Local Labor Supply Responses to Immigration Ximena Del Carpioy

Ça¼ glar Özdenz

Mauro Testaverdex

Mathis Wagner{ September 2014

Abstract How natives adjust is central to an understanding of the impact of immigration in destination countries. Using detailed labor force data for Malaysia for 1990 – 2010, this paper provides estimates of native responses to immigration on multiple extensive margins and rare evidence for a developing country. IV estimates show that increased immigration to a state causes substantial internal inward migration, consistent with immigration increasing the demand for native workers. Relocating Malaysian workers are accompanied by their spouses (three-quarters of who are housewives) and children who attend school. We …nd that these e¤ects are concentrated among middle and lower-skilled Malaysians. Keywords: Migration, native adjustment, labor supply. JEL Classi…cation: F22, J61, R23 We are grateful to Amir Omar and his team at the Institute of Labor Market Information Analysis (ILMIA) for their continuous support, Department of Statistics of Malaysia for data availability and participants at the 6th International Conference on Migration and Development, 4th TEMPO Conference on Migration, and the 9th IZA/World Bank Conference on Employment and Development for comments. The …ndings, conclusions and views expressed are entirely those of the authors and should not be attributed to the World Bank, its executive directors and the countries they represent. y World Bank, Washington, DC 20433, USA; [email protected]. z World Bank, Washington, DC 20433, USA; [email protected]. x World Bank, Washington, DC 20433, USA; [email protected]. { Boston College, Chestnut Hill, MA 02467, USA; [email protected].

1

I. Introduction Labor market behavior of natives in response to immigration is central to an understanding of the impact and associated welfare consequences of international labor mobility. Potential adjustment mechanisms are numerous, but the primary focus of the literature has been on how immigration impacts relative wages in local labor markets as well as the resulting in‡ows and out‡ows of natives (Card, 2009, Borjas 2014). This paper, using detailed labor force survey data from Malaysia, extends the analysis to consider a number of additional margins along which native labor supply adjusts to immigration shocks. We decompose the total causal e¤ect of immigration on changes in the population of a local area into the propensity of those individuals to be employed (full-time or part-time), unemployed, or choose alternatives, such as continue with their education, retire or work in household activities. Our results show there is adjustment on all of these dimensions, highlighting the value of a comprehensive analysis. The paper contributes to a growing literature on native worker (and …rm) responses to immigration. The link between arrival of foreign migrants and internal mobility of native workers across geographic areas has been extensively studied, primarily using U.S. data. Early work by Filer (1992), Frey (1995), and White and Liang (1998) …nds evidence that immigration into a locality results in outward migration of natives to other areas. Subsequent papers, following the in‡uential work by Altonji and Card (1991), by Wright, Ellis and Reibel (1997), Card and DiNardo (2000), Card (2001), and Kritz and Gurak (2001) found the magnitude of these ‡ows to be small. On the other hand, Borjas (2006) …nds evidence consistent with substantial native responses to immigration. Card and Lewis (2007) …nd little evidence of such responses looking at changes in industry composition, a …nding supported by Gonzalez and Ortega (2011) using Spanish data and Dustmann and Glitz (2012) using German data. Shifting their focus to labor market responses of women, Cortes and Pan (2013) and Cortes and Tessada (2011) …nd increased hours worked and employment in Hong Kong and the U.S. respectively. Borjas, Grogger and Hanson (2010) 2

…nd declining employment and rising incarceration rates African-American men. Plausible adjustment mechanisms to immigration are numerous and recent important work by Lewis (2011, 2013) has looked at changes in technology, and Peri and Sparber (2009) and Ottaviano, Peri and Wright (2013) at changes in native task specialization. We consider a broader set of native extensive margin responses to immigration than previous work, providing new insights into how natives adapt to immigration shocks. A further contribution of the paper is to help address a pronounced imbalance in the literature, where existing evidence is nearly exclusively for OECD destinations, even though almost half of global immigration takes place between non-OECD countries (Ozden et al., 2011; Artuc et al., forthcoming).1 We use detailed data from the Malaysian Labour Force Survey (LFS) for the years 1990 to 2010 for the analysis. Over the past twenty years, Malaysia has experienced a rapid increase in the share of immigrants in the working-age population, going from 3.2 percent in 1990 to 10.4 percent in 2010. Immigrants are primarily from Indonesia (55 percent) and the Philippines (20 percent) with the remainder coming from other East and South Asian countries. South-South migration is likely to become more important as diverging demographic patterns and increasing regional economic integration are leading to increased migration ‡ows in many parts of the developing world. Whether the e¤ects of South-South are di¤erent from those of South-North migration remains an open question. We identify the impact of immigration on a number of native labor market outcomes using variation in the in‡ow of immigrants across states and over time in Malaysia. In order to deal with the endogeneity of immigration ‡ows, we instrument for these using changes in the size and age structure of the populations in migrant source countries and the di¤erential propensity of these groups to migrate to particular states. The instruments rely both on cross-sectional and time-series variation. This structure allows us to include state-speci…c linear time trends to control for unobserved local labor market speci…c long1

Exceptions are Gindling (2009) for Nicaragua and Facchini, Mayda and Medola (2013) for South Africa; both papers focus on employment and wages of natives.

3

term demand trends. We …nd that immigration into states in Malaysia causes large in‡ows of natives. The perhaps surprising fact is that we observe an increase in both the number of people employed and out-of-labor force. For every 10 immigrants into a state, there is an additional in‡ow of 7.6 working-age natives, two-thirds of who are employed and one-third of who are out of the labor force. We …nd that the increase in the number of people out-oflabor force is concentrated in the housewife and student categories, with no e¤ect on the number of retirees. Our results are consistent with immigration causing an increase in the demand for native labor which induces both employed individuals and their families to migrate across states. The labor force participation rates of women are low (under 50 percent) and families also bring their children, which explains the rise in the number of housewives and students in a state as a response to immigration. Our …ndings imply that immigration does not only change the distribution of employment and output across states, but has much larger social rami…cations as entire families are induced to move. We …nd that these e¤ects are concentrated among middle and lower-skilled Malaysians. The remainder of this paper is structured as follows. Section 2 describes the data and provides background information on Malaysia. Our empirical strategy and instrument are outlined in Section 3. Section 4 describes our results, and Section 5 concludes.

II. Data and Background The LFS of Malaysia provides annual detailed data on various characteristics of the overall population (Malaysian and immigrant) and the structure of employment at the individual level. The main survey is conducted monthly by the Department of Statistics of Malaysia and data are available for the years 1990 to 2010, with the exceptions of 1991 and 1994 when the survey was not conducted and 2008 where the survey weights were not available. The main survey samples, on average, around 1 percent of the population.

4

The LFS records the state people live in and our main unit of analysis is one of the 15 states of Malaysia in a particular year.2 One of the key variables is an individual’s labor market status: employed, unemployed or out-of-labor force. There is detailed information on the reasons why a person remains out of the labor force which we divide into four main categories: student, housewife, retired, and all other reasons. For those employed, we are able to distinguish between full-time and part-time employment.3 We also have information on an individual’s age, gender, and marital status. We aggregate the various educational classi…cations into six main categories: no formal education, primary education, lower secondary education, upper secondary education (including post-secondary, STPM), certi…cate / diploma (vocational training), and university degree and above. Tables 1 and 2 present summary statistics for three years of the survey: 1992, 2001 and 2010.4 Table 1 describes the characteristics of Malaysians by gender, and Table 2 compares these to those of the immigrants. This is a period of rapid economic growth for Malaysia and an impressive transformation of the Malaysian-born labor force’s educational attainment levels. In 1992, 56 percent of the native labor force had primary school education or less, and only 4.5 percent had education beyond upper secondary education. By 2010, the share of the labor force with at most primary school education declined to 24 percent while 16 percent had a vocational diploma or a university degree. The transformation is most impressive among the new entrants to the labor force (ages of 20-25); as of 2010, around 80 percent of this group had obtained (at least) a high school degree. Labor force participation rates are signi…cantly higher for men (around 80 percent) than women (around 46 percent). However, the long term trend is downward-sloping for men but stable for women. Unemployment rates are low and stable at 3 to 4 percent. The 2

