After more than a decade without appearing in major news headlines, U.S. repo markets did their best last week to catch up and highlight what may happen in money markets in an environment of lesser reserve abundance than previously.
The general story has been told extensively in other places and the deeper final causes are yet to be established. In brief, U.S. Treasury repo rates began climbing broadly last Monday, before spiking notably on Tuesday to, in some segments, levels not seen for decades. Spillovers extended into adjacent markets, the Effective Federal Funds Rate traded far above IOER and its policy band, highly rated overnight financial Commercial Paper rates increased by 25 bps, while its less liquid companions in the non-financial and asset-backed realm increased by more than 200 bps at their peak.
Although slow to heed calls for a standing repo facility to address potential bottlenecks to date, the NY Fed was, after initial technical difficulties, in markets, offering up to USD 75bn in an overnight repurchase agreement operation with primary dealers. USD 53bn was pulled on Tuesday, and the full amounts offered of USD 75bn for the remainder of the week, in repeats of Tuesday’s operation.
As a result, temporary stress in repo (and adjacent) markets has receded, while the NY Fed upped its schedule of liquidity provisions by announcing it “will offer daily overnight repo operations for an aggregate amount of at least $75 billion each, until Thursday, October 10, 2019” and “will offer three 14-day term repo operations for an aggregate amount of at least $30 billion each” this week.
Coincidentally, the size of to date conducted overnight repos of USD 75bn almost mirrors the increase in the Federal Reserve’s foreign repo pool since the start of the year. Since a series on said pool was published three years ago in these pages, it seems an opportune moment to update developments surrounding it and its, at minimum indirect, effects on the repo stress last week.
Foreign Repo Pool – Developments since 2016
From the last discussion in September 2016 until year end 2018, no larger developments of significance regarding the pool seem to have occurred. Japan, the main actor behind the pool’s increase in 2015/2016, maintained its balance of ~USD 120bn and Brazil, a secondary driver in 2016, withdrew around USD 20bn in 2018, explaining most of the pool’s bump last summer.
From its lows about a year ago, the size of the pool then increased almost USD 100bn until today. The rise so far has been more volatile than last round. This seems mostly the result of a broader country participation this time, opportunistically allocating funds. Among the largest contributors over the past year are the Central Bank of the Republic of Turkey and the Swiss National Bank, each increasing allocations by more than USD 10bn.
As far as understating potential drivers is possible, the most commonly heard reason for foreign central bank’s increasing allocation to the pool is the inverted U.S. Treasury curve which, supposedly, allows higher returns by shifting investments to the short end of the curve. The curve view, of course, is a simplified (and sometimes misleading) version of a broader term premia model, but the insight might still be accurate: Given the decline in longer-term U.S. interest rates, it might be more profitable to roll shorter-term instruments than carry longer tenored assets.
Considering FX reserve managers’ high safety standards in their asset composition, in the U.S. this eventually results in choosing among U.S. Treasury bills, lending to highly-rated financial institutions on a short-term basis or allocations to the Fed’s repo pool.
As was the case in 2016, the Fed only engages in limited public communications about the nature of the pool, including its interest rate, which remains available only in its quarterly financial statements. In 2016, i.e. before the introduction of the Secured Overnight Financing Rate (SOFR), the rate paid on balances in the pool was a 70:30 weighted average of tri-party and interdealer GCF repos, both secured by Treasury collateral.
The table below shows the average return during H1 2019 for each of the short-term vehicles available to FX reserve managers in USD markets.
Two points stand out in particular.
One, the rate the Fed pays on the pool is within one basis point of the average SOFR rate. Since a lot of time and effort was invested in the development of SOFR, including robust collection of transactions cleared through the FICC’s DVP repo service, it seems reasonable to believe the Fed now also uses this high-profile version of Treasury repo rates as benchmark for interest rates paid on the pool. Reversing perspectives, it could thus equally be argued that the foreign repo pool rate was a (crude) precursor of SOFR.
The second and more surprising fact is that the pool (or SOFR rather) is the highest rate in the sample. Higher than rolling the shortest-term T-bills is not necessarily surprising, but higher than uncollateralized short-term lending to private sector banks (here proxied by overnight UDS Libor) is. The Fed’s repo pool thus combines the best of two worlds: doubly ensured safety, given the Fed as counterparty and U.S. Treasury securities as collateral, plus higher interest rates than is available in private markets. It is little wonder foreign central banks are increasingly favoring the use of the Fed’s repo pool for allocations at the front-end of the USD curve.
