Who really controls America's US$12 trillion repo market 鈥� and why do regulators not seem to know?
First, to set the stage, I think regulators were keenly aware that the repo market is giant and intertwined with other critical markets even before the latest determination of its size. So while getting a better handle on its size is valuable, I do not think the latest estimate of its size will change regulators鈥� desire to monitor it closely or their toolkit to deal with it.
To answer the question, the short answer is that the critical institutions connecting end users and end providers are banks and broker-dealers that act as credit intermediaries between the former groups.
I should identify these key players in the repo markets and the roles they play.
Broadly, repo market participants can be segmented into end users (typically levered institutions such as hedge funds, real estate investment trusts and the like), end providers (institutions such as money market funds, securities lenders, insurance companies, the Federal Home Loan Bank system, etc.), and credit intermediaries (typically banks and broker-dealers).
End providers are generally highly sensitive to counterparty credit quality. As a result, they are unable or unwilling to face end users directly. That is where the credit intermediaries come in: because they have good credit quality, they can sit between the end users and the end providers, thereby facilitating the smooth functioning of the repo markets. So the banks and broker-dealers that act as credit intermediaries act as principals to end users (in reverse repo transactions) and to end providers (in repo transactions).
The settlement mechanisms for the reverse repo and repo legs are different, and it is this difference that gives rise to the lack of visibility into the size of the repo market.
Reverse repo transactions are typically settled bilaterally among end users and intermediaries, under a DVP model. Repo transactions are typically settled via triparty, using BNY's platform.
Because reverse repo trades are settled bilaterally, there is no central data source to aggregate and track the volumes in this segment. The problem is compounded by the fact that the end user community is much more fragmented than either the end provider or credit intermediary communities. Regulators have introduced rules to require market participants to report their DVP reverse repo activity, but it is still in its infancy.
In contrast, there is good data on the repo leg segment, because the majority of these trades (typically between credit intermediaries and end providers) settle on BNY鈥檚 triparty platform.
One overarching point to make about the size of the repo market: the dominant repo trading models 鈥� historically, bilateral uncleared 鈥榤atched book鈥� trading and more recently sponsored central clearing 鈥� substantially overstate the level of activity between end users and end providers. That is because in these trading models, credit intermediaries act as conduits, executing reverse repos with end users and offsetting repos with end providers, i.e. two trades (鈥榣egs鈥�) to bridge the gap between end users and end providers. These 鈥榯wo leg鈥� conduit transactions are each captured in the assessment of the size of the repo market, effectively doubling gross repo volumes.
The Fed paper does not estimate how much smaller the repo market would be if trades were directly executed between end users and end providers, but the authors do state that 鈥渁 large share of repo activity reflects dealers' role as intermediaries 鈥� channeling funds from cash-rich investors to clients seeking leverage 鈥� rather than simply financing their own positions鈥�. The authors also estimate that while 鈥渇or small dealers only about 43 cents of every dollar repo goes to reverse repo, large dealers do run nearly matched books with about 85 cents of every dollar of repo going to reverse repo鈥�.
Based on this, it seems likely that more than two-thirds of the overall US$12 trillion repo market is double-counted conduit聽activity. So a repo trading model such as Guaranteed Repo that allows end users and end providers to transact directly with one another would result in a smaller repo market, thereby reducing systemic risk.
Are a handful of Wall Street banks extracting billions in fees by sitting between hedge funds and money market funds in what is essentially a toll booth operation?
As I mentioned earlier, credit intermediation by banks and broker-dealers is critical to the smooth functioning of the repo market. These intermediaries get paid to provide this intermediation. And given the concentration in repo markets, the global systemically important banks (G-SIBs) do receive a substantial share of the revenue.
However, I think the fees (really, spreads) charged by the intermediaries are actually very reasonable given the value they provide. The spread for intermediating general collateral transactions is approximately 8-12 basis points (annualised) of the notional amount of the trades. Given the sheer size of the repo market, that translates into billions of dollars. But when one considers the cost of developing and maintaining the sophisticated credit evaluation and monitoring systems the intermediaries have and the capital they hold against potential losses, I think most end providers would actually incur higher costs if they seek to disintermediate the banks and take on these roles themselves.
