After flooding the market with liquidity for the better part of a decade, the US Federal Reserve has begun tightening the spigot. Over the next five years, its asset portfolio will shrink by as much as $1.9 trillion as a result of ‘balance sheet normalisation’, which started in October 2017.
This reduction in the Fed’s balance sheet assets must be matched on the liability side by a decline in reserve balances, which banks maintain for liquidity purposes. To put it another way, the Fed’s normalisation strategy is contingent on banks’ ability to operate with much lower central bank reserves, which currently stand at $2.2 trillion, compared with an average of $8.6 billion in 2007.
The level of the US banking system’s demand for reserve balances is something of a mystery. If reserves drop to $100 billion – the Fed’s baseline expectation as recently as April 2017 – assets in the system open market account (Soma) portfolio shrink from $4.2 trillion currently to $2.3 trillion in the third quarter of 2023. A more recent study conducted by the New York Fed in June 2017 pegged the long-term demand for reserves at $613 billion, which implies the Soma portfolio will ‘normalise’ at $2.8 trillion as early as 2022.
Bankers say those projections are wildly optimistic, and fail to account for the impact of regulations such as the liquidity coverage ratio (LCR), which have inflated the demand for excess reserves in the banking system.
A liquidity specialist at a European bank says the long-term demand for excess reserves is likely to be “in the $1 trillion to $1.5 trillion range, which means the Fed’s portfolio is not going to shrink as much as they’re projecting”.
That assessment was shared by four current and former bank liquidity risk managers that spoke to Risk.net for this article. “The long-term demand for reserves is lower than the current level, but not by much – probably in the $1.5 trillion range,” says a treasurer at another European bank.
At those levels, the Fed’s balance sheet is likely to normalise somewhere in the $3 trillion to $4 trillion range – about $1 trillion short of consensus expectations.
Attempting to shrink the balance sheet beyond the level dictated by the banking system’s demand for reserves could prove to be hugely disruptive, and bring the normalisation effort to a premature end.
The Fed had almost perfect control of supply and demand in the Fed funds market. It was a small market, but by controlling that one small market effectively, they were able to have a strong influence on all US money markets, with very little volatilityBill Nelson, The Clearing House Association
“If banks are straining to meet their liquidity requirements or daily payment obligations due to a shortage of reserves, the Fed will have to stop reducing the Soma portfolio,” says Darrell Duffie, professor of finance at Stanford University’s Graduate School of Business. “If balance sheet normalisation causes disruptions in the daily reserves market, the Fed will moderate its policy.”
So, how did we end up here?
The Fed’s predicament has its roots in the regulatory and monetary policy response to the 2008 financial crisis. Historically, the Fed controlled interest rates by adjusting the supply of reserves – with lower reserves translating to higher rates, and vice versa.
The amount of central bank deposits needed to meet the Fed's minimum reserve requirements was relatively small and stable in the decade leading up to the crisis – fluctuating between $3 billion and $12 billion from 1997 through to the end of 2007. Banks maintained an additional $1 billion to $2 billion of excess reserves for cautionary purposes, to avoid running up an inadvertent overdraft at the Fed.
That system worked well for many years – from 2003 to mid-2007, the effective Federal funds rate held steady at around 100 basis points below the discount rate, at which banks can borrow from the Fed, and benchmark Libor and repo rates were stable.
“The Fed had almost perfect control of supply and demand in the Fed funds market. It was a small market, but by controlling that one small market effectively, they were able to have a strong influence of all US money markets, with very little volatility,” says Bill Nelson, chief economist at The Clearing House Association, and a former deputy director in the division of monetary affairs at the Federal Reserve Board.
That changed in the crisis, when spreads between Fed funds, Libor and repo widened dramatically.
The Fed responded by pumping huge amounts of liquidity into the financial system, which resulted in a large increase in excess reserves and put downward pressure on the Fed funds rate. In an effort to regain control of short-term rates, the Fed began paying interest on excess reserves (IOER), set at the upper band of the target range, which was 0.25% at the time.
IOER was expected to set a floor on Fed funds during quantitative easing, as banks would have no incentive to lend at rates lower then they earned at the Fed.
However, government-sponsored entities (GSEs) such as the Federal Home Loan Banks, which have accounts at the Fed but are not eligible to receive IOER, were able to fund themselves at even lower rates, and placed more downward pressure on short-term rates by providing liquidity to banks at less than IOER.
