Shih-Chuan Tsai

Behavioral Investment Strategy Matters: A Statistical Arbitrage Approach

In this study, we employ a statistical arbitrage approach to demonstrate that momentum strategies work only in longer formation and holding periods, a result more conclusive than standard parametric tests can offer. Disposition and overconfidence effects are important factors contributing to the phenomenon. The overconfidence effect seems to dominate the disposition effect, especially in an up market. Moreover, the overconfidence investment behavior of institutional investors is the main cause for significant momentum returns observed in an up market. In a down market, the institutional investors tend to adopt a contrarian strategy while the individuals are still maintaining momentum behavior within shorter periods.

Does Trading Remove or Cause Friction

This study shows that trading causes friction in the market. However, when the market opens, trading of individuals removes market friction, while that of institutional trading does not. The situation during the rest of the day is just the opposite. The uneven behavior of trading noise across investors and time of day makes it a specific, rather than general, transaction cost, contrary to Stolls (2000) finding. Intraday trading activity suppresses both order width and depth, as proxies for trading intensity, and therefore creates noise or friction in the market. Our findings support the proposed financial transaction tax in the European Union.

Search Costs and Investor Trading Activity: Evidences from Limit Order Book

In this study, we analyze investor trading behavior based not on information-related assumptions but on the search model of Vayanos and Wang (2007). Our study shows that search cost dictates trading polarization across investors, firm size, and time of day. We find that individual investors prefer to trade at market open, while institutional investors trade more heavily near market close. Trading costs indicate that it is less costly for institutional investors to trade large cap stocks at market close than at open. Search cost is related significantly to order-based market liquidity measures depending on time of day, market capitalizations, and investor type.

Liquidity and Yield Curve Estimation

This paper examines the effect of incorporating liquidity into the Nelson-Siegel-Svensson model from the perspective of out-of-sample forecasting ability and trading performance. The liquidity consideration reduces the distortion from concentrated trading activities and significantly increases the forecasting ability of the dynamic curve-fitting model. Taking liquidity into account can improve the risk-adjusted returns of a trading strategy developed from the forecasting error series. The improvement in the forecasting ability and the trading performance is consistent across different sets of parameter values and different bond issues, and it becomes more substantial when the forecasting period expands.

Decomposing Risks in Bond Portfolios: International Evidence

This study recalibrates corporate bond idiosyncratic risks in an international context. By applying a statistically powerful risk decomposition scheme, the authors show that diversification is significantly improved by the addition of a global risk benchmark. They construct a long-run stationary yield spread decomposition scheme that further provides a better diversification effect. In addition to global liquidity and default risk factors, a country-specific default risk component is included, and all of them are free of measurement or availability issues. The idiosyncratic risk component is estimated as a fixed effect along with all the parameter estimates, rather than separately from an exogenous generating process. The linear model is simple, yet it can be easily and promptly applied by practitioners.

Does Trading Remove or Bring Frictions

We explore in this paper how trading noise, when considered as a market friction, reacts to trading activity. Transactions cost is a good explanation for intraday trading behavior in the market according to our data. Particularly, we show that in general trading brings friction to market. However, trading friction at market open is the lowest during the day, as trading causes less friction then relatively. This is due to the behavioral difference among investors. When market opens, individual trading removes, while institutional trading brings, market friction. Situation in the rest of the day is just the opposite, where individual, instead of institutional, trading brings friction. The uneven behavior of trading noise across investors and time of day makes it a specific, rather than general, transactions cost, as opposed to Stoll (2000). Intraday trading activity suppresses both order width and depth, as proxies for trading intensity, therefore creates more noise or friction in the market. Width and depth contribute to trading noise in a polarized way, so that individual trading hurts friction in small cap stocks at open, but benefits it at close. Institutional trading brings extremely strong friction to large cap stocks, but less so at market close. So trading noise as a specific, rather than general, transactions cost is prominent only to certain investors, at certain time and for certain stocks in the market. Our findings lend itself to the justification of the new financial transactions tax proposed by the European Union.

Searching Out of Trading Noise: A Study of Intraday Transactions Cost

We attempt to identify in this paper the role of trading noise as a transactions cost to market participant in the sense of Stoll (2000), especially in the presence of trading concentration. Applying the measures of Hu (2006) and Kang and Yeo (2008), we analyze the noise proportion in intraday stock returns and its interaction with investor herding and search cost. Although this noise is high on individual orders and low on institutional orders, its behavior at market open is entirely different from the rest of the day. Noises for small cap stocks, unlike volatilities, are lower than those for large cap stocks. We also found that noise relates positively to trading volume, but inversely to holdings and turnover ratio of institutional investors. Responses from institutional and individuals are quite the opposite. The noise proportion generated by individual order rises with institutional turnover and search cost encountered, while that of institutional order behaves just oppositely. At market open, behaviors of noise from institutional and individual orders just switch mutually, and then switch back afterwards. Also, noise from high-cap stocks is actually more responsive than that from low-cap ones across investors. So trading noise is a specific transactions cost, prominent to only certain investors, at certain time and for certain stocks in the market, rather than a general market friction as argued in Stoll (2000). This transactions cost is inversely related to search costs encountered in trading, which depends on investor, trading hour of day and market capitalization of stocks.