In the ongoing evaluation of what has been happening in the repo market, market participants have identified post-crisis regulations as a potential source of the problem. In particular, these regulations (including the Liquidity Coverage Ratio) require behemoth banks like JP Morgan and Citi to hold large amounts of reserves, and makes them reluctant to lend them out even when repo rates spike.
Having long said that the various liquidity regulations intended to prevent a recurrence of the last crisis could be the cause of a new one, I am certainly quite sympathetic to this view. However, information that is coming out now suggests another potentially complementary and aggravating factor.
In particular, reserve holdings are very concentrated:
Fed data show large banks are keeping a disproportionate amount in reserves, relative to their assets. The 25 largest US banks held an average of 8 per cent of their total assets in reserves at the end of the second quarter, versus 6 per cent for all other banks.
Meanwhile, the four largest US banks — JPMorgan Chase, Bank of America, Citigroup and Wells Fargo — together held $377bn in cash reserves at the end of the second quarter this year, far more than the remaining 21 banks in the top 25
Moreover, the big banks have been reducing their reserves:
Analysts and bank rivals said big changes JPMorgan made in its balance sheet played a role in the spike in the repo market, which is an important adjunct to the Fed Funds market and used by the Fed to influence interest rates.
Without reliable sources of loans through the repo market, the financial system risks losing a valuable source of liquidity. Hedge funds, for example, use it to finance investments in U.S. Treasury securities and banks turn to it as option for raising suddenly-needed cash for clients.
Publicly-filed data shows JPMorgan reduced the cash it has on deposit at the Federal Reserve, from which it might have lent, by $158 billion in the year through June, a 57% decline.
Although JPMorgan’s moves appear to have been logical responses to interest rate trends and post-crisis banking regulations, which have limited it more than other banks, the data shows its switch accounted for about a third of the drop in all banking reserves at the Fed during the period.
“It was a very big move,” said one person who watches bank positions at the Fed but did not want to be named. An executive at a competing bank called the shift “massive”.
Other banks brought down their cash, too, but by only half the percentage, on average.
For example, Bank of America Corp (BAC.N), the second-biggest U.S. bank by assets, with a $2.4 trillion balance sheet, took down 30% of its deposits, a $29 billion reduction.
So . . . substantial concentrations of reserves, and declining levels of reserves. Yes, these are all potential consequences of Frankendodd. But they also are potentially symptomatic of market power and the exercise thereof.
This triggered a synapse, which led me to recall a 1993 article from the Journal of Monetary Economics by R. Glen Donaldson. Donaldson’s article was motivated by a study of the Panic of 1907, when a “cash syndicate” (led by . . . J.P. Morgan, in person and through his eponymous bank) lent to cash strapped trust companies facing depositor runs at very high rates.
Donaldson presents a model in which a spike in the need for cash by a set of market participants (trust companies facing depositor outflow, in his model) makes the funds held by a group of other institutions pivotal: these institutions face a downward sloping demand curve for their funds because of constraints on competitive suppliers of funds. The pivotal institutions supply funds (through a repo-like transaction in which they buy securities from the trusts) at a supercompetitive price (by buying the trusts’ securities at subcompetitive prices). In his model, collusion between the pivotal institutions exacerbates the rate spike.
The main implication of the model is that spikes in the demand for funds lead to spikes in interest rates that are bigger than would prevail in competitive conditions.
There is an element of non-linearity in the model, because the big suppliers’ funds are not pivotal in normal conditions, but become so when the demand becomes sufficiently large. This leads to a switch from competitive to monopoly pricing, which in turn causes a spike in rates.
I should note that the regulatory and market power stories are not mutually exclusive, and are indeed complementary. Regulatory constraints can increase the demand for funds (making it more likely that the big suppliers will be pivotal) and can reduce the supply of funds from the smaller suppliers (which lowers the threshold for the switch from competitive to monopoly pricing, and makes the demand curves for the big suppliers funds steeper, leading to a higher monopoly rate).
I therefore consider it a plausible hypothesis that market power contributed to the repo market spike, and that one channel by which regulations contributed to the spike was through its effect on market power.
How can this hypothesis be tested? Conceptually, if regulatory constraints alone caused the spike, then those in possession of large quantities of reserves (e.g., Morgan) were absolutely constrained in their ability to lend additional reserves: the difference between the repo rate and the Fed Funds rate would represent the shadow price on this regulatory constraint.
If a big bank or banks exercised market power, this constraint would not be binding.
Operationalizing this test is likely to be complex, however. Big holders of reserves will inevitably make all sorts of arguments to say that they couldn’t have lent more.
This brings to mind the California electricity crisis in 1999-2000, when generators operated below various capacity measures, but pleaded that constraints (by unplanned outages, or NOX regulations, etc.) reduced their effective capacity below these nominal capacity measures. Given the complexity of operating a power plant, it was very difficult to determine whether the generators were withholding capacity, or in fact offered as much as they were capable of doing.
Despite the difficulty of operationalizing the test, I think it is something for regulators to attempt. There is a colorable case that the repo rate rise was exacerbated by market power, and given the importance of this market, this possibility should be investigated rigorously.
As an aside, the Donaldson model appeared only a few months before my Journal of Business article on market power manipulation. The two articles have a lot in common, despite the fact that they were developed totally independently, and seemingly involve completely unrelated markets (money vs. physical commodities). However, the core arguments are similar: economic frictions can periodically create market power in markets that are usually competitive.