Printer Friendly

Migration, unemployment, and wages: the case of the California San Joaquin Valley.

I. INTRODUCTION

The recent experience of the California San Joaquin Valley (SJV) labor market appears to be at odds with basic economic principles. Labor market theory suggests that we should observe out-migration from regions experiencing relatively high unemployment levels and relatively low wage levels, and in-migration to regions experiencing relatively low unemployment levels and relatively high wage levels. The argument is that through labor mobility, regional disparities should disappear as wage and unemployment levels converge across regions in the long run. Contrary to this optimistic expectation, however, the SJV has continually received large numbers of in-migrants for the past two decades despite two distinct facts. First, the unemployment rate in the region has not only been chronically high, but it also has exceeded the one for California. Second, the average wage level in the SJV has significantly lagged behind the State's average.

By examining county level data for the past two decades, the analysis in this paper is built around two main questions that aim to gain a better understanding of the SJV labor market and the migrants' decisions to move there. First, in what proportion does local employment growth reduce local unemployment, increase labor force participation, and attract outsiders who will likely take the newly created jobs? This issue is explored through a labor supply decomposition of employment growth that measures labor supply sources. Second, to what extent regional migration rates respond to regional relative wages and unemployment differentials? In addition to factors suggested by traditional theoretical models and in order to better capture differences in the standard of living, the analysis also includes factors likely to attract migrants such as relative housing price and government benefits, which might increase immigrants' utility, and factors likely to drive migrants away such as crime and pollution rates, which might decrease immigrants' utility. This empirical issue is examined by estimating a migration equation. Understanding the labor market dynamics of the SJV might shed light in the design and implementation of development policies aimed at reducing unemployment.

Results provide evidence that market forces alone are insufficient to correct regional unemployment and wage disparities and that non-market factors play a significant role in the decision to migrate. Three main findings are offered. First, in-migrant workers fill most of the newly created jobs. Second, migration seems unresponsive to the unemployment level but responsive to changes in farm income which, it is shown, is the relevant measurement of earnings differentials in a heavily agricultural driven region such as the SJV. Third, migration is sensitive to government-based benefits, property crime rates, and housing prices. The paper is organized as follows. As a general background, Section II illustrates some major trends and highlights a number of salient features of the SJV labor market during the last two decades. Section III discusses the labor supply responses to changes in employment in terms of changes in population, unemployment, and labor force participation rates. Section IV examines the impact of market and nonmarket factors on regional migration rates. Section V discusses some implications for regional economic policy and concludes the paper.

II. THE CALIFORNIA SJV AND ITS UNIQUE LABOR MARKET

The SJV is known nationally and internationally for the breadth and productivity of its mostly agricultural-based economy. Underlying this clear comparative advantage however, for decades this regional economy has faced an assorted list of shocks--from increased international competition to technological advances to in-migration--which have forced it to be in constant flux and have produced significant and sometimes long-lasting effects, particularly on its labor market. (1) Two indicators where this is most evident are unemployment, characterized for being among the highest in California and in the U.S., and the wage level, which has significantly lagged the one for the State. Following Pissarides and McMaster (1990) and in order to show unemployment differentials over time, an inequality index (II) is computed as follows:

II = [([8.summation over (i=1)][[LF.sub.i]/LF][(ln[[UN.sub.i]/UN].sup.2]).sup.[1/2]],

where [LF.sub.i], is the labor force in county i, [UN.sub.i] is the unemployment level in county i, and LF and UN take their respective values for the whole state of California. The index takes the value of 0 when the regional unemployment levels are equal to the California average in all counties. Figure 1 shows that although unemployment inequality between California and the SJV declined in the past few years, it has remained significantly high for the past couple of decades. During this period, the unemployment rate in the SJV has exceeded the one for California by an annual average of 5.6 percentage points.

[FIGURE 1 OMITTED]

A similar index was computed to gauge wage differentials between California and the SJV, but using employment shares as weights. (2) Although wage inequality is less dramatic than unemployment inequality, the index shows a similar pattern. Around 1999, wage inequality started to decline; yet, recent wage inequality is still higher than it was 20 years ago. During this period, the real wage growth rate in California has exceeded the one for the SJV by an annual average of 0.35 percentage points. Thus data show that the SJV has consistently exhibited higher than average unemployment rates and lower than average wage rates for the past two decades.

The high employment share on agricultural activities in the SJV--close to 10% in average for the past couple of decades--has been frequently blamed not only for the high and persistent unemployment rate, but also for the relatively high poverty incidence in the region due to low wages paid by the agricultural sector (Martin and Taylor, 2000).

However, despite evidence that indicates agriculture as an important contributor to high unemployment rates (particularly during the cold seasons), there is also evidence that suggests nonagricultural factors contribute to migration decisions. For instance, as Feasel and Rodini (2002) have shown, once the agricultural seasonal elements are removed, unemployment rates across counties in the SJV are still substantially high. Therefore, in addition to agriculture, there must be some region-specific factors influencing the high and persistent unemployment rate as well as other variables. Perhaps one of their most insightful findings of Feasel and Rodini (2002) is that individuals living in higher unemployment areas also have a lower propensity to migrate, which is explained by social and demographic factors. Or as the authors explain, even within a well-defined labor market--such as the SJV--labor mobility does not ensure that regional disparities in unemployment will disappear. (3) Furthermore, although the overall relative wage level in the SJV shows a gap in relation to the whole state, it is possible that given the significant size of the agricultural sector in the region, the relevant variable for migration decisions is earnings derived from agricultural activities. Since no agricultural wage data are available, relative farm income is calculated instead. As showed in Figure 2, this indicator reveals that on average, farm income in the SJV has consistently exceeded the one for California by almost one-third for the past two decades. Among the nonmarket factors possibly influencing migration decisions, relative housing price is a strong candidate. Figure 2 shows that on average, median house prices have consistently been below the prices for California by almost a half for the past two decades. Both, relative farm income and relative housing prices are empirically tested later.

