 per capita deficit shock raises borrowing costs by 5.2 basis points more in states with tax limitation laws than in states without such limits. Thus, CaliforniaÕs debt costs would have been higher in the recession of the early 1990s if revenue limits had been in place instead of the Gann amendment. This would translate into an additional $520,000 in interest costs per billion dollars of debt issued. Supermajority rules that limit a stateÕs ability to raise taxes also increase borrowing costs. A $10 per capita deficit shock increases borrowing costs by 5.6 basis points. This result indicates that borrowing costs during the California recession of the early 1990s would have been higher than now; specifically, an additional billion dollars of debt would have cost an extra $560,000. 32 The next two rows in Table 6 present our findings on the interaction between local fiscal limits and the effect of fiscal news on state borrowing costs. The effects associated with these variables are small. So, although there was a direct effect of local tax limits on the total cost of debt (see Table 4), it appears that local fiscal limits do not affect the relationship between fiscal news and bond costs. Finally, the last row of Table 6 presents the cumulative results of CaliforniaÕs fiscal rules. Our estimates suggest that if California had issued new debt in 1992, when the state was running a deficit, California would have had to pay an additional 9.2 basis points to issue new debt. Thus, the interaction of CaliforniaÕs deficit and its fiscal rules would have meant an additional $920,000 in interest costs per billion dollars of debt. The results in Tables 6 and B.2 provide clear support for our earlier claim that bond market participants view revenue limits and expenditure limits quite differently. They also set the stage for our analysis in the next section of the sensitivity of state borrowing costs to news about state fiscal conditions. Interstate Differences in Bond Market Reactions to Fiscal News The analysis underlying our results in Table 6 assumes that different states exhibit different bond yield sensitivities to deficit shocks only because they have different fiscal rules. This is likely to be a substantial oversimplification. To study this issue, we related the change in bond yields to the unexpected state deficit for all states and then included an indicator for specific states. We discovered substantial differences across states in the effect of fiscal news on bond yields. Table 7 reports the effect on bond yields of changes in fiscal health, as measured by unexpected deficit shocks, in specific states. We have 33 Table 7 Interstate Variability in Effect of Fiscal News on Borrowing Costs (in Basis Points) StateEffect of $100 Per Capita Deficit Shock on Bond Yield for Large States California 14.1 (3.6) Texas Ð2.2 (1.9) New York 14.1 (4.7) Florida Ð4.2 (7.8) Pennsylvania 18.3 (6.9) Illinois 2.1 (6.5) Ohio Ð3.0 (5.6) Michigan 0.7 (5.7) Georgia 3.6 (5.1) Massachusetts 8.5 (1.2) Connecticut 3.0 (1.3) NOTES: Estimates are based on regressions relating the change in bond yields from one year to the next to the unexpected fiscal deficit reported within the year. Larger deficits are reported as positive values. States are included in the sample if they are large in population terms or have a large outstanding stock of debt. Standard errors are reported in parentheses 34 chosen the largest states, as well as those states with the largest amount of outstanding debt, because these states are most responsive to such changes. California is the most populous state and also had the second largest amount of outstanding debt in 1995. CaliforniaÕs borrowing costs are more responsive to unexpected budget deficits than are the borrowing costs of most other large states. Given the size of the stateÕs municipal bond market and the Orange County bankruptcy at the end of 1994, it is not surprising that California is more responsive to fiscal shocks. A $100 per capita unexpected increase in CaliforniaÕs budget deficit is predicted to raise borrowing costs by 14 basis points. The only other large states with comparable estimates