FRBSF Economic Letter
1996-38 | December 27, 1996
California residents flocked to other U.S. states during California’s long, deep economic downturn of the early 1990s. The departure of large numbers of Californians contrasted starkly with the longstanding norm of sizable net in-flows of population to the Golden State. Substantial numbers of foreign and domestic migrants made their way to California during the first four decades of the post-WWII period. However, during the first half of the 1990s, a continued sizable net in-flow of population from abroad only partly offset the large-scale net out-migration of Californians to other states.
Most recently, the tide appears to have turned. This Economic Letter discusses preliminary statistics that suggest that the unprecedented net outflow of population from California slowed substantially in 1995 and early 1996. In addition, it reviews simulations of a statistical model of domestic migration which suggest that the recent pickup in California job opportunities will further stem movement from the state.
No single source provides estimates of California migration patterns that are both reliable and up to date. The U.S. Bureau of the Census’s estimates of the components of population change are based on Internal Revenue Service tax returns and are relatively reliable but are published with some time lag. In January of this year, the Census Bureau issued a report indicating that the State of California lost about 1 1/2 million people to other U.S. states during the five-year period ending in July 1995. The annual pattern shows a peak net domestic outflow of about 425,000 residents in fiscal year 1993-94, with the net out-migration slowing to about 380,000 in the twelve months ended in July 1995.
Address changes from California Department of Motor Vehicles (DMV) driver’s license records provide a more timely, but less reliable, indicator of state-to-state migration patterns. The raw driver’s license statistics substantially understate the outflow from California, largely because of incomplete reporting by motor vehicle bureaus in states other than California and also because not all members of households possess driver’s licenses. Our interpretation of calculations in Johnson (1996) suggests that the California driver’s license outflow should be multiplied by roughly a factor of two to convert it to the units of population change; for similar reasons of undercoverage, the driver’s license inflow should be adjusted upward by about one-third. Using such adjustments, the implied net migration pattern from the driver’s license data in the early 1990s is roughly comparable to the pattern in the Census estimates, showing a peak net outflow of about 425,000 persons between July 1993 and July 1994 (Figure 1). The most recent DMV report suggests that the net outflow slowed to about 275,000 persons in the twelve months ending in mid-1996.
Data from the IRS and the California State Department of Motor Vehicles also provide insight as to the origins and destinations of California migrants. The large industrial states of the Midwest and Northeast, including New York, Illinois, and Michigan, typically have sent substantial numbers of migrants to California. During the early 1990s, even as large numbers of Californians were leaving the state, sizable net in-migration to California from these same states was recorded. In contrast, net out-migration from California to other western states accelerated during the 1990-1994 period, most notably to Nevada, Oregon, and Washington. The more favorable economic opportunities available in neighboring states during the early 1990s represented a significant “pull” for many Californians.
Californians departing the state during the early 1990s tended to be young, low-skilled white adults possessing less than a college education (see Johnson and Lovelady (1995)). Domestic migrants entering California were fewer but tended to be significantly better educated than those departing the state, so that, on net, California experienced only a slight loss of college graduates during that period. Almost three-fourths of the domestic migrants entering the state during the early 1990s were employed in white collar occupations.
As is evident from Figure 2, California domestic migration appears to move in tandem with differences in employment opportunity between California and other states. For example, the sharp increase in net domestic out-migration from California during the early 1990s was coincident with a significant fall-off in employment opportunity in California relative to that of other states, as measured by the US-California unemployment rate differential. The unemployment rate gap grew to about 2 1/2 percentage points in 1993 and 1994, narrowed a bit in 1995, and narrowed further to about 1 3/4 percentage points by mid-1996.
In our statistical model of interstate migration (Gabriel, Mattey, and Wascher 1995), we estimated the extent to which migration between two states depends on such a measure of relative job opportunities, along with measures of wage and housing expense differentials. The model also allowed differences in locational amenities to affect migration. Furthermore, regardless of locational amenities, people in the various states of origin were assumed to have different propensities to migrate, depending on, for example, the age structure of the state population, which is important to life-cycle effects on migration. For instance, research shows that moves are more common for individuals in their 20s and at times of change in employment or marital status. Mobility declines as individuals mature, have families of their own, and develop job, community, and social ties. Further, migration between any particular pair of states should vary inversely with the distance between those areas, as do the transactions costs, informational costs, and psychic costs of moving.
Our statistical estimates confirmed the strong role of higher expected economic returns in inducing net in-migration to an area. More specifically, we found that low unemployment rates in the destination state relative to the origin state tended to prompt strong migration between those areas, particularly if the states were near each other. Accordingly, given the widespread fall-off in labor demand in California during the early 1990s and booming economies in other western states, the model explains the rapid out-migration from California relatively well. In terms of the relative importance of migration determinants, changes over time in unemployment differentials dominated the fit of the model for California net migration over the sample period. In contrast, movements in relative state wage or house price differentials had less of an estimated effect on California migration flows.
Figure 3 plots actual California net migration flows over the 1981-1993 period, together with fitted values for California net migration over the same period as derived from our estimated state-to-state migration model. The estimated population flows fit the actual pattern of net domestic migration relatively well, particularly in the years since 1988. The model did well at predicting the acceleration in net out-migration from California during the 1989-1993 period. During those years, the unemployment rate differential between California and other western states widened substantially. Also, wage growth in California was relatively weak, whereas other western states were bidding more aggressively for the labor force. Out-of-sample simulation of the migration model for the subsequent three years suggests a moderate easing in net out-migration from the state. The implied reduction in the net out-flow of people from California is largely driven by the observed narrowing in unemployment rate differentials between California and other states. Given recent historical values of the economic variables, the fit of the model implies that net out-migration from California decelerated to about 250,000 persons in 1996, which is broadly consistent with the latest reading from the adjusted DMV data.
Figure 3 also presents a simulation of California net migration flows under the assumption of a reversion of unemployment rate, wage, and house price differentials between California and other states to average levels observed during the 1981-1992 period. During those years, the average California unemployment rate was about 1/4 percentage point above the national average, whereas in mid-1996 the California unemployment rate still was about 1 3/4 percentage points above the national average. The model suggests that a relative strengthening in the California economy–on the order of a
1 1/2 percentage point narrowing of the unemployment rate differential–would imply a further substantial slowing in net out-migration from California to less than one-half of the simulated 1996 pace.
In summary, better job opportunities in neighboring states appear to have prompted the outflow of substantial numbers of young, low-skill California residents during the first half of the 1990s. The more recent strengthening of California labor markets has contributed to some turnaround in those migration flows.
Stuart A. Gabriel
Professor of Finance and Business Economics
University of Southern California
Joe P. Mattey
Gabriel, Stuart A., Joe P. Mattey, and William L. Wascher. 1996. “Compensating Differentials and Evolution in the Quality of Life among U. S. States.” Federal Reserve Bank of San Francisco Working Paper 96-07.
______, ______, and ______. 1995. “The Demise of California Reconsidered: Interstate Migration over the Economic Cycle.” FRBSF Economic Review 2, pp. 30-45.
Johnson, Hans P. 1996. Undocumented Immigration to California: 1980-93. San Francisco: Public Policy Institute of California.
______, and Richard Lovelady. 1995. “Migration between California and Other States: 1985-1994.” Report of the California State Library and the California Department of Finance.
Opinions expressed in FRBSF Economic Letter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System. This publication is edited by Sam Zuckerman and Anita Todd. Permission to reprint must be obtained in writing.
Please send editorial comments and requests for reprint permission to
Attn: Research publications, MS 1140
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120