You may recall from my last Alt-M post that in September 2008 the failure of Lehman Brothers caused sizable losses to a large money-market mutual fund, The Reserve Primary Fund, which held hundreds of millions in Lehman IOUs. These losses reduced the fund's asset portfolio value below its total share value at the pegged redemption rate of $1 per share. For two business days the fund's management neither reduced the share claims nor injected capital to restore the assets. Alert institutional shareholders saw that there wasn't enough asset value to pay everyone $1 per share, but the first to redeem could get that much, and so quite rationally ran to redeem. On the second day Reserve Primary "broke the buck," re-pricing its NAV to 97 cents. It was eventually liquidated. Other "prime" mutual funds experienced heavy redemptions during those two and the following two days, although none broke the buck. (Many received capital injections from their sponsoring firms.) On the fifth day the Treasury stepped in to guarantee the $1 share price of all money market mutual funds.
Many regulators and economic analysts have inferred from these events that money-market mutual funds (MMMFs) are inherently run-prone. The fact that this was the first run in MMMF history, however, should give us pause. There is a more plausible reading of the evidence. Although the Reserve Primary Fund did invite a run by letting its total shareholder claims exceed its total assets in value for a bit more than a day, MMMFs can be structured and managed so that this never happens.
In a nutshell, the stage is set for a run on a bank or mutual fund when claims can be redeemed on demand, and claimants have reason to believe that their total claims exceed total assets (plus any off-balance-sheet funds) available for paying them. Bank deposits are debt claims to fixed dollar sums, so a sudden drop in the market value of assets can trigger a run if the bank is so thinly capitalized that market net worth becomes negative. Mutual fund shares are not fixed-dollar debts, but equity claims to percentages of the asset portfolio, such that total claims are supposed to add up to 100% of the asset portfolio and no more. (Neglect of the distinction between debt and equity funding, by the way, is embodied in the obfuscatory language according to which MMMFs and hedge funds are part of a "shadow banking" system.) When assets drop in value, share accounts are to be marked down correspondingly, so that they never over-claim the assets.
Ordinary open-end mutual funds accomplish this by continuously adjusting the "net asset value" or NAV, the dollar price at which one share in the portfolio can be purchased or redeemed. Money-market mutual funds typically operate with the accounting convention of a fixed $1 NAV. To keep $1 shares from over-claiming the assets in the event of asset losses, such a MMMF needs only to continuously adjust the total number of shares. That is, its contractual arrangement with shareholders provides that a proportional number of shares will be immediately subtracted from account balances.
As J. P. Koning has pointed out, this arrangement is not just a conceptual possibility, it has been put in place by some MMMFs in Europe under the rubric of a "reverse distribution" or "share-reduction" mechanism. In September 2014 Bloomberg.com reported that "the world's biggest money manager," BlackRock Inc., was adopting a share-reduction mechanism as a way of coping with negative yields on money-market instruments. Or, as a spokesman put it, BlackRock "determined that it is in the best interests of shareholders to implement a form of share-reduction mechanism, which enables a stable NAV to be maintained on days when the net yield on the fund is negative."
Neglecting this practicable method for run-proofing stable-NAV mutual fund claims, many would-be reformers after 2008 arrived at a diagnosis of inherent run-proneness with systemic negative spillovers, and were inspired to propose additional legal restrictions on MMMFs to make them less fragile. ("Additional" restrictions because mutual funds were already regulated by the SEC under the Investment Company Act of 1940.) In a useful recent review of this literature on run-proofing proposals, Harvard finance professors Hanson, Sharfstein, and Sunderam accordingly declare that "MMF regulation should attempt to both reduce the ex ante incentives of MMFs to take excessive risks and increase the ex post ability of MMFs to absorb losses without setting off runs." They emphasize three proposals popular in the literature:
- "requiring MMFs to adopt a floating NAV structure," that is, outlawing the fixed $1 share value (they oddly characterize this as a way "to subject MMFs more fully to market forces," it is unclear why they do not recognize a conflict between market forces and legal restrictions);
- "requiring MMFs to have a 1% capital buffer combined with a "Minimum Balance at Risk" (MBR) provision whereby investors cannot immediately redeem all of their shares," a proposal from four economists at the Federal Reserve Board and FRB New York; and
- "requiring MMFs to have a 3% subordinated capital buffer."