We include Putrajaya in Selangor throughout the analysis. The LFS does not contain questions on an individual’s income except for a smaller subsample for the years 2007-2010. 4 We choose 1992, rather than 1990, since it is the …rst survey conducted after the 1990 census and hence uses more accurate survey weights. 2001 is the …rst survey after the 2000 census. 3

5

share part-time employed has been declining over time, from 11 to 6 percent for women and from 6 to 4 percent for men. The self-reported fraction of the retired has been rising rapidly over time for men, reaching 18 percent of those out of the labor force; while it continues to be close to zero for women.5 There are two types of formally registered immigrants in Malaysia: expatriates and foreign workers.6 Expatriates are highly skilled managerial, professional, and technical workers who are able to obtain long-term visas and enjoy privileges, such as the ability to bring their families, admission under di¤erent visa regimes, and exemption from certain taxes. These highly-skilled or educated professionals make up only 2 percent of the total visas issued to immigrants. The remaining 98 percent of immigrant workers only receive temporary work permits that are valid for at most a year and renewable for at most …ve years. These workers are not allowed to bring any dependents and are required to exit Malaysia upon termination of their contracts. There is almost no pathway to citizenship. Formal employment of foreign workers is regulated by quotas assigned to speci…c sectors, which are adjusted annually if there are extraordinary circumstances. Foreign workers are hired either directly by a company, or through a recruitment agency that handles the administrative burden of obtaining work permits in return for a fee. Such hiring costs are comparatively low by international standards, but hiring and work permit procedures can be arduous and include medical checks. In addition, annual levies (payable by the employer up until 2013) are charged for the employment of foreign workers. There is a substantial number of irregular or undocumented foreign workers in the labor force. Many of them may have entered Malaysia legally but overstayed their permits. Reliable estimates are not available, but a 1996/97 regularization exercise resulted in almost one million unregistered migrants being legalized. Another new program, labelled the 6P, implemented in 2011 registered over half a million undocumented foreign workers. 5

It is an interesting question why women do not self-report being retired; but that is an issue beyond the scope of this paper. 6 The material in this section draws on Del Carpio et al. (2013) who provide an extensive discussion of the Malaysian immigration system.

6

This suggests that as many as half might be employed without proper documentation. In principle our data capture these undocumented immigrants. Immigrant workers are disproportionately employed in agriculture and unskilled labor intensive service sectors, such as, construction. Their share in manufacturing also increased rapidly during this period, while they are under-represented in relatively skillintensive service sectors such as health, education and public administration. Immigrants are signi…cantly less educated than Malaysians, 86 percent have (at most) primary school education in 1992. Even though this number had fallen to 66 percent in 2010, only 19 percent had completed high school. A closer look at the LFS reveals that a large portion of these high school educated immigrant workers are also employed in low-skilled occupation categories indicating that their actual human capital is lower than Malaysian high-school graduates. Around 55 percent of all immigrants are from Indonesia, 20 percent from the Philippines and the remainder from countries such as Bangladesh, Cambodia, India, Laos, Myanmar, Sri Lanka, Thailand, and Vietnam. In Figure 1 we depict the distribution of immigrants across states, in 1992 and 2010.7 Even as the number of immigrants has tripled over this period there is enormous persistence in the distribution of immigrants across states. Sabah is by far the most popular destination for immigrants, 43 and 40 percent of immigrants in 1992 and 2010, respectively, live in that state; followed by Selangor, Johor and Kuala Lumpur in both 1992 and 2010. Figure 2 depicts the net population gains from 1992 to 2010 due to immigration for each state. Sabah’s population grew by 22 percent due to immigration, followed by Labuan with 12 percent; on average Malaysia’s population grew by 5.5 percent due to immigration. 7

Just like Malaysians, immigrants over this period increasingly locate in urban areas. In 1992 47 percent were in urban areas, that fraction increased to 62 percent by 2010.

7

III. Empirical Strategy Empirical Framework The central contribution of this paper is in understanding di¤erential native responses to immigration. While the literature has focused mainly on employment, we consider a large set of extensive margin choices by natives. The population in state r in year t, P oprt , is the sum of the people in the labor force, Lrt , and out-of-labor force, N Lrt :

P oprt = Lrt + N Lrt :

(1)

P F , or unemployed , part-time Ert Those in the labor force can either by employed full-time Ert

Urt :

F P Lrt = Ert + Ert + Urt :

(2)

Out-of-labor force individuals can be students Srt , housewives Hrt , retired Rrt , or in the other reasons category, Ort :

N Lrt = Srt + Hrt + Rrt + Ort :

(3)

The focus in the literature has been on the impact of immigration, Mrt , on the employment of native workers. Our focus is on decomposing the impact of immigration on the population in a state into those in and out-of-labor force: dP oprt dLrt dN Lrt = + ; dMrt dMrt dMrt

(4)

We then further decompose the impact on the labor force and out-of-labor force groups,

8

based on equations (2) and (3): F P dLrt dErt dErt dUrt = + + dMrt dMrt dMrt dMrt dN Lrt dSrt dHrt dRrt dDrt dOrt = + + + + dMrt dMrt dMrt dMrt dMrt dMrt

(5) (6)

Estimating Equations We separately estimate the derivatives in equations (4), (5), and (6) in order to understand the various channels through which natives respond to immigration. Our estimating equation to identify the impact of immigration on the various outcomes of interest yrt is given by:

yrt = where

r

are state …xed e¤ects,

y Mrt

t

+

r

+

t

+

rt

(7)

+ "rt ;

are year …xed e¤ects,

rt

are state-speci…c linear time

trends, and "rt is an error term. The outcomes we use as dependent variables, yrt ; are the native population of a state in year P oprt , total native employment Ert , full-time F P employment Ert , part-time employment Ert , unemployment Urt , out-of-labor force N Lrt ,

students Srt , housewives Hrt , and retired Rrt . Note that the decomposition above requires the dependent variables to enter linearly (and consequently we let the independent variable M also enter linearly), which means that we are estimating the (weighted) conditional mean impact of immigration. Hence, the impact of immigration on, for example, total employment is simply the sum of the impact on part-time and full-time employment. This additive property will be very helpful in interpreting the results in a straightforward manner, see Section 4 below. The …xed e¤ects and the error term capture shocks to output prices, both factorneutral and factor-enhancing di¤erences in technology, product quality, transportation costs and other factors that a¤ect demand. The inclusion of

r

means that the e¤ect of

immigration is identi…ed from changes over time in the number of immigrants in each 9

state. The inclusion of

t

and

rt

implies that only deviations from the year-speci…c

national averages and state-speci…c linear trends in immigration ‡ows to a state are used for identi…cation. Instrument The central challenge in estimating the equations de…ned by (7) is the endogeneity of immigrant location decisions, which are likely to be correlated with unobserved shocks (positive or negative) to the demand for labor in a state. The likelihood of biased OLS estimates makes it important to instrument for the in‡ow of immigrants to a state.8 Our instrumenting strategy is closely related to Ozden and Wagner (2014). A valid instrument for immigration ‡ows needs to be uncorrelated with any demand shocks, caused by changes in technology or output prices, that may a¤ect the demand for native or immigrant labor in a given locality at a given time. We use changes in the population and age structure of immigrant source countries over time to construct such an instrument. The source countries are Bangladesh, Cambodia, India, Laos, Myanmar, Sri Lanka, Thailand, Vietnam, and most importantly Indonesia and Philippines. Using the data from the United Nations Population Division, we calculate the number of individuals in each of 7 age-groups in each of these source countries in every year during 1990-2010.9 These population numbers form the potential pool of immigrants to Malaysia. This is our measure of the supply of immigrants Stac to Malaysia from source country c in age-group a and year t.10 Since data on immigrants’nationality are grouped into Indonesians, Filipinos and the rest of the world, we add our measure of the supply of immigrants for all other countries into a single category, such that e¤ectively we have three source countries: Indonesia, the 8

An additional advantage of the IV approach is that it helps deal with measurement problems. For example, undocumented immigrants are likely undercounted by the LFS, resulting in attenuation bias in the OLS estimates. For the IV estimates to be consistent, it is only necessary that - conditional on the …xed e¤ects - the ‡ows of illegally employed immigrants are uncorrelated with the instrument. 9 The age groups are 15-19, 20-24, 25-29, 30-34, 35-39, 40-44, and 45 and above. 10 We multiply these population numbers by the average propensity of people from each country to migrate to Malaysia. This is so as to ensure that the magnitudes of the coe¢ cients on each instrument are broadly comparable. These propensities are calculated from data provided by the Ministry of Home A¤airs of Malaysia and are: Bangladesh 1.96%, Cambodia 1.03%, India 0.11%, Lao 0.01%, Myanmar 2.18%, Sri Lanka 0.16%, Thailand 0.22%, Vietnam 0.78%, Indonesia 5.56% and Philippines 0.38%.

