옵션시장 투자자심리지수의 독립성에 관한 연구

This paper aims to measure the investor sentiment using the trading behaviors of individual investors in the derivatives market and then test whether this sentiment index has an independent information from extant investor sentiment indices in the KOSPI200 options market. By defining the sentiment index as the positive trading volumes (long in calls, short in puts) divided by the negative volumes (short in calls, long in puts) of option trading, we investigate its relationship with the conventional four risk factors including Fama-French (1993) three factors and Carhart (1997) one more factor. More specifically, we test whether the sentiment is independent from the information contained in four risk factors (market, size, value, and momentum factors) by conducting a VAR(Vector Auto-regression) and forecasting regressions for KOSPI returns over the horizon of 1, 2, 3, and 6 months. In addition, we carry out a PCA (Principal Component Analysis) of the option investment sentiment and conventional seven sentiment proxies, which are believed to reflect the investor sentiment in Baker and Wurgler (2007), to verify the informative independence of the newly proposed sentiment index. According to the results, we find that first, a VAR analysis seems that option investor sentiment estimated from KOSPI200 option prices relates to market factor but is not related to size, value and momentum factors in our sample period. However, the forecasting regressions, option investment sentiment can forecast one, two, three and six month future returns of value portfolio as well as the market excess returns. These results can be interpreted as follows. The option investor sentiment index tends to explain individual investors’ irrational trading behavior, which is in particular reflected in value stocks and growth stock portfolios more than size and momentum portfolios. Second, the BW sentiment index used worldwide, which is derived from the PCA analysis using six to seven proxies, forecasts only one month size portfolio returns. In the Korean financial markets, BW sentiment index has less informative power than option investor sentiment index, newly proposed in Lee and Yoon (2017). To compare the informative superiority between option sentiment and BW sentiment index, next, we calculate the adjusted BW sentiment index containing information from option sentiment index. The adjusted sentiment index has the information from 8 sentiment proxy variables, consisting of 7 conventional variables and option sentiment index. In this analysis, the adjusted sentiment index containing option information appears to be superior to the BW sentiment index not including option information. One thing to notice here is, however, the sole option sentiment index is better than the other BW investor sentiment indices containing 7 to 8 sentiment proxy variables in explaining future returns of four future factor returns, regardless of the inclusion of option sentiment index as an input of PCA. In addition, for the robustness, we construct the orthogonalized investor sentiment index, but the results are similar with the results for the non-orthogonalized sentiment index.