[FIGURE 2 OMITTED]

In addition, a number of salient and unique features of the SJV labor market are shown in Table 1. First, during the past couple of decades, total population has grown more rapidly in the SJV (2.51%) than in California (1.73%). Although some of this rapid growth is probably due to natural population growth, it is also likely due to relatively higher in-migration, domestic migration in particular. Indeed, while California received an average of two migrants per thousand people, the SJV received eight migrants.
TABLE 1

1985-2006 Average: Selected Labor Market Indicators

 CA SJV

Total population growth rate (%) 1.73 2.51
Labor force participation rate (%) 49.04 44.34
Labor force growth rate (%) 1.63 2.20
Employment growth rate (%) 1.77 2.46
Unemployment rate (%) 6.56 12.10
Employment share on agriculture (%) 1.56 9.46
Real wages (1982-1984 = 100) $21,024 $15,262
Real wages growth rate (%) 0.85 0.50
Net domestic migration/population (000) 2.04 8.19
Total migration/population (000) 8.41 12.89
2006 total population (000) 37,444 6,929

 FRE KER

Total population growth rate (%) 2.14 2.49
Labor force participation rate (%) 48.31 44.67
Labor force growth rate (%) 1.75 2.06
Employment growth rate (%) 1.96 2.28
Unemployment rate (%) 12.19 11.33
Employment share on agriculture (%) 7.92 5.94
Real wages (1982-1984 = 100) $15,178 $17,008
Real wages growth rate (%) 0.62 0.10
Net domestic migration/population (000) 3.13 8.55
Total migration/population (000) 8.36 11.85
2006 total population (000) 909 796

 KIN MAD

Total population growth rate (%) 2.18 3.29
Labor force participation rate (%) 37.17 44.81
Labor force growth rate (%) 2.36 3.05
Employment growth rate (%) 2.52 3.28
Unemployment rate (%) 12.04 11.92
Employment share on agriculture (%) 11.25 14.23
Real wages (1982-1984 = 100) $15,630 $13,879
Real wages growth rate (%) 0.60 0.73
Net domestic migration/population (000) 10.12 16.87
Total migration/population (000) 13.49 19.88
2006 total population (000) 149 147

 MER STA

Total population growth rate (%) 2.27 2.66
Labor force participation rate (%) 42.49 46.49
Labor force growth rate (%) 1.83 2.43
Employment growth rate (%) 2.03 2.87
Unemployment rate (%) 13.16 12.17
Employment share on agriculture (%) 12.10 5.83
Real wages (1982-1984 = 100) $14,207 $15,999
Real wages growth rate (%) 0.42 0.56
Net domestic migration/population (000) 2.89 12.45
Total migration/population (000) 11.85 16.63
2006 total population (000) 249 519

 SJO TUL

Total population growth rate (%) 2.43 2.01
Labor force participation rate (%) 45.00 45.82
Labor force growth rate (%) 2.22 1.95
Employment growth rate (%) 2.51 2.19
Unemployment rate (%) 10.28 13.49
Employment share on agriculture (%) 6.60 11.83
Real wages (1982-1984 = 100) $16,796 $13,397
Real wages growth rate (%) 0.46 0.55
Net domestic migration/population (000) 9.65 1.86
Total migration/population (000) 14.97 6.10
2006 total population (000) 674 425

Sources: California Employment Development Department, California
Department of Finance, Demographic Research Unit, BEA RE1S, US Census
Bureau.


Second, the SJV characterized by relative high unemployment rates and low wages is theoretically supposed to have experienced net out-migration. Yet, the SJV has experienced more in-migration relative to California. Figure 3 shows that net domestic migration in the SJV as a percent of the total population--with the exception of the period 1996-2000--has significantly exceeded the same indicator for California. Actually, for the past 2 years, while the SJV has been a net importer of immigrants California has been a net exporter. It is also worth noticing that in contrast, foreign legal in-migration is higher for California. While the SJV received an average of 4.7 international legal migrants per thousand people, California received 6.3. Unfortunately, due to lack of reliable data, these calculations do not include illegal immigrants. Thus, the contribution of in-migration to population growth--particularly for the SJV--is likely to be underestimated. Complementing this picture though, using County-to-County IRS data, Table 2 shows that almost 75% of in-migrants to the SJV come from other California counties and less than 25% come from a different state within the U.S. In fact, supporting the anecdotal evidence, one out of four Californian in-migrants to the SJV come from either one of two adjacent regions, the Bay Area (defined as Marin, San Francisco, San Mateo, Santa Cruz, Alameda, and Contra Costa counties) and Southern California (defined as San Diego, Imperial, Riverside, and Orange counties). From the Bay Area, Northern SJV counties (Fresno, Madera, Merced, Stanislaus, and Joaquin) draw more than 15% of the total Californian in-migrants. Similarly, from Southern California, Southern SJV counties (Kern, Kings, and Tulare) draw almost 10% of the total Californian in-migrants. Data also show that 22.1% of the total in-migration to the SJV occurs between the same SJV counties, which suggest modest worker mobility within the region. Most Northern counties, Stanislaus and Merced in particular, draw almost 40% of the within SJV migration. Interestingly, Merced, which exhibits the largest in-migration average rate (82.4%), is also the county with the highest average farm income and the fourth lowest median home price.

