For the various combinations of dependent variable and decade Leamer reports three sets of results. One uses just capital and labor as factors, the second uses capital plus labor disaggregated by Leamer’s skills definition, and the third uses capital plus labor disaggregated into nonproduction and production workers (Tables 6 through 8, respectively). The results for wage inequality related to globalization changes in product prices are as follows. For each decade there are four combinations of skill measure (Leamer’s or NP vs. P) and pass-through rate (zero or one) to consider. The 1960s results are somewhat mixed: in two of the cases the estimates warrant rising inequality while the other cases warrant falling inequality. The 1970s cases are more clear: in three of the four cases (all except the NP/P measure combined with zero pass-through) the estimates warrant strongly rising inequality. And for the 1980s the results generally split across skill measures. With Leamer’s measure and both pass-through assumptions the estimates warrant essentially unchanged or falling inequality, while with the NP/P measure and both pass-through assumptions the estimates warrant rising inequality (although given the reported standard errors this rise does not appear to be statistically significant). From the results of these three decades Leamer concludes “that the 1970s was the Stolper-Samuelson decade with product-price changes causing increases in inequality” (p. 31). castle payday loans
The methodology of Leamer closely parallels that of Baldwin and Cain (BC) (1997).4 Their empirical analysis consists of a comprehensive set of descriptive facts (Table 1) followed by a more structural approach. Like Leamer, BC also regress a set of cross-industry zero-profit conditions expressed as changes to estimate mandated factor-price changes that can be compared with actual factor-price changes. Unlike Leamer, however, because of concerns that technology data are poorly measured (p. 15), in their econometric analysis BC do not incorporate tfp measures. Their alternative “is to infer from the general equilibrium trade model the biases in the regression coefficients” (p. 15) that are introduced by omitting technological change. Beyond technology, BC also account for influence of national factor supplies as well. Rather than somehow incorporating these influences in the zero-profit regression analysis, BC evaluate their importance in separate analyses of industry outputs, factor employment ratios, and net exports.