In addition, the substantial balance sheet and capital that repo trades consume and the significant direct and indirect costs associated with them mean that many trades (in excess of 20-25 per cent of gross repo activity) are unprofitable or marginally profitable for the intermediaries. Intermediaries generally support repo trading because their clients demand it as a condition for executing other more profitable business with them.
Sunthay鈥檚 Guaranteed Repo structure retains banks鈥� vital credit intermediation role, just in a substantially more efficient manner. Guaranteed Repo transactions result in lower risk-weighted assets (RWAs) and costs than any other repo trading model, and guaranteed transactions are excluded from the supplementary leverage ratio (SLR), liquidity coverage ratio (LCR), and net stable funding ratio (NSFR). In addition, most elements of the G-SIB Surcharge, particularly the US method 2 calculation, do not apply.
I should mention that Guaranteed Repo transactions are not subject to the US clearing mandate. The mandate only applies to transactions in which at least one of the counterparties (either the repo seller or the repo buyer) is a 鈥渄irect participant鈥� in a central counterparty (CCP) that offers repo clearing services 鈥� which at the current time applies to full netting members of the Fixed Income Clearing Corporation (FICC). We expect Guaranteed Repo transactions to be between entities that are not full netting members, so they will not be subject to the clearing requirement.
How is nearly half of the US repo market operating in complete darkness, with no regulatory oversight whatsoever?
I do not think there is no regulatory oversight. The banks and broker-dealers that act as credit intermediaries for a significant percentage of the repo market are subject to stringent regulation, but I do think it is fair to say that the regulators do not have a great handle on real-time activity in the repo market. There are a number of reasons for that, but one important one is that repo trading is still less electronic than in other critically important markets. As electronic trading becomes more entrenched, transparency will improve, although given the sheer size of the repo market and its reluctance to change, it is likely that a significant percentage of repo trades will continue to be executed via voice and email for the foreseeable future.
The Guaranteed Repo structure Sunthay has developed embeds electronic trading. That is because it is a type of all-to-all trading, which cannot be supported without electronic trading. But a secondary (systemic) benefit is that there is a means to monitor market activity in near-real-time.
What happens when hedge funds can not roll over the trillions they have borrowed to make leveraged bets?
Hedge fund repo trades are a source of liquidity for US Treasury markets, so they perform a valuable function. But it is also true that they contribute to the size of the repo market, increasing systemic risk.
In order to answer this question, I think it is important to distinguish between idiosyncratic stress that individual firms may face, and market-wide stress caused by unexpected events such as the Covid-related 鈥榙ash for cash鈥�.
Idiosyncratic stress can result in contagion, but in general, risk controls and adequate capitalisation of credit intermediaries can contain the risk. So it is market-wide stress that is the real concern.
With respect to market-wide stress, I think the short answer is that dealing with the scenario you describe will require central bank intervention. But that should not be controversial, one of the primary functions of central banks is to act as the lender of last resort. In my opinion, the goal should be to minimise the likelihood that central bank intervention will be necessary, not to eliminate it.
Central banks have taken steps to reduce the likelihood that intervention will be necessary and also to establish facilities and procedures to support intervention when necessary. Backstop repo facilities have been successful in injecting liquidity at short notice, and regulatory relief such as a temporary relaxation of the supplementary leverage ratio have helped calm markets. But regulators understand that existing solutions may not be adequate going forward, and in any case, reducing the likelihood of stress rising to a level where intervention is necessary is better overall. So additional solutions, as well as changes to market structure and regulation, are being introduced.
One solution is to expand central clearing of repos, which allows intermediaries to net their reverse repo and repo trades, thereby eliminating the constraint of the SLR. But central clearing comes with higher costs, greater legal and operational complexity, and increased and concentrated risk, so it is not suitable for all trades or all market participants.
