temporary migration overseas and household labor supply: evidence from urban philippines

17
Temporary Migration Overseas and Housebold Labor Supply= Evidence porn Urban Pbidzppines Edgard R. Rodriguez University of Toronto Erwin R. Tiongson Georgetown University The impact of international migration on the labor supply of workers' nonmigrant relatives has not been well documented in the literature. Using household survey data representing mostly overseas contract work- ers, i.e., temporary migrants, this paper shows that labor supplies of migrants and their nonmigrant relatives are inseparable. Migrants reduce the labor supply of nonmigrant relatives, which translates into lower earnings from local labor markets. Households substitute income for more leisure - a significant and previously little recognized benefit of emigration for Philippine households. This benefit varies by gender of nonmigrants and is generally higher for men. International migration gives hundreds of thousands of Philippine migrants access to high-wage labor markets. Through family ties, international migrants maintain remittance arrangements with their nonmigrant relatives at home. Remittances have thus become a common and significant source of household income. In 1991, for example, 17 percent of the almost 12 million households in the country received some sort of income from abroad, main- ly from migrants. For households receiving remittances from abroad, this source represents a significant share of their income (more than a third of total household income). Since these households are not among the poorest families in the country, remittances have contributed as well to increased income inequality in the short run (Rodriguez, 1998). Migration and international remittances also bring about changes in labor supply and consumption patterns that have been overlooked in the lit- erature. In addition to financing increases in household goods consumption,' remittances also affect the consumption of leisure and, therefore, labor sup- 'Family Income and Expenditure Survey (FIES) data show that remittances from migrants increase expenditures in all categories, especially nondurables and education. 0 2001 by the Center for Migration Studies of New York. All rights reserved. 0198-9183/01/3503.0135 ZMR Volume 35 Number 3 (Fall 2001):709-725 709

Upload: edgard-r-rodriguez

Post on 23-Jul-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

Temporary Migration Overseas and Housebold Labor Supply= Evidence porn Urban Pbidzppines Edgard R. Rodriguez University of Toronto

Erwin R. Tiongson Georgetown University

The impact of international migration on the labor supply of workers' nonmigrant relatives has not been well documented in the literature. Using household survey data representing mostly overseas contract work- ers, i.e., temporary migrants, this paper shows that labor supplies of migrants and their nonmigrant relatives are inseparable. Migrants reduce the labor supply of nonmigrant relatives, which translates into lower earnings from local labor markets. Households substitute income for more leisure - a significant and previously little recognized benefit of emigration for Philippine households. This benefit varies by gender of nonmigrants and is generally higher for men.

International migration gives hundreds of thousands of Philippine migrants access to high-wage labor markets. Through family ties, international migrants maintain remittance arrangements with their nonmigrant relatives at home. Remittances have thus become a common and significant source of household income. In 1991, for example, 17 percent of the almost 12 million households in the country received some sort of income from abroad, main- ly from migrants. For households receiving remittances from abroad, this source represents a significant share of their income (more than a third of total household income). Since these households are not among the poorest families in the country, remittances have contributed as well to increased income inequality in the short run (Rodriguez, 1998).

Migration and international remittances also bring about changes in labor supply and consumption patterns that have been overlooked in the lit- erature. In addition to financing increases in household goods consumption,' remittances also affect the consumption of leisure and, therefore, labor sup-

'Family Income and Expenditure Survey (FIES) data show that remittances from migrants increase expenditures in all categories, especially nondurables and education.

0 2001 by the Center for Migration Studies of New York. All rights reserved. 0198-9183/01/3503.0135

ZMR Volume 35 Number 3 (Fall 2001):709-725 709

Page 2: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

710 INTERNATIONAL MIGRATION REVIEW

ply and wage income. In Metro Manila, for example, households without migrants and households with migrants have similar average incomes (see Appendix Table l), but domestic wages represent 43 percent of total income in the former group and only 28 percent in the latter. This evidence suggests that migrants and their families substitute domestic wage earnings with for- eign earnings. Changes in the labor supply of families with migrants can occur because of two closely related but separate features of migration: the absence of the migrant and the flow of remittances.2 These two features could lead to increased or reduced labor supply among nonmigrant relatives.

Empirical investigations of labor supply indicate that workers’ leisure- work preferences may not be separable from labor supply decisions of other family members. International migration affects the labor supply of nonmi- grants in two ways. First, the effect depends on which tasks the migrant per- formed before departure. If international migrants are substitutes for nonmi- grants in household production, migration would decrease the labor supply of nonmigrants in the local labor market. If migrants are complements in household production, then their departure causes an increase in the labor supply of nonmigrants in the local labor markets.

Second, when migration occurs, nonmigrant relatives receive remit- tances, which they perceive as additional nonlabor income. An increase in nonlabor income then reduces their participation in local labor markets, which consequently reduces locally earned wages.

