Out-of-Sample Equity Premium Prediction: Combination Forecasts and Links to the Real Economy

Testing for Smooth Structural Changes in Time Series Models via Nonparametric Regression

“…Our tests strongly reject the stability of univariate and multivariate return prediction models in the postwar and the post-oil-shocks sample periods. Our findings support the argument of Rapach, Strauss, and Zhou (2010) that "model uncertainty and instability seriously impair the forecasting ability of individual predictive regression models." Our nonparametric estimation suggests that smooth structural changes are a possibility (see, e.g., Figures 8 and 11 in the Supplemental Material).…”

“…We consider a standard predictive regression Y t+1 = α + β X t + ε t+1 where Y t+1 = log[(P t+1 + D t+1 )/P t ] −r t P t is the S&P 500 index, D t is the dividend paid on the S&P 500 index, r t is the 3-month Treasury bill rate, and X t is a predetermined predictor. Following Welch and Goyal (2008) and Rapach, Strauss, and Zhou (2010), we consider 14 financial and economic variables:…”

Do Terrorist Attacks Impact Exchange Rate Behavior? New International Evidence

“…This section, therefore, dwells on outof-sample tests, where we compare the terrorist attack-based exchange rate model (M-T), which amounts to setting β 1 = β 3 = 0 in Equation (1), with a constant-only exchange rate model (M-C), which amounts to setting β 1 = β 2 = β 3 = 0 in Equation (1). Following Rapach, Strauss, and Zhou (2010) and Narayan, Narayan, and Sharma (2013), we utilize a recursive window approach; that is, we estimate the predictive regression model for the in-sample period t 0 to t (50% of the sample) and forecast exchange rate returns for Notes: This table reports the out-of-sample (10-minute) forecast performance results for terrorism-based model against the benchmark historical mean model based. Forecasts are based on using 50% of the in-sample period for recursive forecasting for the remainder of the sample of data.…”

Out‐of‐sample volatility prediction: A new mixed‐frequency approach

“…Similarly, we also predict the volatilities using four popular HAR-RV-type models, including HAR-RV, HAR-RV-J, HAR-RV-CJ, and HAR-RV-TCJ, which all utilize only high-frequency data. Second, to address the issue of model uncertainty 1 (see, e.g., Avramov, 2002;Becker & Clements, 2008;Rapach, Strauss, & Zhou, 2010;Stock & Watson, 2004), we use a mean combination approach to separately generate the GARCH-class and HAR-RV-type forecasts. This is because the simple mean forecasts cannot be outperformed by other complicated combination forecasts (see, e.g., Claeskens, Magnus, Vasnev, & Wang, 2016;Rapach et al, 2010;Stock & Watson, 2004).…”

“…Second, to address the issue of model uncertainty 1 (see, e.g., Avramov, 2002;Becker & Clements, 2008;Rapach, Strauss, & Zhou, 2010;Stock & Watson, 2004), we use a mean combination approach to separately generate the GARCH-class and HAR-RV-type forecasts. This is because the simple mean forecasts cannot be outperformed by other complicated combination forecasts (see, e.g., Claeskens, Magnus, Vasnev, & Wang, 2016;Rapach et al, 2010;Stock & Watson, 2004). At the same time, we also generate the mean combination forecasts based on all the eight individual forecasts of the four GARCH-class models and four HAR-RV-type models, so that the mean combination forecasts also utilize mixed-frequency data.…”