Risk, Migration, and Rural Financial Markets: Evidence from Earthquakes in El Salvador

Dean Yang Risk, Migration, and Rural Financial Markets: Evidence from Earthquakes in El Salvador TWO FACTS MOTIVATE THIS STUDY.1 FIRST, LIFE IN RURAL AREAS IN developing countries is prone to m any kinds of risk, such as illness or the m ortality of household members, crop or other income loss due to natural disasters (weather, insect infestations, or fire, for example), and civil conflict. Second, international m igration is substantial and grow­ ing: betw een 1975 and 2000, the num ber of people worldwide living outside their countries of birth m ore than doubled to 175 million, or 2.9 percent of world population (United Nations 2002).2 For example, m igration from El Salvador to the United States has grown rapidly: betw een 1990 and 2001, th e num ber of Salvadoran-born individuals in the United States grew 69 percent, from 469,000 to 790,000.3These large m igrant inflows in recent decades have become m ajor public policy issues in m igrants’ destination countries. This paper concerns th e intersection of two subject areas in economics: research on th e causes of international m igration and research on the ways households in developing countries cope w ith different types of risk. W hat connections, if any, are there between the pervasiveness of rural risk in developing countries and these substan­ tial outm igration flows? I shed light on this question by examining how m igration from El Salvador responds to economic shocks. The key findsocial researchVol 75 : No 3 : Fall 2008 955 ing of the paper is that the im pact of shocks on outm igration is actu­ ally quite nuanced, and perhaps unexpected: the im pact of a shock on m igration by family mem bers depends on w hether the shock is idiosyn­ cratic (specific to the household) or aggregate (shared w ith other house­ holds in the local area). Salvadoran households become differentially m ore likely to have close relatives who are m igrants in the year following an idiosyncratic shock (a death in the family), compared to households not experiencing such shocks. By contrast, aggregate shocks (proximity to the massive 2001 earthquakes) lead to differential declines in w hether a household has m igrant relatives. Migration typically requires that a substantial fixed cost be paid up front. If credit or assistance from others are im portant m ethods of financing m igration’s fixed cost, in theoiy the im pact of economic shocks on m igration will depend on w hether such shocks also affect access to these financing m echanisms. W hen shocks are idiosyncratic or uncorrelated w ith shocks experienced by others, shocks are likely to raise m igration: they should m ake families m ore willing to send m em bers away for higher-wage work, and there should be no effect on m echanism s of m igration finance. But aggregate shocks may actu­ ally lead to less migration, if such shared shocks m ake it more difficult or costly to access credit or interhousehold assistance networks that normally facilitate migration. (A simple m odel that formalizes this idea is available from the author on request.) Declines in m igration are associated m ainly w ith being located in a quake-affected area (rather than the extent of the individual house­ hold’s earthquake damage), suggesting th at the explanation for the differential decline in m igration lies in changes in general local condi­ tions. The evidence is strongly suggestive that the decline in m igra­ tion in the areas closest to the quakes is due to increased difficulty in obtaining financing for m igration, via formal and inform al credit in particular. Differential declines in m igration in quake-affected areas are accompanied by substantial declines in households’ granted credit (both informal and formal). 956 social research I also present em pirical evidence against alternative explana­ tions for the differential decline in m igration in the areas closest to the quakes. It is not likely to be because of an increase in the dem and for family unity w hen negative shocks occur, as deaths in the family (which presumably would also raise the desire for...