Using household survey data representing mostly overseas contract work- ers, i.e., temporary migrants, this study finds evidence that migration affects labor participation and hours worked by nonmigrant individuals. This results in a higher consumption of leisure and thus increased household welfare, despite lower income from local labor markets. This benefit of migration, previously unaccounted for in the literature, varies by gender and is general- ly higher for men.

This article first examines some theoretical considerations about house- hold labor supply. To estimate the impact of migration on the labor supply of nonmigrants, we use data on migrants (mostly contract workers) and nonmi- grants from the 1991 Survey on Overseas Workers (SOW) and the 1991 Labor Force Survey (LFS), respectively. We use a sample of nonmigrant indi- viduals from households in the National Capital Region (Metro Manila)

2Migration is a necessary but insufficient condition to observe remittances. For example, according to the 1991 Survey on Overseas Workers (SOW), 16.6 percent of migrants had not remitted within the last six months prior to the survey.

Page 3: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

TEMPORARY MIGRATION OVERSEAS AND HOUSEHOLD LABOR SUPPLY 71 1

which has a very high incidence of international migrants. Next, the matched sample is discussed (descriptive statistics are provided in Appendix Table 1). Finally, the econometric results are analyzed and summarized.

A MODEL OF LABOR PARTICIPATION WITH MIGRATION Does family influence the decision to work? There is no generally accepted model of family decision-making. One popular approach has been to con- sider the family as a single decision-making unit (for example, a married cou- ple). Under this approach, the nuclear family has a collective set of prefer- ences and behaves accordingly. Another approach emphasizes that each member of the family has an individual utility constrained by the family’s budget, implying some bargaining among members (Manser and Brown, 1980; McElroy and Horney, 1981).

Migration strongly suggests the interdependence of work decisions with- in a family. Theory, supported by empirical research, has shifted its view of migration from an individual decision-making process to a mutually inter- dependent decision. Migrants and nonmigrants jointly decide migration; costs and returns of migration are shared according to an implicit contractu- al arrangement between the two parties. For example, remittances - a promi- nent, direct benefit of migration - are likely to be ruled by this voluntary arrangement, which can be self-enforcing (Stark and Bloom, 1985: 174-5).

Risk-handling provides another example in which family is collectively responsible for individual migration. Although the family is a small group with which to pool risks, migration overseas - where earnings are either neg- atively correlated, statistically independent, or not highly positively correlat- ed with earnings in the source country - provides coinsurance. Coinsurance also explains nonmigration of the remainder of the family. For instance, acquisition of general human capital is known to provide self-insurance; so does migration, together with specific human capital. Migration eases the risk associated with investment in specific human capital, thus providing effi- ciency gains (Stark and Bloom, 1985175).

Does migration affect the labor supply of nonmigrants? While migration is expected to modify nonmigrant behavior, empirical studies have yet to quantify the magnitude and direction of this impact. We investigate the mag- nitude and sign of this impact using survey data from the Philippines, where international migration is widespread. Since the 1970s, Philippine families have had access to labor markets in the Middle East and Asia. Attractive wages overseas have made families consider “temporary” migration (of up to

Page 4: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

712 INTERNATIONAL MIGRATION REVIEW

several years) as an opportunity to improve their living standards at home. For such households, migration also means changes in family life in a variety of ways, including the labor supply of nonmigrant members.

We estimate labor supply functions for nonmigrants. These functions are conditional on having a migrant relative overseas. The conditional approach offers several advantages. For example, this approach allows us to test whether nonmigrant labor is separable from migration, while assuming general forms of preferences (Browning and Meghir, 1991). In the context of migration, preferences here refer to attitudes of a nonmigrant wife or husband (who has a migrant spouse) towards work in the domestic labor market. If these pref- erences are not taken into account, estimates of household welfare, for instance, may be biased. Ignoring the dependence of nonmigrant labor sup- ply (with respect to international migrants) underestimates migrants’ contri- butions to household welfare if migration does reduce nonmigrant labor sup- ply (and, hence, locally earned income). When one takes this reduction into account, households are better off than first observed because they consume more leisure, despite lower local income.

Participation in paid labor is expected to depend on 1) the opportunity cost of leisure, 2) nonlabor income and 3) family preferences.

Opportunity Cost of Leisure

Labor market participation is represented by the variable L (Lij= 1, when indi- vidual i who belongs to family j participates in paid labor, and L,,=O, other- wise). This dichotomous variable indicates that the decision to participate in labor markets (so hours worked, H,, are positive) depends on the individual’s characteristics (I,) such as the individual’s education and age. These charac- teristics reflect the potential market wage (wIm) of the individual such that older, more educated workers are expected to obtain higher wages given their experience and training. Higher potential earnings are then expected to increase the probability of participating in paid employment. Family attrib- utes (F,), such as number of dependents and their age structure, affect partic- ipation differently, depending on gender and marital status of the individu- als. Although family characteristics may not affect potential market wages, they influence the decision to stay home by increasing or decreasing the low- est wage the person would accept in order to work (ie., his or her reservation wage, w:). The behaviors of men and women are known to differ with respect to forms of participation in family life and responsibilities for child care.

