The current projections agree on some basic issues: Growth rates will be lower than in the past, but absolute levels of growth will remain high. Natural increase will have more effect than net migration on population growth. Domestic migration will be lower, while international migration will remain strong. California will still have higher growth rates than the rest of the country. However, the projections of future population and rates of population growth vary widely. For example:
Planners need to be aware of the range of plausible future population levels of the state, and should, accordingly, consider alternative scenarios in their planning.
The sheer size of the state's population increase has important implications for almost all government services and functions including welfare, education, transportation, and corrections. Large increases in the state's population also have important implications for protection of natural resources, distribution of water, agriculture, and location and nature of development. No less important, but less predictable, is how the changing composition of the state's population will influence the state's economic evolution, its political representation, and its cultural identity or identities. Projections of 50 million Californians by 2025 suggest that policy issues related to population growth will become even more salient to policymakers. However, projecting future populations is an uncertain undertaking. The demographic future of California is very much in doubt. The recession of the early 1990s saw record numbers of domestic migrants leaving California. From 1990 through 1996, between 1 million and 2 million more people left California for other states than moved to California from the rest of the United States. At the same time, immigration (international migration) to the state increased substantially during the 1970s and 1980s and, though slowing, still remained substantial during the recession of the early 1990s. Some argue that the patterns of the early 1990s ushered in a new era of demographic change in California. Some predict that the state, once a magnet for migrants from the rest of the country as well as the world, will be the next demographic New York: a place that receives immigrants and sends out domestic migrants. Projections of the state's population are heavily influenced by assumptions about migration patterns.
In this report, we compare and analyze a number of long-term population projections for California. First, we discuss the projected populations, then compare the various methods and assumptions used to develop the projections, and finally discuss some implications of the comparisons. Additional charts, tables, and analyses are contained in the Appendices.
The length of the projection series and the level of demographic detail vary among the series (see Table 1). The projections produced by the California Department of Finance and the Census Bureau are most directly comparable in terms of demographic detail. The Census Bureau projections extend to 2025, whereas all the other series extend to at least 2040.
The projections show substantial differences even in the short run. Both Census Bureau series project substantially lower population totals than do the other series. For example, the CB series project fewer than 34.5 million California residents in 2005 whereas the DOF projects 37.4 million Californians. This short-term difference is a product of both differing projections of growth and differing current estimates of the state's population at the time the projections were made. During the 1990s, the Census Bureau estimates of the state's population have increasingly diverged from those produced by the DOF: By 1998, the Census Bureau estimates were lower than the DOF estimates by more than 800,000.
As shown in Figure 2, annual growth rates implied by these projections are substantially different--especially the initial projected growth rates, in which the CB projections are three times higher than the DOF projections. The large disparity in initial rates results from differing base years for the projections. The CB projections were developed at a time when the most recent estimates suggested substantial domestic out-migration from California and population growth in the state was quite slow by historic standards. Although the CB projected growth rates are higher than growth rates experienced by the state in the recession, the CB projections did not anticipate the dramatic short-run change in growth rates that occurred between 1994 and 1998 (see Figure 2). Again, if we exclude the alternative series (CCSCE high, CCSCE low, and CB alternative), the range of long-term projections is substantially narrowed.
In absolute terms, the difference between the CB alternative series and the DOF projections is especially pronounced for whites, with the DOF projecting almost 6 million more whites by 2025 than the CB alternative projections (see Table 3). The lower population of whites in the CB series can be attributed to projected continuing net domestic migration losses. Domestic migrants, both to and from California, are more likely to be white than are immigrants or non-migrants in the state.
To the state government, however, the most important component of the dependency ratio is that attributable to children, because the state is the primary provider of services to children (via education) and provides relatively few services to the elderly. After a substantial rise in the child dependency ratio during the 1990s, the DOF and BEA series project a decline to the mid-2010s, followed by an increase (see Figure 5). This is a welcome short-term trend for a state trying to catch up with large increases in public school attendance. The CB projections show an increase in the child dependency ratio from 1995 to the first few years of the next century, before declining to the mid-2010s. After 2015, however, all the series project that the child dependency ratio will increase substantially, rising to levels not seen since the early 1970s according to the DOF projections. [2] The BEA projections show the same long-term pattern, but the increases are substantially lower.
