Individuals as Noise Traders: Evidence from Korean Stock Listings

Abstract

Individual investors are deemed uninformed and are known to cause risks. This dissertation finds empirical evidence that such individuals are the source of noise trader risk that discounts the prices of Korean stock listings from their fundamental values. Using daily observations of 185 pairs of "old-type" preferred and common stocks listed on the Korea Stock Exchange (KSE) and Korea Securities Dealers Automated Quotations (KOSDAQ) from January 2000 until October 2014, I examine the time-varying effect of individual trading weight on the discount of preferred stocks that on average show higher proportions of individual trading than common listings. For a given pair of common and preferred stocks issued on identical assets in place, preferred stock discount has been known in the literature to be due to dual-class differences in dividend yield, voting rights, and liquidity. However, in this dissertation, the excess of the individual trading proportion of preferred stocks compared with that of comparable common listings statistically and economically significantly explains the variation in preferred stock discount after controlling for firm-level characteristics, the aforementioned dual-class factors, size measures, etc. Because this finding is also robust with regard to either negligible or negative preferred stock discounts (premiums), it is specific evidence of noise trader risk from individual traders in preferred stocks. This research sequentially expands the analysis to common stocks in order to investigate whether noise trader risk, when individual trading weight acts as its proxy, is priced in common equity returns in the Korean stock market. Based on a sample of 658 non-financial firms listed on the KSE and surviving during January 2000 to October 2014, the same sample period as before, this research finds that the excess return of sorted portfolios by individual trading weight (dense minus scarce?DMS), in addition to the conventional three factors of market excess return, the size factor (small minus big?SMB), and the book-to-market (B/M) factor (high minus low?HML), meaningfully accounts for common stock returns in both time-series and panel regressions. Interestingly, the size-controlled factor-mimicking portfolios of noise trader risk appear to determine conspicuously the returns of high individual trader density stocks. This means that noise trader risk disappears or is not priced when arbitrageurs are actively present. In order to bolster this finding, three tests are conducted to identify systematic noise. First, the correlation between a value-weighted measure of the change of individual trading weight and the value-weighted market return shows a co-movement pattern. Second, a correlation analysis of the change of individual trading weights of sample stocks yields a significantly positive association. Third, a principal component analysis of the change of individual trading weight confirms that the main principal components occupy sizable proportions of variances. Overall, these results provide Korean evidence of the discount of asset prices due to individuals generating noise trader risk.