[FIGURE 3 OMITTED]
TABLE 2
Sources and Destination of SJV Migrants: 1985-2006 Average

 SJV FRE KIR KIN MAD

In-migration (%)

Same state non-SJV county 74.3 72.1 63.2 58.3 84.2

Different state 23.5 26.2 33.7 33.3 15.2

Foreign 2.2 1.8 3.1 8.4 0.6

Out-migration (%) Same state 65.6 64.7 51.5 54.5 75.7

non-SJV county Different state 32.7 34.5 45.9 38.9 23.6

Foreign 1.8 0.9 2.6 6.7 0.7

 MER SJO STA TUL

In-migration (%)

Same state non-SJV county 82.4 80.8 79.5 74.0

Different state 16.8 18.3 19.7 25.0

Foreign 0.8 0.9 0.8 1.0

Out-migration (%) Same state 72.1 71.1 68.3 66.8

non-SJV county Different state 27.2 28.0 30.9 32.4

Foreign 0.7 0.9 0.8 0.8

Sources: County-to-County Data, US Internal Revenue Service.


In terms of out-migration, more than one-third of migrants leaving the SJV move to other states or other countries. All these observations suggest that differences in market and nonmarket factors between the SJV and the rest of the state are jointly driving migration patterns.

Third, rapid population growth was accompanied by declining labor force participation rates in the SJV and California. Indeed, the decline in labor force participation rates for the past 20 years has been more pronounced for the SJV--where this indicator was 45.04% in 1985 and only 41.8% in 2006 than for California--where it was 49.16% in 1985 and 47.4% in 2006. However, job creation has been relatively more successful in the SJV than in California for the past 20 years. Employment growth in the SJV has exceeded the one for California by an annual average of 0.69 percentage points. An intriguing question here is: Given the relatively large amount of in-migrants to the SJV, who has filled the newly created jobs, local or in-migrant workers? This issue is explored in more detail in the next section.

III. THE COMPOSITION OF LABOR SUPPLY RESPONSES

Over the last couple of decades, state and local governments of regions suffering from high unemployment rates have more aggressively intervened in their labor markets to ameliorate the negative effects of joblessness through either government-run or subsidized economic development programs (Bartik, 1991). However, while there is some agreement about the job-creation effects of economic development policies, there is much less agreement on who fills the newly created jobs: either local or in-migrant workers. This is a question that has been central to the economic development policy debate: In what proportion does local employment growth reduce local unemployment, increase labor force participation, and attract outsiders who will take the newly created jobs? Answering this question would assist in the formulation of regional economic policies, and it would also shed light in understanding migration paths and persistent unemployment levels.

Previous attempts to answer this key question have produced mixed results. (4) Bartik (1991), for example, found that development policies aimed at job creation can reduce long-run unemployment among metropolitan area residents. He argues that even if workers are highly mobile across regions, "currently unemployed residents have a short-run advantage over potential in-migrants because in-migration does take some time to respond to shifts in labor demand." He concludes that state and local economic development policies can achieve their goal of reducing local unemployment and that local workers are likely to fill newly created jobs. Similarly, Eberts and Stone (1992) found that increases in the labor force participation rate account for the largest proportion of the change in the labor supply concluding that it is local workers getting the newly created jobs. Blanchard and Katz (1992), on the other hand, in their examination of regional evolutions for all states in the U.S., found that the dominant adjustment mechanism to a labor demand shock is labor mobility. A corollary of their study is that the newly created jobs at the state level are nearly fully filled by in-migrants, at least in the short run. Similarly, Houseman and Abraham (1990) using annual state-level employment data, found that more than half of the newly created jobs are filled by in-migrant workers. In a less extensive study, Eberts and Stone (1992) found that, after a permanent increase in the number of jobs available--a positive labor demand shock--a new equilibrium characterized by a higher employment level is reached after approximately 10 years, suggesting that the effects of the shock "are felt in local labor markets for long periods." Their study, however, does not analyze its effects on local unemployment or labor force participation rates.

All of these studies, however, were conducted at the state or metropolitan level. The only study available at the county level is the work of Shields and Novak (2000) who examine the dynamic labor responses to employment shocks for 67 counties in Pennsylvania. Their major finding, which resonates with the results obtained by Blanchard and Katz (1992), is that employment growth benefits people originally residing outside the region. In other words, the new jobs created by local economic development efforts mostly go to in-migrant workers. Similarly, their explanation for this finding is labor mobility. According to the authors, the effectiveness to increase employment among local workers is lessened because often, the "best" workers are quite mobile and capable to respond to changing labor conditions even when they are away from the source of new employment. In a couple of relatively recent papers, Partridge and Rickman (1999, 2003) attempted to disentangle the local versus in-migrants issue using a slightly different approach. The subject they examine is whether people follow newly generated jobs into regions, or whether jobs follow newly arrived migrants. Based on an analysis of the lower 48 U.S. states, they conclude that whether people or jobs come first depends upon the state and period under examination, since the numerous factors that are at work shifting labor demand and supply vary over time and space. These studies do not offer conclusive evidence in terms of the local versus in-migrants workers debate.

