I suggest a simpler approach. It would be compatible with other financial stability reforms, but is better understood as a substitute for Dodd-Frank and other measures. The proposed approach would require new legislation consisting of the following specific measures, starting from a pre-Dodd-Frank baseline:
- Prohibit the use of short-term debt funding by all financial firms other than deposit banks. (The financial sector's short-term debt, inclusive of deposits, will be called "broad money" herein; this is a conventional term for measures of the money supply that include various kinds of non-deposit short-term debt.)
- Apply reserve requirements (not to be confused with capital requirements) to all of the broad money issued by deposit banks, thereby giving the Federal Reserve the power to cap the quantity of broad money outstanding.
- Fully insure (i.e., with no coverage caps) all of the broad money issued by deposit banks and phase out insurance of long-term certificates of deposits.
- Charge risk-based fees to the deposit-banking sector for this public backstop. These fees would continue even if the Federal Deposit Insurance Corporation's insurance fund is fully funded, at which point the fees would become a fiscal revenue item. (In the central banking context, this is called "seigniorage": fiscal revenue from money creation.)
- Tighten up existing deposit bank portfolio constraints. Most importantly, implement a "swaps push-out rule" along the lines of Dodd-Frank's.
- Replace the Basel Committee's new liquidity standards with an international accord that prohibits financial institutions from issuing broad money denominated in nondomestic currencies. (In short, wind down the so-called "Eurocurrency" markets.)
This simpler approach—which we can call the "licensed money" approach—obviously centers around the financial sector's short-term debt (think maturities of under a year). The approach confines the issuance of broad money to the existing deposit banking system; it gives the Fed the power to cap the quantity of broad money outstanding and to adjust the cap in the conduct of monetary policy; it wraps broad money with a public backstop, making it sovereign and default-free; and it charges the banking system a fee for this public commitment.
The licensed money approach is designed to render the financial system panic-proof. The term "panic" is used here in a specific sense: to quote Ben Bernanke, it is "a generalized run by providers of short-term funding to a set of financial institutions." The licensed money system is based on the idea that, when it comes to financial stability policy, panics are "the problem" (or the main one anyway). I'll discuss this further, below.
Notice what is not included in the licensed money approach. There is nothing here about "systemic" or "macroprudential" oversight of the financial system; no designations of nonbank financial firms for special regulation; no new resolution authority for nonbanks; nothing about the securitization markets; nothing about derivatives (apart from the push-out); no proprietary trading limits; nothing about breaking up the banks; and no Glass-Steagall-type limitations on affiliations between banks and nonbanks. Again, the licensed money system would be compatible with those measures, but it should reduce the perceived need for them.
THE REGULATORY PERIMETER
The sketch above is silent about what activities can take place outside the licensed banking sector. It only says that those activities must be financed in the capital markets (with equity and/or long-term debt), not the money market. In principle, the licensed money system would allow for a very wide degree of latitude for nonbank financial firms, subject to appropriate standards of disclosure, antifraud, and consumer protection. So nonbanks might be given free rein to engage in structured finance, derivatives, proprietary trading, and so forth. But they would not be allowed to fund short.
It is sometimes suggested that securities dealers and other nonbank financial firms "need" to fund short—that they somehow cannot conduct their businesses without short-term wholesale funding. This argument needs to be put to rest. There is nothing about the broker-dealer business model that requires unstable short-term funding. Dealers could conduct all their current activities while financing themselves entirely with equity and longer-term debt. Naturally, their cost of financing would go up. Some of this cost would be passed through to their customers, mostly in the form of higher bid-ask spreads. But bid-ask spreads today are subsidized; the dealer-funding model has an implicit public backstop. All the more reason to prohibit this funding model.
Once nonbank financial firms are financed in the capital markets (with equity and longer-term debt) instead of the money market, they are amenable to ordinary bankruptcy proceedings—like any other firm. No special insolvency system for nonbanks is required. In this regard, the recognition that "panics are the problem" is immensely clarifying. If panics really are the problem, then financial institution failures do not imperil the broader economy so long as they do not trigger a panic. And the licensed money approach makes the financial system virtually panic-proof.