To counteract this, the Fed established the reverse repo program (RRP) in 2013 as an outlet for excess liquidity. The RRP rate is set at the lower bound of the Fed’s target range and floors the cost of borrowing overnight cash from eligible counterparties, such as GSEs and money market funds (MMFs).
Quantitative easing coupled with IOER and the RRP have proven to be extremely effective tools for conducting monetary policy. It has also left the Fed with a huge balance sheet hangover – a lightning rod for the central bank’s right-wing and libertarian critics, and making it impossible to claim the financial system has yet put the crisis behind it.
As of February 21, the Fed’s total balance sheet assets stood at $4.4 trillion, compared with around $900 billion a decade ago. The Soma portfolio of US Treasuries and agency mortgage bonds account for $4.2 trillion of this, with the remainder made up of liquidity facilities – mainly swap lines with foreign central banks – and residual assets in the Maiden Lane portfolio. The liability side includes $2.2 trillion in reserve balances along with $1.6 trillion of currency in circulation and $215 billion in the US Treasury general account, with reverse repos and other deposits accounting for the remainder.
The Fed’s response to the financial crisis wasn’t limited to monetary policy, of course. With its supervisory hat on, the central bank joined international efforts to strengthen banking regulations and implement the Basel III rules, such as the LCR, net stable funding ratio (NSFR) and supplementary leverage ratio (SLR).
The new rules have vastly increased the banking system’s demand for excess reserves.
The LCR requires banks to hold a sufficient amount of high-quality liquid assets (HQLA) to cover outflows over a 30-day period of acute stress. Essentially, all liabilities – whether deposits, wholesale funding, or derivatives cash outflows – must be supported by reserves or other liquid assets that can be exchanged for cash at short notice, subject to runoff assumptions in stress scenarios.
The eight global systemically important banks (G-Sibs) in the US held a combined $2.4 trillion of HQLA for LCR purposes at the end of the third quarter of 2017. Excess reserves accounted for around $1.1 trillion – or 48% – of total HQLAs at the biggest US banks.
Under the LCR rules, banks must hold at least 60% of their HQLAs in so-called Level 1 assets, which include excess reserves and US Treasuries. Agency mortgage bonds are classed as Level 2A assets and subject to a 15% haircut, while level 2B assets such as liquid corporate bonds come with a 50% haircut.
But excess reserves are by far the most attractive form of HQLA for banks, especially in a low-rate environment.
"You pay up for central bank reserves because they have an ultra-liquidity premium," says Robert Fiedler, who runs Liquidity Risk Corp, a consulting firm that advises several large European banks. “There is nothing more liquid, and it has the lowest possible credit risk.”
The alternatives simply don’t stack up. One-month T-Bills are almost as liquid, but they yield 10–20bp below IOER on average. Ten-year Treasuries offer more income, but even at 3%, the yield barely compensates for the volatility – with rates set to continue rising over the course of the next couple of years, the government bond component of liquidity portfolios will be written down in value every quarter, requiring banks to continually top them up.
US Treasuries also attract an additional cost under the NSFR, the second of Basel III’s twin liquidity ratios. The ratio is calculated by dividing a bank’s available stable funding by its required stable funding (RSF), with a minimum compliance level of 100%. The RSF charge is set at 5% for government securities, compared with 0% for eligible cash. As a result, replacing excess reserves with Treasuries for HQLA purposes will increase a bank’s NSFR costs – a perverse outcome banks say needs to be addressed before the Fed pushes ahead with normalisation.
Returning to a small balance sheet and steering interest rates by controlling the scarcity of reserves is not consistent with the LCR requirementDarrell Duffie, Stanford University
As a result, banks are likely to hold anywhere from 25–50% of their HQLA in the form of excess reserves, says the liquidity specialist at the first European bank. Assuming stressed outflows remain at their current levels, that suggests US G-Sibs alone will demand at least $600 billion of excess reserves on an ongoing basis for LCR compliance.
Stanford’s Duffie says the LCR effectively precludes the Fed from shrinking the balance sheet to anywhere near its pre-crisis levels. This is because the LCR requirements are not only large, but also very volatile.
