| Summary: | This paper investigates the cross-section of expected commodity futures returns in China using a large panel of 13 individual factors. We find that 6 out of 13 individual factors produce positive and significant returns. To aggregate the information among these factors, we apply not only the traditional Fama-MacBeth regression (FM), but also a set of alternative methods, including the forecast combination method (FC), principal component analysis (PCA), principle component regression (PCR) and partial least squares (PLS). It turns out that PLS outperform other methods in forecasting the cross-section of Chinese expected futures returns. The equally weighted combination of 5 methods produces an even higher annualized return and lower standard deviation compared to each single method. The investigation of factor importance reveals that the skewness (SKEW) factor is more important than other factors in predicting expected futures returns in Chinese markets.
|
|---|
