THE EFFECT OF EMIGRATION ON THE LABOR MARKET

International migration is now viewed as the household response to the economic changes of the globalization process. There is now a growing literature that has examined the motives to emigrate and which explores the effects of emigration and remittances within a household decision. While it has long been recognized that international migration and remittances have important effects on households and labor markets in the sender countries, recent improvements in data collection have greatly increased the ability to study these effects. Previously, most researchers relied on cross-sectional data and limited questions on migrants and remittances. With such data, A good review of the motives to remit can be found in Rappaport and Docquier (2005).

THE EFFECT OF EMIGRATION ON THE LABOR MARKET OUTCOMES OF THE SENDER HOUSEHOLD:

A LONGITUDINAL APPROACH USING DATA FROM NICARAGUA

researchers examined the effect of emigration and remittances on household outcomes by comparing outcomes of households from which emigrants left (or that received remittances) with the outcomes of those that did not.2 The key element of this approach is the control group for the comparison or the estimate of outcomes that would have occurred without migration or remittances when characteristics that are unobserved to the researcher are important determinants of outcomes.

Over the past decade, data on migration and remittances has improved in several ways that have benefited researchers. First, statistical institutes conducted surveys have included modules with more questions about migration and remittances.3 Second, more systematic collection of other data for the household allows better controls for other household characteristics. And third, there are more surveys that follow households over time. In this paper, I use data from both before and after emigration to determine the effects of emigration on household labor market outcomes. The innovation of the paper is a difference-indifferences approach that controls for other characteristics of the emigrant household as well as other changes that commonly affect all households. By controlling for the initial situation of the household, the estimates provide an estimate of the effect migration and remittances that is not possible with the cross-sectional approach used in previous studies.

The organization of the paper is as follows. In the next section (Section 1), I discuss the pattern of emigration from Nicaragua. I then describe the data from the 1998 and 2001 Living Standard Measurement Surveys (LSMS) for Nicaragua and the basic patterns from those data. In Section 2, I present the methodological approach to determining the effects of emigration on household outcomes. Section 3 presents the results of the estimation. And Section 4 includes discussion – including a comparison of results with cross-sectional methods – and concluding remarks.

1. Data and Summary Characteristics 1.1 Emigration from Nicaragua Emigration from Nicaragua increased substantially in the 1980s, leveled off during the 1990s, and increased again beginning in the late 1990s. Whereas most of the migration from the 1980s was to One approach to this issue is to use a selection model for the presence of an emigrant to predict household characteristics without an emigrant. Adams (2005) uses this approach for Guatemala with cross-sectional data for 2000 using a selection model for the presence of remittances. The identifying variable for the first stage regression is age of the household head. He finds that the additional effect of remittances is to reduce poverty in Guatemala, especially severe poverty. He finds that remittances reduce poverty, but that households that receive remittances tend to view remittances as temporary sources of income. Many of these studies are summarized in World bank (2006).

See, for example, Adams and Page (2003).

For Mexico, the Mexican Migration Project (http://mmp.opr.princeton.edu/) has combined all of these features for a large number of sender communities. For the Philippines, Yang (2005) and Yang and Martinez (2006), use longitudinal data on households to examine the effect of exchange rate shocks (caused by the Asian financial crisis) to remittances on changes in household outcomes. They find that exchange rate induced increases in remittances lead to higher human capital formation, entrepreneurship, increased child schooling, reduced child labor, and increased hours worked in self-employment. For Central America, studies of the effects of emigration using the more recent national-level data sources are limited.

WELL-BEING AND SOCIAL POLICY

the United States, there was a change towards migration to Costa Rica during the 1990s By the late 1990s, the destination for most emigrants was to Costa Rica and over ten percent of the population of Costa Rica was Nicaraguans. There are now approximately 250,000 Nicaraguans in each of the United States and Costa Rica, representing over 10 percent of the native-born population of Nicaragua.

