For all regressions FH use 446 of the 450 four-digit SIC manufacturing industries. Three industries are excluded because of inadequate data on materials, and computers (SIC 3573) is also excluded because it “was an extreme outlier and had to be excluded to obtain any sensible results … When this industry is pooled with the others … it results in unrealistically large estimates for predicted wage changes” (pp. 21-22). The nonproduction-production classification is used to separate more-skilled from less-skilled workers, and capital is included as a third primary factor. As stated above, FH use two different time periods. Also, FH test the robustness of their results by using alternative measures of computerization. All regressions use weighted least squares with average value of shipments during the relevant period as weights. Electronic Payday Loans Online
The key results for the zero-pass-through case are reported in Tables 5 and 6; the analogous results for the case allowing pass-through are in Tables 7 and 8. In the zero-pass-through case, in both time periods across all different computerization measures both narrow and broad measures of outsourcing yield mandated-wage estimates of rising wage inequality. Indeed, for the 1980s either measure yields a mandated rise in inequality greater than the actual rise. However, most of the parameter estimates are sufficiently imprecise that the mandated rise in inequality is not statistically significantly different from zero. In the case with pass through results are less uniform. During the 1970s both measures of outsourcing lead to parameter estimates of a mandated decline in inequality–although again, all these none of these parameter estimates differs significantly from zero. During the 1980s the narrow outsourcing measure mandates a rise in inequality that is statistically significant across all computer measures. In contrast, the broad outsourcing measure mandates a fall in inequality but not significantly so for two of the three computerization measures.
Overall, FH estimate that the narrow measure of outsourcing mandated rising inequality across all specifications. However, the results for narrow outsourcing during the 1970s and for broad outsourcing in both decades are more sensitive to the pass-through assumption.