The authors favor the third type of restriction as a way to "reduce ex ante incentives for risk-taking" while also allowing a MMMF to "maintain the current fixed NAV structure for ordinary MMF investors and thus preserve any transactional benefits those investors reap from the existing system." Like most contributors to the literature (an exception is a chapter by Agapova), they do not consider a contractual share-reduction mechanism, let alone note that it allows the transactional benefits of a fixed $1 NAV to be combined with 100% capital financing. Possibly this oversight is the result of the mechanism having disappeared from use in the United States, simultaneously with American MMMFs (though not European MMMFs) replacing mark-to-market accounting of short-term assets by amortized cost valuation, and mark-to-market accounting of total share value by "penny-rounding" (not until book-value NAV falls below 99.5 cents does the MMMF have to inject capital or break the buck).
Somewhat surprisingly, proposals for these kinds of legal restrictions have come even from some sources that ostensibly favor greater reliance on free markets. For example, the Shadow Financial Regulatory Committee in February 2012 and again in September 2012 issued a statement that called for outlawing the $1 fixed NAV for MMMFs, with the exception of retail funds whose sponsors provide "an explicit contractual guarantee" to inject enough capital to redeem at par whenever necessary and who meet new "capital and liquidity requirements."
The SEC has adopted many of the proposals for new restrictions on MMMFs, including one of the least popular. In the name of reducing liquidity, credit, and interest-rate risks, a set of restrictions adopted in February 2010 imposes a minimum liquidity ratio (cash or Treasuries to total assets), a AAA rating requirement for 97% of assets, a shorter average portfolio maturity than previously required, and periodic stress tests. In July 2014, the SEC announced additional restrictions, to be phased in over two years. A few months from now, the new rules will be fully in place. The most important two new rules are:
- Institutional prime and tax-free funds, but not institutional government funds, will be required to adopt a variable NAV. (As a reminder, "prime" funds hold mostly paper issued by multinational banks and other financial firms, but also some short-term Treasuries. Tax-free funds hold tax-exempt state and municipal bonds; government funds hold only US Treasury and agency bonds. All three types are divided between retail and institutional versions. The former are now limited by SEC rules to natural persons. The latter have higher minimum investments and lower expense per share ratios and thus are favored by institutional investors and cash managers at large corporations.) There is no good rationale for this requirement given the equally run-proof alternative of a fixed NAV combined with a share-reduction mechanism.
- Funds are given the discretion, whether they want it or not (they cannot contractually bind themselves not to use it), "to impose liquidity fees or to suspend redemptions temporarily, also known as 'gate,' if a fund's level of weekly liquid assets falls below a certain threshold." Hanson, Sharfstein, and Sunderam, and many other analysts, rightly reject such "gating rules" as means to diminish runs. Inability to precommit not to impose a gate can be expected to weaken a MMMF. Knowing that a gate might be imposed, especially if another fund has just done so, investors will likely become more anxious to redeem now rather than wait and see.
In recent years more than half of the MMMF industry's assets ($3.085 trillion at the end of 2015) has been held in prime funds. But the mix is now changing and the total is falling because institutional investors who prefer a fixed NAV free of gates are now compelled to switch out of prime to lower-yielding government funds. Mutual fund providers have begun to respond to institutional investor preferences by converting entire funds. For example, in December 2015 the Fidelity group, "to assure shareholders that they will have daily access to funds used primarily for transactions," converted three funds with shareholder approval. The Fidelity Cash Reserves fund became Fidelity Government Cash Reserves; VIP Money Market became VIP Government Money Market; and Fidelity Retirement Money Market became Fidelity Retirement Government Money Market II. Institutional investors are being limited to options they consider inferior, as shown by the drop in the size of institutional prime funds exceeding the gain in institutional government funds.
