In this repository, I will replicate the chapter 13 liquidity of the book, "Empirical Asset Pricing: The Cross Section of Stock Returns (Bali, Engle and Murray)". I’m very grateful that Dr Christoph Scheuch share the R programming online.
I will report the method to calculate the Amihud (2002) measure of illiquidity over 1-month, 3-month, 6-month, and 12-month window. Moreover, I also use log-transformed versions of the different illiquidity variables.
I will present summary statistics for the measures of illiquidity using our sample of U.S. stocks during the 1960 through 2019 period. Each month, the mean (Mean), standard deviation (SD), skewness (Ske𝑤), excess kurtosis (Kurt), minimum (Min), fifth percentile (5%), 25th percentile (25%), median (Median), 75th percentile (75%), 95th percentile (95%), and maximum (Max) values of the cross-sectional distribution of each variable is calculated. In addition, I will present correlations between different variables and the persistence.
I will report the results of portfolio analysis. I will construct 10 portfolios by ranking the illiquidity each month then make the long-short portfolio. I employ the one-month ahead excess return as the dependent variable to examine the cross-sectional return predictablility. I will show shows the average excess returns, CAPM alphas, Fama and French (1993) three-factor and Carhart (1997) four-factor (FFC) alphas, and alphas relative to the FFC model augmented with the short-term reversal factor (FFCSTR) for the Illiq 10-1 portfolios, along with Newey and West (1987)-adjusted t-statistics testing the null hypothesis that the average excess return or abnormal return of the given 10-1 portfolio is equal to zero.
In the last part, I will present the results of Fama-Macbeth (Fama and MacBeth 1973, FMhereafter) regression analysis. The FM regressions allow us to control for all of the other effects simultaneously, instead of one at a time, in examining the relation between illiquidity and expected stock returns. I begin the FM regression analyses using cross-sectional regression specifications that include illiquidity measure as the lone independent variable, along with one of beta, Size, Mom, or Rev as independent variables, and then all of the variables together. Winsorization is at the 0.5% level on a monthly basis. The dependent variable in all regressions is the one-month-ahead excess stock return.