The economic performance of cities: a Markov-switching approach

This paper examines the determinants of employment growth in metro areas. To obtain growth rates, we use a Markov-switching model that separates a city’s growth path into two distinct phases (high and low), each with its own growth rate. The simple average growth rate over some period is, therefore, the weighted average of the high-phase and low-phase growth rates, with the weight being the frequency of the two phases. We estimate the effects of a variety of factors separately for the high-phase and low-phase growth rates, along with the frequency of the low phase. We find that growth in the high phase is related to human capital, industry mix, and average firm size. In contrast, we find that growth in the low phase is mostly related to industry mix, specifically, the relative importance of manufacturing. Finally, the frequency of the low phase appears to be related to the level of non-education human capital, but to none of the other variables. Overall, our results strongly reject the notion that city-level characteristics influence employment growth equally across the phases of the business cycle.