Empirical investigations of wage determination have often produced autocorrelated residuals from time-series wage equations. Runs of overor underprediction have usually been regarded as weaknesses in specification to be corrected or explained away. In 1980, however, George Perry suggested that such runs represent an important, if neglected, characteristic of American wage setting. He argued that of wage change develop in the labor market. These norms, according to Perry, change discretely; there are periods of more or less wage pushiness. Aggregate wage indexes can be influenced, even if norm shifts are not fully reflected everywhere, providing those sectors that are affected have sufficient weight in the indexes. An obvious division in the labor market is between the union and nonunion sectors. There is reason to believe that while there has been a (downward) shift in wage norms recently, the impact has been concentrated in the union sector (see my 1985 article). Indeed, the union sector is probably inherently more prone to norm shifts than the nonunion.

In another paper (forthcoming), I have contrasted wage setting in the 1920's with that of the post-World War II period. During the 1920's and early 1930's, the U.S. Bureau of Labor Statistics published an incomplete sample of reported wage-change decisions at the establishment level. Perhaps the best way to summarize the results is to direct attention to Table 1, which presents the distribution of manufacturing wagechange decisions during 1924 and 1925, years in which consumer price inflation was, respectively, -.2 and +4.0 percent on a December-to-December basis. The table shows a wide array of wagechange decisions ranging from cuts of over 20 percent to increases of similar magnitude. This dispersion of decisions is remarkable by post-World War II standards. Moreover, the postwar evidence suggests that nominal wage cuts are a rarity, even in periods of low inflation. When they do occur, as in some recent union concessions, the cuts result from a painful negotiations process against a background of threatened or actual mass layoffs. By the 1920's, many features of modern corporate enterprise were present. But were of little significance in most sectors, including manufacturing, the result of a sustained open shop campaign by employers after World War I. There was little labor market intervention by government. Workers resented wage cuts-during periods of generalized wage cutting such reductions became important causes of strikes-but employers implemented them anyway. And when employers did not want to take the blame for wage cuts, they used company unions to negotiate reductions (Robert Dunn, 1927, pp. 21-23). In short, in the absence of or other institutional constraints, the implicit contracts offered by employers in the 1920's provided substantially more wage flexibility than existed after World War II. The wagesetting mechanisms of the 1920's did not approach the flexibility of a classical auction market, a fact of some comfort to implicitcontract theorists. However, it is unclear that one needs to go much beyond simple explanations of how wage cuts (or even relative wage slippage) would lead to worker resentment and management caution.