HOW RTGS KILLED LIQUIDITY: U.S. TRI-PARTY REPO EDITION / THE FINANCIAL TIMES

How RTGS killed liquidity: US tri-party repo edition

By: Guest writer


Daniela Gabor is a professor of economics and macrofinance at the University of West of England, Bristol. In this post she explains how the pressures placed upon intraday liquidity by real-time gross settlement systems led to the creation of the tri-party repo market, which itself proved a central point of failure during the global financial crisis.


In a recent post, Izabella Kaminska argued that Real Time Gross Settlement (RTGS) changed the way in which central banks managed systemic liquidity.

RTGS and expensive intraday overdrafts at the central bank killed systemic liquidity, she argued, because “these systems do not function smoothly unless commercial banks maintain sufficiently large reserves to cover payment settlement risk”.

There is a critical element to complete the puzzle of RTGS and systemic liquidity: the rapid structural change in the US financial system towards securities trading and financing via repo markets, a trend accelerated in the 1980s which spread quickly to European financial structures (as I documented in my paper on the political economy of repo markets).

Evolutionary changes in finance combined with RTGS created what Chicago Fed economists Marshall and Steigerwald described as time-critical liquidity (emphasis theirs):

...a settlement payment, delivery of securities, or transfer of collateral must be made at a particular location, in a particular currency (or securities issue), and in a precise timeframe measured not in days, but in hours or even minutes.

Caught between RTGS and a Federal Reserve committed to monetarist ideas, market participants shifted time-critical liquidity away from the balance sheet of the Federal Reserve, and into the US tri-party repo market, with profound structural and systemic effects that reverberate even today.

The background

Banks rely on central bank money to manage the complex webs of settlement obligations resulting from their daily interactions with each other on behalf of retail and institutional clients. These obligations can be settled periodically on a net basis, or immediately, in ‘real time’ on a gross basis. Banks prefer netting because it allows them to economise on central bank reserves, an asset with significant opportunity costs (at least before the world of Basel III liquidity requirements).

But netting systems weave complex webs of credit relationships that are susceptible to systemic risks. When German authorities closed the small commercial bank Bankhaus I. D. Herstatt KgaA at the end of the business day in Germany on June 26, 1974, the critical flaws in netting-based settlement systems became quickly apparent.

Chase Manhattan, Herstatt’s correspondent bank in New York, decided to withhold dollar payments it was due to make on behalf of the German bank. Herstatt risk, or settlement risk, came on to the policy agenda, as most central banks – the US Federal Reserve a notable exception - operated netting settlement systems. The answer, central banks agreed collectively, was to force banks to settle transactions in real time and on gross basis.

RTGS, however, brought new problems.

RTGS in the time of monetarism

In the late 1970s RTGS system, US banks that lost deposits (as clients made payments) would typically enter an overdraft position with the central bank, unless they could draw on idle reserves in the day’s opening balance. Deposit inflows reduced intraday overdrafts, and vice versa.

Daily overdrafts could in theory be carried overnight, but the Fed discouraged such practices through high penalties and administrative sanctions. Since the banks’ obligation to hold reserves was an overnight obligation, those with overdrafts would borrow from the interbank market or from the Fed at the end of the day. Throughout the day, the Fed would grant credit to those banks in overdraft without charging them.

By the early 1980s, the Fed became increasingly concerned with the rapid growth in intraday overdrafts. Uncollateralised overdrafts exposed the Fed to the risk that counterparties would default after accumulating a large intraday position.

An arguably more important factor was that the large intraday overdrafts made a mockery of the Fed’s commitment to take back control of its balance sheet.

Under the leadership of Paul Volcker, the Fed had been trying to fight stagflation with monetarism. Nowadays there is broad agreement that the control of bank reserves is virtually impossible in a world of endogenous money, and that Volcker’s flirtations with monetarism were a strategy to create political support for the high interest rates judged necessary to bring inflation down – as the ECB’s Ulrich Bindseil documents in this excellent 2004 paper on what he terms the “Reserves Position Doctrine”. Yet at the time, the Fed embraced money supply targets and viewed banks’ reserve positions as a critical policy lever.

The credibility of the lever was weakened by growing intraday overdrafts.

By the mid 1980s, these overdrafts routinely exceeded $60 billion, driven by a growing transfer of securities via Fedwire. Securities’ sales via Fedwire require delivery (of security) vs payment (of cash). A broker-dealer purchasing securities would see its clearing bank’s account at the Fed debited. Without sufficient reserves, the clearing bank’s daily overdraft would increase.

By 1988, four clearing banks together accumulated 70 per cent of daily overdrafts attributable to movements of securities over Fedwire. These were in turn driven by the growing repo market activities of securities dealers.

Indeed, Fed research at the time noted the rapid growth in dealers’ use of repos. Whereas capital requirements restricted bank dealers’ use of repos, non-bank dealers tripled their repo books to around a $286 billion annual average by 1985, with over half in matched books (a.k.a. repo borrowing and lending against the same security, with equal terms to maturity).


Figure 1 Dealers' use of repos, 1981-1985 (annual averages). © Source: Federal Reserve


Securities dealers and their clearing banks found themselves locked in a battle with the Federal Reserve over time-critical liquidity. For market-making purposes, dealers liked to hold large inventories of securities throughout the day to meet customers’ demand, and to fund these overnight via repos. Repos would be unwound in the morning (dealers repurchase collateral), giving dealers access to securities, and renewed in the evening.

