Liquidity Risk And Its Premium In Chinese Stock Markets

Liquidity is time-varying and always unexpected. So liquidity should be treated as a random economic variable. We need to know not only the level of liquidity but also the risk of liquidity. This paper aims at analyzing theoretically and empirically the liquidity risk and its effect on the stock returns in Shanghai and Shenzhen stock markets, and the contents of this paper is comprised of liquidity risk measure, the evolving feature, the factors that influence liquidity risk, and the risk premium. We hope our analysis could make it clear about the distinction and relationship in concept between the level and the risk of liquidity, and could reveal the important position of liquidity risk among the investment risks.First, market liquidity risk in our paper is that the unpredicted deviation from the expected level of liquidity. Then based on the liquidity-adjusted Markowitz's expectation-variance model we show that the variance-covariance (or the equivalent form such as the well-known beta value) of liquidity level is a good measure of liquidity risk. Furthermore, the variance of aggregate liquidity and the covariance between aggregate liquidity and market portfolio return well reflect the aggregate liquidity risk, while the covariance between individual and aggregate liquidity and the covariance between individual liquidity and market portfolio return well reflect the liquidity risk of individual asset. The illiquidity index is the mainly proxy of liquidity level, and turnover is the secondary proxy of liquidity level.When we apply the variance-covariance method to measure the liquidity risk of overall market and individual asset in our domestic stock markets, we find that:(1) Though the level of liquidity measured by turnover in Shanghai stock markets is much higher than others such as NYSE, but liquidity in Shanghai stock markets is instable greatly, and the variance of the turnover in market trading is the biggest one among others. Furthermore the liquidity risk cannot be diversified well against the market price risk. So, the existing problem in liquidity of Chinese stock markets isn't whether the level of liquidity is higher or not, but is the liquidity risk which is too high.(2) Using illiquidity proxy and the GARCH models in which the shape parameters of the error's distribution remain to be estimated, we find that, the asymmetry in the clustering phenomenon of liquidity conditional variance is different obviously from that of stock return: The more the level of liquidity goes down, the higher liquidity risk will be, which means the more instability of liquidity; the more the level of liquidity goes up, the lower liquidity risk will be, which means the more stability of liquidity. We obtain the same results under turnover proxy. Besides, we find that policy, stock return and the time of no trading affect significantly the conditional variance of liquidity.(3) There is also an asymmetric clustering phenomenon in the conditional covariance between the daily liquidity and market return. The negative innovations of return and liquidity level affect much more on the conditional covariance than the positive innovations, and the conditional covariance is bigger when market price goes down. So, it is indeed an investment risk.(4) Based on a simple liquidity-adjusted VaR model, using our domestic stocks'trading data, we find that the impact on VaR by whether considering liquidity or not is great, and the more the investor holds shares, the greater the influence of liquidity on VaR is. So, liquidity risk is an important component of investment risk in the perspective of risk management.(5) By two empirical means, we find that Shanghai and Shenzhen stock markets have a more significant commonality in liquidity than American stock markets which have market-maker institutions and Chinese Hong Kong stock market which is also a pure order-driven market. Besides, in contrast with Chordia, Roll & Subrahmanyam (2000) who study liquidity commonality in NYSE, the sensitivity of individual stock liquidity to market portfolio liquidity is decreasing in individual stock size and liquidity level in Shanghai and Shenzhen stock markets. Maybe, it is a result of less transparency of information and more susceptibility to policy and market speculative behavior for small companies, or a result of"flight to liquidity".(6) Herd behavior is regarded as an important factor affecting systematic liquidity risk. But, from 2001 on, there is no significant herd phenomenon when market went down. On the contrary, there is a trend of polarization of individual liquidity when market down. It can be interpreted by the more rational investment and investors'"flight to liquidity"after 2001.(7) In the structure of liquidity risk of individual stock, the proportion of systematic risk is about 0.429, and that is similar in the proportion to systematic risk of stock price. We also find that, from 2001 on, this proportion went down significantly (lower about 14 percentage). Evidently it is relative to the decrease of herd behavior. The more liquidity the stock is, the more the proportion of liquidity systematic risk would drop after 2001, and this fact reveals the behavior of"flight to liquidity".Liquidity risk should be a priced risk if it is such an important investment risk in our domestic stock markets as we have revealed.Then, we develop a liquidity risk-adjusted asset pricing model. The difference between our model and present literatures is that we introduce simultaneously investors'endogenous liquidity risk and liquidity demand as a state variable so that model is more consistent with the fact that trading cost increases in executive size, especially in order-driven markets, and model randomizes the stock holding periods.We acquire three propositions under some conditions in our model which have been verified by empirical analysis using Shanghai and Shenzhen stock markets'trading data: (i) Expected return of stock is increasing in its illiquidity level and its exposure to liquidity risk, i.e. there is a significant premium for the level and risk of illiquidity. (ii) The greater the illiquidity level or the elasticity coefficient of price impact is, the more the compensation is for unit illiquidity level and for two liquidity beta (i.e.βt2jand( )This means that the stock expected return is a piece-wise linear increasing and convex function of the expected level of illiquidity. (iii) The premium of unit liquidity (includes its level and risk) is increasing in the probability of the investor's demand for liquidity turning into stress state, i.e. unit price is time-varying.The results of the cross-sectional test and the time-varying test on liquidity premium show that market beta of classical CAPM has no power in interpreting stock returns data. Maybe, this means that liquidity risk has a greater influence on Chinese stock returns than pure price risk. So, it is not enough only considering pure price risk when evaluating stock investment risks and risk premiums.Empirical results show that money supply M1 affects unit premium of liquidity level and liquidity risk consistently and significantly. The compensation for unit liquidity (includes its level and risk) is decreasing and the stock price is increasing in money supply M1. This means there is a mechanism of M1 affecting the stock markets, that is, by influencing the required unit premium of liquidity (includes its level and risk) M1 affects the stock price significantly.