10

Philippines, and Other. What remains to be determined is the distribution across states within Malaysia in which the immigrants choose to work. In order to construct this variable, we calculate the average probability of individuals from a source country and age group to be employed in a certain state ac r

=

X2010

t=1990 X2010 t=1990

Mrtac

;

Mtac

where Mrtac is the number of immigrants from a source country in an age group, state, and year, and Mtac is the total number of immigrants in Malaysia from a source country and in an age group. The source country and age-group speci…c instrument for the immigration ‡ows in a certain state and year is given by:

IVrtac =

ac r

Stac :

(8)

As mentioned above, we have three source countries (Indonesia, Philippines and Other) and seven age groups for a total of 21 instruments. Due to the small number of observations we then sum the age-speci…c instruments for the Philippines and the Other category, e¤ectively restricting the coe¢ cient on these instruments to be the same, such that we end up with 9 instruments (one for Filipino and one for Other immigrants, and 7 age-speci…c instruments for Indonesian immigrants).11 The identifying variation of the instruments comes from the interaction of

ac r

and Stac ,

conditional on the included …xed e¤ects, and is due to changes in the size of cohorts in source countries (which are experiencing their demographic transition at di¤erent rates) and their di¤erential propensity to be employed in certain states in Malaysia. The variation in the instrument generated by the di¤erential propensity of immigrant groups (de…ned by nationality and age) to work in di¤erent local labor markets is related to 11

Qualtitatively this does not a¤ect our results, but it does yield a stronger …rst-stage.

11

the commonly used Altonji-Card instrument (Altonji and Card, 1991; Card, 2001). The variation induced by the demographic changes in source countries is inspired by Hanson and McIntosh (2010). Considering potential threats to the validity of this instrument, it is worth noting that the supply of potential migrants to Malaysia from di¤erent source countries (Stac ) is determined by the demographic patterns and transition in those countries, and hence clearly exogenous with respect to contemporaneous labor market shocks in Malaysia. The average propensity of an immigrant from a source country to be employed in a certain state (

ac r )

depends on permanent di¤erences in the levels of demand across local labor markets,

which is why we include state speci…c …xed e¤ects in all our regression speci…cations. It is of course independent of any transitory shocks that may a¤ect demand for natives (and immigrants) in a particular year. However, the concern is that persistent demand shocks, i.e. long periods of decline or growth in certain states, would result in a correlation between the average distribution of immigrants and current demand shocks. The advantage of our instrument, as compared to the Altonji-Card instrument, is that it uses both time-series and cross-sectional variation. Consequently, we can include local labor market (state) speci…c linear time trends to account for this concern, and use the variation induced by the non-linearities in the demographic transition of source countries for identi…cation. It is worth noting that while the Altonji-Card shift-share instrument is typically constructed using initial immigrant concentration (often 2, 5 or 10 years before the beginning of the main sample), we use the average immigrant stock over the sample period to construct the shift-share component of our instrument. It is worth comparing the two approaches. In the case where shocks to labor demand are transitory it is preferable to improve the precision of the instrument, and average out any transitory demand shocks, by using as long a period as possible to construct the shift-share component of the instrument. In the case where the demand shocks are persistent, the typical implementation of the Altonji-Card shift-share instrument has to hope that these shocks do not last for more

12

than the number of years that separate the initial period from the period of analysis (an unveri…able and arbitrary assumption). In contrast, in our approach we directly control for such long-lasting demand shocks by the inclusion of state-speci…c linear time trends (recall that mean independence of the error terms is required for consistency of the IV estimates). In short, given our ability to control for long-term demand trends there is no disadvantage to using the average rather than initial immigrant distribution across local labor markets to construct the shift-share component of the instrument. Moreover, using averages has the considerable advantage of potentially improving the precision of the instrument, especially since the LFS only samples around 1 percent of the population. Nevertheless, as a robustness check we provide estimates where we construct

ac r

using

data for 1990-93 or 1990-95 and begin our analysis in 1995 and 2000 respectively. Our results are robust to these alternative ways of constructing the instruments, see Section 4.3 below.

IV. Results Main We present OLS and IV estimates of equation (7) with state and year …xed e¤ects, and state-speci…c linear time trends, in the …rst and second columns respectively of Table 3. Standard errors are clustered by state-year and are robust to heteroskedasticity as states in Malaysia are of very di¤erent sizes. The results from the …rst-stage are in the Appendix Table A.1.12 The OLS results are very similar to the IV estimates, suggesting that once our large set of …xed e¤ects is included the remaining variation in immigrant ‡ows is close to uncorrelated with contemporaneous demand shocks.13 The F-statistic in the …rst-stage is 12

Our instruments are highly colinear making the sign and signi…cance of each individual instrument hard to interpret. The relevant statistic for determining whether the instruments are weak is the Fstatistic of the …rst-stage, which is equal to 17.0 and well above any conventional weak instrument criteria. 13 A priori the direction of the bias of the OLS estimates is unclear. Immigrants are more likely to

13

17.0. Focusing on the IV results, we …nd that immigration has a large positive impact on the total number of Malaysians living in a state. Most of that e¤ect is driven by an increase in the native workers employed in a state; for every 10 immigrants there are an additional 5.2 natives employed, with 4.4 in full-time and 0.8 part-time employment. Given that only 6 percent of the total labor force are part-time employed, this suggests part-time employment is a lot more responsive to immigration than full-time employment, with elasticities of 0.16 and 0.06 respectively. The number of unemployed people is una¤ected by immigration, though as employment increases, there is a fall in the unemployment rate with an elasticity of -0.02. The number of people out of the labor force in a state also increases substantially due to immigration. For every 10 immigrants in a state, there are an additional 2.2 individuals who are not in the labor force, and the total native population in a state increases by 7.6. The increase in the out-of-labor force population is driven by an increase in those attending school and the number of housewives. For every 10 immigrants in a state there are an additional 1.6 housewives, 0.7 additional students above the age of 15, and 2.8 children below age 15.14 There is no signi…cant e¤ect on the number of retirees. Notably, the employment-population ratio increases only slightly (with an elasticity of 0.1, or a semi-elasticity of 0.06). An explanation is that immigration encourages the in‡ow of entire households into a state, with a signi…cant fraction of women staying at home and children going to school. We further explore this interpretation in Section 4.4 below. It also suggests that while immigration results in substantial reallocation of people across states, it is unlikely to have substantial e¤ects on employment at the national level. The Malaysian Migration Survey Report helps provide some context for these numlocate in states that are experiencing positive shocks to the demand for labor. In that case, the OLS estimate of the e¤ect of immigration on native employment will be upward biased. It is also possible that declining industries make a special e¤ort to attract immigrant labor to lower their wage bill. For example, immigrants require a work permit to legally work in Malaysia; one way that declining industries may respond is by exerting political pressure that more work permits be issued to their industry. In that instance there is a negative correlation between the in‡ow of immigrants and shocks to the employment of native labor, and the OLS estimates would be downward biased. 14 These are excluded from the main regressions which only include the working-age population.

