Author(s): Massimo Guidolin and Stuart Hyde
We systematically assess the recursive performance costs–both ex-ante and ex-post–in recursive real time out-of-sample experiments of implementing diversification strategies that allow occupational investment vehicles (OIVs, like pension funds) to allocate wealth across available assets (equities) by taking into account the presence of regimes and non-stationarities (i.e., structural change in parameters) in the correlation between sector-specific earnings/wages dynamics and stock returns. We find that ex-post, the cost of creating OIVs is negligible and, to the contrary, often negative over our evaluation period: this means that OIVs that exploit and forecast bull and bear regimes end up producing realized performance that are better than those of strategies that do not. The origins of such gains lie in the fact that conditioning on sectorial dynamics, may lead to a more accurate identification and forecasting of regime shifts. Contrary to standard intuition, both ex-ante and ex-post, we find evidence that often an OIV ought to optimally invest in stocks issued either by firms that belong to the same sector that characterizes the OIV or at least from the same country as the OIV.