A. Supply Sources of Employment Growth

To quantify the newly created jobs going to the unemployed, changes in population and new entrants and reentrants to the labor force in the SJV for the period under examination, a basic model of decomposition of employment growth is used. This approach is intended to reveal the labor supply responses to changes in employment, that is, to estimate the relative contribution of the three labor supply components to short-run changes in employment. Following Houseman and Abraham (1990) and Houseman (1995), employment changes can be separated into three labor supply components according to the following identity:

(1) [E.sub.i] [equivalent to] [P*.sub.i] [LFPR*.sub.i] (1 - [UR.sub.i]).

Where [E.sub.i] is the employment level in county i, which is equal to its population [P.sub.i] times its labor force participation rate [LFPR.sub.i] multiplied by one minus the unemployment rate [UR.sub.i]. Therefore, employment changes can be expressed as the sum of the changes in the population (interpreted in the literature as a proxy for the growth of in-migrants), changes in the labor force participation rate (new entrants and reentrants), and changes unemployment (interpreted as the growth of local workers getting employment):

(2) [DELTA][P*.sub.i][LFPR.sub.i]91 - [UR.sub.i]) + [DELTA][LFPR*.sub.i][P.sub.i](1 - [UR.sub.i]) - [DELTA][UR*.sub.i][LFPR.sub.i] [approximately equal to][E.sub.i].

Dividing both sides of the equation by employment [E.sub.i] and renaming terms:

(3) [p.sub.i] + 1f[pr.sub.i] - [ur.sub.i] [approximately equal to] [e.sub.i].

Where [e.sub.i] is employment growth, [p.sub.i] is population growth, [lfpr.sub.i] is the growth in the labor force participation, and [ur.sub.i] is the change in the unemployment rate. The labor supply decomposition of the SJV employment growth is given in Table 3.
TABLE 3

Labor Supply Sources of SJV Employment Growth

 Employment Population Labor Force Unemployment
 (e) (P) (lfpr) (u)

1986 3.20 2.53 -0.11 -0.67
1987 2.66 2.85 -2.03 -1.66
1988 4.69 3.12 1.48 -0.04
1989 2.89 3.01 -0.48 -0.33
1990 5.79 3.75 3.30 1.16
1991 -0.27 3.76 -1.16 2.44
1992 1.60 2.72 1.29 2.03
1993 1.31 1.98 -0.54 0.10
1994 0.58 1.52 -2.32 -1.20
1995 1.82 1.52 -0.15 -0.37
1996 1.70 1.07 -0.31 -0.80
1997 1.35 1.51 -0.67 -0.45
1998 1.17 1.09 -0.02 -0.08
1999 1.47 1.77 -1.31 -0.89
2000 8.08 2.01 2.11 -3.30
2001 1.15 2.67 -0.95 0.49
2002 2.18 2.47 0.91 1.07
2003 1.40 2.76 -0.98 0.31
2004 1.54 2.65 -1.92 -0.75
2005 3.39 2.46 -0.56 -1.33
2006 1.73 2.38 -1.45 -0.76

Source: Calculations based on data from the California Employment
Development Department.


From the first column, it is clear that employment on average grew significantly in the late 1980s and, concurrent with the U.S. recovery after the recession of 1990-1991, employment in the SJV has shown positive growth rates since.

For the past 20 years, employment growth has been supplied by sizeable increases in the population. In fact, since the average growth rate of the population (2.36% annual) has exceeded the one for the labor force (2.08% annual), the labor force participation rate has decreased slightly--notice the negative growth rate of the labor force participation rate in 16 of the past 20 years. Furthermore, for the past two decades, total migration in the SJV (which includes foreign legal migration and net domestic migration) has accounted for almost 50% of the population increases. That is, for every two new persons in the SJV, one is an immigrant. It is also worth drawing attention to the fact that it is not population growth that the labor market needs to absorb to reduce unemployment but rather, the growth in the labor force. The average annual employment growth for the period (2.35%) has exceeded the growth of labor force (2.08%). Thus, with the exception of the recession years (1990-1993 and 2001-2003), the unemployment rate has indeed decreased.