14

bers.15 The average inter-state migration rate for the years 2009-13 is 0.65 percent of the population.16 On average net international migration to Malaysia between 1990 and 2010 is equivalent to around 0.35 percent of the Malaysian population per year; which our IV estimates suggest results in about an equal number of Malaysians migrating across states (three-quarters of those above age 15 and one-quarter below age 15). Hence, immigration to Malaysia resulted in inter-state migration rates equal to around half those currently observed. The report also …nds that about one-quarter of all Malaysian migrants are below age 15, which is nearly exactly our estimated fraction of the immigration induced inter-state migrants who are below age 15 (27 percent). Economic Interpretation The …nding that immigration increases the demand for native workers, even low-skilled Malaysian as we show in Section 4.5 below, speaks to the central question of an extensive literature on immigration: whether immigration harms or improves the labor market outcomes (wages and employment) of native workers. Ozden and Wagner (2014) point out that immigration a¤ects the demand for native workers through two distinct two economic mechanisms: the substitution and scale (output expansion) e¤ects. An outward shift of the immigrant labor supply curve causes a reduction in immigrant wages. This results in two countervailing economic forces. First, for a given level of output, …rms will substitute immigrant for native labor, the standard substitution e¤ect analyzed extensively in the literature. Second, for a given relative wage, …rms will employ more native workers as output expands (the scale e¤ect). Using data from the Malaysian LFS the authors provide estimates of both these e¤ects. They …nd an elasticity of labor demand (the scale e¤ect) of 3.4, which is signi…cantly larger than the elasticity of substitution between immigrant and native labor of 2.5.17 Thus the scale e¤ect outweighs the substitution e¤ect in Malaysia. 15

The report is available for the years 2011, 2012, and 2013. Inter-state migration accounts for about one-quarter of total migration; more than 60 percent of migrants are intra-state, and about 10 percent are international migrants. 17 The estimates in Ozden and Wagner (2014) suggest that immigrants and natives are highly substitutable in Malaysia (especially low-skilled natives), and certainly far from complements in production (which would require a non-positive elasticity of substitution) or even gross complements (which would 16

15

Our …ndings are readily explained the same way. Immigration results in an increase in the demand for Malaysian workers because the reduction in the cost of production due to immigration allows …rms to increase output. This more than o¤sets their incentive to substitute natives workers with cheaper immigrants. Immigration generates a positive shock to demand for native workers; in response workers migrate to that state and they are accompanied by their spouses and children. Robustness Checks We present numerous robustness checks in Tables 4 and 5. In Table 4 the …rst column presents estimates when we include spatially weighted immigration ‡ows to other states as a covariate to allow for spillover e¤ects across states.18 The inclusion of this covariate barely a¤ects our IV estimates of the impact of immigration. The coe¢ cient on the spatially lagged immigration ‡ows is positive in every regression and usually statistically signi…cant, though notably not in the speci…cation with full-time employment or the number of housewives as the outcome variable. Interestingly, the fact that both the spatial spillovers and the direct causal e¤ect of immigration are positive suggests that immigration also creates jobs in neighboring states. The magnitude of this e¤ect is modest though. For example, an additional 1000 immigrants in other states on average increases employment in a state by 2 Malaysians. In the second column of Table 4 we weight observations by the average of the dependent variable in a state over this period, so that they are representative of the impact of immigration on that group (for example, part-time workers or the unemployed). Most of our estimates are practically una¤ected by this weighting scheme.19 In the third column require an elasticity of substitution below one). 18 The estimating equation is: X yrt = Mrt + wr;r0 Mr0 t + 0 r

r

+

t

+

rt

+ "rt ;

where W = [wr;r0 ]r=1;:::R;r0 =1;:::;R is a row-normalized spatial weights matrix. The weights are the inverse of the distance between the state capitals. 19 The exception is unemployment, where we now …nd a signi…cant positive impact of immigration on the number of unemployed in state.

16

we cluster the standard errors by state to allow for serial correlation of the error term over time within state. A priori this is unlikely to be a good idea since we only have 15 clusters. Moreover, in nearly every case clustering at the state-level actually slightly decreases the standard errors, which suggests that conditional on our extensive …xed e¤ects the serial correlation is negative. The fourth column shows the results of a log-log speci…cation, which qualitatively provides the same results as our linear speci…cation. However, the estimates are less precise and it is harder to interpret the elasticities since the number of Malaysians in each category varies enormously.20 In Table 5 we provide IV estimates where we construct the shift-share component of our instrument (

ac r )

using data for 1990-93 or 1990-95 and begin our analysis in 1995 and

2000 respectively. This is in the spirit of the typical Altonji-Card instrument that uses an initial period to construct the instrument. In the …rst column we reproduce our main results from Table 3 for comparison. In the second and third columns we use 1990-93 to construct

ac r

and begin the analysis in 1995, in the fourth and …fth columns we use

1990-95 to construct

ac r

and begin the analysis in 2000.

Comparing the speci…cations that include state and year …xed e¤ects and state-speci…c linear trends (…rst, second and fourth columns) we …nd that qualitatively our results are remarkably robust to changes in how

ac r

is constructed and the years used for our analysis.

However, using less years and only the initial distribution of immigrants substantially reduces the F-statistic in the …rst-stage and the precision of our estimates. If we do not include state-speci…c linear time trends (third and …fth columns), as is typical for papers that use the Altonji-Card instrument, we obtain upward biased estimates. This suggests that unobserved, persistent demand shocks are a feature of the data over this period. The upward bias is very pronounced when the immigrant concentration is measured only two years before the start of the main sample period (third column), but modest when it is 20

Due to the small sample size of the LFS around 45 percent of observations contain zeroes for typically either the number of immigrants or one of the instruments. To deal with this issue we add one to every observation. As an alternative we transform all variables with the inverse hyperbolic sine, which is de…ned at zero (Burbidge et al., 1988; MacKinnon and Magee, 1990). This yields near identical results.

17

constructed using 1990-95 data and the analysis starts in 2000 (…fth column). Gender and Age Tables 6 and 7 report IV results by gender and for four age groups (ages 15-19, 20-29, 30-49, and 50-64). Appendix Tables A2 and A3 report the corresponding OLS results. Immigration has a larger impact on the number of Malaysian men in a state than women, 4.4 additional men and 3.2 additional women for every 10 immigrants. Strikingly, immigration has a far larger impact on male than female employment, where an additional 10 immigrants lead to 4.0 additional male and 1.2 female employed Malaysians. The di¤erence is driven by full-time employment, where the impact on part-time employment is of a similar magnitude across genders. In contrast, the impact of immigration on people in a state who are not in the labor force is far larger for women than men; 10 additional immigrants imply 1.9 additional female and 0.3 male Malaysians out-of-labor force. This di¤erence is explained by the large increase in the number of housewives in a state as a consequence of immigration, an additional 1.6 for every 10 immigrants. The increase in the number of people in school is evenly divided among men and women. In terms of the age distribution of the impact, the positive employment e¤ect of immigration is especially concentrated among those aged between 20 and 49, with a small and positive e¤ect on teenagers (ages 15 to 19) and no real e¤ect on workers above age 50, except for a small increase in those employed part-time. Among the out-of-labor force group, there is an increase for all age groups between 15 to 49, but for very di¤erent reasons. For the 15-19 age group the main increase is for those attending school. For the 20-29 age group the largest increase is among housewives and a smaller increase among students. Finally, for the 30-49 group the only e¤ect is in the number of housewives. The Malaysian Migration Survey Report is informative about the plausibility of these results. It …nds that the labor force participation rate among female internal migrants is about 50 percent (slightly higher than in the overall population, see Table 1). We …nd that 10 additional immigrants in a state leads to the internal movement of 3.2 female

18

Malaysians between the ages of 15 and 64, of who 1.2 are employed, suggesting a labor force participation rate of a little under 40 percent. For male internal migrants the report …nds a labor force participation rate of 87 percent (substantially above the overall average). Our estimates suggest that about 95 percent of men who are induced to move due to immigration work. The Migration Survey Report also …nds that the marriage rate among internal migrants is around 62 percent (ages 15 to 64). This suggests that 2.1 of the 3.2 Malaysian women who move due to the arrival of 10 immigrants are married. We estimate that 1.6 of these female internal migrants are housewives, and if we assume that those move due to their husband’s relocation, our estimates suggest that about three-quarters of all internal female migrants move due to their husband. This back of the envelope calculation is broadly consistently with the Migration Survey Report which reports that around twothirds of all migrating Malaysian women do so on the account of their husband.21 Education The IV results for the …ve education groups are reported in Table 7; OLS results are in Appendix Table A.4. We see an increase in employment at the state level due to immigration for all education categories, except for those with a university degree. The gains have an inverse u-shape; the impact of immigration is positive for those without primary education, and largest for those with primary and lower and upper secondary education. At the top-end of the education distribution immigration has no signi…cant impact on those with a vocational degree or certi…cate and actually reduces the employment of those natives with a university degree. The implication is that even very low-skilled natives have bene…ted economically from low-skilled immigration in these two decades.22 The increase 21

Our back of the envelope calculation assumes that marriage rates of natives in a state are una¤ected by immigration. However, around 5 percent of all internal migration decisions are due to marriage according to the Malaysian Migration Survey Report. Immigration, by improving the economic prospects of natives, may also result in natives of a state marrying someone out-of-state who then relocates to join them. 22 This …nding contrasts somewhat with Ozden and Wagner (2014) for those with minimum education. They …nd that immigrants displace low-skilled natives and primarily bene…t middle-skilled natives. An explanation for the discrepancy is that their analysis was focused on the period 2003-10, while we consider the period 1990-2010. Since Malaysians and immigrants improved their educational attainment over time,

19

in the number of Malaysians out-of-labor force in a state, particularly those who are housewives, is also concentrated at the lower-end of the skill distribution. Provided there is assortative matching on education among Malaysians, this …nding is consistent with our interpretation that immigration causes households to move together.

