Free PDF Ebooks Files @AcrobatPlanet.Com

One of the most active areas of research on liquidity over the past two decades has been an examination of its effect on asset prices. Beginning with the pioneering work of Amihud and Mendelson (1986), much of the earlier research has focused on whether the level of liquidity is an attribute of individual securities that affects their required rate of returns. Consistent with the notion that investors must be compensated for the higher transaction costs that they bear in less liquid markets, these studies have generally found a positive illiquidity-return relation across stocks using a variety of liquidity measures. In addition to the level of liquidity, other aspects of liquidity are also found to influence expected returns. Chordia, Subrahmanyam, and Anshman (2001) find that the variability of dollar volume and share turnover has a significant negative effect on stock returns and Chan (2002) documents that stocks with greater persistence in illiquidity have higher average returns.

The recent discovery of commonality in liquidity by Chordia, Roll, and Subrahmanyam (2000), Hasbrouck and Seppi (2001), and Huberman and Halka (2001) has raised a new question about the role of liquidity in asset pricing. Namely, their findings have initiated researchers to seek whether market-wide liquidity is an important factor in explaining the cross-section of stock returns. Pastor and Stambaugh (2003) create a return reversal measure, which captures order-flow induced temporary price fluctuations, and find that expected stock returns are cross sectionally related to liquidity risk. Acharya and Pedersen (2003) use a scaled version of Amihud’s (2002) illiquidity ratio, which is a price impact proxy given by the ratio of absolute return to dollar volume, and find that liquidity risk is indeed a priced factor. Wang (2003) confirms these results using institutional equity flow as a measure of aggregate liquidity. Eckbo and Norli (2002) provide a comprehensive analysis on this issue and find that all factors, except for the return reversal measure, significantly affect the cross section of portfolio returns.

The effect of market liquidity is also studied in time series. Acharya and Pedersen (2003) and Amihud (2002) propose that the persistence of liquidity implies its ability to forecast market returns. Intuitively, if liquidity is persistent, higher illiquidity today predicts higher illiquidity next period and results in a higher required rate of return. The liquidity persistence also implies a negative contemporaneous return-illiquidity relationship. This is because, if there is a positive illiquidity shock today, investors will anticipate higher illiquidity in the following period and depress current prices in order to earn higher expected returns. The predictive ability of market liquidity is empirically documented by Amihud (2002), Baker and Stein (2002), and Jones (2002). Amihud (2002) uses monthly and annual illiquidity ratios and finds that expected market illiquidity positively affects ex-ante stock excess return over the period 1964 to 1997. Jones (2002) uses the proportional spread of Dow Jones stocks and the share turnover of NYSE stocks over the last century and finds that both spread and turnover predict annual excess market returns up to three years ahead. Baker and Stein (2002) obtain a similar result using annual aggregate NYSE turnover over the period 1932 to 1998, but interpret the result from a behavioral-model perspective. The negative contemporaneous relation between the market’s return and illiquidity is reported by Amihud (2002), Chordia, Roll, and Subrahmanyam (2001), and Pastor and Stambaugh (2003). Amihud, Mendelson, and Wood (1990) suggest that the stock market crash of 1987 can be interpreted as a realization of higher than expected illiquidity, which led to a change in investors’ perception about future liquidity and contributed to the decline in stock prices.

This paper provides an alternative test of market return predictability using the following four liquidity measures that are employed in recent cross sectional studies: (i) the scaled proportional quoted spread of Eckbo and Norli (2002), (ii) the scaled version of Amihud’s (2002) illiquidity ratio used in Acharya and Pedersen (2003), (iii) the return reversal measure of Pastor and Stambaugh (2003), and (iv) the share turnover. Using these measures, I test the two hypotheses proposed by Acharya and Pedersen (2003) and Amihud (2002): (hypothesis 1) expected excess market return is an increasing function of expected market illiquidity, and (hypothesis 2) unexpected market illiquidity has a negative effect on the contemporaneous excess market return.

There are several reasons why this issue deserves further investigation. First, since the identification of factors that predict market returns has continuously been an interest and challenge to financial economists and practitioners, a study that examines the robustness of past empirical findings using different liquidity measures and forecast horizons is important. Second, as with any empirical research involving liquidity, it is desirable to examine a specific issue using a variety of liquidity measures. This is because, unlike other financial variables such as price and volume, liquidity is unobservable and has multiple facets that cannot be captured in a single measure. Third, an additional study on the illiquidity-return relation is necessary given the fact that illiquidity series constructed from daily data tend to be coarser and less accurate when compared to those created from the high-frequency data (Amihud (2002), Eckbo and Norli (2002), and Hasbrouck (2003)). The resulting series are sensitive to specifications and different measures have substantially different statistical properties. It is crucial to use well specified liquidity series that have confirming properties to those obtained from the microstructure data.