Here, cultural norms and institutions play an important role (Brown,

Page 5: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

TEMPORARY MIGRATION OVERSEAS AND HOUSEHOLD LABOR SUPPLY 71 3

1985). In a “traditional society,” married women participate less in paid employment; instead, they undertake all household production. The presence of children reduces female labor participation, although this depends on the particular social arrangements in place to provide child care. By definition, individuals participate more in labor markets as market wages exceed reserva- tion wages (wim>w:).

Nonlabor Income

Wealth or nonlabor income (Yj) also affect the labor supply of nonmigrants. Nonlabor income reduces labor participation as families demand more leisure, together with other goods and services. Given the magnitude of international migrants’ remittances - a major source of nonlabor income for nonmigrants - a strong income effect is expected, with reduced labor supply of nonmigrants. This has been referred to as a “dependence” effect of international migration. The presence and strength of this effect, however, depends on family ties and dynamics; thus, strong family ties imply large and stable flows of remittances from the migrant worker and tend to produce the dependence effect. If migra- tion is perceived as weakening family ties, so that the flow of remittances is not secure, spouses of migrant workers may decide, for example, to increase their labor supply. Evidence from the United States indicates a similar effect on female labor supply when family ties weaken. American married women facing the prospects of divorce gradually increase their hours of work and labor force participation prior to separation; once separation occurs, the changes are even more dramatic (Johnson and Skinner, 1986).

FumiZy Preferences

Participation of other family members in foreign labor markets influence or “condition” a nonmigrant individual’s labor supply (ie., Mi= 1, when family j has a migrant) via effects on the allocation of household production. Deter- mining the cross-effect of one family member on the productivity of another family member remains an empirical question. The size and signs of these effects between husbands and wives have neither been clearly established (McElroy, 1981), nor have the size and signs of these effects between migrants and nonmigrants. Migration affects the productivity of nonmigrants in pro- duction-related and consumption-related activities. For instance, migration of a spouse raises the nonmigrant spouse’s productivity for a given number of hours at home because the nonmigrant spouse takes over tasks once per- formed by the migrant worker. If spouses are substitutes in household pro-

Page 6: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

714 INTERNATIONAL MIGRATION REVIEW

duction, the nonmigrant‘s supply of labor decreases when the spouse leaves for work overseas.

Sociologists have long emphasized the changing roles of wives during the overseas migration of their husbands. First, migration clearly affects their household production activity. In the Philippines, wives of migrant workers undertake both parents’ roles in raising children during their husbands’ absence. Furthermore, because remittances need to be administered, wives take over a more central role in family budgeting, and in cases where the fam- ily operates a small business, wives frequently take charge of its daily opera- tion. Second, wives’ roles in household consumption also change. For instance, wives’ supply of labor increases if they and their husbands were com- plementary in household consumption before husbands’ emigration. With their husbands overseas, family life becomes less enjoyable (ie., have less util- ity), leading to an increase in their labor supply. This offsets any decline of labor supply resulting from the fact that husbands and wives are substitutes in household production. The dominant cross-effect will depend on family circumstances.

In short, family preferences determine the direction of migration’s effect on nonmigrant labor supply. The effect of migration reduces or increases labor supply among nonmigrant relatives, together with the relevant produc- tion functions, depending on the relationship with the migrant, his or her education, and gender. We represent this interdependence with migrants’ characteristics (ZM1) by adding an interactive dummy (expressed by M,Z “i).

Participation in labor markets by nonmigrant relatives (of migrants) is thus summarized as follows:

DATA In October 1991, the National Statistics Office (NSO) undertook the Survey on Overseas Workers (SOW) together with the quarterly Labor Force Survey (LFS). These surveys give information on both migrants and nonmigrants. From the SOW, we obtain data on migrants who were working abroad between April and September 199 1. Most of them are male contract workers, usually in the Middle East. O n average, they have stayed overseas for over two years (more details of this survey are found in Rodriguez and Horton, 1996). Beside migrants’ characteristics, the survey also contains information on

Page 7: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

TEMPORARY MIGRATION OVERSEAS AND HOUSEHOLD LABOR SUPPLY 715

remittances within the last six months prior to the survey (for the determi- nants of these remittances, see Rodriguez, 1996).

Because of its structure and time frame, the October 1991 SOW sample includes mostly contract workers, ie., temporary migrants, rather than per- manent migrants. First, the survey’s screening question asks for migrants who are currently working overseas, thus excluding permanent migrants who do not work, such as retirees, children or students. Second, the time frame of the screening question - which is limited to migrants working overseas between April and September 1991 - reduces the coverage of permanent migrants, particularly those who left several years earlier and who now live in the Unit- ed States, one of the larger host countries for Filipinos overseas.3 In addition, migrants require the presence of relatives in the Philippines to be included in the survey. Due to the nature of permanent migration, permanent migrants are less likely to have immediate family residing in the Philippines.

These features explain the high proportion of contract workers in the 1991 SOW sample, despite the large number of permanent Philippine migrants we observe overseas. More recent NSO surveys have extended the coverage of Filipinos overseas to include permanent migrants.