Another increase in emigration following 1998 occurred after Hurricane Mitch in October 1998.5 Most of these emigrants went to Costa Rica, which, in response to the hurricane, granted amnesty to emigrants that had entered the country legally (but did not have permanent status) prior to the hurricane. 1.2 Data — 1998 and 2001 LSMS The Instituto Nacional de Estadistica y Censo (INEC) and the World Bank conducted Living Standards Measurement Surveys in Nicaragua in 1998, and 2001.7 Each survey contains detailed demographic and labor market information for a national sample of households. In addition, the 2001 survey (only) includes a module with household members residing outside the country. The 1998 survey was conducted before Hurricane Mitch and, following the hurricane, a subset of the 1998 households in areas most affected by the hurricane was re-interviewed in 1999 to assess damage.

The 2001 LSMS was designed to provide a longitudinal panel with the 1998 LSMS. Of the 4,209 households in the 1998 survey, 3,018 can be matched in the 2001 survey. The final sample contains 2,994 matched households. Within the households that can be matched between and 2001, there are 17,475 individuals enumerated in the 1998 LSMS.8 Of these, 12,319 can be matched directly with the line numbers for 1998 provided in the 2001 survey household roster. Other persons can be matched using information on age and sex. The number of persons in the 1998 who were matched to the 2001 data is shown in Appendix Table.

Following the Hurricane in October/November 1998, a subset of households were reinterviewed in 1999. Many, but not all, of the questions from 1998 were repeated and additional questions on the impact of the hurricane were included. Selection for inclusion in the 1999 survey was based on residence in areas affected by the hurricane. For purposes of this paper, I use inclusion in the 1999 survey (or non-inclusion) as an instrument for impact of the hurricane.

In Nicaragua, approximately 3,000 people died and nearly one million people were affected. The areas most affected by the hurricane were in the north, especially Chinandega and Leon. See for example IADB (2000).

There are now several descriptive studies of emigrants and remittances from Nicaragua, including Funkhouser (1992, 1995), Pritchard (1999), Blanco, del Carmen, and Hernandez (2002), and the ILO (2001). Previous studies of the effects of emigration from Nicaragua and remittances on labor market outcomes have used crosssectional data. Funkhouser (1992) found that migration and remittances affect labor force participation and self-employment of remaining household members using data for Managua in 1989. And Barham and Boucher (1998) examined the impact of remittances on income inequality using a small sample of households in the Atlantic Region Funkhouser, Perez Sainz, and Soto (2003) examined the integration of Nicaraguans into the Costa Rican labor market.

A LSMS was also conducted in 1993.

These are out of the total of 23,643 individuals in the 1998 LSMS.

There are persons in the 2001 LSMS that are reported to have been included in the 1998 LSMS, but for which the 1998 line numbers do not match.

THE EFFECT OF EMIGRATION ON THE LABOR MARKET OUTCOMES OF THE SENDER HOUSEHOLD:

A LONGITUDINAL APPROACH USING DATA FROM NICARAGUA

1.3 Emigrants in the LSMS The unique aspect of this study is the observation of emigrants before and after migration.

In the 2001 data, there are 352 persons that emigrated between 1998 and 2001 and can be matched from the line numbers for 1998 provided in the emigrant file in 2001. These persons resided outside the country in 2001. In addition, there are 430 persons from the matched households that are included in the emigrant file for 2001 that emigrated prior to the 1998 LSMS and continued to reside outside the country in 2001.

The pattern of emigration found in the LSMS is shown in Figure 1 using weights. The increase in the proportion of emigrants to Costa Rica begins in the mid-1990s. The increase in migration after 1998 is also quite striking.

There is not strong evidence that areas most affected by the hurricane were associated with larger out-migrations following 1998, at least at the level of the Department. In Appendix Table 2, I report the proportion of households that were included in the 1999 survey, the proportion of the population that migrated prior to 1998, and the proportion of the population of the Department that emigrated between 1998 and 2001. Overall, the department most affected by the hurricane had migration rates similar to other departments before and after 1998.

WELL-BEING AND SOCIAL POLICY

1.4 Summary information Summary information on individuals using the LSMS data is provided in Tables 1 through 3. In Table 1, descriptive information about emigrants and non-migrants are shown for the different cuts of the data. In Table 2, household means are shown. And in Table 3, the labor market insertion of households according to migration status is presented.