Dealer Behavior In Highly Illiquid Risky Assets

HFA Padded
HFA Staff
Published on
Updated on
Illiquid Risky Assets

Dealer Behavior In Highly Illiquid Risky Assets

Michael A. Goldstein

Babson College

320 Tomasso Hall

Babson Park, MA 02457

781-239-4402

Edith S. Hotchkiss

Boston College

Fulton Hall, Room 340

Chestnut Hill, MA 02467

(617) 552-3240

September 15, 2015

Abstract

This study examines dealer behavior in a sample of 14,749 corporate bonds that vary in credit rating and liquidity. Our unique data set allows us to identify purchases and sales by individual dealers, enabling us to determine how long a dealer holds a bond purchase in inventory, how much of that initial purchase is sold to customers, spreads on those sales to customers, and how these vary with the credit rating and liquidity of the bond in the prior 30 days. We find that previous liquidity has little effect on the spreads dealers charge customers; for some rating categories, observed spreads are higher for the most actively traded bonds. Consistent with this finding, dealers’ holding periods do not necessarily decline as liquidity increases; in fact, dealer’s holding periods are lowest for some of the most illiquid bonds. Dealers are also more likely to sell all of an initial purchase of bonds on the same day for the less liquid bonds. These effects become stronger as credit quality decreases. Overall, our results suggest that dealers endogenously adjust their behavior to mitigate inventory risk from trading in illiquid securities.

Dealer Behavior In Highly Illiquid Risky Assets – Introduction

Dealers face a variety of challenges when trying to make markets in relatively illiquid risky assets such as corporate bonds. In particular, dealers may assume inventory risk upon purchasing an asset from a customer if they must wait for a counterparty to arrive to offset the original purchase. Such risks are magnified when dealers face price movements on riskier assets and for illiquid assets where fewer buyers for the asset may arrive.

Standard market microstructure models such as Glosten and Milgrom (1985) generally assume that dealers stand by relatively passively and await the arrival of liquidity traders, who arrive via some external Poisson process. These models were generally created to describe US equity markets, which compared to markets such as that for corporate bonds are relatively liquid. Therefore, these theoretical models may be more appropriate as models of equity market dealers or other dealers facing reasonably large natural demand.

Faced with significant inventory risk, dealers may be less willing to commit capital to trading in more illiquid securities. At the same time, dealers may follow other strategies to mitigate this increased liquidity risk. One likely response is to not stand by passively but rather to search actively for counterparty offers. Duffie, Garleanu, and Pedersen (2005) create a model that suggests that as illiquidity increases, agents increasingly engage in costly search mechanisms in order to find the opposite side of a trade. Market makers and agents endogenously increase their search as liquidity decreases. In contrast, as liquidity increases, more orders come to the market makers and they naturally avoid costly search. Therefore, dealer behavior may vary as liquidity changes.

This paper attempts examines the question of how dealers’ behavior changes for increasingly illiquid assets by focusing on trading in the over-the-counter dealer market for U.S. Corporate bonds. As demonstrated by Goldstein, Hotchkiss, and Sirri (2007) and others, many corporate bonds are very illiquid, with a substantial portion of bonds in the market trading infrequently or not at all. We examine a large sample of 14,749 corporate bonds traded on TRACE that vary considerably in credit rating and liquidity.1 Our unique data set allows us to identify purchases and sales by individual dealers, enabling us to determine how long a dealer holds a bond purchase in inventory, how that initial purchase is sold to customers, the spread on those sales to customers, and how these vary with credit rating and prior measured liquidity of the bond.

We first provide a description of the overall liquidity for U.S. corporate bonds. Well over half of the bonds trade less than once a day on average, with many trading only once a month. In order to be able to construct standard market microstructure measures of liquidity, prior research largely excludes more illiquid bonds which are an important part of the market and of our study. We divide our sample into different subgroups based on trading activity observed in a prior 30 day window, often zero for many bonds, and limit ourselves to basic liquidity measures that do not require frequent trading trades to calculate – specifically, trade count and trading volume (our revision finds similar results using the number of days traded in the prior window). Other measures, such as the measure Amihud (2002) measure of price impact or measured based on Roll (1984) implicitly assume and require multiple trades per day for calculation. Using these measures would preclude the analysis from examining a large number of more illiquid bonds. It would also preclude us from understanding the behavior of dealers in periods where market-wide liquidity is abnormally low.

We focus on initial institutional sized trades by examining individual dealer trades involving an initial purchase of 100 bonds or more from a single customer. Since our data set allows us to examine the trading of individual (anonymous) dealers, we then identify subsequent sales of the same bond from this dealer to one or more other customers. We identify such dealer purchases followed by customer sales as a “dealer-round trip.” Our empirical tests examine how the spreads, holding period, and other characteristics of such round-trips vary with bond credit risk and trading activity in the 30 day window prior to the purchase of bonds.

Illiquid Risky Assets

Illiquid Risky Assets

Illiquid Risky Assets

See full PDF below.

HFA Padded

The post above is drafted by the collaboration of the Hedge Fund Alpha Team.

Leave a Comment