To criticize most generally the reform strategy of seeking to improve financial institutions by imposing restrictions on them, I suggest that reformers and policy-makers who seek what Cecchetti and Schoenholtz call "the optimal mechanism for securing the safety and soundness of MMMFs" need to consider that we can approach optimal arrangements only through a wide-open market competition among alternatives. To think that we can derive optimal financial institutions (those that most fully exhaust gains from voluntary trade) by theorizing about them is an unwarranted pretense. We need to allow different strategies for providing prime MMMFs that are safe (and have other desirable features, which we may trade off against safety) to compete head to head. To affirm that a "subordinated capital buffer," for example, is the optimal or efficient arrangement, for example, we must see it pass the market test by out-competing share-reduction mechanisms with fixed $1 share values, and other arrangements, for the favor of investors. Quite likely, there is no single type of prime MMMF that best serves all potential users, but rather a variety of types would attract customers. If some well-informed investors prefer a fixed-NAV fund that is not run-proof, because it offers them other advantages, there is a heavy burden of proof to show that such a fund should not be allowed. The goal of a robust financial system calls for a diverse ecosystem of mutual funds, not a monoculture that is susceptible to a single disease. Top-down restrictions promote a monoculture.
A case for not allowing free competition among a variety of contractual MMMF arrangements must presumably appeal not just to theoretically possible non-pecuniary externalities from allowing run-prone funds, but to evidence of serious non-pecuniary externalities. This burden has not been met. First, it has not been shown that the run on The Reserve Primary Fund was the shock that prompted heavy redemptions at other funds, rather than both being consequences of a common shock, namely the failure of Lehman Brothers and the associated decline in the market value of short-term claims on other financial institutions. We will never know, because the Treasury intervened, how long the redemptions at other funds would have continued. Second, given the singularity of the run on Reserve Primary, it is far from clear that run-prone funds would be dangerously common in the financial marketplace emerging from unrestricted institutional competition.
Some proponents of restrictions on MMMFs have argued that the possibility of declining asset prices due to heavy redemptions makes even floating-NAV funds vulnerable to me-first runs. Thus Hanson, Sharfstein, and Sunderam speak of an "impetus to Diamond-Dybvig style runs" arising from the possibility that "MMFs forced to liquidate [illiquid] assets may have to sell them at heavily discounted, 'fire-sale' prices." Those who wait to redeem will face a lower NAV because early redemptions will push down the value of the fund's assets. The Diamond-Dybvig model of a self-justifying bank run, however, depicts an intermediary issuing debt claims whose total always exceeds the liquidation ("fire-sale") value of its assets during the middle period of its three-period life. It is not surprising that an insolvent bank is run-prone. It is hard to see how this scenario is relevant to a MMMF that promptly marks down its total claims to the market value of its portfolio. As a matter of fact, "Diamond-Dybvig style runs" among uninsured banks (prompted by mutually-validating fears that others will run, causing insolvency via fire-sale losses), as distinct from "bad-news" runs prompted by pre-run insolvency, are very hard to find in the historical record.
If me-first runs due to feared fire-sale losses were an actual problem among MMMFs, a fund could provide safety against over-claiming by valuing the fund's assets (and correspondingly investor account balances) using its own actually realized sale prices, not observed transaction prices elsewhere. Thus if selling assets to replenish the fund's cash buffer after settling a $1000 check costs the fund more than $1000 in book value of assets, it could reduce the customer's remaining balance accordingly. This would eliminate any beat-the-crowd dynamic in which shareholders rush to sell because they rationally anticipate declining payoffs from fire-sale losses as other shareholders liquidate. Whether this mechanism would have more-than-offsetting disadvantages in the eyes of customers, the financial market would show us – provided we leave the market free to operate.
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.