are New York (14 basis points) and Pennsylvania (18 basis points). Our earlier analysis in Table 6 suggests that if municipal bond markets were only responding to fiscal rules in place, California should have to pay nine basis points more during periods of fiscal stress. We investigated whether the apparently large ÒCalifornia effectÓ could be explained by differences in fiscal rules and found that these rules seem to explain a little less than two-thirds of the California effect. We modified the regression equation that generated the estimates in Table 6 to allow for a separate California effect. The resulting estimates, which are shown in the second column of Table B.2, confirm that the California effect is not simply the result of the stateÕs fiscal rules. The effects of the fiscal variable interactions are not substantively affected by allowing for a California interaction. These results indicate that the bond market is particularly sensitive to deficit news about California and other large states. One possible explanation is that bond market participants pay closer attention to fiscal developments in states with relatively large quantities of outstanding tax- 35 exempt debt than to similar developments in states with limited debt markets. This explanation implies that deficit news has a larger effect in these states because it is more visible and it affects more investors. A second possibility is that the political circumstances in California and several other large states make it more difficult for the states to return to fiscal balance in the years immediately following an unexpected deficit. Unexpected deficits would therefore have a larger effect in these states because they provide more information about the future of state fiscal conditions. Both of these explanations are likely to contribute to our empirical findings. 37 5. Conclusions and Implications This study finds that there is a clear link between state economic circumstances and the interest rate at which a state borrows when it sells general obligation debt. Because lenders expect states with more robust economies to have less difficulty servicing their debt, states with lower unemployment rates face lower interest rates in the capital market. Changes in state economic conditions explain some of the variation over time in the relative borrowing costs of different states. State fiscal institutions, such as CaliforniaÕs Gann amendment or Proposition 218, also affect borrowing costs. Whether and how the stateÕs constitution limits the discretion of policymakers to run fiscal deficits, and whether the state has a tax limit or an expenditure limit in place, are strongly correlated with borrowing costs. These findings suggest that policymakers need to recognize that changes in fiscal policies, and in the budget process more generally, affect these costs. In California, policymakers and citizen groups have historically paid little 38 attention to the effects new fiscal restrictions can have on borrowing costs. Nor has the topic received much research attention. Rosenblum (1999) remarks that one effect of Proposition 218 and its predecessors is that CaliforniaÕs bond rating has not returned to AAA, the highest possible level. Davey (1995) describes how Ohio legislators and others who opposed a revenue limit argued that adopting it might raise the stateÕs borrowing cost. Beyond the finding that fiscal rules matter for borrowing costs, two more specific conclusions emerge from our study. First, the bond market reacts very differently to revenue limits and expenditure limits. Expenditure limits have small effects and probably reduce borrowing costs, but states with tax limitation laws face substantially higher borrowing costs than other states. States that require legislative supermajorities to pass new taxes also face higher borrowing costs. Our estimates suggest that if California had enacted a tax limitation law for state revenues instead of an expenditure limit, the stateÕs borrowing costs would be about 20 basis points higher than they currently are. That difference translates into approximately $3 of additional annual state