As dealers repurchased collateral in the morning, and instructed their clearing banks to pay cash, the clearing banks saw their intraday overdraft positions with the Fed increase. That intraday overdraft would automatically close in the afternoon when dealers contracted new overnight repos (selling securities).

The Fed moves on intraday overdrafts

By 1985, the Fed decided to act. It first imposed a cap at three times the level of regulatory capital. But the cap failed because it did not cover overdrafts generated by transfers of securities through Fedwire. The Fed exempted those as it believed that the liquidity of the US Treasury market, joined at the hip with the repo market, relied on free intraday overdrafts. Even in monetarist times, concerns about the liquidity of the US Treasury market shaped the central bank’s systemic liquidity decisions.

As securities-related overdrafts doubled between 1986 and 1993, the Fed considered other options. The most obvious one was also the least consistent with monetarist ideas. The Fed could have prohibited overdrafts altogether, instead forcing banks to purchase federal funds via open market operations. But this would have dramatically increased the Fed’s balance sheet, rendering visible the fact that financial systems which are organised around securities markets which structurally require large central bank balance sheets.

Rather, the Fed decided to charge clearing banks a fee on daily overdrafts in 1994, imposing an annual 10 basis points charge on April 14. Within six months, overdrafts contracted by 40 per cent as dealers embraced the Salomon Brothers solution to time-critical liquidity – the tri-party repo market.

You can see the collapse in the overdraft use here:



The Salomon Brothers solution: the tri-party repo market

In the late 1970s, Salomon began to notice that it was regularly paying twice to finance its UST securities portfolio.

This is because as market-maker, Salomon received and posted securities throughout the day.

Often, however, it would receive those securities too late to (re)fund them through new bilateral repos.

For example, Salomon funded USTs through an overnight repo with UBS. When UBS no longer wanted to rollover the repo, Salomon would need a new repo lender. But Salomon needed to first receive the security back from UBS and then send it to the new lender via the Fedwire Securities Service.

If UBS sent securities too close to Fedwire closing time, then Salomon would not be able to deliver them in the new repo and would have to use its clearing bank (then Manufacturers Hanover). Like other securities dealers, Salomon Brothers used MH to buy and sell securities, maintaining book entry securities accounts and demand deposit accounts. If delivery to the new repo lender could not be made in time, Salomon would have to pay the interest rate twice, to MH for funding the securities ‘stranded’ overnight in the clearing account and to the new repo lender.

Salomon’s solution was to circumvent Fedwire altogether. Its clearing bank, MH, would instead co-ordinate the exchange of cash and collateral, in what would become a tri-party repo. If Salomon and the new repo lender both had accounts with MH, then MH could simply settle the new repo by transferring the securities from Salomon’s general account to a segregation account without going via Fedwire. This would eliminate the constraints imposed by the time of the Fedwire closure, and saved Salomon Brothers money.

The Salomon solution offered a convenient way out for dealers seeking to circumvent Fed’s expensive intraday liquidity after 1994. You could say, it was the blockchain private sector-led solution of its day.

Dealers could have chosen to reduce the time between unwinding in the morning and the new repo. But the Salomon tri-party solution allowed securities dealers to have control over securities during the day, which was critical to their market-making activities. Indeed, clearing banks would unwind all tri-party repo trades in the morning, between 8 and 8.30am, including those with maturities longer than overnight, and re-wind repos in the evening.

As Copeland et al noted in 2015:


A complete unwind of all repos, and not merely of those maturing, is an operationally simple process. An alternative would be a process by which dealers could substitute collateral (including cash) into repo deals without unwinding them, in order to extract a needed security, possibly at multiple points throughout the business day. Throughout the day collateral substitution is prevalent in European tri-party repo markets. By contrast, the US clearing banks have offered some automated collateral substitution capabilities to US tri-party repo market participants only since June 2011

The US tri-party repo market thus provides us with one of the most extravagant episodes of monetary history: long-term repo agreements were in fact a series of overnight repos. Without the ability to substitute collateral pledged in long-term repos, securities dealers bended time so they could minimise the costs of their market-making activities, and circumvent the balance sheet of the central bank.

Yet the Salomon solution did not solve the problem of time-critical intraday liquidity. Rather, the problem shifted from the Fed to the tri-party clearing banks. The clearing bank would instead provide funding for the securities inventories of the broker-dealers throughout the day.

Tri-party agents become a sort of shadow central bank

As a result of the set-up, tri-party agents replaced the Fed in providing intraday liquidity to securities dealers. Put differently, dealers financed their securities through unsecured loans from tri-party dealers during the day, and by repos during the night. This increased concentration in the tri-party segment. Economising on intraday Fed liquidity required dealers to trade with each other, and with repo lenders via the same tri-party agent.

By the time of Lehman Brothers’ collapse, there remained two US tri-party agents: JPMorgan Chase and Bank of New York Mellon. Although the Fed regulated them as depository institutions, it did not impose additional regulations for such large intraday lending positions, in gross terms equal to assets generated from all other activities. By mid 2008, the two were lending intraday USD 2.8 trillion in their tri-party repo business.

Post-Lehman, the Federal Reserve introduced a series of measures to reduce reliance on intraday credit. These prompted JPMorgan to exit from its tri-party repo business in 2018, leaving BNY Mellon alone in the market just as tri-party repo volumes were increasing again.


Figure 2 US tri-party repo volumes © Source: Federal Reserve


Given the above, it’s hard to imagine how the story of US repo market tensions in September 2019 isn’t connected to the evolution of the tri-party repo market and its role in channelling time-critical liquidity.

The question yet to be answered, is how?