Meanwhile, the LFS provides data on the number of hours worked and wages of those members who resided in the Philippines and worked between July and September 1991. With this information, we estimate the labor par- ticipation of nonmigrants, as described in equation 1 using data on nonlabor income from the 1991 Family Income and Expenditure Survey (FIES). This survey, undertaken in July 1991 and January 1992, is the only source of data on nonlabor income. The difference in timing explains why households in this survey are not exactly the same as those in the LFS or SOW, although households interviewed for the FIES come from the same NSO pool of households. Given the computational costs of matching three surveys, we have limited our sample to the National Capital Region (Metro Manila), which has a very high incidence of international migrants (some 10 percent of all households report having migrants overseas) and accounts for most of the migrants overseas.

RESULTS Table 1 shows the results of a probit analysis, which identifies the effects of individual and family characteristics on the probability of participating in

3According to the 1990 U.S. census, almost 1 million residents are Philippine-born.

Page 8: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

716 INTERNATIONAL MIGRATION REVIEW

labor markets, using the model corresponding to equation 1. We define par- ticipation in labor markets as paid employment, including self-employment and excluding unemployed or unpaid family workers.4 We expect women and men to have different responses in terms of labor participation: 66.7 percent of all men worked for wages, while only 45.6 percent of women participated in paid labor.

TABLE 1 PROBABILITY OF LABOR PARTICIPATION AMONG NONMIGRANTS: lMEN AND WOMEN OF WOWING AGE

(15-64 YEARS), NATIONAL CAPIIXL REGION, THE PHILIPPINES, 1991 ( W S AND MAXIMUM LIKELIHOOD ESTIM.4")

Men Women Means Probit Marginal Means Probit Marginal

Estimates Effects Estimates Effects Participation (Last Quarter) 66.7% 45.6% No. oiNonmigrant individuals 6,438 Constant -4.17"

Individual Characteristics (17)

Complete High School 29.2% 0.13' (0.05)

Incomplete College 23.8% -0.26' (0.05)

University 15.3% 0.33" (0.07)

Head 45.8% 0.41" (0.08)

Married 52.2% 0.53" (0.06)

32.3 years 0.26' (0.01)

Age Squared/ 100 12.2 -0.33' (0.01)

Children (0-14 years) 1.8 0.03b (0.01)

(0.01) Adult Men (1 5-64 years) 2.4 -0.02

(0.02)

(0.05)

If family has 1 or more Migrants 9.0% -0.29' (0.13)

If Migrant is Male 5.9% 0.21

Age

Nonmigrant Family Composition:

Adult Women (15-64 years) 2.1 -0.0006

Seniors (over 64 years) 0.2 -0.04

Migrant-related Characteristics:

4.2

-8.5

10.7

13.3

17.2

8.5

-10.7

1 .o 0.0

-0.7

-1.3

-9.4

6.8 - (0.13)

If Mierant belones to nuclear familv 3.9% -0.28b -9.1

7,294 -3.35"

(15)

28.2% -0.07b (0.04)

18.8% -0.34"

17.5% (0.05)

9.0% 0.06 (0.07)

(0.05) 32.5 years 0.19'

51.1% -0.70"

12.3

1.8

2.6

1.8

0.2

11 .O%

8.6%

5.6%

(0.01) -0.24" (0.01)

-0.04" (0.01) 0.09" (0.01) -0.02 (0.01) 0.09'

(0.04)

-0.48" (0.13) 0.01

(0.13) 0.09

-2.6

-12.18 (0.05) 18.1

2.3

-26.4

7.2

-9.0

-1.5

3.4

-0.8

3.4

-18.1

0.4

3.4 (spoule or childkn of head of HH) (0.13) (0.11)

*Both groups (unemployed and unpaid workers) only account for 6.5 percent of individuals ofworking age residing in the National Capital Region, which is the sample used in this paper. For details on the sample, see Appendix Tables 1, 2 and 3.

Page 9: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

TEMPORARY MIGRATION OVERSEAS AND HOUSEHOLD LABOR SUPPLY 717

TABLE 1 (CONTINUED) PROBABILITY OF LABOR PARTICIPATION AMONG NONMIGRANTS: MEN AND WOMEN OF WORKING AGE

(15-64 YEARS), NATIONAL CAPITAL REGION, THE PHILIPPINES, 1991 (MEANS AND MAXIMUM LIKELIHOOD ESTIMATES)

Men Women Means Probit Mareinal Means Probit Marginal

Estimates Efficts Estimates Effzcts If Migrant has Tertiary Education 5.2% -0.09 -2.9 6.5% 0.33" 12.4

(0.13) (0.1) Nonlabor Income:

Remittances within HH 1.8 -0.01" -0.3 2.7 -0.005" -0.2

migrant family member) Imputed Rent 3.3 -0.002 -0.1 4.7 0.008" 0.3

migrant family member)

(thousands of pesos per non- (0.003) (0.002)

