The majority of federally insured savings and loans failed in the 1980s, wiping out the Federal Savings and Loan Insurance Corporation in 1989. The fiasco ultimately cost taxpayers around $150 billion to make savings depositors whole. Two years later, the failures of hundreds of commercial banks put the Federal Deposit Insurance Corporation in the red. (The FDIC got a bridge loan from the US Treasury, which it eventually repaid.) It became clear that deposit insurance had fostered immense moral hazard, enabling the growth of unsound S&Ls and commercial banks.
For many reformers these events raised the question of how the core services of banks (intermediation and payments) might be provided without the expense of tax-funded guarantees, and yet without the danger of runs that had prompted the creation of the FSLIC and FDIC. A number of economists (myself included) pointed to checkable money-market mutual funds (MMMFs) as an alternative to bank deposits that are not run-prone and therefore have no need for taxpayer-funded guarantees.
MMMFs, like other mutual funds and unlike banks, offer savers not debt claims promising specified dollar payouts on specified dates but rather equity claims (shares) in the dollar value of a portfolio. Like other mutual funds, a MMMF buys back shares on demand at the current "net asset value" or NAV. The modifier "money-market" means that a fund invests only in fixed-income securities with less than a year in remaining maturity, which means that present-value losses will be negligible from a rise in interest rates. A fund can keep default and liquidity risks low by maintaining a diversified portfolio of highly rated securities with active secondary markets.
In 1976 Merrill Lynch introduced a MMMF that allowed customers to write checks against their account balances, an innovation which was quickly copied by other funds. Money-market share accounts now combined the services of checking accounts with much higher returns, because they were not subject to the binding interest-rate ceiling (under the Fed's Regulation Q) then constraining bank accounts. To make them seem more like bank accounts, fund providers adopted the convention of pegging the share redemption value or NAV at $1, and varying the number of shares in an account, rather than varying the share price to reflect changes in the value of portfolio assets. The popularity of MMMFs soared. MMMFs that hold only Treasury obligations are called "government" funds. Those that hold mostly commercial paper and jumbo bank CDs are called "prime" funds.
J. Huston McCulloch put the case for MMMFs not needing government guarantees well in a 1993 article: "[E]ven though MMMFs invest in financial instruments that may not come due for many weeks or months, they are entirely run-proof. Should the volume of withdrawals be high enough" to require net sales that shrink the asset portfolio, "the fund's liability to its remaining depositors simply falls in the same proportion." That is, each MMMF share is a claim not for a fixed dollar sum, but only for a fixed percentage of the portfolio's value. A fall in the total value of the asset portfolio, whether from redemptions or from bad-news events that reduce assets' market prices, immediately reduces the total value of shares so that they never over-claim the available assets. Any bad-news net market value loss is immediately spread evenly over shareholders rather than being concentrated "on the last unlucky depositors in line, as occurs in a run on a traditional bank." With no greater losses falling on the person last in line to withdraw, there is no incentive to run to withdraw ahead of others. Thus, "as long as MMMFs behave like true mutual funds," continuously marking portfolio assets and shares to market value, the problem of the me-first incentive to run "cannot arise."
I made essentially the same argument in chapter 6 of my text The Theory of Monetary Institutions. There I argued that a run arises from the combination of three conditions: (1) claims are redeemable in pre-specified dollar amounts (i.e. are debts), (2) redemption is unconditionally available on demand, with a first-come first-served rule for meeting redemption demands, and (3) the last claim in line has a lower expected value. Mutual funds eliminate the first element (claims are equity rather than debt), which is sufficient to eliminate the run problem. It's no use rushing to redeem when bad news about the asset portfolio arrives, because your account balance has already been marked down. They also eliminate the third element (because every share redemption receives the same percentage of the portfolio value) when assets are liquid enough or the fund is small enough to make "fire-sale" losses from asset sales negligible.
But wait — doesn't this argument assume that MMMFs vary the price of their shares like ordinary mutual funds? Doesn't it matter that the share redemption value is pegged at $1? McCulloch explained why it should not matter: "Some MMMFs offer investors a variable number of shares of fixed value instead of a fixed number of shares of variable value. This is merely a cosmetic difference with no substance, however." The problem of claims exceeding portfolio value "arises [only] when funds try to offer investors a fixed number of shares of fixed value." In other words, so long as $1 shares are promptly subtracted from each account in proportion to any decline in total portfolio value, or alternatively promptly marked below $1 (an event called "breaking the buck"), there remains no incentive to run.