V. Conclusions This paper provides an analysis of native responses to immigration on multiple margins, and presents rare evidence for a developing country. We …nd that immigration causes substantial internal movement of natives. Immigration increases the demand for Malaysian workers in a state, and many of these workers bring their spouses, three-quarters of married female migrants are occupied as housewives, and school-age children. The results speak to a broader economic and social impact of immigration than is typically consider in the literature. A line of inquiry that naturally follows from this work is understanding the e¤ect of immigration on women, who are frequently tied-movers, and the implications for their labor force attachment and fertility decisions. Similarly important is the impact immigration has on the educational decisions of natives, both as the returns to education change and as parents uproot their children to take advantage of the opportunities provided by immigration.

References [1] Altonji, J. and Card, D. (1991), The e¤ects of immigration on the labor market outcomes of less-skilled natives, in J. Abowd and R. Freeman (eds.), Immigration, Trade, and the Labor Market, University of Chicago Press, Chicago. the lowest Malaysian education groups had to increasingly compete with immigrants, something that was not yet true during the 1990s.

20

[2] Artuc, E., Docquier, F., Ozden, C. and Parsons, C. (forthcoming), A Global Assessment of Human Capital Mobility, World Development. [3] Borjas, G. (2006), Native Internal Migration and the Labor Market Impact of Immigration, Journal of Human Resources, 41, 221-258. [4] Borjas, G. (2014), Immigration Economics, Harvard University Press, Cambridge, USA. [5] Borjas, G., Grogger, J. and Hanson, G. (2010), Immigration and the Economic Status of Black Men, Economica, 77, 255-282. [6] Burbidge, J., Magee, L., and Robb, L. (1988), Alternative Transformations to Handle Extreme Values of the Dependent Variable, Journal of the American Statistical Association, 83, 123–127. [7] Card, D. (2001), Immigrant In‡ows, Native Out‡ows, and the Local Labor Market Impacts of Higher Immigration, Journal of Labor Economics, 19, 22-64. [8] Card, D. (2009), Immigration and Inequality, American Economic Review, 99, 1-21. [9] Card, D. and DiNardo, J. (2000), Do Immigrant In‡ows Lead to Native Out‡ows?, American Economic Review, 90, 360-367. [10] Card, D. and Lewis, E. (2007), The Di¤usion of Mexican Immigrants During the 1990s: Explanations and Impacts, in G. Borjas, (ed.), Mexican Immigration to the United States, University of Chicago Press, Chicago. [11] Cortes, P. and Pan, J. (2013), Outsourcing Household Production: Demand for Foreign Domestic Helpers and Native Labor Supply in Hong Kong, Journal of Labor Economics, 31, 327-371. [12] Cortes, P. and Tessada, J. (2011), Low-skilled Immigration and the Labor Supply of Highly Skilled Women, American Economic Journal: Applied Economics, 3, 88-123. 21

[13] Del Carpio, X., Karupiah, R., Marouani, M. , Ozden, C. , Testaverde, M., and Wagner, M. (2013), Immigration in Malaysia: Assessment of its Economic E¤ects, and a Review of the Policy and System, World Bank, Washington DC. [14] Department of Statistics, Malaysia. (2012), Migration Survey Report, Putrajaya, Malaysia. [15] Dustmann, C. and Glitz, A. (2012), How do Industries and Firms Respond to Changes in Local Labor Supply? IZA Discussion Paper no. 6257. [16] Facchini, G., Mayda, A. M. and Mendola, M. (2013), South-South Migration and the Labor Market: Evidence from South Africa, mimeo. [17] Filer, R. (1992), The Impact of Immigrant Arrivals on Migratory Patterns of Native Workers, in G. Borjas and R. Freeman (eds.), Immigration and the Work Force: Economic Consequences for the United States and Source Areas, University of Chicago Press, Chicago. [18] Frey, W. (1995), Immigration and Internal Migration Flight from US Metropolitan Areas: Toward a New Demographic Balkanization, Urban Studies, 32, 733 - 757. [19] Gindling, T. (2009), South-South Migration: the Impact of Nicaraguan Immigrants on Earnings, Inequality and Poverty in Costa Rica, World Development 37, 116-126. [20] González Luna, L. and Ortega, F. (2011), How do Very Open Economies Absorb Large Immigration Flows? Evidence from Spanish Regions, Labour Economics, 18, 57-70. [21] Hanson, G. and McIntosh, C. (2010), The Great Mexican Emigration, Review of Economics and Statistics, 92, 798-810. [22] Kritz, M. and Gurak, D. (2001), The Impact of Immigration on the Internal Migration of Natives and Immigrants, Demography, 38, 133-145. 22

[23] Lewis, E. (2011), Immigration, Skill Mix, and Capital Skill Complementarity, Quarterly Journal of Economics,126,1029-1069. [24] Lewis, E. (2013), Immigration and Production Technology, Annual Review of Economics, 5, 165-191. [25] MacKinnon, J. G. and Magee, L. (1990), Transforming the Dependent Variable in Regression Models, International Economic Review, 31, 315–339. [26] Ottaviano, G., Peri, G., and Wright, G. (2013), Immigration, O¤shoring and American Jobs, American Economic Review, 103, 1925-1959. [27] Ozden, C., Parsons, C., Schi¤, M. and Walmsley, T. (2011), Where on Earth is Everybody? The Evolution of Global Bilateral Migration 1960-2000,” World Bank Economic Review, 25, 12-56 [28] Ozden, C. and Wagner, M. (2014), Immigrants versus Natives? Displacement and Job Creation, mimeo, Boston College. [29] Peri, G. and Sparber, C. (2009), Task Specialization, Immigration, and Wages, American Economic Journal: Applied Economics, 1, 135-169. [30] White, M. and Liang, Z. (1998), The E¤ect of Immigration on the Internal Migration of the Native-Born Population, 1981-1990, Population Research And Policy Review, 17, 141 - 66. [31] Wright, R., Ellis, M. and Reibel, M. (1997), The Linkage Between Immigration and Internal Migration in Large Metropolitan Areas in the United States, Economic Geography, 73, 234 - 254.

23

Figure 1: Distribution of Immigrants across Malaysian States in 1992 and 2010

Figure 2: Net Population Gain by State due to Immigration, 1992 to 2010.

