NYFed economists: Tri-party was surprisingly stable

Fear for the safety of the tri-party repo market was a key reason for the Bush Administration’s bailout of the financial markets.

Now economists at the Federal Reserve Bank of New York have issued a report on how that market worked from July 2008 to the beginning of 2010. It was surprisingly stable, they report.

Here’s the introduction from “The Tri-Party Repo Market before the 2010 Reforms” by Adam Copeland, Antoine Martin, and Michael Walker, published in November 2010.

This paper aims to shed some light on the US tri-party repo market, an important funding market that played a role in some of the key events associated with the recent financial crisis. The Task Force on Tri-Party Repo Infrastructure (Task Force 2010) notes that “At several points during the financial crisis of 2007-2009, the tri-party repo market took on particular importance in relation to the failures and near-failures of Countrywide Securities, Bear Stearns, and Lehman Brothers. The potential for the tri-party repo market to cease functioning, with impacts to securities firms, money market mutual funds, major banks involved in payment and settlements globally, and even to the liquidity of the U.S. Treasury and Agency securities, has been cited by policy makers as a key concern behind aggressive interventions to contain the financial crisis.”

We provide a descriptive and quantitative account of the tri-party repo market before the reforms  proposed in 2010 by the Task Force. In particular, we describe in some detail the mechanics of this market and some of its vulnerabilities. We also use data collected by the Federal Reserve Bank of New York (FRBNY) to document quantitative features of this market. Our data covers the period from July 2008 to the beginning of 2010. The tri-party repo market is a large funding market in which dealers fund their portfolios of securities through repurchase agreements (repos). The largest cash providers in this market are money market mutual funds and securities lenders that seek a short-term investment for their available cash. Two tri-party clearing banks, JPMorgan Chase and the Bank of New York
Mellon, provide intermediation services to the  dealers and the cash investors. The vulnerabilities of the tri-party repo market are magnified by its size, the fact that the share of less liquid collateral was growing before the crisis, and the fact that the majority of funding was shortterm, usually overnight. The size of the market reached $2.8 trillion and the size of the largest portfolios financed by dealers exceeded $400 billion at the peak of the market. The share of less liquid collateral  approached 30 percent of the collateral funded in the market at the peak and has decreased to less than 20 percent in 2010.

We find that both the level of haircuts and the amount of funding were surprisingly stable in this market during the period for which we have data, from July 2008 to early 2010. The stability of the margins is in contrast to evidence from other repo markets. Of course, this apparent stability did not prevent the tri-party repo market from contributing to the problems experienced by Lehman Brothers, on which we provide some evidence. The available  evidence suggests that runs in the tri-party repo market may occur precipitously, more like  traditional bank runs, rather than manifest  themselves in the form of large increases in margins.
Haircuts in the tri-party repo market barely moved during the crisis. This is in stark contrast with the study of Gorton and Metrick (2009) of an interdealer market for less liquid collateral and the experience of some other repo markets. The stability of haircuts in tri-party repo seems to be due, in part, to the behavior of some large cash investors who look at the counterparty to a repo first and to the collateral second. For these investors, withdrawing funding from a troubled counterparty altogether seems preferable to increasing margins. The stability of haircuts suggest that while the “margin spirals” described in Geanakoplos (2003) or Brunnermeier and Pedersen (2009) seem to characterize the experience of some repo markets, the tri-party repo
market is very different.

The data shows quite stable relationships between cash investors and the dealers they perceive to be creditworthy. While the amount of funds provided by some large cash investors does fluctuate somewhat, dealers can generally count on the same set of counterparties providing a minimum amount of funding. In particular, we find few examples of interruptions in the relationship between a cash investor and a dealer. The total amount of collateral funded in the tri-party repo market decreased between July 2008 and early 2010, but anecdotal evidence suggests this could be due, in large part, to dealers’s desire to reduce their leverage. While
the market may have been stressed, it appears that most dealers were able to maintain stable funding in the tri-party repo market from July 2008 to early 2010. It is a challenge to reconcile the apparent stability of the tri-party repo market with the dramatic events related to the failure of Lehman Brothers, to which the tri-party repo market seems to have contributed, as we document in section 5.1.

Another puzzle is the difference in the behavior of haircuts between the tri-party repo market and other repo markets. While we lack the data to provide a definitive answer, several features of the market appear to be relevant: 1) Cash investors and dealers appear to establish long-term relationships. If  investors attach value to their relationships with dealers, they may be willing to continue to provide tri-party repo lending to a dealer without raising margins, even in stressed circumstances. 2) Some cash investors appear to be reluctant or unprepared to take possession of the collateral and prefer to withdraw funding if they think a dealer is not credit worthy. The level of haircuts and the type of collateral may be unimportant for such investors and they may not manage either carefully. 3) Tri-party repos have very short terms, usually overnight. This allows investors to pull their funding at a moment’s notice if a dealer becomes too weak. Additional data on different repo markets would help ascertain the importance of each of these features. Lack of data also prevents us from considering the differences between secured and unsecured funding. Anecdotal evidence suggests that unsecured funding behaves differently from both bilateral and tri-party repos. In particular, it appears that tri-party repo funding is typically more stable than unsecured funding, despite the unwillingness of some cash investors to take possession of the collateral.

The features described in the last paragraph also help explain why “runs” in the tri-party repo market are likely to happen precipitously, rather than through a gradual process. In that respect, tri-party repo runs resemble traditional bank runs. The fragility of the tri-party repo market is also exacerbated by some institutional features, such as the morning “unwind” of repos, as we describe in section 4. Hence, the apparent stability of the tri-party repo market during the crisis should not obscure the fact that runs, when they occur, can have devastating and systemic consequences, as was the case with the failure of Lehman. This risk  underscores the need for reform in this market.

Another potentially important factor in the apparent stability of the tri-party repo market was the presence of the Primary Dealer Credit Facility (PDCF) during the period for which we have data (see Adrian, Burke, and McAndrews 2009). Many market participants have mentioned that the PDCF played an important role in the stability of the market.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s