跨期報酬風險抵換關係

Abstract

我們認為不平衡關係似乎是為什麼我們經常會發現股票市場風險與報酬之間不具顯著關係的原因。由於這個因素，為了此不平衡回歸，我們提出新一類的估計方法（取名為GCO-AR）能讓回歸兩側為I(0),I(1)或I(d) 但不盡相同。此估計方法十分方便在於它不需要先行估計長期記憶的係數。我們的模擬結果證實了這新估計方法的有效性。接著，我們只使用此新估計方法與市場波動指數，而非條件變異數模型作為條件變異數的估計，重新檢視跨期風險報酬抵換關係。從GCO-AR的估計，我們發現在六大國際股票市場確實存在顯著的風險與報酬抵換關係，而這些相對風險趨避指數在經濟上合理。

We argue that the imbalance issue seems to be the reason why we often find insignif- icant relation between risk and return in the equity market. In light of this problem, we propose a new class of estimation, named GCO-AR, for the imbalanced regression where we allow the orders of both sides of the regression to be I(0), I(1) or I(d) processes (long memory processes) but are not equal to each other. The implementation of our estimator is convenient since it does not necessitate the estimation of fraction integer. All simulations provide the evidence on the usefulness of our methodology. We then employ our new estimation to re-examine the intertemporal risk- return tradeoff with market volatility indices rather than conditional heteroskedasticity models as a proxy to the conditional variance. With GCO-AR estimation, we shed new lights on the relation between conditional risk premium and conditional variance, where we do find significantly positive and reasonable estimates of the coefficient of relative risk aversion across six major equity indices.