24

Table 1: Descriptive Statistics I, Malaysians by gender (Fractions) Male 1992

Female

2001

2010

1992

2001

2010 0.45

Labor Force / Population

0.84

0.82

0.77

0.47

0.46

Employed / Population

0.81

0.79

0.75

0.45

0.44

0.44

Unemployed / Labor Force

0.04

0.04

0.03

0.04

0.04

0.04

Part-time / All Employed

0.06

0.05

0.04

0.11

0.08

0.07

In School / Out of Labor Force

0.67

0.68

0.65

0.19

0.25

0.29

Housewife / Out of Labor Force

0.03

0.03

0.02

0.77

0.71

0.66

Retired / Out of Labor Force

0.13

0.14

0.18

0.01

0.01

0.01

Education distribution: No formal

0.06

0.03

0.02

0.15

0.09

0.05

Primary

0.46

0.34

0.20

0.43

0.33

0.20

Lower secondary

0.18

0.20

0.20

0.15

0.17

0.18

Upper secondary

0.25

0.33

0.41

0.23

0.33

0.42

Certificate/Diploma

0.03

0.05

0.09

0.02

0.05

0.08

Degree and above

0.03

0.05

0.07

0.01

0.03

0.07

Married

0.56

0.57

0.54

0.61

0.61

0.59

70,943

67,555

119,961

73,655

68,508

122,315

5,283,050

7,169,698

8,551,266

5,350,709

6,822,395

8,428,412

No. Observations Total

25

Table 2: Descriptive Statistics II, Malaysians and Immigrants (Fractions) Malaysians

Immigrants

1992

2001

2010

1992

2001

2010

Ages 15 to 19

0.17

0.16

0.15

0.12

0.16

0.07

Ages 20 to 29

0.30

0.29

0.27

0.42

0.29

0.19

Ages 30 to 49

0.39

0.41

0.39

0.39

0.41

0.62

Ages 50 to 64

0.14

0.14

0.19

0.06

0.14

0.12

0.14

Age distribution:

Education distribution: No formal

0.11

0.06

0.04

0.22

0.18

Primary

0.45

0.33

0.20

0.64

0.66

0.51

Lower secondary

0.17

0.18

0.19

0.05

0.05

0.14

Upper secondary

0.24

0.33

0.42

0.05

0.07

0.14

Diploma

0.02

0.05

0.09

0.02

0.01

0.02

Degree and above

0.02

0.04

0.07

0.03

0.03

0.04

Industry distribution: Agriculture, Mining

0.21

0.15

0.12

0.40

0.33

0.33

Manufacturing

0.24

0.22

0.17

0.15

0.24

0.18

Construction

0.07

0.09

0.09

0.17

0.12

0.14

Services

0.33

0.39

0.43

0.26

0.29

0.34

Public Administration, Health, Education

0.15

0.16

0.19

0.01

0.02

0.01

0.50

0.49

0.50

0.42

0.47

0.43

Gender distribution: Female

No. Observations Total

144,598

136,063

242,276

7,967

9,267

16,224

10,633,759

13,992,093

16,979,678

461,861

1,162,942

1,395,003

26

Table 3: Main Results OLS

IV

(1)

(2)

A: Dependent Variable = Native Population Ages 15-64 Immigrant Employment

0.747∗∗∗

0.757∗∗∗

(0.185)

(0.210)

B: Dependent Variable = Total Native Employment Immigrant Employment

0.499∗∗∗

0.519∗∗∗

(0.129)

(0.156)

C: Dependent Variable = Full-Time Native Employment Immigrant Employment

0.425∗∗∗

0.441∗∗∗

(0.126)

(0.151)

D: Dependent Variable = Part-Time Native Employment Immigrant Employment

0.074∗∗∗

0.077∗∗∗

(0.022)

(0.023)

E: Dependent Variable = Native Unemployment Immigrant Employment

0.011

0.014

(0.015)

(0.018)

F: Dependent Variable = Native Out of the Labor Force Immigrant Employment

0.236∗∗∗

0.224∗∗∗

(0.070)

(0.065)

G: Dependent Variable = Native Out of the Labor Force - Students Immigrant Employment

0.080∗∗∗

0.069∗∗∗

(0.024)

(0.021)

H: Dependent Variable = Native Out of the Labor Force - Housewife Immigrant Employment

0.150∗∗∗

0.163∗∗∗

(0.053)

(0.046)

I: Dependent Variable = Native Out of the Labor Force - Retired Immigrant Employment

-0.002

-0.013

(0.008)

(0.013)

J: Dependent Variable = Native Population Ages 0-14 Immigrant Employment

0.292∗∗

0.281∗∗

(0.119)

(0.116)

State and Year Fixed Effects

Yes

Yes

State Linear Time Trend

Yes

Yes

.

17.0

270

270

F-statistic Observations

Note: Each coefficient is from a separate regression. Standard errors are clustered by state-year and robust to heteroskedasticity. *** p<0.01, ** p<0.05, * p<0.1.

27

Table 4: Robustness Checks 1 Spatial weights

Weighted Regressions

SEs Clustered by State

Log-Log

A: Dependent Variable = Native Population Ages 15-64 Immigrant Employment

0.788∗∗∗

0.794∗∗∗

0.757∗∗∗

0.113∗

(0.213)

(0.234)

(0.201)

(0.065)

B: Dependent Variable = Total Native Employment Immigrant Employment

0.533∗∗∗

0.543∗∗∗

0.519∗∗∗

0.092

(0.158)

(0.169)

(0.114)

(0.075)

C: Dependent Variable = Full-Time Native Employment Immigrant Employment

0.451∗∗∗

0.450∗∗∗

0.441∗∗∗

0.055

(0.153)

(0.168)

(0.117)

(0.072)

D: Dependent Variable = Part-Time Native Employment Immigrant Employment

0.082∗∗∗

0.088∗∗∗

0.077∗∗∗

-0.030

(0.024)

(0.027)

(0.016)

(0.351)

E: Dependent Variable = Native Unemployment Immigrant Employment

0.016

0.036∗∗

0.014

0.001

(0.018)

(0.016)

(0.016)

(0.164)

F: Dependent Variable = Native Out of the Labor Force Immigrant Employment

0.238∗∗∗

0.224∗∗∗

0.224∗∗∗

0.155∗∗∗

(0.065)

(0.076)

(0.078)

(0.060)

G: Dependent Variable = Native Out of the Labor Force - Students Immigrant Employment

0.076∗∗∗

0.064∗∗∗

0.069∗∗∗

0.162∗∗

(0.022)

(0.023)

(0.025)

(0.077)

H: Dependent Variable = Native Out of the Labor Force - Housewife Immigrant Employment

0.164∗∗∗

0.170∗∗∗

0.163∗∗∗

0.088∗

(0.047)

(0.050)

(0.044)

(0.051)

I: Dependent Variable = Native Out of the Labor Force - Retired Immigrant Employment

-0.012

-0.038

-0.013

0.117

(0.013)

(0.025)

(0.026)

(0.333)

J: Dependent Variable = Native Population Ages 0-14 Immigrant Employment

0.282∗∗

0.320∗∗

0.281∗∗

-0.003

(0.118)

(0.125)

(0.134)

(0.044)

State and Year Fixed Effects

Yes

Yes

Yes

Yes

State Linear Time Trend

Yes

Yes

Yes

Yes

F-statistic

16.8

See note

16.8

7.5

Observations

270

270

270

270

Note: Each coefficient is from a separate regression. Regressions in the second column are weighted by average value of the dependent variable in a state over time. First stage F-statistics in the weighted regressions are between 14.7 and 21.3. Standard errors are clustered by state-year, except in Column 3 where they are clustered by state, and robust to heteroskedasticity. *** p<0.01, ** p<0.05, * p<0.1.

28

Table 5: Robustness checks 2 - Instruments Sample

1990-2010

1995-2010

1995-2010

2000-2010

2000-2010

Instruments

1990-2010

1990-1993

1990-1993

1990-1995

1990-1995

(1)

(2)

(3)

(4)

(5)

A: Dependent Variable = Native Population Immigrant Employment

0.757∗∗∗

0.973∗∗

4.530∗∗∗

1.021∗∗

1.262

(0.210)

(0.398)

(1.610)

(0.466)

(0.978)

B: Dependent Variable = Total Native Employment Immigrant Employment

0.519∗∗∗

0.518∗

2.431∗∗∗

0.762∗

0.664

(0.156)

(0.303)

(0.895)

(0.419)

(0.633)

C: Dependent Variable = Full-Time Native Employment Immigrant Employment

0.441∗∗∗

0.393

2.376∗∗∗

0.548

0.566

(0.151)

(0.290)

(0.896)

(0.369)

(0.604)

D: Dependent Variable = Part-Time Native Employment Immigrant Employment

0.077∗∗∗

0.125∗∗

0.056

0.214∗∗∗

0.099

(0.023)

(0.053)

(0.047)

(0.065)

(0.066)

E: Dependent Variable = Native Unemployment Immigrant Employment

0.014

0.050

0.145∗∗

-0.003

-0.018

(0.018)

(0.035)

(0.062)

(0.034)

(0.042)

F: Dependent Variable = Native Out of the Labor Force Immigrant Employment

0.224∗∗∗

0.404∗∗∗

1.953∗∗∗

0.261∗∗∗

0.616

(0.065)

(0.098)