Another solution is for central banks to expand membership of their backstop repo facilities beyond banks. One example is the Bank of England鈥檚 Contingent Non-Bank Financial Institution Repo Facility (CNRF). This facility would provide funding directly to eligible insurance companies, liability-driven investment (LDI) funds, and pension funds in the event of market-wide stress, such as the gilt market鈥檚 reaction to the Liz Truss government鈥檚 鈥榤ini budget鈥�. The limitation of this model is that central banks are restricted in the types of counterparties they can face, so they will not be much use if stress arises in excluded entities.
Regulators are also reviewing regulations to assess whether relief on that front may help market liquidity. For example, US regulators have proposed conforming the US enhanced SLR for G-SIBs to the Basel consensus standard, rather than maintaining the higher 鈥榞old plated鈥� standard currently in effect, thereby reducing the capital required to support low-return activities such as repo.
And finally, there is Sunthay鈥檚 Guaranteed Repo solution, which is an alternative or complement to central clearing that achieves similar balance sheet efficiency, while also reducing costs, capital requirements and operational complexity for market participants. Guaranteed Repo can also facilitate access to central bank liquidity facilities by having guarantors (who are eligible for direct access to these facilities) intermediate for entities that are ineligible on a standalone basis. So the structure can help rapidly inject liquidity where needed, without exposing central banks to credit risk from counterparties they would be unable or unwilling to face directly, and this can happen without the constraints imposed by the SLR.
Guaranteed Repo also reduces system-wide leverage (which neither central clearing nor bilateral trading models achieve), thereby reducing systemic risk.
Is the Federal Reserve's 2019 repo crisis intervention now permanently backstopping Wall Street's gambling addiction?
I would not characterise it as a gambling addiction. The cash-futures basis trade often gets characterised as such, but it also serves to maintain liquidity in US Treasury markets. Ending or severely curtailing the trade may have adverse effects on Treasury markets.
In my opinion, a bigger risk is that regulatory changes 鈥� enacted as well as proposed 鈥� could actually reduce liquidity rather than enhance it. Central clearing is an example: the costs of clearing may make certain transactions uneconomical or cause some market participants to drop out. That could reduce liquidity, increasing the need for central bank intervention.
I think regulatory support for additional solutions such as Guaranteed Repo that enhance market liquidity and reduce systemic risk will be beneficial.
Why are banks allowed to use accounting tricks that hide 58 per cent of their actual repo exposures from stress tests and capital requirements?
To be clear, accounting netting (whereby offsetting reverse repo and repo trades cancel each other out on the balance sheet) does not reduce risk-based capital requirements. Risk-based capital (reflected in RWAs) is applied on gross repo exposures (i.e. ignoring the accounting netting).
As for stress tests, there are a number of metrics that look at gross exposures as the appropriate stress point. I do not believe that accounting netting will cause these stress points to be obscured.
Accounting netting does reduce SLR-based capital, making it less likely that the SLR will act as a binding constraint during stress periods, thereby (hopefully) enhancing repo market liquidity.
A point to mention here is that an important reason for introducing the repo clearing mandate in the US was to expand netting opportunities. One condition for netting is that only trades between the same counterparties can be netted. In sponsored central clearing, intermediary banks execute reverse repos with end users and offsetting repos with end providers and then novate both transactions to the CCP. Post-novation, the CCP becomes the seller to every buyer and the buyer to every seller, thereby replacing the original counterparties with the CCP and satisfying the netting condition. As a result, intermediaries can net transactions that they previously could not, freeing up balance sheet to support repo markets.
How are foreign banks potentially running massive repo operations on US soil without American regulators being aware of the scale?
What is excluded from the recent revision of the size of the repo market is activity conducted by the US branches of foreign banks. The Fed researchers did account for the repo activity of the US broker-dealer subsidiaries of foreign banks as data on those institutions are available. Historically, foreign banks have conducted the majority of their client-focused repo trading activities out of their broker-dealer subsidiaries. It is likely that regulators are aware of the scale of repo activity of the foreign banks鈥� US branches and are monitoring it, so I do not think they are clueless on this topic.