A reasonable question to ask is why is the repo pool’s rate so high and why does the Fed chose to pay it nonetheless. The Fed’s answer to these questions would surely be interesting to hear; the first one at least can be answered from the outside. SOFR represents not ‘the’ U.S. Treasury repo rate, but an average of several market segments. As in many financial markets, dealers intermediate financing from the demand to the supply side, while living off the bid-ask spread. In repo markets this broadly means borrowing in the tri-party market from cash rich investors (e.g. money market funds, security leaders, FHLBs) and lending to leveraged vehicles (e.g. hedge funds & aggressive fixed income funds), mostly on a bilateral basis.
Since SOFR captures both legs of this intermediation, its rate is naturally higher than actual market rates a regular cash investor (like a MMF) can receive. If the Fed, quite reasonably, intends to remunerate pool balances with market interest rates, a clearer definition of the rate as the ‘cash lending’ rate might be appropriate. Data-wise, the adjustment would be trivial. The 25th percentile of SOFR captured transactions very well captures the rate paid on average on the ‘cash lender-dealer’ transaction, the most logical equivalent private sector alternative to the pool.
Further along these lines, it could equally be questioned why the Fed should even pay private repo rates on pool balances. These can, depending on market dynamics, trade quite a bit above Treasury bills, further increasing the appeal of the pool against regular U.S. safe assets – and in the process increasing the interest payments to foreign central banks by the Fed.
While the above situation is at least in part a political decision, adjudicating what level of attractiveness overseas FX mangers are granted by the Fed, the current design has another unfavorable design feature which, returning to last week, at the margin contributes to future tightness in repo markets.
The pool’s ‘inverted’ & pro-cyclical supply response
Whenever a policy unrelated spike in money market rates occurs, a supply and demand response follows. On the supply side, this means leveraged vehicles reconsider whether financed assets are still positively carrying at higher funding rates. No larger liquidations will occur due to one or two days of higher rates, but should tightness persist for longer, unwinding of levered positions should be expected.
On the supply side, this means reallocation of cash balances currently deployed elsewhere into repo markets. Trading in the Fed Funds market slowed last week, as FHLBs redirected funds into repo markets, less regulated foreign banks increase borrowing to fund repo arbitrage trades and U.S. banks consider whether higher repo rates justify more lending in these markets, despite the less favorable regulatory treatment.
What unites both responses is that they will lead to a rebalancing in the opposite direction of the spike, as more supply is made ready while demand is scaled back.
Users of the foreign repo pool act according to the same incentives as their private counterparts, by increasing balances as rates increase. In their case however, the supply effects on markets is propulsive rather than absorptive, as funds allocated to the Fed’s pool act as exogenous factor in reducing banks’ excess reserves. Lower excess reserves in turn reduce banks’ excess liquidity which could otherwise be mobilized to counter higher repo rates by lending into repo markets.
To put it another way, the private repo market every day produces an interest rate, signaling market participants either tightness or abundance ahead of next day’s operations. In case of tightness, foreign central banks logically choose to supply funds to the now higher yielding repo pool. Unfortunately, when they, say, sell U.S. Treasuries or divert deposits previously held with commercial banks into the pool, excess reserves banks hold decline, lowering their ability to counter market tightness naturally. This way, the initial tightness will be procyclically prolonged due to an imperfect incentive structure at the moment.
The repo pool itself was not the immediate cause of the spike in repo rates last week, but balances nonetheless increased by USD 15.5bn the week ending Wednesday September 18th, bringing the above dynamics into play. In addition, while it’s a coincidence that the increase in the pool this year matches the Fed’s current level of overnight liquidity provisions, the absence of a rise in the past months could realistically have cushioned the spike last week.
Central bank deposit/repo facilities for FX managers – a global perspective
Zooming out from the Fed specific matters discussed so far, a general preference by FX mangers for deposit or repo allocations directly with other central banks is observable across the globe.
The Federal Reserve’s repo pool, as well as deposit facilities for other central banks by the European Central Bank and the Bank of Japan have grown substantially since 2016, each now exceeding USD 230bn in size. While increases in the ECB’s and BoJ’s deposits during 2008/2009 were rather collateral reinvestments of local currency funds obtained by the Fed as part of FX swap lines, increases in recent years are in line with financial statements by FX reserve managers, disclosing their deposits held with fellow central banks (including the BIS, which itself manages another ~USD 200bn pool not shown here).
In total, the USD 800bn FX managers now hold on deposit or in repo facilities with other central banks almost matches their USD 1tn currently held with commercial banks globally. Together, these non-security liquidity tranches now account for 15.5% of global FX reserves as of March 2019 and are increasingly important to include when analyzing global money markets. Their ultimately almost unconstrained theoretical size could in the future act as natural habitat for FX mangers pushed out of sovereign bond markets by further rounds of quantitative easing.