بررسی مجدد بازده اجتماعی برای آموزش و پرورش: شواهدی از شهرهای ایالات متحده

This paper re-examines the impact of city educational composition on wages, often interpreted as human capital externalities. Using U.S. Census data, I find large, positive spillovers from college education in the 1980s, as documented by Moretti (2004a). In contrast, in the 1990s, the supply of skilled workers has no impact on average wages and may even negatively impact the wages of low-skill workers. These findings invite reinterpretation of previous studies on social returns to education, as shifts in the impact of city education composition on wages are not consistent with standard models of technological human capital externalities.

There is little empirical consensus regarding either the existence or magnitude of human capital externalities. Studies using variation in lower educational attainment to identify external returns to education find little evidence of education spillovers Acemoglu (1999), while studies using variation in the proportion of college graduates in localities find external returns Moretti (2004a). In this paper, I demonstrate that the effect of the share of college graduates in a city on wages is remarkably unstable over time. Using city level census data, I find large positive spillovers from college education in the 1980s, consistent with the previous literature. However, in the 1990s this result disappears. To gain further insight into the pattern of education spillovers, I examine the relation between college share and wages at different levels of education. The pattern of spillovers is difficult to reconcile with technological theories of spillovers, where externalities are built into the production function. Using data from the 1980, 1990 and 2000 U.S. Censuses for 286 consistently defined metropolitan areas, I examine decadal changes in city-level wages and their relation to the change in the fraction of a city's labor force with at least a college degree. After controlling for the effect of own education, the coefficient on city college share measures the presence of human capital spillovers under conditions described later. Working in differences allows me to control for possible correlations between city time-invariant factors and changes in college share by using within-city variation. However, there are still potential endogeneity concerns arising from omitted time-varying variables that are correlated with changes in the fraction of college graduates in a city. I address these concerns by using an instrumental variable strategy to predict changes in the local proportion of college graduates. I also address the potentially non-random location of workers across cities using a method first developed by Dahl (2002). Despite addressing several econometric concerns, the instability of the estimated education spillovers remains. These findings invite reinterpretation of previous studies on social returns to education, as parameter instability is not a general feature of models of human capital spillovers. The remaining sections of this paper are organized as follows. In Section 2, I give a brief review of the literature on the social returns to education. In Section 3, I re-evaluate the evidence on education spillovers in U.S. cities from 1980–2000. Section 4 concludes.

The empirical literature on human capital externalities has often come to conflicting conclusions regarding the magnitude and existence of education spillovers. One emerging view reconciles previous mixed results by suggesting that positive externalities only accrue to post-secondary education. In this paper, I show that this view of human capital externalities is far from complete. Even when focusing on a city's share of college graduates as the measure of local human capital, spillovers are remarkably unstable between the 1980s and the 1990s. I document the instability of the over-time pattern of education spillovers using U.S. Census data from 1980 to 2000, focusing on a set of 286 consistently defined metropolitan areas. Using an empirical specification that is commonly employed in the literature on social returns to education, I find that the external return to college education is large, positive, and benefits all workers in the 1980s. However, in the 1990s this relationship changes dramatically; the overall impact of changes in college share on average city wages is essentially zero. This result hides heterogeneous impacts by education level. In particular, I find that more high-skill workers in a locality negatively affect the wages of low-skill workers in the 1990s. I address the potential endogeneity of city college share by using an instrumental variable approach and possible non-random self-selection of workers across cities by using an approach developed by Dahl (2002). Despite addressing these econometric concerns, the instability of the estimated spillovers remains. What has changed between a city's skill level and its effect on wages between the 1980s and the 1990s? Are the effects spurious or has there been a fundamental shift in the relationship of city-skill composition and wages? The fact that three instruments, using very different sources of variation, largely come to the same conclusions suggests that the observed turn around in the impact of college share may be real. In this case, it is worth reconsidering the economics and asking what type of model could potentially be consistent with these observations. There are essentially two types of models in the literature that suggest a causal relationship between the fraction of skill in an economy and wages: models of technological spillovers and models with pecuniary externalities. In models of technological spillovers, the externality is often built into the production function by specifying aggregate productivity to be a function of aggregate human capital. Spillovers in these types of models are often hypothesized to come from worker interaction and the exchange of ideas (see Moretti, 2004b for a review). These models are not likely to be consistent with the patterns in Section 3 unless the production function changed in an odd way, and the sharing of knowledge suddenly became unimportant. In models pecuniary externalities, spillovers from education arise through market interaction and changes in underlying parameters can produce significant differences in outcomes. Most empirical studies of the relationship between an economy's aggregate skill and wages refer to these models, without always taking their precise implications seriously due to the fact that they imply otherwise similar empirical relationships (Acemoglu and Angrist, 2001 and Ciccone and Peri, 2006). One possibility is that instead of models of technological spillovers, models of pecuniary externalities that incorporate structural breaks, such as Acemoglu (1999), may be a relevant way to interpret the patterns in the data and examining the additional implications of such models is an area for future research.