Erratum: The Aggregate Implications of Regional Business Cycles

We thank Man Chon Iao—a Ph.D. student at NYU—for bringing to our attention that we had a mistake in our code that generated the results in the published version of our paper. In this erratum, we: (1) discuss the mistake, (2) highlight the changes we made to our code in response to the mistake, and (3) reproduce all the relevant tables and figures of the paper after correcting the mistake. In particular, Section 2 of this erratum discusses the mistake, Section 3 updates the paper's core tables and figures, and Section 4 updates all remaining motivating and robustness tables and figures. Any table or figure we did not reproduce means the table/figure was unchanged compared to the original.

In summary, the magnitudes of the reported estimates change, although the qualitative results remain.

At the heart of the empirical component of our paper is the creation of state level wage measures during the period surrounding the Great Recession. When we initially made our composition adjusted state level wage measures, we summed over the wages for those working in each of our detailed demographic groups within each state for each year using repeated cross-sectional data from the American Community Survey. We then divided the total wages paid in each state-demographic group-year cell by the total number of individuals within each state-demographic group-year cell. This step produced a measure of the average wage for each demographic group in each state in each year. We then aggregated the state level demographic groups in each year—holding the group weights fixed at some initial time period level—to make our measure of demographically adjusted state wages in each year. Our mistake stems from the fact that we should have divided by the total number of “working” individuals within each group instead of the total number of individuals (unconditional on work status) within each group.

The main empirical result in the paper is the estimation of a state level New Keynesian Wage Phillips Curve (Table V, Section 5). The main quantitative results are the implications for aggregate business cycles of incorporating regional data when estimating a DSGE model (Figures 4 and 5, Section 7). We update these results below.

Below, we present the updated results for Figure 1, Figures 3, 3, 4, 5, Appendix A5–A6, and Tables I, II, and IV, V, VI, VII, VIII of the main paper. All other tables and figures are unaffected by our changes.