(0.673)

(0.068)

(0.385)

G: Dependent Variable = Native Out of the Labor Force - Students Immigrant Employment

0.069∗∗∗

0.121∗∗∗

0.845∗∗∗

0.023

0.246

(0.021)

(0.036)

(0.324)

(0.036)

(0.188)

H: Dependent Variable = Native Out of the Labor Force - Housewife Immigrant Employment

0.163∗∗∗

0.296∗∗∗

0.803∗∗∗

0.218∗∗∗

0.230∗∗

(0.046)

(0.071)

(0.235)

(0.051)

(0.114)

I: Dependent Variable = Native Out of the Labor Force - Retired Immigrant Employment

-0.013

0.010

0.182∗∗

0.046∗

0.099

(0.013)

(0.020)

(0.079)

(0.027)

(0.070)

J: Dependent Variable = Native Population Ages 0-14 0.281∗∗

0.501∗∗∗

1.914∗∗∗

0.259∗

0.465

(0.116)

(0.178)

(0.57)

(0.138)

(0.316)

State and Year Fixed Effects

Yes

Yes

Yes

Yes

Yes

State Linear Time Trend

Yes

Yes

No

Yes

No

F-statistic

17.0

4.0

8.8

2.0

3.8

Observations

270

225

225

150

150

Immigrant Employment

Note: Each coefficient is from a separate regression. Standard errors are clustered by state-year and robust to heteroskedasticity. *** p<0.01, ** p<0.05, * p<0.1.

29

Table 6: IV Regressions by Gender Male

Female

A: Dependent Variable = Native Population Ages 15-64 Immigrant Employment

0.442∗∗∗

0.315∗∗∗

(0.130)

(0.087)

B: Dependent Variable = Total Native Employment Immigrant Employment

0.399∗∗∗

0.120∗∗

(0.102)

(0.058)

C: Dependent Variable = Full-Time Native Employment Immigrant Employment

0.357∗∗∗

0.084

(0.101)

(0.054)

D: Dependent Variable = Part-Time Native Employment Immigrant Employment

0.042∗∗∗

0.035∗∗∗

(0.013)

(0.011)

E: Dependent Variable = Native Unemployment Immigrant Employment

0.012

0.002

(0.013)

(0.008)

F: Dependent Variable = Native Out of the Labor Force Immigrant Employment

0.031

0.193∗∗∗

(0.022)

(0.049)

G: Dependent Variable = Native Out of the Labor Force - Students Immigrant Employment

0.037∗∗∗

0.032∗∗

(0.013)

(0.012)

H: Dependent Variable = Native Out of the Labor Force - Housewife Immigrant Employment

-0.003

0.165∗∗∗

(0.003)

(0.045)

I: Dependent Variable = Native Out of the Labor Force - Retired Immigrant Employment

-0.009

-0.004

(0.011)

(0.002)

J: Dependent Variable = Native Population Ages 0-14 0.168∗∗∗

0.113∗

(0.056)

(0.067)

State and Year Fixed Effects

Yes

Yes

State Linear Time Trend

Yes

Yes

F-statistic

17.0

17.0

Observations

270

270

Immigrant Employment

Note: Each coefficient is from a separate regression. Standard errors are clustered by state-year and robust to heteroskedasticity. *** p<0.01, ** p<0.05, * p<0.1.

30

Table 7: IV Regressions by Age Group 15-19

20-29

30-49

50-64

A: Dependent Variable = Native Population Immigrant Employment

0.123∗∗∗

0.429∗∗∗

0.229∗∗

-0.025

(0.023)

(0.075)

(0.109)

(0.063)

B: Dependent Variable = Total Native Employment Immigrant Employment

0.046∗∗∗

0.311∗∗∗

0.174∗

-0.013

(0.010)

(0.058)

(0.090)

(0.034)

C: Dependent Variable = Full-Time Native Employment Immigrant Employment

0.039∗∗∗

0.283∗∗∗

0.148∗

-0.028

(0.010)

(0.057)

(0.088)

(0.034)

D: Dependent Variable = Part-Time Native Employment Immigrant Employment

0.007∗∗∗

0.029∗∗∗

0.026∗∗∗

0.016∗∗

(0.002)

(0.008)

(0.010)

(0.007)

E: Dependent Variable = Native Unemployment Immigrant Employment

0.009

0.011

-0.003

-0.003

(0.006)

(0.010)

(0.004)

(0.002)

F: Dependent Variable = Native Out of the Labor Force Immigrant Employment

0.068∗∗∗

0.107∗∗∗

0.057∗∗

-0.009

(0.019)

(0.020)

(0.026)

(0.030)

G: Dependent Variable = Native Out of the Labor Force - Students Immigrant Employment

0.054∗∗∗

0.015

0.0002

-0.0002

(0.017)

(0.014)

(0.0007)

(0.0002)

H: Dependent Variable = Native Out of the Labor Force - Housewife Immigrant Employment

0.015∗∗∗

0.093∗∗∗

0.053∗∗

0.0009

(0.005)

(0.015)

(0.025)

(0.017)

I: Dependent Variable = Native Out of the Labor Force - Retired Immigrant Employment

-0.00003

0.0002

-0.0002

-0.013

(0.0001)

(0.0002)

(0.0008)

(0.013)

State and Year Fixed Effects

Yes

Yes

Yes

Yes

State Linear Time Trend

Yes

Yes

Yes

Yes

F-statistic

17.0

17.0

17.0

17.0

Observations

270

270

270

270

Note: Each coefficient is from a separate regression. Standard errors are clustered by state-year and robust to heteroskedasticity. *** p<0.01, ** p<0.05, * p<0.1.

31

Table 8: IV Regressions by Education Category No formal

Primary

Lower Secondary

Upper Secondary

Certificate/Diploma

Degree and above

A: Dependent Variable = Native Population Ages 15-64 Immigrant Employment

0.082∗∗∗

0.495∗∗∗

0.259∗∗∗

-0.008

-0.020

-0.051∗∗

(0.028)

(0.144)

(0.052)

(0.056)

(0.017)

(0.025)

B: Dependent Variable = Total Native Employment Immigrant Employment

0.050∗∗

0.332∗∗∗

0.168∗∗∗

0.034

-0.018

-0.047∗∗

(0.021)

(0.106)

(0.035)

(0.039)

(0.014)

(0.022)

C: Dependent Variable = Full-Time Native Employment Immigrant Employment

0.040∗

0.301∗∗∗

0.162∗∗∗

0.018

-0.031∗∗

-0.048∗∗

(0.021)

(0.101)

(0.033)

(0.038)

(0.013)

(0.022)

D: Dependent Variable = Part-Time Native Employment Immigrant Employment

0.010∗∗

0.031∗∗∗

0.007

0.016∗∗

0.013∗∗∗

0.001

(0.005)

(0.011)

(0.004)

(0.007)

(0.003)

(0.002)

E: Dependent Variable = Native Unemployment Immigrant Employment

-0.001

0.006

0.016∗∗∗

-0.002

-0.002

-0.003∗∗

(0.001)

(0.007)

(0.005)

(0.010)

(0.002)

(0.002)

F: Dependent Variable = Native Out of the Labor Force Immigrant Employment

0.034∗∗

0.157∗∗∗

0.075∗∗∗

-0.040∗∗

-0.0007

-0.0008

(0.013)

(0.039)

(0.026)

(0.019)

(0.004)

(0.004)

G: Dependent Variable = Native Out of the Labor Force - Students Immigrant Employment

-0.0004

0.013∗

0.032∗∗∗

0.021

0.004

0.0004

(0.0004)

(0.007)

(0.012)

(0.016)

(0.003)

(0.002)

H: Dependent Variable = Native Out of the Labor Force - Housewife Immigrant Employment

0.032∗∗∗

0.131∗∗∗

0.046∗∗∗

-0.041∗∗∗

-0.003

-0.002

(0.011)

(0.025)

(0.015)

(0.016)

(0.002)

(0.001)

I: Dependent Variable = Native Out of the Labor Force - Retired 0.0003

0.0005

0.0008

-0.013∗∗

-0.001

-0.0006

(0.002)

(0.006)

(0.001)

(0.005)

(0.001)

(0.002)

State and Year Fixed Effects

Yes

Yes

Yes

Yes

Yes

Yes

State Linear Time Trend

Yes

Yes

Yes

Yes

Yes

Yes

F-statistic

17.0

17.0

17.0

17.0

17.0

17.0

Observations

270

270

270

270

270

270

Immigrant Employment

Note: Each coefficient is from a separate regression. Standard errors are clustered by state-year and robust to heteroskedasticity. *** p<0.01, ** p<0.05, * p<0.1.

