Evidence of nominal wage stickness from microdata

Abstract

For much of this century, sticky nominal wages have been considered a key reason that nominal shocks to the economy may have real effects. Historical explanations of sticky nominal wages often rely on money illusion, a concept unpopular with neoclassical economists because it implies irrationality. More modem explanations cite costs (for instance, George Akerlof and Janet Yellen, 1985) and imperfect information about the rate of inflation (for instance, Edmund S. Phelps, 1970). Tests of sticky nominal wages have looked at their indirect effects, particularly regarding the countercyclicality of real wages (for instance, see Gary Solon et al., 1994). There has been little direct empirical analysis. This paper addresses that shortcoming by examining longitudinal microeconomic data on the distribution of annual nominal wage and salary changes of workers who remain on the same job. This paper finds that there are some workers whose wages or salaries exhibit nominal stickiness. Specifically, it finds: (1) A significant fraction of workers remaining on the same job over a year receive the same nominal wage/salary in consecutive years. (2) When a given real wage/salary change requires a small nominal change, it is less likely to occur than when it requires a larger nominal change. Over the period studied, between 1 and 2 percent of workers would have received a small pay change in the absence of but instead received none. (3) There is also evidence of downward nominal wage stickiness, but with important differences between wage earners and salary earners. Wage earners receive nominal wage cuts less frequently than would be expected on the basis of distributions of real wage changes. In the period studied, approximately 9.4 percent of wage earners would have received a nominal wage reduction in the absence of downward wage rigidities, but instead do not.' In contrast, salary earners do not receive pay cuts less frequently than would be expected, particularly in later years. The frequency of zero nominal pay changes combined with the relative infrequency of small pay changes provide micro-level evidence of the presence of menu costs, which can lead firms to postpone small pay rate changes. Menu costs in pay rate adjustments may include the administrative costs of changing payrolls and the costs of performance appraisal and negotiations that generally accompany wage/salary changes. While there is considerable debate over whether costs can have a profound impact on aggregate fluctuations and create nonneutrality of money, this paper does not address the macroeconomic implications of costs in wage/salary adjustments.2 Instead, it asks whether the distribution of annual wage and salary adjustments shows microeconomic evidence of a necessary but not sufficient condition for macroeconomic effects. Menu costs are not enough to explain the sharp drop in wage distributions below nominal zero. The phenomenon strongly suggests that either workers or firms resist nominal pay cuts, as would be predicted by traditional Keynesians. Because of this resistance,