What is the best composite liquidity proxy for explaining stock returns? Evidence from the Chinese stock market

Abstract

Accurate measurement of multidimensional liquidity is crucial for effective asset pricing and risk management. We construct 126 multidimensional composite liquidity proxies by using different combinations of individual single-dimensional liquidity proxies and different proxy combining methods. We propose an approach to select the optimal composite liquidity proxy, with both characteristic-level horseraces and systematic-factor-level comparisons among the competing composite proxies. Our results suggest that the Asymptotic Principal Component (APC) method is the suitable combining method, and the Amihud-HL-FHT proxy is the optimal multidimensional liquidity proxy for explaining stock returns in the Chinese stock market. These results remain robust when compared with nested composite proxies, adjusting the significance thresholds, extending the sample period, and using alternative comparison measures.