expenditures per capita, or $95.6 million, if we assume that the level of state indebtedness was not affected by such policies. Second, there are substantial differences across states in the sensitivity of general obligation bond yields to fiscal news. The borrowing costs of states with relatively lax fiscal rules are more sensitive to budget news than are the borrowing costs of states with tighter fiscal rules. Although its fiscal rules are relatively tight, California is among the group of states with the highest responsiveness. It may be that municipal bond analysts and other credit market participants pay more attention to fiscal news in large states with many outstanding bonds than to 39 comparable news in smaller states. Our results suggest that accurate budget projections are more important in California than in many other states, because unfavorable budget news has a substantial and adverse effect on the stateÕs borrowing cost. The need for accurate budgeting is especially important over the next few years as California increasingly turns to the municipal bond market to pay for new infrastructure spending. 41 Appendix A State Fiscal Institutions State State Supermajority TaxLocal Local StateBalanced BudgetDebt RestrictSpending LimitRevenue LimitYear PassedVote RequiredTax LimitDebt Vote Alabama 10 Yes Maj Alaska 6 Yes 1982 1972 Maj Arizona 10 Yes 1978 1992 2/3 1980 Maj Arkansas 9 Yes* 1934 3/4 1981 Maj California 6 Yes* 1979 1978 2/3 1978 2/3 Colorado 10 No 1992 a1992 1992 2/3 1992 Maj Connecticut 5 No 1991 Delaware 10 No 1980 3/5 Florida 10 Yes* 1994 1971 3/5 1995 Maj Georgia 10 Yes Maj Hawaii 10 Yes 1978 Idaho 10 Yes 1980 1992 3/5 Illinois 4 No 1991 Maj Indiana 10 Yes 1973 Iowa 10 Yes 1978 3/5 Kansas 10 Yes* 1970 Maj Kentucky 10 Yes* 1979 Maj Louisiana 4 No 1993 1991 b1966 2/3 1978 Maj Maine 9 Yes* Maj Maryland 6 No Massachusetts 3 No 1986 1980 42 State State Supermajority TaxLocal Local StateBalanced BudgetDebt RestrictSpending LimitRevenue LimitYear PassedVote RequiredTax LimitDebt Vote Michigan 6 Yes* 1978 1978 Maj Minnesota 8 Yes Maj Mississippi 9 Yes 1970 3/5 Maj Missouri 10 Yes* 1980 1996 c1996 2/3 1980 2/3 Montana 10 No 1981 1998 3/4 1987 Maj Nebraska 10 Yes 1990 Maj Nevada 4 Yes 1994 d1996 2/3 1983 Maj New Hamp. 2 No Maj New Jersey 10 Yes* 1990 e New Mexico 10 Yes 1979 Maj New York 3 Yes* Maj North Carolina 10 No 1991 Maj North Dakota 8 Yes 1981 2/3 Ohio 10 Yes 1976 Maj Oklahoma 10 No 1985 1992 3/4 3/5 Oregon 8 Yes* 1979 1996 3/5 1991 Maj Pennsylvania 6 Yes* Maj Rhode Island 10 Yes* 1992 f1985 South Carolina 10 Yes* 1980 Maj South Dakota 10 Yes 1996 2/3 Maj Tennessee 10 No 1978 g Maj Texas 8 Yes 1978g1982 Maj Utah 10 Yes 1989 Maj Vermont 0 Yes Virginia 8 Yes Maj Washington 8 Yes 1993 1979 1993 2/3 1971 3/5 West Virginia 10 Yes 1990 3/5 Wisconsin 6 Yes Maj Wyoming 8 Yes Maj NOTES: Data on budget stringency rules and state and local debt restrictions are from ACIR (1987a) and Rafool (1997). Data on revenue and expenditure limits are from Rueben (1996). Data on local limits are from ACIR (1997). * denotes states that require a popular vote to approve debt issue. In states with nonbinding limits, limits can be overridden by simple legislative majorities. aColorado passed a nonbinding spending limit in 1977.bLouisiana adopted a nonbinding revenue limit in 1979 and a binding one in 1991.cMissouri adopted a nonbinding revenue limit in 1980.dNevada passed a nonbinding spending limit in 1979.eNew Jersey also passed a spending limit in 1976 that expired in 1983.fRhode Island adopted a nonbinding limit in 1977, but it was replaced with a binding limit in 1992. gTennessee and Texas have limits that are not binding. 43 Appendix B Detailed Description of Regression Results The empirical findings summarized in the text are based on two sets of regression equations. The first analyzes the relationship between the level of general obligation bond yields and the structure of state fiscal rules; the second relates the change in the bond yield for a state to the unexpected budget deficit, or surplus, in a given year. Poterba and Rueben (1999, 1998) investigate each of these statistical relationships and present a range of different regression models that are broadly supportive of the general findings reported here. In this appendix, we describe the regression equations that underlie our core findings. 