굿딜 바운드를 이용한 시장이례현상 연구

Market anomalies have usually been evaluated by parametric models that have pricing error for reference assets. But this has the problem of joint hypotheses test for both model specification error test and market anomaly test. We diagnosed the market anomaly effect for KOSPI market using nonparametric stochastic discount factor (SDF) model that has no pricing error for reference assets. Specifically, we extracted admissible SDFs that satisfy both no arbitrage principle and no good deal opportunity for reference assets. Then, we estimated risk adjusted excess return of market anomaly mimicking portfolio using admissible SDFs. As the result, we derived maximum and minimum value of performance which we call good deal upper bound and lower bound irrespectively. In the last, we diagnosed market anomaly effects as size, book to market (BM) ratio, EPS to price (EP) ratio and dividend to price (DP) ratio effect. Empirical result is summarized as the following. Firstly, we found that market anomaly effects estimated by good deal bounds were not so evident as Sharpe ratio or Jensen’s alpha. This implies that most part of the market anomaly effect is due to model specification error. Secondly, we found that the market anomaly effect was from the positive skewness property of mimicking portfolios. Both Sharpe’s ratio and Jensen’s alpha overpriced portfolios with higher positive skewness. On the other hand, good deal bounds discounted the future cash flow of the same portfolios. Thirdly, we suggested the positive skewness preference of noise traders as the main reason of market anomaly effect. Specifically, we insist that both the high skewness property of market anomaly mimicking portfolios and the positive skewness preference of noise traders caused the market anomaly effect in KOSPI market.