Air Traffic, Internal Migration, and Economics in the United States

Abstract

This thesis studies the causal effects on the local labor market from two types of major changes: air travel service and internal migration. Both direct estimations suffer from endogeneity concerns because of the interdependence between the changes and local economic outcomes. Therefore, I construct instrumental variables to control for the endogeneity problem in the estimation process. In the case of changes in air traffic growth, I constructed instruments in two different ways and have a chapter devoted to each. In the first chapter, I apply a Bartik (1991) shift-share instrument to construct predicted air traffic to address the endogeneity concern and study the causal effects on local economic outcomes in the US metropolitan areas from 1993 to 2016. Instruments for small airports and large hubs are constructed differently because of their different operating systems. For small airports, I multiply the air traffic growth in each hub by their shares in the small airport in the initial year of 1993 and then sum the products across hubs as an instrument for air traffic. I apply the K-Means clustering unsupervised machine learning technique to choose hubs under the hub-and-spoke system. The instrument for air traffic at large hubs is constructed similarly based on the airlines located at the hubs in 1993 and the growth rates of activity for those airlines at the national level. The results for cities with small airports reveal that a 10% increase in air traffic growth leads to a 0.64% increase in local employment growth, implying 3,200 new jobs created in a typical city. No statistically significant effects exist in cities with large hubs. The mechanism of these effects is mainly through labor demand. Financial service sectors are the primary beneficiary of air traffic growth. In the second chapter, I exploit changes in the number of airline routes due to hub activities and airline mergers to construct the instruments for local air traffic growth and estimate the effects on the local population, income, and employment growth at the county level from 1990 to 2014. The results confirm a positive correlation between air service and omitted variables in the OLS estimation. From the IV results, a 1% percent change in air service will significantly increase the population growth rate in the associated county by 0.159% and the local employment growth rate by 0.225%. There is a small and statistically insignificant effect of air service on the local per capita income growth rate, which is consistent with findings from the earlier literature about the rigidity in the per capita income adjustment. In the third chapter, I investigate the causal effect of internal migration on three local labor market outcomes: earnings, employment, and the probability of migrating out. For the endogeneity concern that migrants may be attracted to cities with good economies, I construct predicted in-migration and out-migration rates as instruments for the net-migration rate into the cities in two steps. First, I use the variation in the shift-share measures and weather events to predict the migrant flows to and from U.S. cities. Then I estimate migrants’ settlement patterns by the distance between the sending and receiving areas. The predicted in-migration (out-migration) rate is the sum over all sending (receiving) areas of the products between predicted out-migration (in-migration) flows and the corresponding probabilities of migrating to (from) that destination. The results show that net migration had little effect on the earning outcomes but did influence the hours worked in a usual week. Also, existing residents seem not to be crowded out by the in-migrants. However, net-migration flows into a metropolitan area significantly increase the employment status of existing residents. Relative to being part-time and unemployed, a one percent increase in the net-migration rate led to a 0.335% higher probability of being full-time employed. The probability of being in part-time employment was decreased by 0.171%, and the unemployment probability was reduced by 0.164%, with a one percent increase in the net-migration rate. These results imply that workers are working longer under the same wage payment system in response to internal migration instead of direct wage reduction.