Interpreting the cyclical behavior of the price level in the U.S.

• Author: Spencer, David E.

1. Introduction

   The development of macroeconomic theory is largely guided by so-called "stylized facts." Whether truly facts or myths, the existence of such widely believed empirical regularities tends to constrain the class of acceptable theories. To be taken seriously, a macroeconomic model must be able to explain salient stylized facts in a reasonably satisfactory way. It is therefore imperative that these "facts" be such.

   A key stylized fact that has informed macroeconomic model builders for the last several decades is that the price level has been procyclical in the U.S. Such claims go back at least to Burns and Mitchell [3] and have prevailed as part of the conventional wisdom among macro-economists. For example, among the business cycle facts listed by Lucas [8, 9] is the claim that "[p]rices generally are procyclical." More recently, Mankiw [9, 88] criticizes real business cycle theory because "it cannot explain the procyclical behavior of prices." Perhaps the strongest evidence that the procyclical behavior of prices is taken as a stylized fact is its frequent repetition in standard contemporary macroeconomics textbooks.

   Much of the effort of macroeconomists over the last two decades has been expended to explain the perceived positive correlation between the price level and aggregate output. This is reflected in the substantial literature devoted to developing theories of positively sloped aggregate supply. This general perception of procyclical prices has been largely responsible for the fact that the aggregate demand/aggregate supply (AD/AS) model is the consensus textbook macroeconomic model. The hallmark of these models is that the price level is slow to adjust to nominal (aggregate demand) shocks and so nominal shocks have temporary real effects. Explanations of a positively sloped short-run aggregate supply curve have varied but generally rely on either incomplete information or the existence of nominal wage and/or price rigidities.

   Recently, however, the conventional wisdom has been seriously challenged. Kydland and Prescott [6] reexamine the correlation between (detrended) real GNP and both the GNP Deflator and the Consumer Price Index for the post-World War II U.S. Using quarterly data, they find strong negative correlation over the period 1954-89 and conclude that the existence of a pro-cyclical price level is a myth. On the basis of their results, they draw explicit conclusions about the appropriateness of alternative classes of macroeconomic models. They state: "We hope that the facts reported here will help guide the selection of model economies to study. We caution that any theory in which procyclical prices figure crucially in accounting for postwar business cycle fluctuations is doomed to failure." [6, 17; italics added.]

   These empirical claims have been moderated somewhat by the results of Wolf [15]. Though he agrees that the price level has not been consistently procyclical in the post-World War II U.S., he finds that the clear countercyclical behavior of the price level is principally a post-1973 phenomenon.

   In a very thorough paper, Cooley and Ohanian [4] investigate the cyclical behavior of prices in the U.S. over the period from 1822 to 1987. They conclude that "procyclical behavior of prices is not a consistent feature of the data. For the post-World War II U.S. economy prices are countercyclical..." [4, 26]. Indeed their results suggest that procyclical behavior of the price level has been the exception rather than the rule in the U.S. during the last century and a half.

   Smith [12] finds broadly similar results for ten countries over extended sample periods, but especially for the post-World War II period. Like Kydland and Prescott, Smith suggests that his findings present a challenge to the empirical usefulness of conventional business cycle models.

   The emerging consensus, it appears, is that the price level has not been consistently pro-cyclical and, indeed, it has been countercyclical in the post-World War II period. Does this new stylized fact imply, as some contend, that conventional models relying on short-term stickiness in wages or output prices are inadequate? Does the apparent cyclical behavior of the price level support business cycle explanations that rely on the predominance of productivity shocks and deny real effects of nominal shocks? Is it the case that aggregate demand shocks play a negligible role in the cyclical behavior of output?

   In this paper, I will argue that the answer to these questions is no. I will discuss theoretical and empirical evidence that the apparent positive correlation between the price level and output can be explained very well by conventional (textbook) macroeconomic models. In fact, such models do not necessarily imply positive correlation between the price level and aggregate output.

   Conventional explanations of fluctuations in the price level and aggregate output rely on some version of the aggregate demand/aggregate supply (AD/AS) model, the dominant model of current major textbooks. It is characterized by a negatively sloped aggregate demand curve, a short-run aggregate supply curve that is positively sloped due to some "friction" in the economy, and a vertical long-run aggregate supply curve.(1)

   In the AD/AS model, fluctuations in the price level and aggregate output are explained as responses to shocks either to aggregate demand or aggregate supply. In textbook language, a shock to aggregate demand causes the aggregate demand curve to shift along a positively sloped short-run aggregate supply curve so that the price level and output initially move in the same direction. Thus, the AD/AS model predicts that the price level will initially be procyclical in the face of aggregate demand shocks. This, however, is not the end of the story. As the frictions are overcome (through, for example, acquisition of information or renegotiation of nominal contracts), output reverses direction and the price level continues moving in the same direction. Accordingly, during the adjustment period, which may be protracted, the price level will be countercyclical.(2) Thus, even in an economy in which fluctuations are driven solely by aggregate demand shocks, the sign of the historical correlation between time series for the price level and output is ambiguous.(3)

   Furthermore, shocks to aggregate supply will cause movements along the aggregate demand curve which implies negative correlation between the price level and output. Thus, the AD/AS model does not make an unambiguous prediction about the cyclical behavior of the price level since the correlation between the price level and output depends not only on the dominant source of shocks but also on the nature of the dynamic responses. What the conventional model does predict is that, in the face of a positive aggregate demand shock, output will first rise and then fall while the price...