臺大管理論叢 NTU Management Review VOL.29 NO.1

11 NTU Management Review Vol. 29 No. 1 Apr. 2019 3.1 Illiquidity Premium in Portfolio Returns 3.1.1 Construction of the Illiquidity Portfolios and Common Factors We start by sorting all stocks in each market by their illiquidity ( ILLIQ ) and stock return volatility. Similar to the computation of ILLIQ , we calculate the return standard deviation using daily returns in a three-month period from t- 3 to t -1. For each month t , we divide stocks within a market into three groups based on their volatility of daily returns. Then within each volatility portfolio, we further group stocks into five illiquidity portfolios based on ILLIQ in the same three-month period. Thus, we have 3x5 = 15 portfolios at the end of month t -1 in each market. We require each portfolio to have at least five stocks and the portfolios are rebalanced every three months. We skip month t and examine the average stock returns in each portfolio in month t +1 to t +3 in order to control for possible short-term return reversals associated with return volatility (Ang, Hodrick, Xing, and Zhang, 2009). We use three weighting methods to calculate the portfolio returns: return-weighted average using stocks’ prior-month gross return as the weight (Asparouhova, Bessembinder, and Kalcheva, 2010, 2013), value-weighted average using stocks’ market capitalization at the end of the preceding month as the weight, and volume-weighted average using the US dollar trading volume over the portfolio formation period as the weight. Using these different weighting schemes establishes robustness of our findings on illiquidity premium to microstructure related biases in computing returns, relative weights on small and large stocks and a small free float in some stocks in which there are large blockholders. Lastly, we define the liquidity premium as the return difference between the high- ILLIQ portfolio and the low- ILLIQ portfolio which we denote IML – Illiquid-minus- liquid – both of which consist of three portfolios averaged across different levels of volatility. The IML returns may be related to common risk factors. To obtain the risk-adjusted illiquidity premium, we construct global and regional factors of market, size-based and value-based returns. The global market factor is proxied by the return on the Morgan Stanley Capital International (MSCI) global equity index in excess of US 1-month T-Bill rate. For the global size and value factors, we first construct these factors within each market and then take a value-weighted average of market-level factors to construct the global factors. The size and value factors in each market are constructed in the same way as Fama and French (1993). Specifically, at the end of June in each year, stocks are divided into two size groups based on the median market capitalization. Also, stocks are