Incidentally, the proposal would mean the end of the money market mutual fund sector as it exists today. The idea is to have only one set of licensed issuers of broad money, operating under the purview of the money and banking authorities. The licensed money system would get the Securities and Exchange Commission out of the monetary business, which falls outside its mission and core competency.
It may be useful here to visualize, at a very high level, what the licensed money approach would mean for a giant financial conglomerate like J.P. Morgan, Bank of America, or Citigroup. In simple terms, we can picture the conglomerate as consisting of a holding company with two subsidiaries: a big deposit bank and a big securities firm. Under the licensed money system, the securities firm would be required to "term out"—that is, end its reliance on short-term funding. Nor would the conglomerate be able to simply move its securities business into its deposit bank: deposit banks have long been expressly prohibited from engaging in securities dealing, subject to very narrow exceptions (see 12 U.S.C. § 24). Nor could the conglomerate use its deposit bank to fund its securities business: the Federal Reserve Act already imposes strict limitations on such affiliate transactions (see 12 U.S.C. §§ 371c–371c-1).
But, you might ask, aren't regulators already taking steps to panic-proof the financial system? I am referring here to the new liquidity rules that U.S. regulators proposed in October 2013 for the biggest financial firms—rules that are designed to implement the global "Basel III" liquidity standards. Essentially, the new rules, if and when implemented, will require the largest financial firms to hold enough liquid securities to cover expected near-term outflows in a liquidity squeeze. Isn't this a reasonable approach to panic-proofing?
The new rules are probably much better than nothing, but there are reasons to question whether this is a good strategy. First, the new rules are not generally applicable; they apply only to certain large financial firms. Short-term funding should therefore be expected to migrate outside the purview of the new rules. Second, the new rules are necessarily quite complicated. They require a specification of what kinds of assets will remain liquid in a crisis, as well as an estimation of firms' expected cash inflows and outflows under a stress scenario. Needless to say, such estimates are quite speculative. As Federal Reserve Governor Daniel Tarullo emphasized in a recent speech, the new rules will not immunize the financial system against damaging runs and panics.
Finally, and most importantly, the new liquidity rules are not part of an integrated institutional design. In this regard, observe that the licensed money system accomplishes several related objectives: it makes the entire broad money supply sovereign and default-free; it gives the Fed the power to cap the quantity of broad money outstanding; it generates seigniorage revenues from the entire banking system, not just the central bank; and it re-establishes money creation as a matter of national sovereignty (more on this in the next section). The new liquidity rules accomplish none of those things. They represent an ad hoc patch, not a carefully thought-through design.
If panics really are the problem, then financial institution failures do not imperil the broader economy so long as they do not trigger a panic. And the licensed money approach makes the system virtually panic-proof.
INTERNATIONAL CONSIDERATIONS
The licensed money system contains a crucial international component. Today, overseas financial entities issue huge amounts of dollar-denominated short-term debt instruments (called "Eurodollars"). Those instruments are typically issued to U.S.-based institutions, and the proceeds of issuance are typically invested back into the U.S. credit markets. This is classic fractional-reserve banking: it is money creation. It involves the issuance of broad money denominated in dollars, but it takes place outside the reach of U.S. monetary and banking authorities. During the recent crisis, the panic in the Eurodollar market prompted a massive policy response from the Federal Reserve, peaking at a staggering $580 billion in U.S. dollar funding to foreign institutions via liquidity swaps with foreign central banks.
Traditionally, money creation has been viewed as a matter of national sovereignty. We presumably have reason to care when counterfeiters of U.S. dollars are operating overseas, and U.S. authorities are vigilant about combating this activity. Conceptually speaking, the issuance of dollar-denominated broad monies by overseas financial institutions isn't all that different.
The most straightforward way to deal with this would be through an international accord—an addendum to the existing Basel capital accord. Countries (or currency areas) would agree to prohibit domestic financial institutions from issuing broad money denominated in nondomestic currencies. Essentially, the Eurocurrency markets would cease to exist. Money creation would be recognized as a sovereign prerogative.
To be clear, none of this would preclude foreign financial institutions from owning U.S. deposit banks. Foreign institutions could therefore offer dollar-denominated broad money (including checkable deposits) to their customers abroad. But their U.S. bank subsidiaries would obviously be subject to the full panoply of regulatory standards that govern other U.S. banks. Hence, all dollar-denominated money creation would fall within the purview of U.S. banking and monetary authorities.
