US Treasury’s Office of Financial Research
Ben Munyan
01 December 2015
Ben Munyan, researcher for the US Treasury’s Office of Financial Research and author of the report, Regulatory Arbitrage in Repo Markets, discusses his findings on the non-US repo market
Image: Shutterstock
Your report focuses on how banks use repo to strategically ‘window-dress’ their balance sheets during certain times of the year. What do you mean by ‘window dressing’?
By window dressing, I mean a pattern of adjusting behaviour to appear safer and more conservative around a regular reporting date, specifically where the dealer affiliate of a non-US bank may choose to unwind repo transactions and reduce leverage around a quarter-end reporting date, then resume normal (higher) repo activity and leverage when the new quarter begins.
What does your data reveal about the difference between US and European banks?
I don’t find evidence of US banks window dressing, but there is evidence that dealers with non-US bank parents are reducing their repo borrowing in a way that’s consistent with window dressing.
By combining multiple datasets, this study was able to rule out some other potential explanations for this behaviour, such as a lack of available cash or an unwillingness to lend to dealer affiliates of banks at the quarter-end.
Why don’t US banks window dress?
Dealer affiliates of US banks don’t have an incentive to window dress at the quarter-end because their parent bank has to report not just the quarter-end but also the quarter-average balance sheet. A sudden change in repo activity would affect the quarter-end report, but not the quarter-average.
How do the LCR and NSFR affect banks and their use of repo?
Because I observed this quarter-end pattern since July 2008, it is unlikely that the liquidity coverage ratio (LCR) and the net stable funding ratio (NSFR) are driving this phenomenon.
Under the Basel framework, the LCR may be affected by a bank’s judicious use of reverse repo transactions, and that may in the future complicate a bank’s repo book management. In contrast, the companies achieve a quarter-end leverage ratio reduction throughout the period studied.
Why is leverage reduction an important driver for non-US banks to window dress?
Dealer affiliates of non-US banks report at quarter-end but not their quarter-average. Therefore, by reducing repo activity at the end of a quarter, a dealer can improve the appearance of the parent bank’s consolidated balance sheet for the reporting date and then resume normal activity once the new quarter begins.
In the report, why do you conclude that the ‘decline in repo is caused by the non-US bank dealers—not their repo lenders’?
I use a variety of tests to rule out repo lenders as the cause of this decline. In one, I use daily figures from iMoneyNet for assets under management in the US money market fund (MMF) industry to show that the magnitude of cash leaving MMFs is an order smaller than the decline in repo, so it doesn’t appear to be driven by outflows from MMFs and therefore an inability to lend. If the decline in repo was caused by an unobserved (non-MMF) repo cash lender withdrawing from the market at quarter-end, we would expect dealers to rely even more heavily on MMFs for cash financing.
In another, I use monthly disclosures of MMF portfolios from the Securities and Exchange Commission’s (SEC) form N-MFP to show that this does not seem to be the case. MMFs’ un-invested cash holdings are actually higher at quarter-end than a typical month-end (until the introduction of reverse repo agreements, which provides an alternative overnight repo investment for MMFs).
It might alternatively be the case that the network of triparty repo lending happens to be such that only the repo lenders that provide cash to dealers with non-US parent banks are the ones that have outflows or are constrained from lending at quarter-end. This would mean that by looking at the MMF industry in aggregate I am missing out on this driving force. To test this, I use a within-investor specification to control for any investor/MMF-specific shocks that I might otherwise miss. My findings show that an individual investor’s repo lending doesn’t change across the board at the end of a quarter (the regression coefficient called ‘Last Day of a Quarter’ is statistically insignificant), it only significantly declines for dealers with a European bank parent (the coefficient for lending to dealers with a Japanese bank parent is also negative, but they are a much smaller fraction of the market than US and European bank dealers and there may be a sample size issue reducing the power of this test on those dealers).
Could it not be the case that MMFs simply don’t want to lend to non-US bank dealers at the end of a quarter?
If this were the case, I would expect non-US bank dealers’ cost of funding to either stay the same or increase. However, my findings show that among MMFs (again using their complete monthly portfolio disclosures), lending declines, but so does the cost of lending/borrowing, given by the repo rate.
I’ll note here that for dealers with Japanese bank parents, there is no significance on the quantity of lending they get from MMFs, but a positive coefficient in the rate they receive, however, that coefficient when combined with the ‘quarter-end’ industry-wide coefficient still means an overall decline in cost of borrowing.
These analyses, when taken together, seem to rule out most explanations for this quarter-end decline from the side of repo lenders. That being said, I’m always interested in exploring another testable hypothesis for this quarter-end phenomenon. My hope with this research is that regulators and market participants can use the identification strategies in this paper to more effectively monitor the safety and smooth functioning of this market whenever they encounter a new phenomenon, and act appropriately.
← Previous interview
State of Illinois securities lending
Robert Crossen
Next interview →
Eurex Clearing
Matthias Graulich
