Spatial Features of Okun\u27s Law Using U.S. Data

This dissertation addresses issues related to the regional labor market interactions within the United States using the Okun\u27s Law as theoretical framework. In chapter one I estimate the state-level Okun\u27s law after accounting for national changes and spatial spillovers. The estimates show that state-specific growth has a small effect on state unemployment rate changes. State growth experiences that are part of national or regional growths have a substantially bigger effect on state unemployment rates. I compare my results with international data which show a much larger association after accounting for time effects and spatial spillovers; implying that uncoordinated state-level demand management policies may not have substantial effects on unemployment rates in an integrated labor market. In chapter two I use 358 Metropolitan Statistical Areas (MSA) during the period 2002-2009 to examine the relationship between the change in the unemployment rate and output growth. My main finding is that urban unemployment rate is highly dependent on national and regional conditions, implying the existence of a national labor market rather than urban labor markets. State level data shows a similar pattern, although their dependence on national conditions is lower compared with Metropolitan Statistical Areas. Finally chapter three uses pooled data and several linear models, to compare the out-of-sample forecast performance for the unemployment rate for 48 U.S. states; using the root mean square forecast error RMSFE to choose the best model. I also run different tests to select between the best candidates. My research shows that the predictive accuracy of the forecast improves when spatially weighted variables are included. I also found that an AR(1) with a spatial autoregressive lag seems to be the best model in terms of a lower RMSFE; results also moves toward keeping the parsimony of the model