44 Fiscal Rules and the Level of General Obligation Yields The basic regression model that we estimate to test for a relationship between the level of borrowing costs, and the structure of state fiscal institutions, is RR XX ZZ it jt it jt it jt t i it jt-= - + - +++-()*()* ()abqkee (1) where R it denotes the nominal interest rate on bonds issued by state i at time t, X it denotes the set of state-specific economic conditions that may affect borrowing costs, and Z it represents the set of state fiscal institutions that we are interested in studying. We estimate models for the cross-state differences in yields because the Chubb survey data are collected in this way. To make the explanatory variables consistent with the dependent variables, they must be computed as differences relative to the corresponding values for New Jersey. The variables that we include as control variables for borrowing costs are state outstanding general obligation debt per capita, the state unemployment rate, the level of real per capita income in the state, state general fund revenues as a fraction of per capita income, and the top state marginal tax rate on interest income. The first three variables are drawn from publications of the Bureau of Labor Statistics, the National Income and Product Accounts, and the Census of Governments. The marginal tax rate variable was computed using the State Tax Module of the National Bureau of Economic Research TAXSIM program and augmented through downloaded information on state income tax forms. We also include variables that proxy for the political climate in the state, on the grounds that such variables may provide information on the future evolution of state deficits and therefore on the creditworthiness of 45 the state. Our principal variable of this type is the Americans for Democratic Action (ADA) score for the stateÕs senate delegation; this should provide a general indication of the political ideology of the state. We include seven variables ( Z it) to measure the fiscal institutions in each state. The first is an indicator variable for states with weak anti- deficit rules, defined as a score of 5 or less on the ACIR index of fiscal stringency. The second is an indicator variable for legislative or constitutional limits on the legislatureÕs power to issue long-term debt. The third and fourth variables are indicator variables for binding tax limits, and binding expenditure limits, respectively. The fifth variable indicates whether a supermajority is required to enact a new revenue source. The sixth and seventh variables measure restrictions on local governmentÕs ability to raise revenues or to issue debt. Previous studies have related bond yields as reported in the Chubb survey to various state fiscal rules, although none of the previous studies have considered as broad a range of fiscal institutions, or as long a time span of bond yields, as the analysis reported here. Eichengreen (1992) and Goldstein and Woglom (1992) focus on single cross-sections of bond yields. Even with much smaller datasets than the panel dataset that underlies our study, they present evidence that borrowing costs are higher for states with weak anti-deficit fiscal rules. Bayoumi, Goldstein, and Woglom (1995) is most similar to our analysis. It is based on panel data for a period that ends before the state fiscal crises of the early 1990s, and it develops a careful model of how the outstanding stock of state debt affects borrowing costs. That study is also limited to anti-deficit rules, and it does not consider the effect of tax or expenditure limits. Our results are consistent with theirs. 