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

The Illiquidity Premium: Further Evidence from Global and Asia-Pacific Markets 10 After employing the above filters, we have 48,960 stocks and 5,328,712 stock months in our final sample. Table 1 reports the descriptive statistics for market level returns and illiquidity. The final sample consists of 45 markets with various sample periods across markets between 1990 and 2015. The average number of stocks per month in a market ranges between 59 (Argentina) and 2,208 (Japan). Consistent with prior empirical work, we find higher volatility of stock returns in emerging markets relative to developed markets. The monthly standard deviation of market returns averages 9.8% in emerging markets and 6.9% in developed markets. More volatile stock markets also have higher average market illiquidity. The time series mean (median) of market level ILLIQ is higher for emerging markets at 0.70 (0.43) compared with 0.12 (0.07) for developed markets. Among the Asia-Pacific markets, many exhibit high return volatility with average market volatility of 9.0%, and are relatively illiquid (average ILLIQ is 0.52). The most illiquid market in the sample is India with average ILLIQ of 3.0. There are two markets in the Asia-Pacific region that are relatively liquid based on ILLIQ measure: China and Taiwan. One possibility is that these markets are dominated by small (retail) traders who are uninformed. By Kyle’s (1985) model, greater volatility of trading of uninformed investors reduces price impact and illiquidity. 3. Illiquidity Premium in International Markets Following the analyses in Amihud et al. (2015) we use two approaches to estimate the illiquidity premium in international financial markets. The first approach involves constructing liquidity-sorted portfolios and estimating the illiquidity premium as the difference in returns on the high and low illiquidity portfolios, within each market. If liquidity is priced we expect illiquid portfolios to earn higher expected returns than liquid portfolios after adjusting for the differences in exposure to risk factors, producing a positive illiquidity premium. The second approach relies on estimating within market cross-sectional Fama-MacBeth regressions of individual stock returns on lagged stock illiquidity, controlling for other firm characteristics that may affect stock returns. A positive regression coefficient associated with illiquidity indicates a premium for illiquidity. We expand the sample period in Amihud et al. (2015) from 2011 to 2015, and find that adding four more years to the data does not qualitatively affect the main results. We find significant evidence of a positive premium for illiquidity in international stock markets.