The use of the scaled proportional spread is justified since it exhibits similar time-series dynamics to Hasbrouck’s (2003) effective-spread measure, which he finds to be the best daily proxy. The use of the illiquidity ratio is also supported by Hasbrouck’s (2003) finding that it has the highest correlation with the price impact estimate obtained from the microstructure data. Although the return reversal measure does not yield as high a correlation with the microstructure-based price impact as does the illiquidity ratio, Pastor and Stambaugh (2003) show that it adequately captures many of the known historical properties of market liquidity. The use of the share turnover measure provides a robustness test of Baker and Stein (2002) and Jones (2002), who find return predictability using turnover measures obtained from different data. The liquidity series used in this paper also possess properties that are consistent with those found in past research. I find that the liquidity level is generally higher when the market return is positive, market volatility is low, and economic conditions are sound. Lastly, given that these four liquidity measures are found to be the factor representations of pervasive liquidity risk that significantly affects the cross section of stock returns (Acharya and Pedersen (2003), Eckbo and Norli (2002), and Pastor and Stambaugh (2003)), it is natural to extend the study and examine how these factors affect stock returns on the aggregate level.

I test the hypotheses following the methodology of Amihud (2002). In contrast to past empirical findings, I find only weak evidence of return predictability over the period 1966 to 2002 using different liquidity measures and forecast horizons. An increase in the proportional spread predicts a higher excess market return in the following period based on monthly and quarterly data, but the predictability is confined to the latter half of the sample period. A decrease in quarterly return reversal also predicts a higher market return one quarter ahead, but the predictability is significant only during the first half of the sample. For all the rest of the liquidity series and forecast horizons, I do not find any significant return predictability, suggesting that past evidence on return predictability may be a result of specific sample periods or forecast horizons being examined or the liquidity series being used. Although there is little evidence that the current level of market liquidity is useful in predicting future excess market returns, there is a strong contemporaneous relationship between market return and market liquidity. Specifically, over the entire sample period, I find that there is a significant negative effect of illiquidity shocks on current excess market returns using all illiquidity series and test horizons with a single exception of annual proportional spread.

Given the significant effect that illiquidity shocks have on current market returns, I investigate how the market return responds to different illiquidity shocks simultaneously. I find that various forms of illiquidity shocks significantly affect excess market returns simultaneously, especially over monthly and quarterly horizons. The simultaneous effect of illiquidity shocks remains significant even after taking into account the effect of market volatility on the market return.

I also examine whether the response of market returns to illiquidity shocks differs across economic states using monthly liquidity series. Given that illiquidity shocks result in lower contemporaneous returns because investors anticipate higher illiquidity in the next period and depress current prices to earn higher expected returns, I conjecture that illiquidity has a greater effect on market returns when higher future illiquidity is particularly unwelcome by investors. One such possibility is when investors face greater need to liquidate their financial assets in the future. When such liquidation must occur at a time it is expected to be costly to do so, investors will react to the illiquidity shock by depressing prices by a greater amount. Since the probability of liquidation is expected to be higher when the economic condition turns sour, I assume that investors assess their need for future liquidation based on variables that predict future economic growth and react to this period’s illiquidity shock accordingly. I define economic states using various financial and macroeconomic variables and find some evidence consistent with this conjecture. I find that there is a stronger effect of illiquidity shocks on market returns when the economy is in recessions than when it is in expansions for the first half of the sample period. The effect of illiquidity shocks on market returns is also greater when the short-term interest rate is higher. The result is robust using all four measures of liquidity, suggesting that the short rate has the greatest influence on investors’ perception about their need for future liquidation. There is also some evidence that illiquidity shocks have a stronger effect on market returns when the term spread is lower and the growth in the leading economic indicator falls below the trend in its growth.

The rest of the paper is organized as follows. Section 2 explains the construction of the four liquidity measures and discusses their time series properties. Section 3 tests the hypotheses and reports the regression results. The section also studies asymmetric illiquidity-return relations across different economic states. Section 4 concludes.

Download
PDF Ebook Liquidity and Expected Market Returns: An Alternative Test