“Returning to a small balance sheet and steering interest rates by controlling the scarcity of reserves is not consistent with the LCR requirement,” says Duffie. “If the Fed raises interest rates by reducing the supply of reserves, and some random event increases the demand for HQLA at banks, it would cause rates to spike a lot more than the Fed intended. And conversely, if banks see a surge of liquidity, and they don’t need as much HQLA, the scarcity of reserves will be lessened, and rates could drop sharply in a very short period of time.”
The Fed is likely to hit the brakes on normalisation long before that point. “If banks are lurching in and out of the Fed funds market to meet LCR requirements, it will cause a lot of volatility in short-term rates,” says Duffie. “If the Fed wants to maintain the LCR, which was a good innovation, then it will also want to adjust its balance sheet accordingly, and that means maintaining a significant amount of reserves.”
But banks could also see their liquidity, funding and capital metrics change materially as rates rise and the Fed shrinks its balance sheet.
The first question is whether banks will see a deposit flight as the Fed normalises policy. US dollar deposits at commercial banks have almost doubled from $6.7 trillion at the start of 2008 to nearly $12 trillion currently. Banks now have more deposits than they need and have been aggressively turning away non-operating deposits in an effort to ease their regulatory costs.
As rates rise, new dollars are more likely to go elsewhere, with MMFs a likely beneficiary.
“In an environment where interest rates are going up quarterly, we expect to see more cash flowing into products that earn market rates, rather than deposits,” says Deborah Cunningham, chief investment officer at Federated Investors. “That means MMF assets, which have held steady at around $3 trillion since 2009, could rise to $4 trillion–$5 trillion over the next year or two.”
In that scenario, deposits – and, by extension, LCR requirements – would remain at around their current levels, while the amount of cash available to borrow in the short-term wholesale funding markets would increase.
That, coupled with increased short-term debt issuance from the US Treasury, could prompt firms to reconstitute their HQLA buffers. “The big problem is scarcity. When Treasuries are in short supply, and pricing is artificial, banks prefer to hold excess reserves,” says Karen Petrou, managing partner at Federal Financial Analytics, a Washington, DC-based consultancy. “If that starts to ease, a number of banks will make significant HQLA decisions, and excess reserves could lose [out].”
There’s no question this is going to cause a change in how HQLA buffers are managedChristian Rasmussen, UBS
Scarcity of US Treasuries is unlikely to be a problem for much longer. While the Fed curtails its bond buying, the US Treasury department is poised to issue more than $1 trillion of debt in 2018 – the most since 2010, and up from $550 billion last year. Of this, more than $500 billion will be in the form of one- and three-month T-Bills, up from $137 billion in the 2017 fiscal year, which will help absorb some of the demand from banks seeking to replace excess reserves as the Fed reduces its balance sheet.
As supply increases, and spreads widen, banks will reassess their HQLA choices. “There’s no question this is going to cause a change in how HQLA buffers are managed,” says Christian Rasmussen, global head of liability creation, structuring and flow management at UBS. “The large G-Sibs will probably shift some existing collateral into short maturity profiles, and stay short because of the spread widening.”
But increased T-Bill issuance – which began ramping up in February – is a double-edged sword for liquidity-seeking banks. The Treasury will deposit the proceeds of the issuance in its account at the Fed, which currently stands at around $200 billion. This is projected to increase to $360 billion at the end of June, and to more than $400 billion by year-end.
As that balance increases – running counter to the goals of normalisation – even deeper cuts will be needed in excess reserves. So, if the Soma portfolio shrinks by $370 billion – in line with the Fed’s projections for 2018 – and the Treasury’s balance increases by $200 billion, excess reserves will need to decline by $570 billion to around $1.5 trillion this year.
Rising T-Bill issuance is also likely to increase short-term funding costs. “When Treasury’s cash balance goes up, it removes cash from short-term lending,” says Josh Younger, head of US interest rate derivatives strategy at JP Morgan. “We think the recent widening in FRA/OIS and most other measures of short-term bank funding costs has been driven in large part by the variation in Treasury’s cash operating balance.”
At the margin, reserve draining increases the cost of short-term funding by removing cash from the system, but when the Treasury grosses up its cash balance, that’s when things really start movingJosh Younger, JP Morgan
The FRA/OIS basis – between forward-rate agreements and overnight indexed swaps – widened to more than 40bp in the last week of February, from around 20bp at the start of 2018. Meanwhile, the GCF repo rate hit 1.575% on March 1, the highest level since the last quarter-end spike.