WHY PANIC-PROOFING?
The licensed money system is based on the idea that financial stability policy should concern itself with panic-proofing—and perhaps not much else. This claim often meets with fierce resistance, generally on two grounds. The first objection is that the problem of financial instability is about much more than panics. (Panics, to repeat, are widespread redemptions of short-term debt, period.) Panic-proofing, it is said, would not necessarily mitigate problems like "asset-price bubbles," "overleverage," "excessive risk-taking," and so on. And those other problems are taken to pose a serious danger to the broader economy in and of themselves, irrespective of their propensity to trigger panics.
The second objection complements the first. It holds that, even if panics were indeed "the problem" (so to speak), panic-proofing—eliminating or greatly limiting run-prone funding, in one way or another—would not be desirable. There would be costs to such an approach, and enforcement would be challenging. Besides, the argument goes, there are other ways to deal with panics. For example, we might leave fragile short-term funding untouched while seeking to forestall the types of events that trigger panics (collapsed bubbles and so forth). Or we could just deal with panics as they arise; that's what the lender of last resort is for, after all.
These are fair objections, and addressing them in detail would require a much longer article. I offer only a summary response here. Start with the first objection: the claim that things like "bubbles" and "overleverage" pose a danger to the broader economy in and of themselves, regardless of whether they might trigger a panic. Fair enough, but when it comes to institutional design, we do have to prioritize. Few would deny that panics do serious damage to the broader economy; they arguably dwarf other financial phenomena in their destructiveness. The case against bubbles is somewhat less conclusive. As many others have pointed out, the dot-com bust of the early 2000s destroyed more wealth than the recent housing bust, yet the dot-com episode was followed by only a brief and mild recession.
Now, maybe what we really care about is debt-fueled bubbles—in other words, cycles of debt and deleveraging. A prominent set of theories takes this basic tack. These include Irving Fisher's "debt deflation" theory, Hyman Minsky's "financial instability hypothesis," Richard Koo's "balance-sheet recession" theory, Ben Bernanke's "financial accelerator" idea, and John Geanakoplos's "leverage cycle" theory. Generally speaking, the theories do not ascribe any particular significance to short-term debt. They are about deleveraging in a generic sense, not panics per se.
Even if we accept those theories, we then face some thorny policy questions. The theories tend to stress not just financial sector leverage, but also (or even mostly) overall levels of household and business debt. Such matters take us far beyond the traditional perimeter of financial regulation. Furthermore, there is reason to think that overleverage becomes a serious problem only when the economy has been severely damaged by some kind of shock. If this is right, and if panics are a principal source of major trauma, then panic-proofing might render "overleverage" somewhat less problematic. Finally, it is worth considering whether our modern answer to panics—basically, an open-ended commitment of public support to much of the financial sector—might not be a major source of "debt-fueled bubbles" and "overleverage." If so, then this only strengthens the case for rethinking how we deal with panics.
In the end, there are no conclusive answers to these questions and we are left with judgment calls. But there appears to be a strong argument that panics should be viewed as the central problem for financial stability policy. Incidentally, this discussion highlights the problem with lumping various phenomena together into a complex "financial crisis" and then treating it as the unit of analysis and the thing to be prevented. This methodological tendency—which is practically ubiquitous—has major consequences for policy analysis. It inevitably draws attention away from the panic and toward the purported "excesses" that preceded it. As a result, panic-proofing rarely makes it onto the policy radar.
This brings us to the second objection mentioned above: the claim that even if panics are taken to be "the problem," panic-proofing is not a good answer. As noted above, other strategies are available: we might tackle panics indirectly by seeking to prevent the occurrence of triggering events (like collapsed bubbles), or we might just deal with panics on the back end with public liquidity support. These arguments raise basic questions about the optimal locus of intervention. The notion that we should try to forestall triggering events presupposes that we can successfully identify them (avoiding both false negatives and false positives) and successfully defuse them at reasonable cost. This presumably means fighting bubbles, which is sure to be difficult and controversial in the moment. Relying on the lender of last resort to deal with panics is problematic, too. The very existence of this commitment introduces potentially severe distortions into the financial system: it encourages the growth of individual financial firms and the financial sector as a whole, and it perversely subsidizes the financial sector through artificially low funding costs.