46 We estimate regression equation (1) for the 40 states covered in the Chubb survey over the 1973Ð1995 period. The dependent variable is measured in basis points, which permits easy interpretation of the coefficients on the indicator variables for different fiscal rules. The coefficient estimates are reported in Table B.1. Unexpected Fiscal News and the Reaction of Bond Market Yields In our second specification, we compute the fiscal ÒsurpriseÓ that bond market participants learned about between two Chubb surveys 12 months apart. We label this variable DEFSHOCK, for deficit shock, but it can take positive or negative values depending on whether the fiscal news is unfavorable (higher deficit) or favorable (lower deficit). DEFSHOCK is measured in dollars, but we scale this variable to per capita terms and deflate to constant dollars for our regression analysis. We then compute the difference between the reported yields on a given stateÕs bonds in the two surveys. Given the way the survey data are collected, this is actually a Òdifference in differences.Ó Since each survey reports the difference between the yield on a stateÕs bonds and the yield on New JerseyÕs bonds, the difference between responses in two surveys is the change in the yield on a stateÕs bonds, minus the change in the yield on New JerseyÕs bonds. The dependent variable is therefore D()RR it jt-. DEFSHOCK is differenced from New JerseyÕs unexpected per capita deficit. In addition to DEFSHOCK, we include the change in the stateÕs unemployment rate between the two Chubb surveys relative to the change in New JerseyÕs unemployment rate, and the change in each stateÕs per capita outstanding debt relative to the change in New JerseyÕs 47 Table B.1 Regression Estimates of How Fiscal Institutions Affect State Borrowing Costs, 1973Ð1995 Explanatory Variable1973Ð 1995 Sample Outstanding debt to personal income 1.93 (1.20) Unemployment rate 5.60 (0.57) Per capita income 1.21 (0.52) State revenue/personal income Ð0.23 (0.12) Top marginal tax rate on interest income 0.47 (0.38) ADA score 4.41 (4.96) Weak anti-deficit rules 8.99 (6.58) Limit on legislatureÕs power to issue long-term debt Ð4.72 (5.51) Binding expenditure limit Ð3.65 (2.12) Binding revenue limit 17.45 (5.33) Supermajority required to pass new taxes 3.15 (3.94) Limit on local revenues (Proposition-13-type limits) 4.56 (2.54) Supermajority required to pass local bonds Ð4.04 (3.00) Adjusted R2 0.42 NOTE: Values in parentheses are standard errors. 48 debt level. These variables should capture the fact that state economic conditions are an important determinant of general obligation bond yields. Our principal hypothesis is that the effect of unexpected fiscal news on bond yields is mediated by the structure of the stateÕs fiscal institutions. In addition to the actual DEFSHOCK variable, we therefore include interactions between DEFSHOCK and the seven fiscal institutions described above. The coefficients on the interaction terms measure the effect of a stateÕs fiscal institution on the way its borrowing costs respond to fiscal news, normalized by how similar fiscal rules in New Jersey affect the response of that stateÕs bond yield to similar news. The actual estimating equation is therefore: DD D DD DD ()( )* ()*( )* ( * *)* R R DEBTSTOCK DEBTSTOCK UNEMP UNEMP DEFSHOCK DEFSHOCK DEFSHOCK Z DEFSHOCK Z it jt it jt it jt it jt it it jt jt it jt-= - +- + -+ -++-a a a aa e e1 2 3 45 (2) In this case, a 4 is a vector consisting of seven distinct coefficients on the seven fiscal institutions. Conventional F-tests allow us to reject the null hypothesis that these coefficients are jointly zero at very high confidence levels. Equation (2) is estimated on annual data for 1988Ð1997. Although we have nearly 15 years of prior data on the dependent variable, we cannot construct the DEFSHOCK variable before 1988, because that is when the National Association of State Budget Officers began collecting the full range of fiscal data that our DEFSHOCK calculations require. Table B.2 reports our estimates of equation (2). 