A comparison of forecasting performance and systematic risk across different political environments

In this study, we investigate whether: 1) there is a substantial difference in out-of-sample predictability US stock market returns under different political environments (and why the difference may occur); 2) whether an ICAPM risk factor is more prevalent under these environments. Traditional predictors, typically found to perform poorly compared to the historical average of market returns, work quite well under certain political environments. We find evidence that returns are more forecastable and exhibit more autocorrelation when the president is a republican or in his second-term, with the best forecasting performance occurring when the president is a second-term republican. We then examine the results from an ICAPM perspective: if returns are more predictable and exhibit more autocorrelation, then a shock to current market returns will have a larger impact on future investment opportunities, resulting in additional risk. We show that systematic risk is indeed higher under these environments.