Inflation, Fixed Cost of Price Adjustment, and Measurement of Relative-Price Variability: Theory and Evidence
指出，现有文献中用于衡量相对价格变动的指标（连续观测期间价格变动率的方差）不能正确反映由价格调整固定成本引起的相对价格变动，并提出了更合适的度量方法，为理论提供了实证支持。
In an interesting paper, Julio Rotemberg (1983) extends the works of Eitan Sheshinski and Yoram Weiss (1977) and Michael Mussa (1981) to examine the aggregate consequences of monetary growth in an economy in which sellers incur a fixed cost of price adjustment. The fixed cost prevents a continual adjustment of individual prices, so each seller keeps his price constant in periods of discrete length, increasing his price in discrete jumps at the end of these periods. A central feature of the inflationary process is therefore that relative prices become dispersed around their means, and the theory predicts that even a fully anticipated inflation increases the variability of relative prices.' The empirical evidence on this point does, however, appear to be mixed. On the one hand, Richard Parks (1978) and Stanley Fischer (1981a,b; 1982) conclude that anticipated inflation does increase relative-price variability (which Fischer interprets as providing empirical support for the theory); on the other hand, Mario Blejer (1981, 1983) and Blejer and Leonardo Leiderman (1982) find that anticipated inflation has no effect, or only a very weak effect on relative-price variability (which makes John Taylor, 1981, doubt the quantitative significance of the theory).2 This paper uses Rotemberg's theoretical framework to demonstrate that the proxy for relative-price variability adopted in the empirical literature does not properly capture the relative-price variability that is caused by a fixed cost of price adjustment. The methodology of the above cited studies is therefore not appropriate for evaluating the empirical relevance of the fixed-cost theory, and the divergence of their results is not surprising. Specifically, the proxy which consists of the variance of the rates of price change between successive observations does not always increase with the anticipated inflation. Even with disaggregate data,3 the proxy decreases with the anticipated inflation in approximately half of any interval of inflation rates for which a seller adjusts his price no more than a given number of times between successive observations. Moreover, the relationship between the anticipated inflation and the proxy depends on such extraneous factors as the timing of the observations. A change in the timing may change a positive relationship between the anticipated inflation and the proxy to a negative one, and vice versa. The paper also proposes alternative, more satisfactory measures of relative-price variability and provides empirical support for the theoretical conclusions.