(thousands of pesos per non- (0.002) (0.002)

Chi-squared for covariates (17 d.f.) 2,756' 1,312' Source: Marched Sample, SOW, LFS and FIES "Significant at 95% or higher bSignificanr between 90% and 95%

More educated and older individuals are likely to have higher labor par- ticipation. Gender differences exist. First, when marginal effects are calculat- ed (following Greene, 1993: 369), a completed university degree raises the probability of having a paid job 10.7 percentage points above that of work- ing-age men with incomplete secondary education or less. For women (who are, on average, more likely to hold a university degree), a complete universi- ty education raises the probability of participating in paid work by even more -18.1 percentage points. For both men and women, incomplete college edu- cation reduces their participation, reflecting the fact that nonmigrants with incomplete education are still in school. Second, age profiles show the expect- ed pattern of increasing participation up to a certain age and then decreasing participation after. The lower age coefficients for women also reflect the fact that returns to age (or experience) are lower for women because of the greater frequency of interrupted careers.

Family characteristics (headship, marital status and number of family members) also affect labor participation and indicate differences between men and women. First, headship of households increases men's participation in paid employment significantly (by 13.3 percentage points), while the effect of headship among women is insignificant. Less than one tenth of women in the sample were heads of their families, while almost half the men were. From a traditional perspective, men were more likely to assume the responsibility of providing for the family. Men may be considered more productive in under- taking work outside the household, while women are left in charge of home

Page 10: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

718 INTERNATIONAL MIGRATION REVIEW

production. Second, marital status affects men and women differently. The probability of participation is 17.2 percentage points higher for married than for unmarried men, while the probability decreases for married women by 26.4 percentage points. The latter indicates that, on average, married couples perceive women (traditionally or economically) as being more productive in household production. Recall, however, that married women with a universi- ty degree are much more likely to participate than married women without higher education. Third, the number and age composition of other family members appear to influence female labor participation more than that of males. Each additional child in the household reduces the probability of a woman's labor participation by only 1.5 percentage points. Adult women and older relatives (over 64 years) in the household release women from house- work, increasing the probability of participating in paid labor by 3.4 per- centage points per additional person in these two categories of relatives living at home.

Does having migrant relatives affect the behavior of nonmigrants? It does appear that migrant and nonmigrant labor participation are not separable. This nonseparability affects men and women differently (see Table 1). For men, having a migrant household member reduces the probability of labor participation by 9.4 percentage points. If the migrant is a member of the nuclear family (the head's spouse or children), then the total probability declines by 18.5 percentage points. For women, having a relative who is a migrant also has a negative impact on participation (a decrease of 18.1 per- centage points), regardless of gender or relationship to their household. How- ever, if the migrant has tertiary education, labor participation of nonmigrant women rises significantly by 12.4 percentage points. The overall effect of a migrant with higher education is still negative, that is, a decrease of 5.7 per- centage points in the labor participation of nonmigrant women.

A series of factors (such as measurement error, heterogeneity and simul- taneity) can exaggerate or diminish the effect of having a migrant as part of the household (that is, M;=l). Factors such as incomplete coverage, unob- served household preferences, or household-specific labor constraints cannot be addressed with the cross-sectional data at hand, and they may produce a number of biases in our results. Under certain circumstances, these factors may indicate a downward bias.

First, the variable Mj may be measured with error. Because of the high proportion of contract workers in the 1991 SOW, M, is only capturing part of the total number of international migrants in our sample of Philippine

Page 11: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

TEMPORARY MIGRATION OVERSEAS AND HOUSEHOLD LABOR SUPPLY 719

households. The direction of the bias depends on the correlation between having a temporary migrant (which we observe) and having a permanent migrant. If having a temporary migrant is negatively correlated to having a permanent migrant, then the effect of M, may be understated.5 A priori, we expect this negative correlation to hold if permanent migration is the pre- ferred option among families in the Philippines. Permanent migration tends to thwart temporary migration. Host countries that encourage permanent settlement also emphasize family reunification, which reduces the need or the desire to become temporary migrants among prospective migrants who already have relatives as permanent migrants overseas. Hence, households with no temporary migrants may well have permanent migrants who do not show in the sample, that is, M,=O, although the family j has permanent migrants. In this case, our coefficient for Mj is downward-biased. The effect on labor supply indicates that only households with temporary migrants have reduced labor supply, although this reduction may also be common in house- holds with permanent migrants. This effect of migration on labor supply appears smaller than the actual effect if permanent migration were accounted for since we assume that both types of migration have the same negative effect on labor supply.

A second factor that affects the coefficient of M, is heterogeneity. House- holds are different from each other, and some of these differences explain why households have migrants in the first place. These differences may or may not be observed, In particular, family size and composition are observable charac- teristics, and they arc correlated with having migrants. For instance, larger fam- ilies with working-age adults are expected to have more migrants than smaller families with many dependents (either children or seniors). Hence, without accounting for the presence of working-age adults in the family, the effect of M, tends to be exaggerated. Our results take heterogeneity into account by adding observable family characteristics in the regression analysis, but they do not account for any other family characteristics that are not observable.