IV. THE DETERMINANTS OF MIGRATION TO THE SJV

Given the relatively high domestic in-migra-tion numbers that the SJV has experienced in the past 20 years, it is worth exploring the determinants of such labor mobility. In particular, it would be important to examine the factors explaining in-migrants' decision to move into the SJV in the first place; and then, to examine why migrants seem reluctant to move out of region despite its relative higher unemployment rates and relatively lower wage rates. Based on traditional labor market theory, the simplest model tested in this section assumes that in-migration should negatively respond to relative unemployment and positively to relative wage rates. It is also worth noticing that as Armstrong and Taylor (2000) explain, the classical model of migration is constructed under the extreme assumptions of perfect information, price flexibility, and costless labor migration--among others--when in fact we know that the determinants of migration are more complex. Therefore, given these weaknesses and the evidence shown in the previous sections, the basic model is expanded based on the following three considerations. First, given the significant size of the agricultural sector in the region, the model tests not only the relative overall wage ratio, but also tests a proxy that intends to capture relative earnings in the agricultural sector. Second, in addition to employment opportunities and better paid jobs, as labor market theory indicates, migrants might also consider additional factors that better capture differences in the standard of living. These include, for example, housing prices, government services, quality of life, and other economics benefits. Hsing (1998) proposes a model where potential migrants maximize utility from not only employment and earnings but also from government services, and minimize disutility from negative regional factors. This approach has been applied extensively in examining interstate migration. For example Greenwood (1975) argues that the availability of public goods should be included as an important determinant in the migration decisions. In addition, Cebula (2005) reports that in-migration is an increasing function of the numbers of sunshine days, which suggests that the quality of life can be a determinant of migration decisions, at least at the state level. Third, but related to the second point, most migration decisions are not made by single workers, but by families. As Mincer (1978) has argued, the migration decision should not be based on whether a particular member of the household is better off at the destination than at the origin, but on whether the whole family is better off. In this regard, using survey data, DaVanzo (1978) finds that families whose heads are unemployed or are dissatisfied with their jobs are more likely to move than those whose heads are not searching for jobs. However, as Bartik (1991) points out, resistance to move out despite adverse economic circumstances can be influenced by moving costs that are not necessarily monetary. For example, the development of a 'sense of place'--a complex of habitual buildings, businesses, social relationships, and family--can produce large enough benefits to become a deterrent to move out. This last consideration is particularly relevant for the SJV since large numbers of migrants are His-panics who, as Winters et al. (1999) explain, have developed community and family networks that provide direct assistance in the form of money, housing, transportation, and food. Although Muller (2003) has also tied discrimination to migration and unemployment, this issue falls beyond the scope of this paper. Therefore, following Pissarides and McMaster (1990), the model to test is the following:

[m.sub.[i,t]] = [OMEGA]([m.sub.[i,t-j]], [u.sub.[i,t-j]], [w.sub.[i,t-j]], [g.sub.it], [hp.sub.it], [n.sub.it], [7.summation over (i=1)] [D.sub.i]),

where [m.sub.[i,t]] is the migration rate of county i at year t defined as the ratio of net domestic migration to total population lagged 1 year. [m.sub.[i,t]-j] is the migration rate of county i at year t - j, where j are the number of lags. [u.sub.i,[t-j]] is the unemployment ratio of county i at year t -j, and is defined as unemployment rate in county i divided by the unemployment rate in California. [w.sub.i,[t-j]] is the wage ratio of county i at year t -j, and is defined as the average wage per job in county i divided by the average wage per job in California. To capture the effect of the heavily agricultural-based regional economy of the SJV on migrations patterns, the model also tests for the relative farm income, [fi.sub.[i,t-j]], which is defined as the farm income in county i divided by farm income in California. The data are from the Bureau of Economic Analysis's (BEA) Regional Economic Information System (REIS). [hp.sub.it] is defined as the median housing price in county i divided by the median housing price in California. The data are from the California Association of Realtors. [g.sub.it] is a set of public policy variables intended to capture the potential government benefits that migrants may obtain by moving to those areas that provide them more generously. These variables, also based on the BEA's REIS are: ui, unemployment insurance compensation; imb, income maintenance benefits, and tt, total current transfers receipts of individuals from governments at all levels. All these variables are ratios of the indicator for the county to the same indicator for California. [n.sub.i,t] is a set of variables that capture negative regional factors which, presumably reduce the incentives to migrate to those regions where they are more prevalent. These variables are: ozone, the number of days per year that exceed 8-h national standards for ozone levels as reported in the California Air Resources Board; rvc, the rate of violent crime and rpc, the rate of property crime. Both are variables per 100,000 people as reported in the California Department of Justice. [D.sub.i] are county dummy variables. Several versions of this function are estimated by ordinary least squares based on pooled time series cross-section data for the eight SJV counties for the period 1985-2005. Each model includes dummy effects.

A. Empirical Results

The overall fit is good with an [[bar].R.sup.2] of near 0.65 for the full model (Table 4). Significance tests for the county dummies were conducted and it was found that joint equality to 0 of county-specific effects is rejected in each case. Although the table only reports models with one lag for the migration, unemployment, and wage variables, two and three lags were tested but they were not significant. Also, to check for robustness, the ratio of net domestic migration to the labor force lagged 1 year (rather than to the population lagged 1 year as defined earlier) was tested as an alternative measure for the migration rate. Although the magnitude of the estimates differs slightly, the results are not qualitatively sensitive to this alternative measure. In all models, heteroskedasticity-consistent robust standard errors are computed. The F-sta-tistic is significant at the .99 level and the Dur-bin-Watson statistic indicates that there is no first-order autocorrelation.
TABLE 4
The Determinants of Migration in the San Joaquin Valley

 Dependent Variable: Migration Rate

Explanatory (1) (2) (3) (4)
Variables

m(-l) 0.632(8.75) 0.537(6.55) 0.559(6.88) 0.524(6.56)

u(-1) 0.202(0.69) 0.318(0.23)

w(-l) 1.678(0.88) 3.05(1.18)

Change 0.101(0.12) 0.911(0.51) 0.908(1.01)
in u(-l)

Change 16.664 (2.89) 13.46(2.94) 14.25(3.22)
in w(-l)

ui 0.517(2.28)

imb 2.32(2.39)

tt 7.420(2.10)

rvc

rpc

Ozone

Housing

Farm
income(-l)

Change in
farm
income(-1)

County YES YES YES YES
effects

Prob > F 0.000 0.000 0.000 0.000

[[bar.R].sup.2] 0.45 0.55 0.58 0.57

DW 1.77 1.71 1.85 1.84

Obs. 152 152 144 144

 Dependent Variable: Migration Rate

Explanatory Variables (5) (6)

m(-l) 0.533(6.45) 0.521(5.97)

u(-1)

w(-l)

