Momentum Strategies for Long-Memory Processes

"Several papers analyse the issue of serial correlation in real estate markets and the impact of this stylised fact on asset allocation choices. However, while momentum and contrarian strategies have been thoroughly tested for equities, their application to real estate markets has been mainly restricted to REITs and other investment vehicles (e.g.property companies). The finance literature suggests that serial correlation in asset returns represents only one of the possible explanations of underreaction and overreaction behaviours in equity markets, other main factors being cross-serial covariances and systematic overperformance. We test the impact of longmemory processes on momentum strategies. We choose to explore direct real estate investments because their monthly returns show a high pattern of autocorrelation. In line with previous empirical studies ñ e.g. Lo-MacKinlay [RFS,1990], Lewellen [RFS, 2002], Pan et al [JEF, 2004] ñ we find that the ""smoothing factor"" is not sufficient to explain the presence of significant overperformances. We also prove that momentum strategies still earn a higher performance when returns are unsmoothed. "