Business cycle volatility and economic growth: a reassessment

One of the important implications of real business cycle theory is that there should be a positive relationship between economic fluctuations and economic growth, or, to be more specific, that the growth of real gross domestic product per capita should fall as the business cycle becomes less volatile. This is, of course, a conjecture that is open to empirical verification, at least in principle, but there are a variety of approaches that may be taken which will not necessarily yield consistent sets of findings. Altman (1995), for example, uses primarily the Maddison (1991) data set of some 13 countries from 1870 to 1986 and detrends using log-linear trend lines across benchmark years to obtain the cyclical component of output. He then uses rank and linear correlation techniques to assess the strength of a possible positive relationship between cyclical volatility, measured as the standard deviation of the cyclical component, and output growth across a variety of sub-periods. Altman finds little evidence in favour of this implication of real business cycles models, but it may be argued that his conclusion relies heavily on two questionable features of his empirical approach. Both the method used to construct the cyclical components and the correlation techniques employed to assess the strength of any relationship between cyclical volatility and growth may be criticised as being less than statistical ‘best practice’. We therefore aim to reassess this evidence by extending Altman’s analysis in three directions. The first is to use a more extensive output per capita data set - we employ Maddison’s (1995) updated set of 22 countries, which now ends in 1994. We then employ several statistical techniques that are explicitly designed to extract business cycle components from annual economic time series and, finally, we use robust non-parametric methods to investigate the relationship between cyclical volatility and growth.