Seasonal fluctuations and business cycles.

• Author: Miron, Jeffrey A.

Since the first studies of aggregate economic fluctuations, empirical researchers have struggled over the proper treatment of seasonal fluctuations. Many researchers initially found seasonal fluctuations to be of significant interest,(1) but over time this attitude changed. Beginning during the worldwide contraction of the 1920s and 1930s, researchers came to regard seasonal cycles as a kind of "noise," to be removed from the data before analysis of the underlying business cycle could begin. Arthur F. Burns, Wesley C. Mitchell, and Frederick R. Macaulay were not only pioneers in business cycle research; they also founded the practice of using seasonally adjusted data to study short-term, aggregate fluctuations.(2) Their practice, of abstracting from seasonal fluctuations through use of adjusted or annual data, is still generally accepted as the appropriate treatment of seasonal fluctuations.

My research of the last 15 years challenges the conventional practice and returns to the older tradition of examining seasonal fluctuations explicitly and treating them as economically interesting in their own right. Within this general framework, my work has had two main goals. The first is to convince macroeconomists that seasonal fluctuations are interesting. My research suggests that seasonal cycles are interesting in and of themselves, and that examining seasonal cycles can reveal much about the nature of business cycles.

Second, I explicitly analyze seasonal fluctuations and use them to gain increased understanding of aggregate fluctuations in general. I have completed a number of pieces of research in which the use of seasonal fluctuations provides substantial information about particular economic phenomena, both seasonal and nonseasonal.

My research focuses exclusively on the deterministic, "seasonal dummy" representation of seasonality, rather than on more general models. With Joseph J. Beaulieu, I show that the seasonal dummy model provides a good approximation to the behavior of aggregate economic time series. We argue that, under this condition, seasonal fluctuations hold particular interest for the study of aggregate fluctuations.(3) That is because identifying assumptions about fluctuations in the seasonal dummy - such as the assumption that Christmas shopping represents a shift in demand - are often more plausible than analogous assumptions about other kinds of seasonal fluctuations, or about conventional business cycle fluctuations.(4) These identifying restrictions can be used both to show that seasonal fluctuations themselves provide compelling examples of various macroeconomic phenomena and to analyze the behavior of business cycle fluctuations in general.

As a first step in this task, Robert B. Barsky, Beaulieu, and I demonstrate that seasonal fluctuations in aggregate variables are quantitatively important. We document the seasonal patterns in these variables for the United States and other countries.(5) Examination of these patterns indicates that seasonal fluctuations provide examples of a number of interesting macroeconomic phenomena, including shifts in preferences (Christmas), production bunching (summer vacation periods), and endogenous monetary policy (seasonal interest rate smoothing). The patterns also suggest a number of identifying restrictions about seasonal...