32

Table A1: First Stage Estimates for Baseline Specification (1) IV - Indonesia 15-19

9847.170∗∗

IV - Indonesia 20-24

-4626.762∗

(4633.504) (2568.440) IV - Indonesia 25-29

-507.018 (8800.175)

IV - Indonesia 30-34

-15607.780 (17057.710)

IV - Indonesia 35-39

-5716.655 (11328.680)

IV - Indonesia 40-44

-6126.210 (7158.136)

IV - Indonesia 45+

-5738.468∗∗

IV - Philippines

100842.100

(2576.580) (80931.680) 4953.933∗∗

IV - Other

(2125.089) State and Year Fixed Effects

Yes

State Linear Time Trend

Yes

F-statistic

17.0

Observations

270

Note: Standard errors are clustered by state-year and robust to heteroskedasticity. *** p<0.01, ** p<0.05, * p<0.1.

33

Table A2: OLS Regressions by Gender Male

Female

A: Dependent Variable = Native Population Ages 15-64 Immigrant Employment

0.402∗∗∗

0.345∗∗∗

(0.108)

(0.082)

B: Dependent Variable = Total Native Employment Immigrant Employment

0.352∗∗∗

0.147∗∗∗

(0.085)

(0.051)

C: Dependent Variable = Full-Time Native Employment Immigrant Employment

0.311∗∗∗

0.114∗∗

(0.084)

(0.048)

D: Dependent Variable = Part-Time Native Employment Immigrant Employment

0.041∗∗∗

0.033∗∗∗

(0.013)

(0.010)

E: Dependent Variable = Native Unemployment Immigrant Employment

0.009

0.002

(0.010)

(0.007)

F: Dependent Variable = Native Out of the Labor Force Immigrant Employment

0.041∗∗

0.195∗∗∗

(0.020)

(0.056)

G: Dependent Variable = Native Out of the Labor Force - Students Immigrant Employment

0.038∗∗∗

0.042∗∗∗

(0.014)

(0.012)

H: Dependent Variable = Native Out of the Labor Force - Housewife Immigrant Employment

-0.003

0.153∗∗∗

(0.003)

(0.051)

I: Dependent Variable = Native Out of the Labor Force - Retired Immigrant Employment

-0.0003

-0.002

(0.007)

(0.002)

J: Dependent Variable = Native Population Ages 0-14 Immigrant Employment

0.160∗∗∗

0.132∗∗

(0.058)

(0.064)

State and Year Fixed Effects

Yes

Yes

State Linear Time Trend

Yes

Yes

Observations

270

270

Note: Each coefficient is from a separate regression. Standard errors are clustered by state-year and robust to heteroskedasticity. *** p<0.01, ** p<0.05, * p<0.1.

34

Table A3: OLS Regressions by Age Group 15-19

20-29

30-49

50-64

A: Dependent Variable = Native Population Immigrant Employment

0.122∗∗∗

0.388∗∗∗

0.224∗∗∗

0.013

(0.026)

(0.082)

(0.086)

(0.042)

B: Dependent Variable = Total Native Employment Immigrant Employment

0.041∗∗∗

0.279∗∗∗

0.171∗∗

0.008

(0.011)

(0.060)

(0.068)

(0.024)

C: Dependent Variable = Full-Time Native Employment Immigrant Employment

0.035∗∗∗

0.254∗∗∗

0.142∗∗

-0.006

(0.011)

(0.060)

(0.066)

(0.023)

D: Dependent Variable = Part-Time Native Employment Immigrant Employment

0.006∗∗∗

0.025∗∗∗

0.028∗∗∗

0.014∗∗

(0.002)

(0.007)

(0.010)

(0.006)

E: Dependent Variable = Native Unemployment Immigrant Employment

0.002

0.013

-0.001

-0.002

(0.006)

(0.009)

(0.003)

(0.002)

F: Dependent Variable = Native Out of the Labor Force Immigrant Employment

0.079∗∗∗

0.095∗∗∗

0.055∗∗

0.007

(0.02)

(0.023)

(0.028)

(0.020)

G: Dependent Variable = Native Out of the Labor Force - Students Immigrant Employment

0.062∗∗∗

0.018

-0.00006

-0.00008

(0.018)

(0.014)

(0.0006)

(0.0002)

H: Dependent Variable = Native Out of the Labor Force - Housewife Immigrant Employment

0.014∗∗

0.076∗∗∗

0.051∗

0.010

(0.007)

(0.018)

(0.027)

(0.012)

I: Dependent Variable = Native Out of the Labor Force - Retired Immigrant Employment

-0.0001

0.0003

-0.0002

-0.002

(0.0001)

(0.0002)

(0.0007)

(0.008)

State and Year Fixed Effects

Yes

Yes

Yes

Yes

State Linear Time Trend

Yes

Yes

Yes

Yes

Observations

270

270

270

270

Note: Each coefficient is from a separate regression. Standard errors are clustered by state-year and robust to heteroskedasticity. *** p<0.01, ** p<0.05, * p<0.1.

35

Table A4: OLS Regressions by Education Category No formal

Primary

Lower Secondary

Upper Secondary

Certificate/Diploma

Degree and above

A: Dependent Variable = Native Population Ages 15-64 Immigrant Employment

0.062∗∗

0.410∗∗∗

0.235∗∗∗

0.051

0.013

-0.025

(0.026)

(0.116)

(0.047)

(0.052)

(0.015)

(0.018)

B: Dependent Variable = Total Native Employment Immigrant Employment

0.036∗∗

0.271∗∗∗

0.140∗∗∗

0.069∗

0.008

-0.026∗

(0.018)

(0.085)

(0.028)

(0.037)

(0.012)

(0.015)

C: Dependent Variable = Full-Time Native Employment Immigrant Employment

0.028∗

0.243∗∗∗

0.135∗∗∗

0.051

-0.003

-0.028∗

(0.017)

(0.080)

(0.028)

(0.036)

(0.011)

(0.015)

D: Dependent Variable = Part-Time Native Employment Immigrant Employment

0.008

0.028∗∗∗

0.006∗

0.019∗∗∗

0.012∗∗∗

0.002

(0.005)

(0.010)

(0.003)

(0.007)

(0.003)

(0.001)

E: Dependent Variable = Native Unemployment Immigrant Employment

-0.001

0.005

0.011∗∗∗

-0.003

-0.0001

-0.0009

(0.001)

(0.006)

(0.004)

(0.010)

(0.002)

(0.001)

F: Dependent Variable = Native Out of the Labor Force Immigrant Employment

0.028∗

0.134∗∗∗

0.084∗∗∗

-0.015

0.005

0.001

(0.014)

(0.037)

(0.024)

(0.018)

(0.004)

(0.003)

G: Dependent Variable = Native Out of the Labor Force - Students Immigrant Employment

-0.0002

0.008

0.038∗∗∗

0.027

0.005

0.002

(0.0003)

(0.006)

(0.011)

(0.018)

(0.003)

(0.002)

H: Dependent Variable = Native Out of the Labor Force - Housewife Immigrant Employment

0.026∗∗

0.110∗∗∗

0.043∗∗∗

-0.029∗∗

0.0006

-0.0007

(0.013)

(0.029)

(0.015)

(0.013)

(0.001)

(0.001)

I: Dependent Variable = Native Out of the Labor Force - Retired 0.002∗

0.004

0.002

-0.009∗∗

-0.0003

-0.0006

(0.001)

(0.005)

(0.001)

(0.004)

(0.0008)

(0.002)

State and Year Fixed Effects

Yes

Yes

Yes

Yes

Yes

Yes

State Linear Time Trend

Yes

Yes

Yes

Yes

Yes

Yes

Observations

270

270

270

270

270

270

Immigrant Employment

Note: Each coefficient is from a separate regression. Standard errors are clustered by state-year and robust to heteroskedasticity. *** p<0.01, ** p<0.05, * p<0.1.

36