Preferences provide an important example of an unobserved factor that influences labor supply (and that can therefore affect our estimate of M,). Households with preference for lower labor participation tend to have more migrants than families who participate more actively in paid labor. In other

>If having a temporary migrant in a family is positively correlated to having a permanent migrant, the effect of Mj in our results may be upward-biased. The total effect on labor sup- ply would seem due to temporary migration alone, although there may be other migrants who also reduce the labor supply of nonmigrants but are ignored in the sample.

Page 12: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

720 INTERNATIONAL MIGRATION REVIEW

words, the error term of the labor participation equation is negatively correlat- ed with Mj (it., having a migrant). In this case, we understate the effect of Mj on labor supply when we estimate labor participation using ordinary least squares (OLS).

A third factor that affects our estimates of Mj is simultaneity. While our estimates assume that only Mj determines Lj, labor supply of nonmigrants also determines migration; that is, Lij determines M,. A common belief is that migrants tend to come from households where members‘ labor is lower than the labor they would have supplied had they not encountered constraints. These constraints in the domestic labor market may be, for example, expressed in the difficulty to find work for pay or in working longer hours. Families facing these constraints therefore have lower-than-desired labor supply, and they may choose to have more migrants than families who face no constraints in the domestic labor market. For example, unemployment among educated urban youth is quite common in developing countries such as the Philippines. This lack of opportunities for the young and educated illustrates a constraint for some families, and this constraint results in more emigration. Thus, constrained families (it., families with involuntary low labor supply) are also likely to have a migrant in the family. If this is the case, and low labor supply explains emi- gration (and emigration also affects labor supply of nonmigrants), then our OLS estimate of Mj is likely to be upward-biased because of simultaneity.

There are some grounds to believe that the effect of simultaneity is some- what reduced in our sample.6 For instance, we only take a sample of house- holds in the largest urban center of the Philippines, which implies, in princi- ple, that all individuals in the sample face the same high unemployment rate. In fact, the National Capital Region, or Metropolitan Manila, has the high- est unemployment rate in the country. In 1991, for example, its unemploy- ment rate was over 15 percent, while the national average was about 9 per- cent (Asian Development Bank, 1998).

In addition to having migrants, our results indicate that nonlabor income of nonmigrants also affects their participation in paid employment. One plausible indicator of nonlabor income is imputed rent; however, in the sur- vey, this category of “income” (which gives the market value of the dwelling) may not be different from using any other indicator of consumption because wealthier families are expected to seek more expensive accommodations.

6Not all individuals in Metropolitan Manila face the same labor market constraints since con- straints may depend on their individual characteristics, so that wen using a sample of indi- viduals residing in the same city can still produce upward-biased OLS estimates of Mj.

Page 13: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

TEMPORARY MIGRATION OVERSEAS AND HOUSEHOLD LABOR SUPPLY 72 1

Another possible indicator of nonlabor income are remittances from migrants. In the SOW, we observe that having a migrant does not automati- cally yield remittances for the household (of all Philippine migrants, 83.4 per- cent remit), so the effect of remittances can be identified separately from the effect of having a migrant. If remittances are perceived as nonlabor income by nonrnigrant family members, we would obtain some evidence to support a general model in which joint utility is maximized (that is, remittances and nonlabor income yield the same coefficient in the labor supply regressions.)

Results in Table 1 show the expected sign for nonlabor income variables. They are significant and suggest stronger income effects for men than for women when per-capita remittances are used. An additional thousand pesos (about U.S.$40) of remittances per nonmigrant family member reduces labor participation of nonmigrant men by 0.3 percentage points, while women’s probability of engaging in paid work only declines by 0.2 percentage points. The larger income effect for men reflects their traditional high involvement in paid work (if they are already supplying close to their maximum potential labor); at that point, their labor supply is less responsive to changes in wages than to changes in income. These results agree with cross-sectional studies of other countries, which suggest that men’s labor supply is more responsive to income, while women’s labor supply is more responsive to changes in wages (Mincer, 1985; Killingsworth, 1983).

It is unclear whether migrants’ remittances have a similar effect on labor supply as other nonlabor income. The coefficient for imputed rent, the other observed component of nonlabor income, was significant only for women, and the estimated sign was opposite to the expected sign. Signs for imputed rent are upward-biased because individuals who work more can also afford better housing and therefore have higher imputed rent. The coefficients of other components of nonlabor income, such as net winnings from gambling (not reported here), had the expected sign, but were also either insignificant or different from the coefficient of remittances.

CONCLUSIONS International migration influences labor supply of nonmigrants, and there- fore, the labor supplies of nonmigrants and migrants are not separable. Migrants reduce the labor supply of nonmigrants (their labor participation and their hours worked) who substitute income for more leisure. This repre- sents an additional benefit of emigration that has not received enough atten- tion in the literature.

