The reality and myth of business cycles : the nature and representation of short-run economic fluctuations
Business cycles as a distinct type of economic behaviour originated in the severe instabilities experienced by European banking systems in the nineteenth century. A large number of divergent explanations of the phenomenon were proposed by contemporary economists; but by 1900 a consensus had emerged about how they were propagated, if not about causes. Business cycles were thought to be induced by disequilibrium relations among real and monetary variables. This quantity-theoretic view was formulated as self-sustaining sequences of phases of prosperity, recession and depression in 'general conditions' by Wesley Mitchell in 1913. Mitchell, with Arthur Burns, attempted to document these 'comovements' between the wars, but found that actual behaviour was complex and that all episodes were effectively unique. Their results were taken as 'proof of the comovement hypothesis by later economists, and most current research assumes such behaviour. Econometric research proposes an a priori decomposition into 'trend' and 'cycle' on the identifying assumption of separate data generating processes for each component, following the standard interpretation of Burns and Mitchell. Most empirical studies find that such decompositions either are rejected by the data or else fail to capture important empirical properties. Theoretical research assumes comovements to be the effects of random shocks propagated through moving average processes. This model is not in general supported empirically owing to the difficulties in identifying shocks from time-series data. The current literature mostly describes growth-rate rather than levels fluctuations, and models are increasingly being formulated explicitly in terms of growth. Evidence from undecomposed time series in levels suggests that the comovement hypothesis is not supported and further, that timing relations among economic variables are not stable.