Book Description
This dissertation analyzes how local labor markets respond to macroeconomic shocks, such as changes in the global price of commodities or international migration flows, based on their predetermined characteristics. I further investigate how short-run elasticities of employment, population and earnings to such economic shocks differ from the long-run ones. Last, I analyze which econometrics methods are best suited to estimate such responses over time. Chapter 1 analyzes how resource-rich local labor markets adjust to large swings in the global price of natural resources. I provide a novel dynamic analysis of the consequences of natural resource price cycles by measuring the economic performance of oil-rich areas in the U.S. over three decades (1970-2000). I find that the consequences of booms are different from the consequences of busts. Oil-rich economies adjust quickly to the new long-run equilibrium during booms by increasing local employment, nominal wages and income from capital. Migration responses are limited, which is consistent with the small real wage gains that occur because of contemporaneous large increases in local prices. The negative impact of a bust is borne locally through higher nonemployment and exacerbated by a large reduction in human capital investments observed during the preceding boom. The adverse long-run effects of boom-bust cycles mainly depress the lower end of the income distribution. Chapter 2, joint with Giovanni Peri, analyzes important correlations between immigration and labor market outcomes of native workers in the US. Using data on local labor markets, states and regions we first look at simple correlations and then we use regression analysis with an increasing number of controls for observed and unobserved factors. We review the potential methods to separate the part of this correlation that captures the causal link from immigrants to native labor outcomes and we show estimates obtained with 2SLS method using the popular shift-share instrument. One fact emerging from all the specifications is that the net growth of immigrant labor has a zero to positive correlation with changes in native wages and native employment, in aggregate and by skill group. We review the literature on the channels and the mechanisms that allow local economies to absorb immigrants with no negative (and possibly positive) impact on the labor demand for natives. Finally, Chapter 3 analyzes the identification of treatment effects delayed in time. Applied economists are increasingly interested in recovering the causal effect of a treatment over time, although the methods they use often ignore that outcomes and regressors may have internal propagation dynamics. Typical applications include geographic unit-by-time panel data in nonexperimental settings, such those analyzed in Chapters 1 and 2. In those contexts the elasticities of labor market outcomes to macro shocks in the short run may differ substantially from those in the long run. This chapter illustrates by means of two simple Monte Carlo simulations that traditional regression methods fail to recover such delayed effects. I find that distributed lags models in panel settings are substantially biased unless under trivial conditions with no internal propagation dynamics in both outcomes and regressors. In contrast, local projections methods perform significantly better. Most importantly, they can easily accommodate Arellano and Bond (1991) instruments thus reducing the bias further even in the presence of lagged dependent variables.