Page 14: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

722 INTERNATIONP~L MIGRATION REVIEW

The effect varies according to the gender of nonmigrants. Male labor par- ticipation declines when migrants belong to the nuclear family. Among women, having a migrant reduces their chances of working, but educated migrants raise this probability. Estimates indicate that participation declines by up to 18.5 percentage points for men (and only 5.7 percentage points for women if migrants have tertiary education). These reductions in labor par- ticipation represent declines of 27.7 percent for men (and 12.5 percent for women). A decline in the probability to work implies lower labor earnings. Besides the effect of migration per se, remittances also reduce labor supply of nonmigrants. This income effect is small, but relatively stronger for men than for women. An additional thousand pesos in per-capita remittances reduces the chances of working by 0.2 percentage points for women and 0.3 percent- age points for men.

APPENDIX 1 HOUSEHOLD INCOME AND EXPENDITURES:

HOUSEHOLDS WITH AND WITHOUT MIGRANTS, NATIONAL CAPITAL REGION, THE PHILIPPINES, OCTOBER 1991

Households without Households with migrants migrants Mean S.D. Mean S.D.

No. of observations 3496 409 Family Size

A. Nonmigrant Family Members (1)+(2)+(3)+(4) (1) Children (0-5 years old) (2) Youngsters (6-14 years old) (3) Adults (15-64 years old) (4) Seniors (over 65 years old) B. International Migrants (Adults)

Annual Household Income (pesos) (1)+(2)+(3)+(4) (1) Entrepreneurial Income (2) Wage Income (3) International Remittances Remittances (April-September 199 1) (4) Other Income Imputed Rent

Annual Household Expenses (pesos)

Income

Expenditure

(1)+(2)+(31+(4)+(5)+(6) (1) Food Expenses Food Expenses ( % of total expenses) (2) Household Operations (3) Education (4) Nondurables (5) Durables

5.56 0.75 1.12 3.51 0.18

0

159064

33593 77258 6101

0 421 12 20094

100211

45138 57.24 3413 3340 311 2656

2.29 0.99 1.23 1.87 0.48

393259

286422 162060 40050

181228 59093

183250

35951 15.74 8934 8680 2152 18285

5.25 0.67 1.06 3.33 0.17 1.14

151808

11212 46914 61875 79887 3 1806 17206

105722

48547 51.81 2734 4643 377 3860

2.25 0.89 118 1.85 0.43 0.40

103053

26585 562

62678 51 1277 51052 34900

76536

23938 15.53 3357 6629 913

18372 (6) Disbursements 10302 69971 13626 39223 (7) Other Expenses 35050 98301 31935 25658 Source: Matched Sample, FlES, and SOW

Page 15: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

TEMPORARY MIGRATION OVERSEAS AND HOUSEHOLD LABOR SUPPLY 723

APPENDIX 2 NONMIGRANTS’ DEMOGRAPHIC CHARACTERISTICS:

NATIONAL CAPITAL REGION, THE PHILIPPINES, OCTOBER 1991 (14,670 NONMIGRANTS)

Mean S.D. Minimum Maximum Relationship with Head

Head 0.262 0.440 0 1 Wife 0.206 0.404 0 1 Son/Daughter 0.330 0.470 0 1 SonlDaughter-in-law 0.029 0.167 0 1 Grandchildren 0.012 0.108 0 1 Parents 0.006 0.079 0 1 Other Relatives 0.091 0.287 0 1 Nonrelatives 0.063 0.244 0 1

Age 32.42 13.05 15 65 Male 0.464 0.499 0 1

No Education” 0.009 0.093 0 1 Incomplete Primary 0.044 0.205 0 1 Complete Primary 0.128 0.335 0 1 Incomplete High School 0.152 0.359 0 1 Complete High School 0.287 0.452 0 1 Incomplete College 0.209 0.407 0 1 Complete College 0.170 0.376 0 1

Single 0.445 0.497 0 1

Demographic Characteristics

Education

Marital Status

Married 0.515 0.500 0 1 Other 0.040 0.196 0 1

Source: LFS “0.3% did not report highest grade obtained.

APPENDIX 3 NONMIGWS’ LABOR W CHARA~~EIUSTICS:

(14,670 NONMIGRANTS) NATIONAL CAPITAL REGION, THE PHILIPPINES, OCTOBER 1991

Reference Period N Mean S.D. Min Max Last Quarter 14670 0.540 0.498 0 1

1. Paid Employment a. Total Wages (pesos) 7923 10398.47 19164.58 80 702000 h. Total Hours (hours) 7923 593.17 128.2 12 1472

Class of Worker (%) a. Private Employee 7923 0.684 0.465 0 1 b. Public Employee 7923 0.097 0.296 0 1 c. Employer 7923 0.205 0.404 0 1 d. Other 7923 0.014 0.115 0 1

Permanent Job 7923 0.871 0.336 0 1 Seasonal Job 7923 0.116 0.321 0 1 Odd Job 7923 0.013 0.114 0 1 2. Unpaid Employmenr 14670 0.013 0.112 0 1 3. Unemployment 14670 0.052 0.221 0 1 4. Out of Labor Force 14670 0.396 0.489 0 1

