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...