This paper demonstrate the importance of liquidity to asset pricing. It shows that liquidity is strongly related to the persistence of the momentum anomaly, which has not been explained by standard asset-pricing models to date. Most of these models take the stand that expected returns vary across assets because of variations in risk (see, e.g., Ferson and Jagannathan (1996)). Typically, the effects of market frictions, such as transaction costs, are ignored.
From a theoretical standpoint, one might argue that transaction costs can be ignored in the pricing of financial assets because investors can choose to trade only in liquid assets with low transaction costs and hold higher transaction-costs assets for longer periods (see, e.g., Constantinides (1986). See also Heaton and Lucas (1996), and Vayanos (1998)). Hence, when transaction costs are amortized over the expected holding period they become rather small and of second order. This argument assumes that transaction costs are constant and that investor are free to choose when to trade. However, these two assumptions may not hold in practice. First, this paper shows empirically that liquidity varies over time, which raises the possibility of a premium associated with liquidity risk. For example, when considering whether to undertake a large investment, an arbitrageur may demand a premium for bearing the risk of incurring large costs when closing out the position in the future. Second, investors may be impatient to execute their trades or they might be subject to liquidity shocks, forcing them to liquidate their positions. This paper finds that transaction costs can impose a first order effect on prices.
This study focuses on the relationship between liquidity and momentum. The momentum anomaly (see, Jegadeesh and Titman (1993)) is recognized as one of the biggest challenges to asset pricing (see, e.g., Fama and French (1996), and Fama (1998)). Momentum strategies exhibit high abnormal returns that cannot be explained by measures of risk to date (see, e.g., Grundy and Martin (2001), and Jegadeesh and Titman (2001)). Hence, behavioral explanations, based on some type of bounded rationality of investors, such as overconfidence or underreaction of investors to information, have been developed to explain momentum continuation (see, e.g., Barberis, Shleifer, and Vishny (1998), Daniel, Hirshleifer, and Subrahmanyam (1998), and Hong and Stein (1999)). However, exploiting momentum-based strategies involves high turnover (see, e.g., Moskowitz and Grinblatt (1999), and Grundy and Martin (2001)). Therefore, one should take transaction costs explicitly into account while evaluating whether such strategies provide post-transaction-cost returns, to adequately compensate for the risk inherent in them (see, e.g., Korajczyk and Sadka (2002)).
Liquidity and transaction costs have received much attention in recent studies. However, a careful look reveals that liquidity can be defined in many different ways (see O’Hara (1995)). Even while limiting the discussion to execution costs of trades in equity markets, one would naturally distinguish between direct costs, such as brokerage fees, and indirect or invisible costs, such as price impacts of trades (see, e.g., Treynor (1994)). In this paper, the derivation of invisible liquidity costs relies on concepts from the market microstructure literature.
Starting with the work of Demsetz (1968) and Garman (1976), the market microstructure literature has evolved both theoretically and empirically, contingent on the availability of intraday data. Intraday data, such as provided by the Institute for the Study of Securities Markets (ISSM), became available only in the late 1980s. Until then, microstructure research has focused mainly on developing models to explain the role of the bid-ask spread as part of the trading activity (see, e.g., Amihud and Mendelson (1986), Copeland and Galai (1983), Kyle (1985), Glosten and Milgrom (1985), Easley and O’Hara (1987), and Admati and Pfleiderer (1988)). Another front focused on inferring spreads from interday data, such as daily and monthly security returns (see, e.g., Roll (1984)). Currently, tick-by-tick data for a period of almost two decades and for a large cross section of firms is available to researchers of financial markets. We therefore no longer need to use daily/monthly data to infer the bid-ask spreads–we can observe them directly, and we can now apply the core theoretical models of microstructure theory to a large panel of data.
Download
PDF Ebook Momentum, Liquidity Risk, and Limits to Arbitrage
