Self-Selection and the Earnings of Immigrants: Reply
回应了Jasso和Rosenzweig对作者1987年移民自我选择模型的批评，指出他们忽略了移民回流等选择性偏差，但未提供系统性解决方案。
My 1987 paper addresses a central question in the economics of immigration: Who are the immigrants? It has been traditionally assumed that immigrants are positively selected-that is, immigrants are more skilled, on average, than the typical person in the source country. One contribution of my theoretical analysis was to show that this assumption is not necessarily consistent with wealth-maximizing behavior. In fact, under some conditions the immigrant flow may be negatively selected.' My original study focused only on the selection bias caused by immigrants endogenously sorting themselves out of the source country's population.2 Guillermina Jasso and Mark Rosenzweig point out that I ignored other types of selection biases in both my theoretical development and empirical framework. These additional biases are due to: (1) the selective emigration of the foreign-born in the United States; and (2) the fact that my empirical work was limited to immigrants originating in 41 source countries. Regarding the first of these points, Jasso and Rosenzweig restate the well-known fact that emigration (or return migration) is an important characteristic of the immigrant experience. Robert Warren and Jennifer Marks Peck (1980), for instance, estimate that perhaps as many as 30 percent of immigrants leave the United States within a decade or two after their arrival. Jasso and Rosenzweig, however, do not examine systematically how the selective emigration of the foreign-born affects either the theory or the empirical analysis. In essence, their criticism simply asserts that selective emigration is likely to cause problems. My 1987 paper, however, already made note of that fact (p. 538). Moreover, the biases caused by emigration were an integral part of my earlier critique of the econometric methodology used by most studies of immigrant earnings determination (Borjas 1985, pp. 466-67). What is obviously needed is the generalization of both my theoretical and empirical frameworks to account for the emigration of the foreign-born. Such a generalization, however, is not provided by Jasso and Rosenzweig and does not yet exist.3 The second point made by Jasso and Rosenzweig deals with the empirical results presented in my paper. Because I limited my analysis to immigrants originating in 41 countries, the sample is not representative and serious selection biases may be at work. Obviously, it is always better to use a more representative sample in the empirical analysis. The data were limited to 41 source countries because I was concerned not only with immigrant wage levels, but also with immigrant wage growth as recorded by the 2/10