Methods
Methods used to project California's population range from the computationally complex "multi-state cohort component projection" technique employed by the Census Bureau to the simple derivative approach employed by UCLA. The projections can be classified into three types: (1) purely demographic (DOF, CB preferred); (2) demographic and economic (CB alternative, BEA, and CCSCE); and (3) derivative extrapolations (UCLA). The purely demographic approaches of the DOF and the CB preferred projections rely on an analysis of historical trends in the components of population change: births, deaths, and migration. This approach implicitly assumes that factors that led to such trends in the past will continue in the future. The methods that incorporate economic factors do so because employment is assumed to drive population growth through migration. Migration, especially domestic migration, to or from California responds strongly to employment opportunities in California versus the nation. The methods employed by CCSCE, BEA, and the CB alternative series use projections of employment to determine future populations of the state. UCLA's projections are primarily based on and extrapolated from an earlier set of projections produced by the Department of Finance, but also include a subjective assessment of recent economic conditions in the state. Finally, some of the state projections consider national population projections (CCSCE, BEA, CB), while others do not (DOF, UCLA).
More complex methods do not necessarily produce more accurate projections, although they might provide details (such as populations by age and gender) that are necessary to the user. Key sources of uncertainty are future employment, domestic migration, and fertility. For example, a decline in fertility rates could lead to substantially lower population projections over the long term for California than the DOF and CB projections currently envision.
Assumptions
When projections are developed partly explains differences between them. The most recent economic and demographic trends available at the time a projection is made are important factors in projecting future trends. In California, domestic migration has been especially volatile over the past ten years. The CB and BEA projections were developed when the most recent data showed massive domestic out-migration from California. Since those projections were developed, however, the large domestic migration outflows have ceased. The DOF, CCSCE, and UCLA projections were all developed after this turnaround, and thus show higher short-term projections. Figure 6 shows the importance of timing for the projections by comparing the DOF and CB migration estimates and projections.
It is not clear which of the current projections for California are most accurate. Based on when the projections were developed, we can conclude that the CB and BEA projections are too low, at least in the short run. In the long run, the wide range of current population projections reflects the uncertainty of California's demographic future. Although the short-term projections of the CB and BEA are too low, any of the long-term population projections could be realized. None of the scenarios represented by the projections are unrealistic. The highest projection series (UCLA) implies annual growth rates for California that are lower than for any comparable historic period. The lowest projection series (CCSCE's low series) assumes that California's share of national employment growth will be only moderately lower than in the past.
Despite their wide disparities, the current projections agree on some basic issues:
Table A.1 | |||
Methodologies
of Long-Term Projections for California
|
|||
Source | Method | Assumptions | Controls |
California Department of Finance | Cohort component | Historical trends in fertility, mortality, and migration used to project future levels | None |
United States Census Bureau, Series A ("Preferred Series") | Cohort component | Historical trends in fertility, mortality, and migration used to project future levels | Sum of states controlled to U.S. Census Bureau middle series national projections |
United States Census Bureau, Series B ("Alternative Series") | Composite of cohort component and economic model | Historical trends in fertility and mortality used to project future levels. Migration determined from BEA model of future employment. | Sum of states controlled to U.S. Census Bureau middle series national projections |
United States Bureau of Economic Analysis | Economic model | Historical trends in state share of U.S. employment in basic industries used to project future employment. Population determined from employment. | Sum of states controlled to U.S. Census Bureau middle series national projections |
UCLA |
Derivative, economic considerations
|
Adjustment of DOF 97 interim series growth rates to 2010 based on recent economic trends; DOF 97 projected growth rates used for 2010-2020. Projected trends in rates of growth to 2020 used to generate extrapolations to the year 2050. | None |
CCSCE | Composite demographic component and economic model | Uses DOF and CB projected fertility and mortality rates. Migration driven by employment projections. Analyses of trends in state share of U.S. employment in basic industries used to project future employment. | Middle series uses U.S. Census Bureau middle series national projections to develop state employment projections. |
The purely demographic approaches of the DOF and the CB preferred projections rely on an analysis of the components of population change: births, deaths, and migration. The DOF and the CB use cohort component population projection methods in which the population is projected over time by applying projected age, ethnic, and gender-specific rates of migration, fertility, and mortality to a base year population disaggregated by age. Both use historical series of estimated trends in mortality, fertility, and migration rates to inform projected rates. This approach implicitly assumes that factors that led to such trends in the past will continue in the future, but does not attempt to model underlying determinants of the specific components. Determination of the appropriate historical time period and the importance placed on the most recent data introduce some subjectivity into the process.