In practice, subtracting $1 shares is not done (for reasons not immediately obvious), and breaking the buck has become an occasion to liquidate the fund. Accordingly parent companies, to keep a MMMF alive and preserve its brand-name capital, almost always choose to eat losses and maintain the $1 share value. A 2010 report by Moody's identified 147 occasions over the period 1980-2007 when a MMMF suffered a net decline in portfolio assets that, without a rescue, would require breaking the buck. Only one fund actually broke the buck. (It was then liquidated, with shareholders receiving 96.1 cents per share.) In 146 cases the parent firm stepped in, absorbing losses to keep the share value at $1. If a parent firm acts immediately, upon news of critical asset losses, either to break the buck or instead to pitch in to preserve the par value, then running to get a better payoff than other shareholders remains either impossible or pointless.
Fast-forward to September 2008. On Sunday the 14th, Lehman Brothers found itself insolvent and without a rescuer. Early Monday morning Lehman filed for bankruptcy. A money-market fund called The Reserve Primary Fund was at that point holding $785 million in Lehman debt instruments, slightly above 1% of its $62.5 billion in assets under management. (This size put it in the top twenty, but outside the top ten.) An immediate 20% write-down on Lehman paper meant a $157 million drop in the value of the Fund's asset portfolio. Under the SEC's "penny-rounding" rules then in effect, the fund's NAV was allowed to remain at $1.00 per share in the face of a 0.2% loss (20% on 1% of assets) because such a loss leaves the unrounded value above 99.5 cents. When the market opened on Monday morning, shareholders quickly ran on the fund. They worried, for good reason as it turned out, that current assets were insufficient to redeem all shares at $1.00, and that the parent company would not pitch in to support the price. And indeed the parent company did not pitch in. By 1pm Monday (the 16th) shareholders had redeemed a bit more than a quarter of their claims at $1 per share. The fund's custodian State Street Bank refused to make further payouts, and on Tuesday the fund broke the buck. The Reserve also imposed daily withdrawal limits on its other funds.
During that Monday, and again on Tuesday and Wednesday, other prime funds experienced heavier than normal redemption outflows. Other MMMF parent firms, by contrast to The Reserve, immediately supported their prime funds that had Lehman-related losses, and continued to redeem at $1 per share. No other funds broke the buck. By the 19th the industry-wide dollar value of assets under management by MMMFs was down by $247 billion, a bit less than 7 percent of the value held ten days earlier.
After these three days of relatively heavy net redemptions following the Lehman bankruptcy and Reserve Primary buck-breaking, on Thursday the 19th, the US Treasury stepped in to stanch the redemptions, which it considered equivalent to runs, with something that it considered equivalent to federal deposit insurance. It announced what Secretary Hank Paulson described as a "temporary guaranty program for the U.S. money market mutual fund industry," assuring shareholders in participating funds that their shares would be redeemed at $1 even if their fund's net asset value fell below par. The Federal Reserve pitched in on September 22 by creating a special "Asset-Backed Commercial Paper Money Market Mutual Fund Liquidity Facility" to lend funds to banks for acquiring the commercial paper assets that MMMFs were shedding.
As later described by Philip Swagel, who was a Treasury official at the time, the MMMF guarantee program was initially funded, in an unprecedented and legally dubious move, from the Treasury's Exchange Stabilization Fund:
The US Department of the Treasury (2008) used the $50 billion Exchange Stabilization Fund—originally established back in the 1930s to address issues affecting the exchange rate of the US dollar—to set up an insurance program to insure depositors in money market funds. … Use of the Exchange Stabilization Fund for this purpose was plausibly legal—after all, a panicked flight from US dollar-denominated securities could be seen as posing a threat to the exchange value of the dollar—but its use in this way was without precedent.
It should be noted that there was in fact no panicked flight from US dollar-denominated securities in general. US Treasury securities rose in value during the crisis as investors worldwide considered them a safe haven. The trade-weighted US dollar index actually rose sharply in the six months after Lehman fell and the Primary Reserve Fund broke the buck. In its indifference to the rule of law, the US Treasury acted much like the Federal Reserve System did during the crisis.
