The Impact of Liquidity on the Cross Section of Equity Returns. Van Heerden (2016)

The authors are looking to see how does liquidity effect the factor returns of stocks in the Johannesburg Stock Exchange over the 1994 - 2011 period.  To do so, they setup a univariate regression for factors that are commonly used in prior studies (e.g., value, size, momentum, and volatility); then they compare the factor coefficients of those regressions to the coefficients using a population of large cap stocks only (with market capitalization being a proxy for liquidity).

They find cash-flow-to-price to be the most significant to explain the cross section returns of the population; and its effect actually increases when limiting the sample to large cap companies only.  The size effect is somewhat muted when limiting the companies to large cap.  Finally, the momentum effect is more pronounced with found to be more pronounced with smaller companies, in line with prior studies.  Interesting to note, the market beta in the CAPM formula is statistically insignificant, but is more significant with larger companies.

Next, the authors built a bivariate regression for each of the factors, which includes the dummy variable with one of the coefficients to see what is the net effect to the regression by limiting the population to large caps only.  They find that the only statistically significant change is for momentum and value factors (i.e., moving averages, dividend yield, and cash flow to price); for example, by adding liquidity to the sample, the momentum returns of the sample decrease, and by adding liquidity to the sample, the value-related returns increase. 

Finally, the authors formed a univariate regression for each of the factors by ranking the stocks by the top 30% and bottom 30% of their factors, and forming a zero-cost portfolio as if buying the top-ranked 30% and shorting the bottom-ranked 30%.  Again they find the cash-flow-to-price factor to be the highest coefficient (to the tune of 2% per month excess return) and with the highest t-statistic.  Size is found to be negatively related to returns, in line with other studies.  In addition, the momentum returns are found to be about 1% return per month, similar to other studies.  As was done in the prior regressions in the paper, the authors then limited the population to large cap companies only, and found in all cases the factor-related returns are muted when limiting to large cap only; although, the cash-flow-to-price, 12-month momentum, and book-value-to-market factors still exhibit outperformance at a statistically significant level.

VAN HEERDEN, J. D., & VAN RENSBURG, P. (2016). The Impact of Liquidity on the Cross Section of Equity Returns on the Johannesburg Securities Exchange. Economics, Management & Financial Markets, 11(2), 59–86.