Change in u(-l) 0.440(0.48) 0.343(0.22)

Change in w(-l) 13.618(2.77) 9.01(2.12)

ui 0.385(1.92) 0.287(1.91)

imb 1.71(1.97) 1.44(1.95)

tt

rvc 0.001(0.01) 0.001(0.01)

rpc -0.001(-1.91) -0.001(-1.88)

Ozone 0.003(0.76) 0.002(0.55)

Housing -0.444(-3.51)

Farm income(-l)

Change in farm income(-1)

County effects YES YES

Prob > F 0.000 0.000

[[ber.R].sup.2] 0.59 0.64

DW 1.84 1.84

Obs. 144 122

 Dependent Variable: Migration Rate

Explanatory Variables (7) (8)

m(-l) 0.282(3.30) 0.199(2.97)


u(-1) 0.127(0.14)

w(-l)

Change in u(-l) 0.043(0.19) 0.022 (0.09)

Change in w(-l)

ui 0.112(1.90)

imb 1.51(1.95)

tt

rvc 0.001(0.01)

rpc 0.001(-1.85)

Ozone 0.001(0.29)

Housing -0.386(-2.42)

Farm income(-l) 4.25(3.22) 3.83(2.77)

Change in farm income(-1) 0.004(0.073)

County effects YES YES

Prob > F 0.000 0.000

[[ber.R].sup.2] 0.49 0.65

DW 1.82 1.84

Obs. 152 122

Notes: Numbers in parentheses are t-statistics computed with robust
standard errors.


Models 1 through 6 reveal interesting patterns about the way migration responds to wage and unemployment. Results indicate that neither the change in the unemployment differential nor the level of the unemployment differential influences migration. Also, Models 1 and 2 indicate that the change in the regional wage ratio influences migration rates but not the level of the wage ratio. Thus, evidence suggests that although there are no unemployment-induced net emigration flows, wage-induced net in-migration flows may exist--the estimated coefficient for the change in the regional wage ratio variable is consistently significant for all six models. The finding that migration does not respond to high unemployment rates seems not only counterintuitive, but as discussed earlier, at odds with economic theory. A plausible explanation to the seeming resistance of workers to leave the region is that some of the conditions implied by the classical model are not fulfilled in practice, specifically the assumption that there are no costs associated to interregional mobility, neither financial nor social. In other words, as Polese (1981) has argued, the role of migration in reducing regional disparities may be less certain than is often assumed. Other authors have suggested an alternative explanation. DaVanzo (1978), for example, argues that the inability to find a significant relationship between migration and unemployment may be due to the way unemployment is measured. Unfortunately, most alternative measures rely on microdata, and therefore studies such as DaVanzo (1976) and Fields (1976), which find a positive relationship between unemployment and emigration, are not comparable to findings in this study.

As discussed before, the approach here is to examine whether additional migration determinants are strong enough to induce residents to stay and nonresidents to in-migrate. Models 3 through 6 show the results of testing additional variables. Model 3 indicates that relatively higher unemployment insurance compensation (ui) and higher income maintenance benefits (imb) induce in-migration (both coefficients are statistically significant at the 95% level). It is worth noticing that the average unemployment insurance compensation for the past 20 years in the SJV is $186 per 1,000 people, while the same indicator for California is only $103, which suggests more generous public assistance programs in the SJV than in California as a whole. This result corroborates the findings of Goss and Paul (1990) according to which federal discretionary unemployment-compensation programs likely serve to retard emigration of those who are involuntarily unemployed. Interestingly, similar evidence has also been found at the international level. For example, Heitmueller (2005) for a group of European nations finds that unemployment benefits work as some sort of insurance devise for (risk averse) migrants, retarding their decision to move. Alternative measurements of government benefits that include federal education and training assistance, as well as medical benefits were tested, but results were not significant and thus are not reported. However, as reported in Model 4, total current transfers receipts of individuals from governments at all levels tt--which includes all government benefits--is significantly strong with a coefficient of 7.74 and t = 2.10. Confirming the finding in Model 3, this result suggests that government-based benefits in the SJV, which are more generous relative to the rest of the State of California, significantly induce in-migration (or deter out-migration), controlling for the unemployment and wage ratios.

To examine whether other migration determinants related to the standard of living are strong enough to induce residents to leav-e--and perhaps discourage nonresidents to in-migrate to the SJV--Models 5 and 6 show the results of testing additional variables. Model 5 shows that only the rate of property crime (rpc) has the expected sign and is significant--the coefficient for violent crime (rvc) is not significant though. Nevertheless, this coefficient is noticeably small, which indicates a very low response of migrants to property crime. In contrast to this finding, Cebula and Payne (2005) find that in-migration is a decreasing function of violent crime. The report, however, is different in two regards. First, it does not test for property crime and second, it uses gross state in-migration data, thus it is not fully comparable. In addition, the models test for the number of days per year exceeding the 8-h national standards for ozone levels (ozone) and for the relative median housing price (housing). The coefficient for ozone is not significant indicating that the quality of air (as measured by this indicator) is not a deterrent to in-migration. However, as shown in Model 6, the coefficient for housing indicates that, as expected, migration responds negatively (positively) to high (low) relative housing prices. (5) It is worth noticing that on average, median house prices have consistently been below the prices for California by almost a half for the past two decades. This result concurs with the findings of Berger and Blomquist (1992) who, using county level data, report housing prices as significant determinants of migrants when choosing a destination. Finally, Models 7 and 8 are similar to Models 2 and 6, respectively, except that instead of testing for the relative overall wage, they test for the relative farm income. Results indicate that the level of the relative farm income influences migration rates but not the changes in relative farm income. The coefficient for relative farm income is positive and statistically significant in both models, which shows that the in-migration rate increases as the average farm income in the SJV increases relative to the one for California.