The methods that incorporate economic factors do so because employment is seen as a driver of population growth through migration. Migration, especially domestic migration, to or from California responds strongly to employment opportunities in California versus the nation. The methods employed by CCSCE, BEA, and the CB alternative series use projections of employment to determine future populations of the state. The employment projections are based on historic trends in the states share of employment in "basic" industries.[3] These trends are projected into the future to determine future employment. The CB alternative series and CCSCE then use these employment projections to determine migration flows in a cohort component framework. BEA uses them to determine the population of working age and then uses historic trends in state-specific ratios of the population not of working age to the population of working age to determine nonworking age populations. BEAs projections for the states are controlled to the CB middle series national projections.
The method used in the UCLA projections relies largely on projections developed by the DOF. The UCLA projections are derived primarily from a set of population projections developed by the DOF in 1997. Termed the "interim" series by DOF, these 1997 DOF projections have been superseded by the 1998 cohort component projections developed by DOF. The older interim projections and the current DOF projections are fairly close, with the older series suggesting slightly higher growth so that by 2020 the older projections are almost 5 percent higher than the current DOF series. UCLA takes the interim DOF projections and adjusts them in the short run according to current economic conditions and extends them by linear extrapolation to extend beyond 2020. The UCLA projections note the following economic factors for adjusting 1997 DOF projections: "Because of the strong employment and income growth in California in the last two years, we assume that population growth, already on the upswing, will accelerate over the next few years in response to the positive economic climate in California" (Bolton 1999).
Below, we assess critical assumptions in each of the projection series. The most uncertain components of population change are migration and fertility, and thus we focus our discussion on those assumptions. Mortality assumptions differ little between the series. Table B.1 provides a brief description of the assumptions.
CB and DOF Projections
The CB projections suggest that California will become the next demographic New York: a place that receives international migrants and sends out domestic migrants. In the short run (the first five years) the CB projections are highly dependent on the interstate migration patterns of the early 1990s. The base year estimates are determined largely by a time series approach that slightly moderates migration in the prior year. Projections for the medium term (2000-2010) are based on both patterns during the early 1990s as well as longer-term averages, while projections for the long term (2010-2025) are based on long-term averages in interstate migration. Because California experienced unprecedented net interstate migration outflows during the early 1990s, the CB approach leads to tremendous and continuing, albeit slightly moderated, net outflows.
Already, however, the CB projections of migration are tracking far below the latest estimates of migration to the state. The DOF and, to a lesser extent, the CB estimates suggest that by 1997 domestic migration to California had rebounded from the dramatic losses of the early to mid 1990s, so that the number of persons leaving California for other states was about the same as the number arriving from other states. The CB preferred series projected annual net domestic migration losses exceeding 400,000 per year to continue from 1995 to 2000, far greater losses than even the most recent CB estimates suggest. Total net migration to the state (which includes net domestic and net international migration) had increased to about 200,000 by 1997 according to both DOF and CB estimates.
