1.Introduction 
1Although it has long been an officially held view in the Soviet Union that a centrally planned economy cannot be subject to cyclical fluctuations, some East European economists have already detected fluctuations in the rates of growth of their economies 10  15 years ago. Since that time the socalled "growth cycles" or "quasicycles" attracted considerable attention and several attempts have been made to find a theoretical explanation for their existence. For various reasons, however, most of the suggested hypotheses have never been sufficiently empirically tested. Some authors were concerned primarily with purely theoretical models (Lange) and never attempted to verify their theoretical propositions statistically. Others tried to support their theories by empirical evidence, but either the data base was too sketchy or the econometric tools and computational facilities they had at hand were too crude to deal with the problem adequately. 
The purpose of this paper is to use the more extensive data base which is now available for Czechoslovakia and standard contemporary econometric methods to test the crucial building blocks of some growth cycle theories. By doing so we intend to show that the new empirical evidence may suggest a somewhat different explanation of growth cycles than was usually accepted. 
A well known problem in the investigation of growth cycles is that they cannot be detected by working with absolute figures, since there are rarely any absolute declines in East European economic time series. Most of the authors, therefore, work with rates of growth. But the rates of growth have two obvious disadvantages. First, they fluctuate too much and may therefore conceal the regularities. Second, rates of growth are always sensitive to the situation in the previous year For example, if a quite normal year was for some reasons preceded by an extremely bad year, then, measured by rates of growth, it would appear as an unusually successful year 
For these reasons we prefer to work with the deviations of the investigated variable from its "normal level". In a growing economy such a. "normal level" cannot be constant, it must contain a trend. This can be done, for example, by taking for a "normal level" of the variable its fitted value from the functional relation which is explaining it. For example, we can investigate the deviations of the actual output from the "normal" or "capacity" output defined as a fitted value of the production function. Similarly, we can compare the actual consumption or investment with their "normal levels" as explained by the consumption or investment functions. Or more simply, we can take the time trend as a proxy for the "normal level" and investigate the deviations of the actual values from the trend values. We must, however, be aware of the fact that the fluctuations of the residual around the trend are sensitive to the type of trend selected. The simplest exponential trend may not work well incase in postwar cases where there were long term changes in rates of growth, as was clearly the Czechoslovakia. 4But then one can add into the regression equation the second or third power of time and check whether the coefficients are statistically significant. 