49 Table B.2 Regression Estimates of How Fiscal Institutions Affect Bond Market Reaction to State Fiscal News, 1988Ð1997 Explanatory Variable Estimated Coefficients Change in outstanding debt 0.0024 (0.0014)0.0022 (0.0014) Change in unemployment rate 1.4870 (0.3057)1.4754 (0.3148) DEFSHOCK Ð0.0210 (0.0178)Ð0.0096 (0.0168) DEFSHOCK*(weak anti-deficit rules) 0.0608 (0.0214)0.0521 (0.0224) DEFSHOCK*(limit on legislatureÕs power to issue long-term debt)0.0299 (0.0186)0.0215 (0.0176) DEFSHOCK*(binding expenditure limit) Ð0.0006 (0.0062)Ð0.0033 (0.0062) DEFSHOCK*(binding revenue limit) 0.0518 (0.0222)0.0501 (0.0226) DEFSHOCK* (supermajority required to pass new taxes)0.0557 (0.0220) 0.0357 (0.0224) DEFSHOCK* (limit on local revenues) (Proposition-13-type limits)Ð0.0101 (0.0076)Ð0.0095 (0.0079) DEFSHOCK* (supermajority required to pass local bonds )0.0161 (0.0193)Ð0.0052 (0.0123) DEFSHOCK* (California) Ñ 0.1165 (0.0276) Adjusted R2 0.2963 0.3089 NOTES: Values in parentheses are standard errors. All equations are estimated for 1988Ð1997 using data for the 40 states included in the Chubb survey. Estimates are given for a state response to a per capita deficit. Unexpected deficits are expressed as positive values in the regression results. 50 Unlike equation (1), which resembles the estimating equations in several previous studies, only one study has explored how bond yields change in reaction to fiscal developments. Lowry and Alt (1997) find that actual deficits have a larger positive effect on borrowing costs when states have weaker anti-deficit rules. Our findings based on fiscal ÒnewsÓ rather that the level of fiscal variables are consistent with this result. 51 References Advisory Council on Intergovernmental Relations, Fiscal Discipline in the Federal System: National Reform and the Experience of the States, Washington, D.C.: Advisory Council on Intergovernmental Relations, 1987a. Advisory Council on Intergovernmental Relations, Significant Features of Fiscal Federalism: 1988. Volume 1, Budget Processes and Tax Systems, Washington, D.C.: Advisory Council on Intergovernmental Relations, 1987b. Advisory Council on Intergovernmental Relations, State Laws Governing Local Government Structure and Administration, Washington, D.C.: Advisory Council on Intergovernmental Relations, 1993. Advisory Council on Intergovernmental Relations, A Profile of Tax and Expenditure Limits in the Fifty States, Washington, D.C.: Advisory Council on Intergovernmental Relations, 1997. Advisory Council on Intergovernmental Relations, Significant Features of Fiscal Federalism: 1995. Volume 1, Budget Processes and Tax Systems and Volume 2: Revenues and Expenditures, Washington, D.C.: Advisory Council on Intergovernmental Relations, 1998. 52 Alt, James E., and Robert C. Lowry, ÒDivided Government, Fiscal Institutions, and Budget Deficits: Evidence from the States,Ó American Political Science Review, Vol. 88, 1994, pp. 811Ð828. Bayoumi, Tamim, Morris Goldstein, and Geoffrey Woglom, ÒDo Credit Markets Discipline Sovereign Borrowers: Evidence from U.S. States,Ó Journal of Money Credit and Banking, Vol. 27, 1995, pp. 1046Ð1059. Bohn, Henning, and Robert P. Inman, ÒConstitutional Limits and Public Deficits: Evidence from the U.S. States,Ó Carnegie - Rochester Conference Series on Public Policy, December 1995. Briffault, Richard, Balancing Acts: The Reality Behind State Balanced Budget Requirements, New York: Twentieth Century Fund Press, 1996. Bunche, Beverly S., ÒThe Effect of Constitutional Debt Limits on State GovernmentsÕ Use of Public Authorities,Ó Public Choice, Vol. 68, 1991, pp. 57Ð69. Bureau of Labor Statistics, ÒLocal Area Unemployment StatisticsÓ Washington, D.C., 1999, http://stats.bls.gov/cgi-bin/surveymost. California Business Roundtable, Building a Legacy for the Next Generation, Sacramento, California, 1999. California State Treasurer, Smart Investments: Special Edition of CaliforniaÕs Debt Affordability Report, Sacramento, California, 1999. Davey, C., ÒPlan to Make Tax Increase Tougher Gets Cool Reception in Senate,Ó Cincinnati Enquirer, September 24, 1995, p. C4. Eichengreen, Barry, Should the Maastricht Treaty Be Saved? Princeton Studies in International Finance No. 74, Princeton, New Jersey: Princeton University, 1992. Goldstein, Morris, and Geoffrey Woglom, ÒMarket-Based Fiscal Discipline in Monetary Unions: Evidence from the U.S. Municipal Bond Market,Ó in M. B. Canzoneri, V. Grilli, and P. R. Masson, eds., Establishing a Central Bank: Issues in Europe and Lessons from the United States, Cambridge: Cambridge University Press, 1992. 