A predictive comparison of some simple long memory and short memory models of daily US stock returns, with emphasis on business cycle effects

This chapter builds on previous work by Bhardwaj and Swanson (2004) who address the notion that many fractional I(d) processes may fall into the empty box" category, as discussed in Granger (1999). However, rather than focusing primarily on linear models, as do Bhardwaj and Swanson, we analyze the business cycle effects on the forecasting performance of these ARFIMA, AR, MA, ARMA, GARCH, and STAR models. This is done via examination of ex ante forecasting evidence based on an updated version of the absolute returns series examined by Ding, Granger and Engle (1993); and via the use of Diebold and Mariano (1995) and Clark and McCracken (2001) predictive accuracy tests. Results are presented for a variety of forecast horizons and for recursive and rolling estimation schemes. We find that the business cycle does not seem to have an effect on the relative forecasting performance of ARFIMA models.