Because the NBER Productivity Data Base covering all four-digit manufacturing industries now extends through 1994, this can be checked. I have used these data to replicate Krueger’s findings for the sample of finished-processor industries and then to see what results obtain when the sample is expanded to include all industries. Because I do not have all of Krueger’s data to replicate his construction of factor cost shares I follow Leamer’s approach for constructing cost shares for nonproduction labor, production labor, capital, energy, and materials. These alternative measures seem reasonable in that they replicate Krueger’s results. The cost-share regressors are for 1989 and the dependent variable is the annualized rate of change in industry price. For all sets of industries I regress product-price changes on the cost shares to estimate the mandated factor-price changes.
|(value of shipments)||(2.091)||(1.793)|
|(value of shipments)||(0.082)||(2.002)|
|(value of shipments)||(0.098)||(1.970)|
Table 1 reports the results. The first three rows replicate Krueger’s finding that for the sample of finished-processor industries the mandated wage increase for nonproduction labor exceeds that for production labor. The second three rows expand the sample to all manufacturing industries. Qualitatively, the parameter estimates between the two labor types have flipped. Now production labor has a larger mandated wage increase than nonproduction labor–this implies a mandated decline, not increase, in wage inequality. The final three rows show that the non-finished-processor industries are driving the different results for overall manufacturing. For this sample of industries the annualized mandated decline in inequality is larger than that of manufacturing overall. Notice that the results are robust to alternative weighting schemes for industries.
On balance, then, it seems that Krueger’s results for the 1990s are particular to his sample of industries. Table 1 suggests that for the full sample of all manufacturing industries the product-price changes during the early 1990s were not mandating increased wage inequality.td