Ebook Liquidity Risk in the Corporate Bond Markets
Any investor holding a security or a portfolio of securities or considering purchasing a security is exposed to liquidity risk. As in Chacko & Stafford (2004), we use the following reduced-form definition of liquidity: liquidity is simply the gap between the fundamental value of a security and the price at which the security is actually transacted at; high liquidity means this gap is small and vice versa. Thus, liquidity risk is the uncertainty of how wide or narrow this gap will be at any point in time. For all investors and potential investors, liquidity risk is a real risk that they bear. Every transaction is essentially a negative NPV project for the buy-side investor. If the investor knew how negative the NPV would be, then this would not be a risk - the investor could simply perform his asset allocation optimization by factoring in the transaction costs. The risk comes in not knowing how far off the investor will transact from the fundamental value of the asset he is buying or selling. Furthermore, this is a risk that is not fully diversifiable. The natural next question then is whether the systematic portion of liquidity risk is priced, i.e., do investors command a risk premium for bearing liquidity risk? In this paper, we address this question.
We will call the gap between the transactable price of an asset and its fundamental value as the half-spread. The bid-ask spread is simply the sum of the buy half-spread (the gap between the price for which a buy-side investor pays to purchase an asset and the asset’s fundamental value) and the sell half-spread (fundamental value minus the sell price). There are two factors that give rise to half-spreads. The first is that for a transaction to occur a match must be made between a buyer and a seller. It is very likely however that when one wants to sell a particular quantity of a specific asset, there will not be a buyer who wants precisely that same asset in the same quantity at the same point in time. This gives rise to a financial intermediary, namely a market maker. The market maker bridges the asset-type, quantity, and time gap between buyers and sellers by using his balance sheet to store assets. For this inventory service, the market maker requires a fee, which he collects through the half spreads. This notion of a market maker was initially applied by Demsetz (1968) in the context of market specialists and expanded on by a number of other papers [Garman (1976), Stoll (1978), Ho & Stoll (1981), O’Hara & Oldfield (1986), Amihud & Mendelson (1986, 1988), Grossman & Miller (1988), Biais (1993), and Madhavan & Smidt (1993)]. The second factor that gives rise to half-spreads is asymmetric information. In a transaction, one can never be sure whether the counterparty in a transaction is informed. To compensate for the possibility of transacting with an informed trader, market makers charge an additional fee to all traders, which is once again collected through the half-spreads. This notion of information-based transaction costs has been examined in many papers as well - see, for example, Glosten & Milgrom (1985), Easley & O’Hara (1987, 2001), and Easley, Hvidkjaer, & O’Hara (2002).
While the question of whether a bid-ask spread exists has an easy answer, of much more contention is the question of whether the bid-ask spread is time-varying and whether this time variation is systematic, i.e., whether it is priced. Intuitively, one might expect liquidity risk to be priced. Major "liquidity" shocks such as during the LTCM crisis or just after the bursting of the technology bubble, resulted in both low liquidity (very high half-spreads) and poor (stock) market performance. A security that experiences low returns precisely when an investor’s wealth drops must offer a premium to investors to induce them to hold the security. While these are only two datapoints, they are suggestive of liquidity risk being priced. However, the theoretical counterargument to this naive inference has been posed by a number of papers, including Constatinides (1986), Aiyagari & Gertler (1991), Vayanos & Vila (1991), and Vayanos (1998), who argue that liquidity costs can only be a second-order determinant of asset prices because half-spreads are too small relative to the equilibrium risk premium.
Whether liquidity risk is an important source of systematic risk is extremely important to practioners and academics alike. However, there has been relatively little empirical research devoted to this topic and the results appear to be mixed. The main problem is that liquidity is correlated to trading, and the downside risk associated with liqudity is that very little trading exists in a particular security that an investor wishes to buy or sell. With limited trading, any given buy or sell order will have a larger deviation from fundamental price - therefore, a wider effective half-spread. Consequently, it is precisely securities that are thinly traded where we would like to study liquidity and liquidity risk. However, thinly traded securities by definition have very little trade data associated them. Therefore, much of the empirical and theoretical research that has been done on liquidity risk has been done with highly liquid securities; namely, US equities. So it is not surprising that the question of whether liquidity is priced has not been settled. Amihud & Mendelson (1986), Brennan & Subrahmanyam (1996), Brennan, Chordia & Subrahmanyam (1998), Datar, Naik, & Radcliffe (1998), Chordia, Roll, & Subrahmanyam (2000) have all found positive relationships between stock returns and overall liquidity as measured by spreads, depth, and volume. Meanwhile Chordia, Subrahmanyam & Anshuman (2001) find a negative relationship between liquidity and expected returns, while Hasbrouck & Seppi (2001) find no relationship. Finally, Huberman & Halka (2001) and Pastor & Stambaugh (2003) have examined the more relevant question of whether liquidity risk is a systematic factor.
In this paper, we look at the issue of whether liquidity risk is priced using data from the US corporate bond markets. Compared to US equities, corporate bonds are extremely illiquid. While the median stock trades once every few minutes, the median US corporate bond trades approximately once every two months. In this market, liquidity is a problem for most market participants. Therefore, to the extent investors command a premium for liquidity risk it should be more easily discerned in this market.
Unlike equities, corporate bonds are traded in a dealer market. Therefore, the US coporate bond market is essentially an over-the-counter market. Obtaining data on this market is difficult. No single dealer has enough share and therefore sees enough transactions for a meaningful analysis to be conducted. For this reason, our dataset will come from one of the world’s largest custody banks. As part of the custody process, custody banks record the transactions conducted by their clients; thus, the largest custody banks essentially see across the transactions databases of multiple dealers. While not being able to see all of the transactions of the corporate bond market, custodians can see a substantial part of it. Thus, as long as their view of the market is not biased, this should provide a satisfactory database for analyzing the question of liquidity.
A substantial problem still remains. Even if we could look across the whole market a measure of liquidity would not adequately capture the difference in liquidity between most bonds. Because trading volume is so low for most bonds, measured differences in liquidity using traditional methods would only measure small differences between most bonds.
To address the question of a liquidity measure we construct a new liquidity measure that assesses the accessibility of a bond, rather than its trading volume. Because corporate bonds trade in a dealer network, dealers rely on being able to access their buy-side clients’ inventories either to purchase or sell bonds. If a bond is readily accessible, meaning a dealer can call up one
of a number of buy-side clients and obtain the bond easily, the bond can be thought of as liquid even though it may not actually trade very much. Specifically, if a bond issue is held primarily by a set of investors with high portfolio turnover, the bond may be thought of as more accessible essentially, it is easier for a dealer to call up one of the investors holding this bond and convince them to sell it. On the other hand, if a bond issue is held primarily by investors with extremely low portfolio turnover (long term buy-and-hold investors, such as insurance companies) it is more difficult for dealers to call up the typical holder of one of these bonds and convince them to sell it. Thus, our measure of liquidity is a bond’s accessibility. To utilize this concept, we construct a statistic known as latent liquidity, which measures the accessibility of a bond to dealers based on the aggregate trading characteristics of investors holding bonds.
Download
PDF Ebook Liquidity Risk in the Corporate Bond Markets
Posted in :