After one year, the Treasury ended its MMMF guarantee program. It has since imposed new pricing restrictions, liquidity requirements, and accounting rules on the funds in the name of reducing the problem of runs. (I will discuss these regulatory changes in my next Alt-M post.)
So what happened in September 2008? Is the run on Reserve Primary and heavy redemptions at other prime funds evidence that, contrary to McCulloch's and my argument, prime MMMFs with a fixed $1 share price are in fact inherently fragile?
Stephen G. Cecchetti, former Director of Research at the Federal Reserve Bank of New York, and co-blogger Kermit L. Schoenholt have said so:
The fundamental problem facing U.S. regulators is that money market funds are banks in everything but their outward legal form. They perform liquidity and credit functions that are identical to those of chartered banks; in particular, they offer the equivalent of bank checking deposits, making them vulnerable to a run.
This argument won't do. It completely fails to engage the basic counter-argument that checkable equity claims (MMMFs) are not run-prone because they distribute portfolio asset losses in an essentially different way from checkable debt claims (bank deposits).
Useful analysis of the run-proneness of MMMFs is provided by a 2013 comment on SEC rule proposals by the Squam Lake Group, a committee of 13 center-left to center-right financial economists. They note that a MMMF (like a bank) will be run-prone whenever the aggregate redemption value of its shares or NAV exceeds the actual market value of the fund's assets, so that early redeemers can expect to get more than late redeemers. Under current accounting rules for money-market mutual funds (which they abbreviate MMFs), they point out, this can happen for two reasons:
First, mutual funds have the option to account for assets at amortized cost if they have a maturity of 60 days or less. With that option, the [total redemption value of shares] is not a true reflection of the fair market value of fund assets. Whenever investors can redeem at a NAV that is higher than the fair value of the assets, investors have incentives to run.
Second, and more fundamentally, prime MMFs invest substantially in assets without a liquid secondary market. This creates an incentive for fund investors to run during a period of financial stress, because even "fair market value" may exceed by a significant amount the value at which the fund can quickly sell assets to meet investor redemptions. Therefore, … the first MMF investors to redeem their shares during a crisis are likely to receive a higher price for their shares than those who follow once the fund is forced to meet redemption demands by selling assets that have not yet matured. … This first-to-redeem advantage, which is exacerbated by amortized cost accounting, creates an incentive for MMF shareholders to run.
In other words, MMMFs in August 2008 did not exhibit the immunity to runs that McCulloch and I expected in cases where the accounting rules did not, as we assumed they generally do, rule out an excess of aggregate share redemption value over actual asset portfolio value. Some funds used accounting rules that allowed them not to mark 60-days-or-fewer assets to market at all, and not to mark other assets to a market price that corresponded to their actual immediate liquidation value.
In summary, we learned in August 2008 that MMMFs using certain accounting rules are not run-proof. For 24 hours The Reserve Primary Fund carried a diminished asset portfolio without either topping it up or diminishing the claims against it, and consequently was rationally run upon. We did not learn that MMMFs are inherently fragile, but rather that run-proneness depends on the accounting practices that a fund uses.
From this diagnosis, no policy intervention is indicated. What follows is rather that in a market where losses remain private, investors can be expected to consider the relative fragility under certain circumstance of funds that opt to use potentially run-incentivizing accounting practices. Such funds, if they do not offer some fully compensating advantage, should be expected to lose their market share. Money-market mutual funds that instead credibly bind themselves to thoroughgoing mark-to-market accounting and other run-proofing practices (such as perhaps a pre-funded commitment by the parent company to shelter shareholders from losses), and advertise that fact, should be expected to flourish in the marketplace. Such MMMFs remain an available payment mechanism that is not susceptible to runs and therefore has no need for guarantees at taxpayer expense.
To come in a later post: What to make of the US Treasury's new restrictions on MMMFs?
*Acknowledgment: I thank Kyle Davidson for research assistance.
Editor's note: After this post was originally published we received a memo from John Dellaportas, a lawyer representing Reserve Primary Fund, pointing out some inaccuracies in it. The present version corrects those inaccuracies. We thank Mr. Dellaportas for the information he supplied.