53 Inman, Robert, ÒDo Balanced Budget Rules Work? U.S. Experience and Possible Lessons for the EMU,Ó NBER Working Paper 5838, Cambridge, Massachusetts, 1996. Kiewiet, D. Roderick, and Kristin Szakaly, ÒThe Efficacy of Constitutional Restrictions on Borrowing, Taxing, and Spending: An Analysis of State Bonded Indebtedness, 1961Ð90,Ó Journal of Law, Economics, and Organization, Vol. 12, 1996, pp. 62Ð97. Knight, Brian, ÒSupermajority Voting Requirements for Tax Increases: Evidence from the States,Ó mimeo, Madison, Wisconsin: University of Wisconsin Department of Economics, 1999. Legislative AnalystÕs Office, CaliforniaÕs Fiscal Outlook: The LAOÕs Economic and Budget Projections, Sacramento, California, 1998. Lowry, Robert C., and James E. Alt, ÒA Visible Hand? Bond Markets, Political Parties, Balanced Budget Laws, and State Government Debt,Ó mimeo, Cambridge, Massachusetts: Harvard University Department of Government, 1997. National Association of State Budget Officers, Fiscal Survey of the States, Washington, D.C., 1986Ð1997. Poterba, James M., ÒState Responses to Fiscal Crises: The Effects of Budgetary Institutions and Politics,Ó Journal of Political Economy, Vol. 102, 1994, pp. 799Ð821. Poterba, James M., ÒDo Budget Rules Work?,Ó in A. Auerbach, ed., Fiscal Policy: Lessons from Economic Research, Cambridge, Massachusetts: MIT Press, 1997, pp. 53Ð86. Poterba, James, and Kim Rueben, ÒFiscal News, State Budget Rules, and Tax-Exempt Bond Yields,Ó mimeo, Cambridge, Massachusetts: MIT Department of Economics, 1998. Poterba, James, and Kim Rueben, ÒState Fiscal Institutions and the Municipal Bond Market,Ó in J. Poterba and J. von Hagen, eds., Fiscal Institutions and Fiscal Performance, Chicago, Illinois: University of Chicago Press, 1999. 54 Rafool, Mandy, ÒState Tax and Expenditure Limits,Ó State Tax Notes, Washington, D.C., January 13, 1997, pp. 115Ð132. Rosenblum, Joseph, The State of California: Update March 1999, New York: Sanford Bernstein Municipal Bond Research, 1999. Rueben, Kim S., ÒTax Limitations and Government Growth: The Effect of State Tax and Expenditure Limits on State and Local Government,Ó mimeo, Public Policy Institute of California, 1996. State of California, GovernorÕs Proposed Budget Highlights 1999Ð2000: Sacramento, California, 1999. U.S. Census Bureau, ÒState Government FinancesÓ Washington, D.C., 1999, http://www.census.gov/govs/www/state.html. 55 About the Authors JAMES M. POTERBA James M. Poterba is a professor of economics at the Massachusetts Institute of Technology who specializes in the analysis of taxation and government expenditure programs. His past research has examined the way state governments respond to unexpected fiscal windfalls and fiscal crises, and it has also considered the factors that determine interest rates in the tax-exempt bond market. He holds a B.A. from Harvard College and a D.Phil. from Oxford University. KIM S. RUEBEN Kim S. Rueben is an economist who is interested in the relationship between fiscal institutions and state and local government budgetary decisions and the relationship between institutional structure and educational resources. She has examined how state tax and expenditure limits have affected government finances, including the effects of budgetary institutions on municipal bonds and public employee salaries. Currently, she is investigating how limitation laws affect student outcomes and is also examining which local governments are successful in passing new taxes when faced with the need for voter approval. She holds a Ph.D. in economics from the Massachusetts Institute of Technology." ["post_date_gmt"]=> string(19) "2017-05-20 09:34:43" ["comment_status"]=> string(4) "open" ["ping_status"]=> string(6) "closed" ["post_password"]=> string(0) "" ["post_name"]=> string(9) "r_1299jpr" ["to_ping"]=> string(0) "" ["pinged"]=> string(0) "" ["post_modified"]=> string(19) "2017-05-20 02:34:43" ["post_modified_gmt"]=> string(19) "2017-05-20 09:34:43" ["post_content_filtered"]=> string(0) "" ["guid"]=> string(51) "http://148.62.4.17/wp-content/uploads/R_1299JPR.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) }