Hopefully future updates and increased reporting requirements will cast some light on the scale of this activity.
Is the SEC's new clearing rule just pushing risk around rather than reducing it?
There is a case to be made that central clearing actually increases system-wide risk, albeit in a regulated manner. Central clearing introduces an additional legal entity 鈥� the CCP 鈥� into the system. That adds a point of potential failure. Stringent regulatory oversight, uniform margin regimes, multilateral netting, loss mutualisation among CCP members, contingent liquidity facilities, and capitalisation of the CCP against losses can mitigate this risk, but it is an increase in system-wide risk.
It also concentrates risk in the CCP. Which means that failure of the CCP 鈥� admittedly a low-probability event 鈥� can have catastrophic consequences. This is a well understood risk, and regulators have to choose between this risk and the risk of a looser system with less transparency and oversight.
The more important issue is that the dominant repo trading models cause system-wide leverage to be twice the level it would be if end users and end providers transacted directly with one another (this outcome is a corollary to the overstatement of the size of the repo market caused by the dominant models). That is because credit intermediaries act as principals to both groups, executing reverse repos with end users (generating one turn of leverage) and offsetting repos with end providers (generating a second turn of leverage). Novation of these transactions to a CCP merely transfers the leverage to the CCP.
In contrast, Guaranteed Repo reduces system-wide leverage, as end users and end providers transact directly with one another. Guarantors are not counterparties to transactions; rather, they are contingent obligors who are required to perform only upon the default of a guaranteed obligor. This structure is a 鈥榙ual default鈥� model, whereby both the original obligor and the guarantor must default near-simultaneously to precipitate a disorderly liquidation of collateral. This model diffuses risk among a large number of market participants, rather than concentrating it in a single entity.
And again, the corollary impact of Guaranteed Repo trades is that they reduce the overall size of the repo market, reducing systemic risk.
It has become increasingly clear that no single solution is possible for a market as large, complex and systemically important as the repo market. Rather, multiple solutions that achieve regulators鈥� and market participants鈥� goals are the most effective path forward.
Are money market funds 鈥� supposedly the safest investments for Main Street 鈥� fuelling Wall Street's 'most dangerous' trades?
No. Money market funds (MMFs) are end providers in the repo markets, and critical to their functioning (and by extension, the functioning of the US Treasury market). But the MMFs do not take direct credit exposure to end users.
I also do not know if it is fair to describe the cash-futures basis trade as Wall Street鈥檚 most dangerous trades. There is no doubt that the sheer size of the repo market and the level of leverage in these trades makes them systemically important, but I think it is also important to recognise that there could be significantly less liquidity in the US Treasury market if they were to be prohibited.
How much of America's financial stability depends on computer algorithms deciding whether to lend trillions overnight?
In the context of repo markets, I am not aware of participants engaging in algorithmic trading, i.e. rapid intraday selling and buying. This is a funding market, in which counterparty credit risk is the paramount consideration. Repo trades are reviewed and approved individually, and there is no secondary market for these trades once they are locked. So they are not in my opinion conducive to algorithmic trading.
The trading flows we have created for Sunthay鈥檚 Guaranteed Repo solution are similar to existing repo protocols, i.e. individual trades are reviewed and approved by all parties.
What is stopping a coordinated attack on repo markets from bringing down the entire US financial system in hours?
Really strong controls, particularly among the custodians that make this market viable. Trillions worth of cash and collateral are exchanged on a daily basis, so any disruption to custody and/or settlement can have massive ripple effects. The custodians are exceedingly well managed and carefully regulated so the controls exist.
I would like to add that Sunthay鈥檚 Guaranteed Repo solution embeds collateral management services provided by Euroclear under an innovative structure that brings the efficiency of triparty settlement to delivery versus payment (DVP) transactions. This model not only improves the efficiency of DVP settlement, it also extends the functionality of triparty settlement to firms that have not historically had access to it, either because they were not eligible for triparty membership or because it is expensive. So this is another systemic benefit of our structure.
Next interview →
London Reporting House
Ben Corrigan
Image: Shiv Rao 