Here, we assess other factors that could influence nonmetro labor markets. First, our analysis is at the county level; therefore, it might not account for commuting of highskilled workers from nearby regions to work in urban centers. i To address this issue, some studies use commuting zones (CZs) as units of analysis; however, heterogeneity within CZs can be problematic-as most nonmetropolitan economic activities happen within counties (Partridge, Rickman, Olfert, & Tan, 2017). It is important to note that while commuting happens, the effect cannot be too large, or the nonmetropolitan counties would have been included in the nearest MSA (i.e., a key determinant of MSAs is commuting thresholds). By using nonmetro counties, this study only considers counties without a city (of at least 50,000 people) and no strong commuting links to metropolitan areas-i.e., precisely, the rural settings to test these hypotheses.Nonetheless, to assess robustness, we consider the role of commuting and other metro-nonmetro technological spillovers by controlling for distance from the populationweighted center of a nonmetro county to the population-weighed center of its nearest MSA (Partridge, Rickman, Ali, & Olfert, 2009). ii As expected, Table A1 shows the estimated results are virtually indistinguishable when the distance variable is added to the base model (i.e., the BB estimator with 2 lags of the unemployment rate). In addition, we also estimate using the natural log of employment-to-population ratio as the dependent variable. Despite imprecise estimation, including the distance to the nearest MSA does not quantitatively alter the benchmark results. Coupled with Table 2, these results suggest that urban