Table B.1 | |||
Fertility, Mortality, and Migration Assumptions of Long-Term Projections for California
|
|||
Source | Fertility | Mortality | Migration |
California Department of Finance |
Very little or no change for Blacks, American Indians, and Asians and Pacific Islanders. Small increase for Whites. Small decline for Hispanics. |
California mortality rates are merged to U.S. middle series projected mortality. |
Projections are based on long-term historical averages of net migration. No distinction is made
between international and domestic migration. State net migration is the sum of net county
migration. In the short run to 2003, DOF assumes "pent-up demand" will lead to
higher than historical average migration of 250,000 per year. From 2003 to 2015, DOF merges this
level to the long term historical average of 180,000 per year, and keeps net migration constant at
the level for the remainder of the projection horizon. The assumptions reflect some modifications of county-specific migration made in consultation with local government planners and demographers.
|
United States Census Bureau, Series A ("Preferred Series") | No change from current race/ethnic specific rates. | In the short run (first five years) the CB projections are highly dependent on interstate migration patterns of the early 1990s. The base year estimates are determined largely by a time series approach that slightly moderates migration in the prior year. Projections for the medium term (2000-2010) are based on patterns during the early 1990s as well as longer-term averages, while projections for the long term (2010-2025) are based on long-term averages in interstate migration. | |
United States Census Bureau, Series B ("Alternative Series") | No change from current race/ethnic specific rates. | The alternative series relies on employment projections developed by the BEA to drive interstate migration. | |
United States Bureau of Economic Analysis | None made in the methodology. | None made in the methodology. | The BEA employment projections are driven by projections in "basic" industries : " [E]mployment in each basic industry was projected on the basis of the historical trend in the States share of employment in that industry nationally. The projections are based on the assumption that the factors that affected the States employment share in the past (for example, relative wage rates and access to inputs and markets) will continue to affect it, but less strongly, in the future, so that the rate of change in employment share will slow" p. M-8. |
CCSCE | Uses DOF projected fertility rates to 2010. Merges California rates downward to CB middle series national projection assumptions after 2010. | Uses DOF mortality rates. | Employment driven based on shifting California share of employment in basic industries. Middle series is based on CB national middle series projections. The high series assumes 20 percent greater growth in U.S. population, with California capturing an additional 20 percent of that growth. The low series assumes 20 percent less growth in U.S. population, with California capturing 20 percent less of that growth. |
UCLA | None made in the methodology. | None made in the methodology. | None made in the methodology. |
The DOF projections of migration are therefore more plausible, at least in the short run, than those of the CB. The DOF projections have the advantage of having been developed after the recovery in domestic migration, and expect "net migration in the period between 1997 and 2003 to rebound from the historically low level of the mid 1990s and to average more than 250,000 per year," and that "following this pent-up demand, migration will slowly decline to about 180,000 after the year 2015." The long term projection "represents a return to a long-term historical annual average, modified by county-specific migration assumptions made in consultation with local government planners and demographers whenever possible."
While the CB projections of migration are too low, at least in the short run, the CB projections of fertility rates are relatively high. The CB leaves Californias comparatively high fertility rates at their current levels throughout the projection period. Even the DOF fertility rates, which assume a slight decrease in fertility for Latinos (see Table B.2), may be too high. Recent data suggest that second generation Hispanic immigrants have substantially lower fertility than first generation immigrants (see Table B.3). As Californias population comes to be increasingly comprised of second and subsequent generation Latinos, we expect that fertility rates could drop substantially.
