Crashes and Recoveries in Illiquid Markets
Ricardo Lagos, Guillaume Rocheteau, Pierre-Olivier Weill
2007
Social Science Research Network
We study the dynamics of liquidity provision by dealers during an asset market crash, described as a temporary negative shock to investors' aggregate asset demand. We consider a class of dynamic market settings where dealers can trade continuously with each other, while trading between dealers and investors is subject to delays and involves bargaining. We derive conditions on fundamentals, such as preferences, market structure and the characteristics of the market crash (e.g., severity,
more »
... nce) under which dealers provide liquidity to investors following the crash. We also characterize the conditions under which dealers' incentives to provide liquidity are consistent with market efficiency. Weill (2007) studies the timing of liquidity provision by dealers in a dynamic version of DGP. He asks under which conditions dealers will, and ought to, lean against the wind in the immediate aftermath of the crash. Weill (2007) and the literature spurred by DGP, however, keep the framework tractable by imposing a stark restriction on asset holdings, namely, that investors can only hold either 0 or 1 unit of the asset. Lagos and Rocheteau (2007) study a version of DGP where investors can hold unrestricted asset positions and ...nd that, as result of this restriction on asset holdings, existing search-based theories of ...nancial liquidity neglect a critical aspect of investor behavior in illiquid markets, namely the fact that market participants can mitigate trading frictions by adjusting their asset positions so as to reduce their trading needs. This e¤ect of trading frictions on the demand for liquidity has been pointed out in a di¤erent context by Constantinides (1986). In this paper, we go beyond previous studies by allowing both dealers and investors to hold unrestricted asset positions. This turns out to generate new implications for both the demand and the supply of liquidity. Absent extraneous upper bounds on asset holdings, in times of crisis, investors with high utility for the asset may absorb the selling pressure coming from investors with low utility by holding positions that are large relative to what they would hold during normal times. In other words, by removing the typical restrictions on investors' asset holdings, we ...nd that investors may provide liquidity to other investors in times of crisis, much like dealers do. These new e¤ects on the supply and demand of liquidity imply that, in contrast to Weill (2007) , dealers may sometimes not ...nd it in their interest to provide liquidity during a crash. Also, it may sometimes be e¢ cient for them not to lean against the wind. Whether or not dealers will provide liquidity, and whether or not they ought to, depends on fundamentals, including the details of market structure and the characteristics of the crash. Our stylized description of a market crash consists of an aggregate negative preference shock to investors' asset demands, followed by a (possibly stochastic) recovery path. 3 We ...nd that the amount of liquidity provided by dealers following the crash varies nonmonotonically with the magnitude of trading frictions. When frictions are small, investors choose to take more 3 This scenario could represent, for instance, an international shock such as the 1997 Asian crisis or the 1998 Russian sovereign default, domestic turbulence such as that triggered by the September 11 terrorist attack, or even some company-speci...c shock, such as the collapse of Enron. Our "crash" follows the spirit of Grossman and Miller's (1988) crash dynamics. In Grossman and Miller, dealers provide liquidity in order to share risk with outside investors. In our model, dealers have no such utility motive for holding assets; instead, they allow investors to trade faster. In related work, Bernardo and Welch (2004) use the feature of nonsequential access of investors to market makers to describe a market crash as a ...nancial run. 3 extreme positions because they know that they can rebalance their asset holdings very quickly. Speci...cally, investors with higher-than-average utility for assets become more willing to hold larger-than-average positions and absorb more of the selling pressure coming from investors whose demands for the asset are lower than normal. In some cases, the former end up supplying so much liquidity to other investors, that dealers don't ...nd it pro...table to step in. If, on the contrary, trading frictions are large enough, dealers do not accumulate inventories either, but for a di¤erent reason: Trading frictions reduce investors'demand for liquidity. Indeed, in order to reduce their exposure to the trading frictions, investors choose to take less extreme asset positions. In fact, it is possible that they demand so little liquidity that dealers don't ...nd it pro...table to accumulate inventories following a crash. Thus, if one considers a spectrum of asset markets ranging from those with very small frictions, such as the New York Stock Exchange (NYSE), to those with large trading frictions, such as the corporate bond market, one would expect to see dealers accumulate more asset inventories during a crash in markets which are in the intermediate range of the spectrum. We also ...nd that, from the standpoint of investors, an increase in dealers' bargaining strength is equivalent to an increase in trading frictions. Hence, just as with trading frictions, dealers are less likely to accumulate inventories if their bargaining strength is either very small or very large. This ...nding contrasts with the commonly held view that the market power of dealers (e.g., NYSE specialists) is what gives them incentives to provide liquidity. In our model, an increase in the dealers'bargaining strength may reduce the aggregate amount of inventory they accumulate, because investors endogenously take less extreme positions and demand less liquidity. Similarly, a market reform that reduces dealers'market power, as observed in equity markets in the 90's, can raise dealers'incentives to provide liquidity during a market crash. Our model can rationalize why dealers intervene in some crises and withdraw in others. In line with Hendershott and Seasholes's (2006) empirical evidence on the inventory strategies of NYSE specialists, in our model, dealers'incentives to provide liquidity are driven by anticipated capital gains. Therefore, dealers are more likely to accumulate inventories when the crisis is severe and expected to be short-lived: A large price drop and the expectation of a quick rebound make it more pro...table for dealers to buy low early in the crash and sell high later, as demand for the asset recovers. From a normative standpoint, we ...nd that the equilibrium asset allocation across investors and the dealers' inventory policies are socially e¢ cient if and only if dealers' bargaining strength is equal to zero. Given the nonmonotonic equilibrium relationship between 4
doi:10.2139/ssrn.1022033
fatcat:7lstop3evfdzrorcj6nw4qeway