Employment Status (%)

(Continue)

Page 16: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

724 INTERNATIONAL MIGRATION REVIEW

APPENDIX 3 (CONTINUED) NONMIGRANTS’ LABOR hhRKET CHARACTERISTICS:

NATIONU CAPITAL REGION, THE mLIPPINEs, OCTOBER 1991 (14,670 NONMIGRANTS)

Reference Period N Mean S.D. Min Max Last Week

Employment (Last Week) 14670 0.543 0.498 0 1 Unemploymentd (Last Week) 14670 0.048 0.215 0 1 Out of Labor Force” (Last Week) 14670 0.409 0.492 0 1

a. Too young, too old or sick 5994 0.078 0.268 0 1 b. Schooling 5994 0.379 0.485 0 1 c. Housekeeping 5994 0.423 0.494 0 1

Reasons to be Out of the Labor Force (Last Week)

d. Other reasons 5994 0.121 0.326 0 1 Source: LFS ‘1.6% of workers reporting a job in the last quarter had no job in the last week

REFERENCES Asian Development Bank 1998 Compendium of Social Statistics in the Philippines. Manila: Asian Development Bank. Arcinas, F. R. and C. R. Banzon-Bautista 1992 “The Philippines.” In The Impact of Labor Migration on Househokh: A Comparative

Study in Seven Asian Countries. Ed. G. Godfrey. Tokyo: United Nations University Press. Pp. 172-226.

Brown, C. 1985 “An Institutional Model of Wives‘ Work Decisions,” Industrial Relations,

24(2): 182-204. Browning, M. and C. Meghir 1991 “The Effects of Male and Female Labor Supply on Commodity Demands,” Economet-

rica, 59(4):925-951. Greene, W. 1993 Econometric Analysis. Toronto: Maxwell MacMillan Canada. Second Edition. Gunatilleke, G., ed. 1992 The Impact ofLabor Migration on Householdc A Comparative Study in Seven Asian Coun-

tries. Tokyo: United Nations University Press.

Johnson, W. and J. Skinner 1986 “Labor Supply and Marital Separation,” The American Economic Review,

76(3):455469. Killingsworth, M. 1983 Labor Supph: Cambridge Survey of Economic Literature. Cambridge: Cambridge Uni-

Lundberg, S. 1988 “Labor Supply of Husbands and Wives: A Simultaneous Equations Approach,” Review

Manser, M. and M. Brown 1980 “Marriage and Household Decision-Making: A Bargaining Analysis,” International Eco-

versity Press.

of Economics and Statistics, 70:224-235.

nomic Review, 21(1):3144.

Page 17: Temporary Migration Overseas and Household Labor Supply: Evidence from Urban Philippines

TEMPOMY MIGRATION OVERSEAS AND HOUSEHOLD LABOR SUPPLY 725

McElroy, M. 1981 “Appendix: Empirical Results from Estimates of Joint Labor Supply Functions of Hus-

bands and Wives.” In Research in Labor Economics: A Research Annual. Ed. R. Ehren- berg. Greenwich, CT: JAI Press. 453-64.

McElroy, M. and M. J. Horney 1981 “Nash-Bargained Household Decisions: Toward a Generalization of the Theory of

Demand,” international Economic Review, 22(2):333-349. Mincer, J. 1985 “Intercountry Comparisons of Labor Force Trends and of Related Developments: An

Mroz, A. 1987 “The Sensitivity of an Empirical Model of Married Women’s Hours of Work to Eco-

Pollak, R. 1969 “Conditional Demand Functions and Consumption Theory,” Quarterly Journal of Eco-

Pudney, S. 1989 Modelling Individual Choice: The Econometrics of Corners, Kinks and Holes. Cambridge,

MA and Oxford: Blackwell. Reichert, J. 1981 “The Migrant Syndrome: Seasonal U.S. Wage Labor and Rural Development in Cen-

tral Mexico,” Human Organisation, 40( 1):56-66. Rodriguez, E. 1998 “International Migration and Income Distribution in the Philippines,” Economic Devel-

Overview,” Journal of Labor Economics, 3( l), Supplement, Sl-S32.

nomic and Statistical Assumptions,” Econometrica, 55(4):765-800.

nomics, 3(1):60-78.

opment and Cultural Change, 46(2):32$350. January.

1996 “International Migrants’ Remittances in the Philippines,” The Canadian Journal ofEco-

Rodriguez, E. and S. Horton 1996 “International Return Migration and Remittances in the Philippines.” In Development

Stratea, Employment and Migration: County fiperiences. Ed. D. O’Connor and L. Farsakh. Paris: OECD Development Centre. Pp. 171-202.

nomics, 29, Special Issue, S427-S432. April.

Stark, 0. and D. Bloom 1985 “The New Economics of Labor Migration,” The American Economic Review,

75 (2) : 1 73- 1 78.