As discussed earlier, the SS theorem involves the prices of only those industries with current domestic production. Price changes for industries not domestically produced simply entail consumption-deflated real-wage changes for all domestic factors. Cheaper foreign t-shirts are a good thing for even less-skilled U.S. workers if no U.S. firms produce t-shirts. Whether or not the United States makes t-shirts depends on forces such as the national endowment mix and level of technology. Thus changes over time in product mixes can matter greatly for the economy’s links between product prices and factor prices.
There is a potential problem with the domestic, export, and import price data used by the studies in this paper. All these data are price indexes designed to measure the prices of unchanging baskets of goods. All three series are produced by the Bureau of Labor Statistics (BLS), and the BLS Handbook of Methods (1992) details the various methods for eliminating product changes from the indexes. This is done both by selecting a fixed basket of goods to price and by systematically eliminating from transaction prices any changes due to quality changes (except in instances of “drastic” introductions and/or eliminations of products).
The price-index methodology suggests that the BLS price data tend not to reflect changes in product mixes either by domestic producers or foreign producers. Price increases because of quality upgrading get reduced while price decreases because of quality reductions get increased. Thus, there is an important tension between the theory of cones of diversification and the empirical reality of price-index construction.