Local Employer Concentration and Wages: Instrumental Variables Estimates Using Mergers and Acquisitions
| Dep. Var.: ΔlogHHI (ind–local–year)t–5→t | ||||||||
|---|---|---|---|---|---|---|---|---|
| SIC3 | ||||||||
| 5% Market Share Threshold | 10% Market Share Threshold | |||||||
| County | CZ | County | CZ | |||||
| M&A: (log N merged) | Plants (1) | Emp. (2) | Plants (3) | Emp. (4) | Plants (5) | Emp. (6) | Plants (7) | Emp. (8) |
| Panel A: First Stage | ||||||||
| M&At–1 | 0.229 | 0.037 | 0.201 | 0.035 | 0.330 | 0.049 | 0.317 | 0.051 |
| 15.33 | 12.20 | 11.45 | 12.04 | 15.69 | 15.35 | 14.21 | 15.20 | |
| M&t–2 | 0.203 | 0.033 | 0.190 | 0.032 | 0.299 | 0.044 | 0.295 | 0.047 |
| 10.19 | 13.13 | 13.26 | 13.63 | 15.61 | 14.91 | 12.69 | 15.97 | |
| M&At–3 | 0.179 | 0.031 | 0.163 | 0.030 | 0.264 | 0.039 | 0.226 | 0.039 |
| 8.95 | 11.89 | 11.85 | 14.48 | 12.29 | 12.51 | 8.00 | 13.25 | |
| M&At–4 | 0.154 | 0.027 | 0.136 | 0.025 | 0.226 | 0.033 | 0.185 | 0.035 |
| 8.11 | 10.10 | 8.70 | 10.72 | 8.62 | 8.40 | 5.57 | 11.17 | |
| M&At–5 | 0.182 | 0.031 | 0.147 | 0.028 | 0.275 | 0.040 | 0.200 | 0.038 |
| 7.37 | 10.88 | 6.94 | 11.03 | 7.58 | 8.68 | 4.50 | 10.90 | |
| Δlog(emp, ind–local–year)t–5→t | -0.167 | -0.167 | -0.214 | -0.214 | -0.167 | -0.167 | -0.214 | -0.214 |
| -28.13 | -28.07 | -23.48 | -23.47 | -28.01 | -28.02 | -23.39 | -23.36 | |
| Δlog(labor productivity)t–5→t | -0.010 | -0.010 | -0.010 | -0.010 | -0.010 | -0.010 | -0.010 | -0.010 |
| -6.02 | -6.03 | -4.85 | -4.82 | -5.97 | -5.96 | -4.78 | -4.81 | |
| Δlog(plants per segment)t–5→t | -0.006 | -0.007 | -0.002 | -0.002 | -0.006 | -0.006 | -0.002 | -0.002 |
| -2.54 | -2.61 | -0.94 | -1.00 | -2.54 | -2.53 | -0.65 | -0.67 | |
| Δlog(plants per firm)t–5→t | 0.009 | 0.009 | 0.009 | 0.009 | 0.009 | 0.009 | 0.009 | 0.009 |
| 5.29 | 5.3 | 5.83 | 5.88 | 5.3 | 5.31 | 5.94 | 5.97 | |
| Industry–year fixed effects | Y | Y | Y | Y | Y | Y | Y | Y |
| F-stat | 53.79 | 55.70 | 40.14 | 51.49 | 65.22 | 66.47 | 50.81 | 69.72 |
| Observations | 468,000 | 468,000 | 468,000 | 468,000 | 468,000 | 468,000 | 468,000 | 468,000 |
| R2 | 0.0939 | 0.0948 | 0.1045 | 0.1061 | 0.0936 | 0.0937 | 0.1027 | 0.1037 |
| Panel B: Second Stage | ||||||||
| ΔlogHHI (ind–local–year)t–5→t, IV | -0.047 | -0.041 | -0.047 | -0.043 | -0.061 | -0.055 | -0.040 | -0.030 |
| -2.29 | -2.36 | -2.52 | -2.97 | -2.62 | -2.67 | -2.26 | -2.02 | |
| Δlog(emp, ind–local–year)t–5→t | -0.012 | -0.011 | -0.014 | -0.013 | -0.014 | -0.013 | -0.012 | -0.010 |
| -3.3 | -3.56 | -3.26 | -3.89 | -3.57 | -3.76 | -3.14 | -3.08 | |
| Δlog(labor productivity)t–5→t | 0.016 | 0.016 | 0.016 | 0.016 | 0.016 | 0.016 | 0.016 | 0.016 |
| 12.31 | 12.42 | 10.91 | 11.05 | 12.15 | 12.27 | 10.83 | 10.98 | |
| Δlog(plants per segment)t–5→t | -0.002 | -0.002 | -0.002 | -0.002 | -0.002 | -0.002 | -0.002 | -0.002 |
| -2.11 | -2.09 | -1.87 | -1.87 | -2.15 | -2.13 | -1.87 | -1.87 | |
| Δlog(plants per firm)t–5→t | 0.004 | 0.004 | 0.004 | 0.004 | 0.004 | 0.004 | 0.004 | 0.004 |
| 6.29 | 6.24 | 6.47 | 6.53 | 6.44 | 6.41 | 6.38 | 6.33 | |
| Industry–year fixed effects | Y | Y | Y | Y | Y | Y | Y | Y |
| Observations | 468,000 | 468,000 | 468,000 | 468,000 | 468,000 | 468,000 | 468,000 | 468,000 |
| R2 | 0.0245 | 0.0256 | 0.0227 | 0.0237 | 0.0211 | 0.0228 | 0.0246 | 0.0266 |
Notes: This table examines the effects of employer concentration in a local labor market on wages using an instrumental variable (IV) approach. The instrument is the lagged (one to five years, M&At–1 to M&At–5) value of the log number of plants or employment involved in mergers of two firms in a given local labor market, defined by three-digit SIC codes and counties or CZs. We define a merger as an ownership change of a given plant between two firms that already had plants in the local labor market one year before the merger using the LBD. To ensure that the instruments for employment concentration are relevant, we require that the combined employment share of both the acquiring and target firms in the local labor market was at least 5 percent (Columns 1–4) or 10 percent (Columns 5–8) one year before a merger. Panel A presents estimates for the first-stage IV regression, in which the dependent variable is ΔlogHHI (ind–local–year)t–5–t, the change in log(HHI) from five years before to the current year. Panel B presents estimates for the second-stage IV regression, in which the dependent variable is Δlog(Avg. wages)t–5–t, the change in average wages from five years before to the current year. All control variables in Panels A and B, such as Δlog(labor productivity)t–5→t, are also changes from five years before to the current year. The t-statistics based on standard errors adjusted for sample clustering at the county or CZ level are reported below the coefficient estimates. The numbers of observations are rounded to the nearest thousands to follow the U.S. Census Bureau’s disclosure rules.