V. SUMMARY AND POLICY IMPLICATIONS

Although theoretical labor market models emphasize the role of unemployment and wage rates in determining regional migration patterns, recent empirical studies have also included nonmarket factors in examining labor mobility. Following a similar approach, this study attempts to disentangle the apparent conflict between the large number of migrants hosted by the California SJV during the past two decades (despite significant high unemployment rated and low wage rates) and the theory. Using data for eight counties composing the SJV labor market, the analysis yields three significant conclusions that increase our understanding of the migration patterns to the SJV. First, the decomposition of employment growth reveals that the supply of in-migrants in the regional economy is the most important determinant for the past 20 years. Both, changes in the population and changes in the labor force participation rate, are significant determinants of employment growth. Thus, results show that in-migrant workers fill most of the newly created jobs. Based on county-to-county data, it is also shown that almost 75% of in-migrants to the SJV come from other California counties. Second, a simple migration equation for the SJV labor market shows that nonmarket factors are relevant in explaining migration patterns to the SJV. Results suggest that migration is unresponsive to the unemployment level but somewhat responsive to changes in the wage rate. In addition, migration responds positively to government-based benefits, negatively to property crime rates, and positively to relative lower housing prices. Although the results in this study seem robust, caution must be exercised in drawing any policy implications. Typically, regional development policies seek to create jobs by attracting business to the region--aiming at increasing local labor demand. However, as Feasel and Rodini (2002) conclude, the high and chronic unemployment in the SJV is not simply due to lack of labor demand (jobs), but is also associated with nonmarket factors. For example, as argued by Borjas (1999) and supported by this study, in-migrants tend to cluster in regions that offer generous public assistance programs--the so-called "welfare magnets" and also in regions characterized by low housing prices. Therefore, regional policy makers should evaluate the implications of supporting such programs--which might aim at reducing poverty or improving income distribution--against the possibility of attracting large numbers of migrants who can also engross the number of unemployed, increase the demand (and price) for housing, etc. In other words, more research is needed on the supply side of the labor market to understand individual's motivation to migrate or to stay in certain regions. Third, the analysis demonstrates that although migration is not responsive to the overall wage differential, it is responsive to differences in farm income. This underlines the significance of using the relevant measurement for earnings. In the case of the SJV, with its vast agricultural sector, farm income produces evidence in complete conformity with labor market theory.

APPENDIX A

Definition of the San Joaquin Valley Labor Market

The table below shows travel to work data from the 2000 census for these counties. Based on the "operational definition" of a labor market as explained in Goetz (1999), with less than 10% of workers traveling to other counties for work, it seems reasonable to consider these eight counties as a separate individual labor market. (6) Although information at the subcounty might be more appropriate in terms of defining the labor market, data for the variables of interest are not available for geographic levels smaller than the county.
San Joaquin Valley Counties Travel to Work Data

 % People who live in this county

 FRE MAD KER KIN MER STA SJO TUL

People work in FRE 92.6 23.8 0.1 9.3 1.8 0.1 0.0 4.9
this county

 MAD 0.9 68.8 0.0 0.2 0.0 0.0 0.0 2.5

 KER 2.6 0.1 93.5 1.7 1.6 0.0 0.0 0.1

 KIN 0.1 0.3 0.1 79.2 0.0 0.0 0.0 2.7

 MER 0.2 2.4 0.0 0.0 75.0 2.9 0.0 0.0

 STA 0.0 0.3 0.0 0.0 12.0 79.0 3.1 0.0

 SJO 0.0 0.1 0.0 0.0 1.3 8.2 76.5 0.0

 TUL 1.8 0.2 0.9 6.5 0.0 0.0 0.0 88.1

 TOTAL 98.2 96.0 94.6 96.9 91.7 90.2 79.6 98.3

Source: Author's calculations based on County-To-County Worker Flow
Files, 2000 U.S. Census Bureau.


REFERENCES

Armstrong, H., and J. Taylor. Regional Economics and Policy. Maiden, MA: Blackwell Publishers, 2000.

Bartik, T. J. Who Benefits from State and Local Economic Development Policies? Kalamazoo, MI: W.E. Upjohn Institute for Employment Research, 1991.

Bartik, T. J., and R. D. Bingham. "Can Economic Development Programs be Evaluated?" Upjohn Institute Staff Working Paper No. 95-29, W.E. Upjohn Institute for Employment Research, Kalamazoo, MI, 1995.

Berger, M. C, and G. C. Blomquist. "Mobility and Destination in Migration Decisions: The roles of Earnings, Quality of Life, and Housing Prices." Journal of Housing Economics, 2, 1992, 37-59.

Blanchard, O., and L. Katz. "Regional Evolutions." Brooking Papers on Economic Activity, 23(1), 1992, 1-75.

Borjas, G. J. "Immigration and Welfare Magnets." Journal of Labor Economics 17, 1999, 607 37.

Cebula, R. "Internal Migration Determinants: Recent Evidence." International Advances in Economic Research 11, 2005, 267-74.

