Comparing Time-Varying Price Impact and Information Impounding Time Across Asian Stock Exchanges: An Adaptive Lasso Approach

Abstract

In this paper we construct a parsimonious vector autoregression model with durations, trades and quote revisions to compare price impact and information impounding time in four Asian stock exchanges. We conduct simultaneous model selection and estimation using the adaptive lasso approach. We measure the price impact and the information impounding time via examining the impulse response of a new trade. Weekly estimation from January 2007 to August 2012 allows us to compare the time-varying price impacts. We find that the price impact and the information impounding time vary considerably across time. Traders in Korean and Japanese markets pay more attention to high price impact trades while traders in Hong Kong and Chinese markets pay less attention. Larger mean price impact has larger impact uncertainty measured by the price impact’s standard deviation. Longer mean information impounding time has higher convergence uncertainty measured by the information impounding time’s standard deviation. For many stocks, price impacts have spikes during the 2008 financial crisis. The price impact is negatively related to liquidity and positively related to systematic risk. The average correlation among price impacts of different stocks also varies across time. In Korean and Chinese markets, when systematic risk is high, the average correlation among price impacts is also high.

Keywords: adaptive lasso, attention, price impact, market microstructure

JEL Classification: G01, G15, C52