Money markets reliant on Fed injections
On 16 and 17 September, the US money markets seized up: excessive demand for cash caused overnight borrowing rates to surge. This was mainly caused by regulatory liquidity requirements which, given insufficient central bank reserves, limited the ability of major banks to absorb the spike in demand[1]. To ease the pressure, the Federal Reserve has since 17 September been injecting central bank money through overnight and term repo[2] operations with primary dealers. Overall, and given the limits set by the Fed, the amount of liquidity injected amounted to USD 237 bn by 18 December, while the demand for cash seen during the nine operations outstanding on that date amounted to USD 300 bn (figure 1). In addition, the Fed has been buying T-bills outright since mid-October, at a rate of USD 60 bn per month.
In the space of three months, from Wednesday 11 September to Wednesday 11 December, the measures taken by the Fed led to USD 328 bn of extra central bank money being injected (USD 213 bn via its repo operations and USD 115 bn via securities purchases). This amount may have reached USD 367 bn on 18 December.
The Fed has so far succeeded in easing the tension in the money markets. However, the specific needs of market participants as they approach their financial year-end could ramp up the pressure again (see below). As a result, the Fed announced on Thursday 12 December that it would increase its support. Overall, if the volume of demand for repo transactions reaches the amount offered by the Fed, USD 283 bn of additional central bank money could be injected by year-end, taking the total amount of support to USD 650 bn[3]. Huge though these numbers may seem, they may not be enough.
A third of the liquidity injected will end up in Treasury’s account with the Fed
First, even if the demand for repo transactions reaches the total amount offered, bank reserves held with the Fed will not increase by USD 650 bn by year-end, because part of the liquidity injections will continue to finance the increase in the Treasury General Account (TGA, figure 2).
Between 11 September and 11 December, banks’ current accounts with the Fed only swelled by USD 213 bn[4]: this was because USD 28 bn of bank deposits were converted into notes and coins and, most importantly, because the Treasury’s general account increased by USD 121 bn[5]. Similarly, between now and year-end, given the ongoing increase in notes and coins in circulation (USD 7 bn) and the likely increase in the TGA (USD 106 bn[6]), bank reserves are only expected to rise by USD 209 bn[7], taking them to around USD 1,880 bn, the same level as in October 2018.
Finally, during the period in question (11 September 2019 to 1 January 2020), the amount of cash held by banks with the Fed is only likely to rise by USD 420 bn. A third of the Fed’s cash injections will therefore have served, indirectly, to increase the TGA.
The impact of regulations on the money markets is underestimated
In addition, beyond liquidity requirements, capital requirements could also be playing a role in disrupting the money markets at the moment that market participants close their annual accounts.
Leverage constraints
Investors’ lack of appetite for Treasury issues has led to an unprecedented increase in primary dealers’ inventories of Treasuries (figure 3). Since mid-September, primary dealers have been able to refinance their net positions through repo operations with the Fed. However, those operations are taking place on the tri-party repo platform, with Bank of New York Mellon playing the role of clearing bank. The use of the tri-party repo market means that positions cannot be netted, unlike operations taking place via the Fixed Income Clearing Corporation (FICC). Ahead of their financial year-end, there is a risk that the liquidity offered by the Fed through its repo operations may not be accessible to the dealers most constrained by their leverage requirements[8]. Neither are the Fed’s T-Bill purchases enabling dealers to offload their large inventories of coupon bonds, which make up 85% of their Treasury portfolios[9].
There is a widespread belief among central bank officials that the tension in US money markets is being caused by excessive concentration of “excess” reserves. It is true that the eight systemically important US banks[10] alone account for almost half of reserves held with the Fed. There is a good reason for that: they are very large institutions (accounting for 53% of total assets) and are therefore subject to much stricter liquidity requirements, which increase their needs for central bank money (on an intraday, not overnight, basis)[11]. They also include the only two institutions (at least until the end of 2017 in JP Morgan’s case) that play the role of clearing bank in the tri-party repo market.
However, we believe that the pressure on liquidity stems from dwindling reserves, and probably also from their lesser concentration. In the fourth quarter of 2018, the same pressure on money market rates was avoided because the largest commercial bank (JP Morgan National Association) met most of the overnight refinancing demand: its reverse repo positions increased by USD 110 bn while its reserves fell by USD 130 bn (figure 4). Currently, however, certain large banks no longer have excess reserves as regard their liquidity requirements. It is only those large banks (in particular JP Morgan) that are (or were) able to absorb potential shocks, in particular at the financial year-end when, for example, foreign banks stop circulating cash borrowed from money market funds to other participants that cannot access those funds, or when dealers seek to make greater use of repo markets that allow netting. These markets need a lender of last resort. But, the Fed’s facilities, given their characteristics in terms of counterparties and types of repo, may not be able to absorb the effect of these year-end adjustments.
The G-SIB surcharge
The effect of these adjustments on the repo and forex swap markets could also be exacerbated by the withdrawal of certain large US banks keen to minimise their G-SIB surcharges[12].
For reference, regulators require banks whose failure could create a global systemic risk (Global Systemically Important Banks or G-SIBs) to carry additional capital. In the US, regulators calculate that additional capital using two methods, and the higher figure is the one adopted.
The FSB method is based on the five criteria used to identify G-SIBs: the size of banks, their interconnectedness, the lack of readily available substitutes or financial institution infrastructure for the services they provide, their cross-jurisdictional activities and their complexity. Based on a system of tranches, a capital surcharge is applied to each institution on the basis of its relative score. The second method replaces the lack-of-substitutes criterion by a measure of a bank’s dependency on short-term market financing, and favours an absolute measurement of each bank’s systemic importance. The second method almost always results in a larger surcharge than the first[13].
Based on the second method and balance sheets of the third quarter of 2019, the G-SIB surcharges for JP Morgan, Bank of America and Goldman Sachs could be increased by 50 basis points at the end of the year (from 3.5% to 4% for JPM, from 2.5% to 3% for BoA and GS, see figure 5)[14]. Past experience shows that banks whose systemic importance scores are close to a cut-off point tend to reduce their complexity, interconnectedness and cross-jurisdictional activity scores in the last quarter of the year in order to minimise their surcharge[15]. An effective way of reducing those scores is by not renewing overnight loans and borrowings in the repo and forex swap markets.