Cebula, R., and J. Payne. "Migration, Economic Opportunity, and the Quality of Life in the U.S., 1999-2000." International Review of Economics and Business 52, 2005, 245-53.

DaVanzo, J. "Docs Unemployment Affect Migration? Evidence from Micro Data." Review of Economics and Statistics, 60, 1978, 504-14.

Eberts, R. W., and J. A. Stone. Wage and Employment Adjustment in Local Labor Markets. Kalamazoo, MI: W.E. Upjohn Institute for Employment Research, 1992.

Feasel, E. M., and M. L. Rodini. "Understanding Unemployment Across California Counties." Economic Inquiry, 40, 2002, 12-30.

Fields, G. "Labor Force Migration, Unemployment and Turnover." Review of Economics and Statistics 58, 1976,407-415.

Goetz, S. "Migration and Labor Markets," in The Web Book of Regional Science, edited by Scott Loveridge. Morgantown, WV: Regional Research Institute, West Virginia Institute, 1999.

Goss, E., and C. Paul. "The Impact of Insurance Benefits on the Probability of Migration of the Unemployed." Journal of Regional Science, 30, 1990, 349-58.

Greenwood, M. "Research on Internal Migration in the United States: A Survey." Journal of Economic Literature, 13, 1975, 397-433.

Heitmueller, A. "Unemployment Benefits, Risk Aversion, and Migration Incentives." Journal of Population Economics, 8, 2005, 93-112.

Houseman, S. N. "Job Growth and the Quality of Jobs in the U.S. Economy." Upjohn Institute Staff Working Paper 95-39, W.E. Upjohn Institute for Employment Research, Kalamazoo, MI, 1995.

Houseman, S. N., and K. Abraham. Regional Market Responses to Demand Shocks: A Comparison of the United States and Germany. Paper presented at the Association for Public Policy Analysis and Management, San Francisco, October 1990.

Hsing, Y. "Public Policy and Regional Migration." Journal of Economic Research, 3, 1998, 41-50.

Martin. P., and E. Taylor. Farm Employment. Immigration, Poverty: A Vicious Circle? University of California, Davis, CA, 2000, mimeographed.

Mincer, J. "Family Migration Decisions." Journal of Political Economy, 86, 1978, 749-73.

Muller, T. "Migration, Unemployment and Discrimination." European Economic Review, 47, 2003, 409-427.

Partridge, M. D., and D. Rickman. "Which Comes First, Jobs or People? An Analysis of the Recent Stylized Facts." Economic Letters, 64, 1999, 117-23.

--. "The Waxing and Waning or Regional Economies: The Chicken-Egg Question of Jobs versus People." Journal of Urban Economics 53, 2003, 76-97.

Pissarides, C, and I. McMaster. "Regional Migration, Wages and Unemployment: Empirical Evidence and Implications for Policy." Oxford Economic Papers, 42, 1990, 812-31.

Polese, M. "Regional Disparity, Migration and Economic Adjustment: A Reappraisal." Canadian Public Policy--Analyse de Politiques, 7, 1981, 519-25.

Shields, M., and J. Novak. Dynamic Labor Responses to Employment Shocks - A County Level Analysis. Harrisburg, PA: Pennsylvania State University, 2000, mimeographed.

Weiler, S. "Unemployment in Regional Labor Markets: Using Structural Theories to Understand Local Jobless Rates in West Virginia." Industrial and Labor Relations Review 54, 2001, 573-91.

Winters, P., A. de Janvry, and E. Sadoulet. "Family and Community Networks in Mexico-US Migration." Working Papers in Agricultural and Resource Economics, No. 99-12, Graduate School of Agricultural and Resource Economics, University of New England, Portland, ME, 1999.

Avalos: Assistant Professor. Department of Economics, California State University-Fresno, 5245 North Backer Avenue, MS PB 20, Fresno, CA 93740. Phone 559-278-8793, Fax 559-278-7234, E-mail aavalos@csufresno.edu

(1.) For the purposes of this paper, the SJV labor market is composed of the following eight counties: Fresno (FRE), Kern (KER). Kings (KIN), Madera (MAD). Merced (MER). San Joaquin (SJO), Stanislaus (STA) and Tulare (TUL). For a technical definition of the SJV labor market see Appendix A.

(2.) Wages here are the average wage per job as reported in BEA's REIS.

(3.) Weiler (2001) offers a good review of structural theories to understand local and regional unemployment.

(4.) The following does not pretend to be an exhaustive review of the literature on the topic. Instead, it only briefly describes and discusses results obtained in the most relevant studies.

(5.) The number of observations in Models 6 and 8 is 122 because median house prices are not available for Kern County from a consistent data source.

(6.) One exception is San Joaquin County, which is the northern most of the San Joaquin Valley counties.

ABBERVIATIONS

REIS: Regional Economic Information System

SJV: San Joaquin Valley

doi: 10.1111/j.1465-7287.2009.00159.x
COPYRIGHT 2010 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2010 Gale, Cengage Learning. All rights reserved.

 Reader Opinion

Title:

Comment:



 

Article Details
Printer friendly Cite/link Email Feedback
Author:Avalos, Antonio
Publication:Contemporary Economic Policy
Geographic Code:1U9CA
Date:Jan 1, 2010
Words:7983
Previous Article:Should there be intraday money markets?
Next Article:Job centers: are finders keepers? A note from Japan.
Topics:

Terms of use | Copyright © 2015 Farlex, Inc. | Feedback | For webmasters