Analysis by JP Morgan’s derivatives strategists shows shifts in the Treasury’s operating balance at the Fed has a far greater impact on FRA/OIS – by a factor of four or five – than other forms of reserve draining.
“At the margin, reserve draining increases the cost of short-term funding by removing cash from the system, but when the Treasury grosses up its cash balance, that’s when things really start moving – and we expect them to get back to around $350 billion over the next two months, up from around $200 billion currently, and a low of around $65 billion last December,” says Younger.
If short-term funding costs rise – with repo rates at or above the upper bound of the Fed’s target range, wider FRA/OIS spreads, and a more negative cross-currency basis – then banks may want to hold more cash at the Fed to compensate.
So how much cash should banks hold for intraday liquidity purposes? Liquidity Risk Corp’s Fiedler says it would be prudent to maintain excess reserves equivalent to around 50% of the daily average interbank payment volumes. “That would be very prudent from a liquidity risk management perspective,” he says (see box: Intraday liquidity: once bitten ...).
Fedwire, which facilitates cash transfers between banks with accounts at the Fed, recorded daily average volumes of $3 trillion in 2017. That suggests the demand for excess reserves to cover intraday payments is $1.5 trillion, in the most conservative scenario, which aligns with the upper range of estimates from liquidity risk managers. “If banks are holding more cash than that, then there is something wrong – they are holding too much,” says Fiedler.
It’s very disruptive for banks to have to bid on reserves if they are overly scarceJosh Younger, JP Morgan
As it happens, banks were holding $2.1 trillion of excess reserves as of the end of January, down from a peak of $2.7 trillion in August 2014. Liquidity risk managers say it is likely the $600 billion surplus reflects buffer reserves held for economic reasons, which can be drained without causing disruptions.
A good portion of that is thought to be held by foreign banks engaged in what is known as the ‘IOER arbitrage’ – essentially borrowing from GSEs at Fed funds and depositing the proceeds at the Fed to earn IOER (see box: End of the IOER arb).
But ending the IOER arbitrage is unlikely to deliver the sort of long-term balance sheet reduction sought by the Fed. “The question is, when are you cutting at the bone with respect to the domestic banking system, and that level is very unclear,” says JP Morgan’s Younger. “There’s a utility to reserves beyond the arbitrage spread – banks need them for daily activities, and occasionally they may be willing to borrow in the Fed funds market at above IOER to meet those needs.”
If that happens, the Fed may be forced to put the brakes on normalisation. “It’s very disruptive for banks to have to bid on reserves if they are overly scarce,” says Younger, “and our view is that the Fed would want to avoid any disruptions to the interbank system.”
Regulatory experts say the Fed can take a number of steps to smooth the process. The first and simplest change would be to recognise required reserve balances – which currently stand at about $125 billion – as HQLA for LCR purposes.
Presently, the Federal Reserve’s Regulation D – which governs reserve requirements for deposit-taking institutions – prevents required reserves from being used to cover outflows, which in turn precludes them from being recognised as HQLA.
“We suggest regulation D should be changed to allow required reserves to be used to meet outflows if a bank comes under pressure. They can use exactly the same language as the LCR, which builds in a high hurdle for using the funds,” says Nelson at TCH. “We think that would be consistent with the historical purpose of required reserves and the safe conduct of monetary policy.”
TCH argues such a change would also be consistent with an amendment to the Federal Reserve Act enacted in 1980, which states that “balances maintained to meet [reserve requirements] … may be used to satisfy liquidity requirements which may be imposed under other provisions of Federal or State law”.
Such a change would reduce the demand for excess reserves to meet LCR requirements by roughly $125 billion.
Another relatively easy change would be to eliminate the RSF factor for US Treasuries in the NSFR, encouraging banks to hold more government securities as HQLA. “There are still some question marks around the NSFR, but that is certainly a change I would like to see,” say the liquidity specialist at the first European bank.
Bankers also want the Fed to adopt recommendations made by the US Treasury Department last year to eliminate the gold plating in the SLR and remove US Treasuries from the ratio’s exposure measure. “As normalisation gets underway, we believe either of those changes would be very helpful,” says a regulatory capital expert at a US bank.
Intraday liquidity: once bitten …
Excess reserves in the US banking system averaged $1.9 billion in 2007. That was too little, as banks – especially European firms – found out the following year.
