Seasonality, leading indicators, and alternative business cycle theories
Rolling regressions and Granger causality tests are used to examine the predictive ability of seasonally adjusted and unadjusted leading economic indicators for the US economy. Many of the unadjusted variables perform better than their adjusted counterparts, but asymmetric behaviour is also evident. Alternative leading indicators such as the bond spread and the Fed Funds rate do not predict cycles as well as real money balances. Unadjusted real business failure liabilities and business formation also appear to perform well.