Table B.2 | ||
DOF Estimated and Projected Total Fertility Rates | ||
1996 |
2040 |
|
White |
1.60 |
1.80 |
Hispanic |
3.49 |
3.10 |
Asian and Pac. Islndr. |
2.03 |
2.06 |
Black |
2.09 |
2.09 |
American Indian |
1.79 |
1.79 |
Note: The total fertility rate is the total number of
children a women is expected to have in her lifetime. |
Table B.3 | |
Average Number of Children Born to Hispanics by Immigrant Generation |
|
1st generation |
2nd generation |
2.19 |
1.34 |
Source: PPIC estimates from the 1995 June CPS, age-standardized for women of childbearing ages. |
The BEA population projections are driven by projection of employment in basic industries. By not explicitly considering fertility, mortality, or migration, the methodology employed in the BEA projections does not accurately model the way a population changes over time. This is a weakness of the series. Although employment opportunities are a key determinant of population growth, the relationship is not necessarily unilateral. Finally, identification of basic industries 40 or more years into the future is suspect.
The UCLA projections represent an adjustment of an earlier DOF projection, and also make no explicit assumptions regarding fertility, mortality, or migration. Instead, the key forecasting variable is the population growth rate, which is subjectively adjusted in the short run according to recent economic conditions, and is simply linearly extrapolated in the long run. The lack of complexity or modeling does not render the UCLA projections wrong; some demographers have argued that the accuracy of population projections produced via complex methods or models is no better than simple extrapolation. However, the lack of detail does limit the usefulness of such projections. The UCLA projections for California are higher than those of any other series, suggesting that a thorough consideration of the drivers of population growth leads to lower population projections for the state. UCLA correctly notes, however, that by historical standards for California, its projected growth rates are relatively low.
The CCSCE projections combine economic models with standard demographic cohort component techniques. The approach is similar to the CB alternative series, though the CCSCE projections enjoy the advantage of being developed after California recovered from the recession of the early 1990s.[4] CCSCE forecasts employment in basic industries only to 2020, and then simply uses shifting state shares of total U.S. employment for longer-term forecasts. This seems prudent, given the questionable ability to identify basic industries several decades into the future. Similarly, CCSCE uses DOF fertility rates to 2020, then merges California fertility rates to U.S. projections for the nation. This results in lower fertility levels at the horizon than in the other series, and again seems plausible if second generation Latinos continue to have substantially lower fertility than first generation Latinos.
CCSCE's high and low series projections show how even a moderate change in assumptions can dramatically influence population projections. CCSCE's low projection assumes slightly slower U.S. growth and that California will capture a slightly lower share, while CCSCE's high projection assumes slightly higher US growth and that California will capture a slightly higher share. The resulting population projections for 2040 show a very wide range, from a projection of 49.8 million in the low series to 67.9 million in the high series.
In general, population projections are notoriously bad at predicting fundamental shifts in demographic phenomena. For example, at the onset of, and even during, the baby boom, demographers did not anticipate the large continuing increases in fertility rates that occurred. Similarly, the continuing decline in fertility rates to well below replacement levels during the baby bust was generally not predicted. California might be at another point in its demographic history when a fundamental demographic shift is occurring. That is, the state might be entering a period when it is no longer a magnet for domestic migrants from other states.
Campbell, Paul R. 1996. Population Projections for States by Age, Sex, Race, and Hispanic Origin: 1995 to 2025, U.S. Bureau of the Census, Population Division, PPL-47.
Center for Continuing Study of the California Economy. 1999. California Economic Growth. Palo Alto, CA.
Johnson, Hans P., and Richard Lovelady. 1995. Migration between California and Other States: 19851994. Sacramento, CA: California Department of Finance and California Research Bureau, California State Library.
Levy, Steve. 1999. Personal communication, March 1999.
State of California, Department of Finance. 1998. County Population Projections with Age, Sex and Race/Ethnic Detail. Sacramento, CA.
U.S. Dept. of Commerce, Bureau of Economic Analysis. 1995. BEA Regional Projections to 2045; Volume 1, States. Washington DC: U.S. Government Printing Office.
[2]The higher child dependency ration projections based on the CB preferred series are due to higher fertility projections. The lower BEA projections are based on a method that does not explicitly consider fertility.
[3]Basic industries are those that provide goods or services for a national market.
[4]The large difference in the CCSCE projections as compared with the CB alternative series suggests that the employment forecasts are highly sensitive to business cycle effects.