“If I had told the board 20 years before [the financial crisis] that it could be impossible to exchange euros for dollars, they would have called me insane. But this is exactly what happened in 2007 and 2008,” says Robert Fiedler, who runs the consulting outfit, Liquidity Risk Corporation.
At that time, a number of European banks struggled to make timely US dollar payments despite having plenty of liquidity in euros.
Bank treasurers have been hording cash ever since. “They experienced this shock – something they thought would never happen, happened. They’re not going back to the old way of basically holding no reserves and assuming they can always go into the foreign exchange market and exchange euros for dollars,” says Fiedler.
Regulators have also worked behind the scenes to improve intraday liquidity risk management. “Banks are under immense pressure to reduce intraday liquidity risk in the system,” says a treasurer at a European bank. “Banks are analysing their intraday exposures more carefully, running stress scenarios, and getting more visibility into those risks, and that has an impact on the balances they keep at the Fed.”
In 2013, the Basel Committee on Banking Supervision issued guidelines on monitoring intraday liquidity risk, known as BCBS 248, although the compliance date has been pushed back in many jurisdictions.
“As technology develops, banks will eventually be able to optimise their intraday liquidity function, and you may see excess reserves coming down at the margin. But at this point in time, banks don’t really feel 100% confident about their ability to manage intraday liquidity, and they are maintaining higher reserves so they don’t get caught short,” says a liquidity risk specialist at a European bank.
Managing intraday liquidity has also become a lot tougher over the past decade. Tri-party repo reform took “a tremendous amount of free daylight out of the system,” says the treasurer at the European bank, and volumes are down from a pre-crisis peak of $2.8 trillion to $1.8 trillion currently.
Meanwhile, daily volumes in the Fed funds market are down from a pre-crisis peak of more than $170 billion to around $100 billion currently.
Others worry the expertise and infrastructure that supports the short-term funding markets has become stale after nearly a decade of plentiful central bank liquidity. “If I was sitting on a bank funding desk, I would be asking myself if we have the technology and staffing to efficiently raise short-term cash in the money markets,” says Glenn Havlicek, chief executive of GMLX, which provides liquidity management technology for financial firms. “After a decade of abundant excess reserves, it would be impossible not to get complacent about liquidity risk management.”
While the likes of GMLX – and peer-to-peer repo platforms DBV-X and Elixium – are applying new technologies to improve efficiency in the short-term funding markets, these initiatives are still in their infancy.
The same can be said of efforts to upgrade the infrastructure of the repo market. The Depository Trust & Clearing Corporation expanded its tri-party repo clearing services for the buy side last year, but take-up has been limited.
Ring-fencing requirements and stricter rules governing exposures to foreign subsidiaries and affiliates have also made intraday liquidity risk management much more complicated, and boosted demand for excess reserves. “There’s a lot of ring-fencing that didn’t exist 10 years ago. With the new regulations, there are limitations on the ability of banks to move cash around the globe to meet intraday liquidity needs,” says the treasurer at the European bank. “We’ve created a lot of pinch-points that didn’t exist before.”
End of the IOER arb
For the third quarter of 2017, the Federal Home Loan Banks (FHLBs) maintained an average balance of Fed funds sold of nearly $100 billion, at an annualised yield of 1.17%. It meant foreign banks could take the FHLB loans, deposit them at the Federal Reserve, and earn interest on excess reserves (IOER), which paid 1.25% over the same period – an 8 basis point spread.
The IOER arbitrage is not economic for US banks, which are subject to the supplementary leverage ratio and must pay Federal Deposit Insurance Corporation premiums on reserve balances.
But as the Fed drains reserves, and the Fed funds rate moves closer to IOER, the arbitrage becomes less attractive. “The foreign banks are typically holding reserves because they can fund them at a lower rate than IOER,” says Josh Younger, head of US interest rate derivatives strategy at JP Morgan. “For them, it’s an arbitrage, and they’re already winding down those books.”
As a result, Younger expects foreign banks, which had $967 billion of cash at the Fed as of January 10, to “bear the brunt of early-stage reserve draining”.
New senior management guidance proposed by the Fed on January 11 – which would require foreign banks to show they are operating their branches and agencies in accordance with US capital requirements – could hasten the process.
“That’s a quiet way to end the arbitrage,” says Karen Petrou, managing partner at Federal Financial Analytics, a Washington, DC-based consultancy, which flagged the proposal in a note to clients.
The week on Risk.net, March 10-16 2018Receive this by email