None of this is to say that panic-proofing would be costless (more on those costs in a moment). The point is simply that the alternatives have serious shortcomings, too, and a robust approach to panic-proofing should be on the policy menu—as a complement to or even a substitute for other financial stability measures.
WHAT ABOUT "NARROW" BANKING?
The licensed money system achieves panic-proofing by making the entire broad money supply sovereign and default-free. Public backstops of this kind raise obvious moral hazard problems. Is there a way to do panic-proofing without creating moral hazard?
One strategy would be to limit issuers of broad money to holdings of super-safe assets like base money and/or short-term Treasuries. This portfolio composition presumably would make public backstops of broad money superfluous. There would be no need for anything like deposit insurance.
Proposals of this kind have been around for many decades. They go by names like "100 percent reserve banking" and "narrow banking." These proposals have typically applied to deposits, but there is no reason why they could not be applied to broad money. In fact, University of Chicago economist John Cochrane recently proposed just that. In a June 24, 2013 Wall Street Journal op-ed titled "Stopping Bank Crises Before They Start," he argues that "the financial system needs to be reformed so that it is not prone to runs." His solution: "Don't let financial institutions issue run-prone liabilities"—in other words, panic-proofing. Under his design, any financial firm that issues run-prone debt, whether or not styled as "deposits," would have to confine its portfolio exclusively to base money and short-term Treasuries. The proposal is radical: it would do away with the existing deposit-banking business model.
As a method of panic-proofing, Cochrane's narrow banking approach has considerable appeal. However, there are some apparent drawbacks. The fiscal and monetary implications would need to be thought through carefully. Under his design, the (broad) money supply would apparently be capped at the amount of Treasury debt outstanding. A long-term balanced budget (as farfetched as that might seem today) could present a serious monetary problem; at some point, paying down the debt would automatically reduce the broad money supply. This raises a complex set of questions about monetary policy independence, fiscal management, tax smoothing, and the implications of sovereign default. By contrast, under the licensed money system, fiscal considerations do not impose limits on the quantity of broad money outstanding, as licensed banks can expand their balance sheets (i.e., issue money) by investing in high-quality private credit.
This is not to say that portfolio constraints are unwarranted. On the contrary, the licensed money system would take the existing system as a starting point. Today, U.S. deposit banks are basically limited to holding diversified portfolios of relatively high-quality credit instruments (loans and investment-grade bonds). They may not own stocks or junk bonds, for example. In recent decades, the major loophole has been in the derivatives area. As noted above, the licensed money system would follow Dodd-Frank by putting in place a swaps push-out. This is not because of any prejudice against derivatives. Rather, it simply reflects the fact that derivatives generally do not advance the system's objectives. It is important to keep in mind that the system is designed to issue the broad money supply—and banks issue money when they acquire credit assets (cash instruments), but generally not when they write derivatives (synthetic instruments). An exception to the push-out is in order for interest-rate hedging; the Dodd-Frank Act includes such an exception.
Of course, the licensed money approach raises challenges of its own—moral hazard in particular. Needless to say, the record of deposit insurance in the United States is not unblemished. The savings and loan (S&L) crisis of the 1980s led to a costly $124 billion taxpayer bailout. However, it is worth noting that in the years preceding the S&L episode, U.S. bank and thrift regulation was characterized by glaring design defects. Portfolio constraints were relaxed in the early 1980s, allowing thrifts to extend into new and riskier asset classes. A reasonably coherent capital regime did not exist until 1988. Risk-based deposit insurance premiums were not introduced until 1991. Finally, prior to 1991, regulators were not legally required to shut down critically undercapitalized banks and thrifts on a prompt basis, so problems were left to fester for years.
The lender-of-last-resort commitment encourages the growth of financial firms and the financial sector as a whole, and it perversely subsidizes the financial sector through artificially low funding costs.
In response to the S&L debacle, meaningful improvements were made to the design of U.S. depository regulation. And it is noteworthy that, despite the staggering magnitude of credit impairments in the United States from 2008 to 2010, no taxpayer bailout of the deposit insurance system was required—not even close. Total bank failure costs to the FDIC's deposit insurance fund as a result of the recent crisis are estimated to be around $100 billion. Those losses are being fully recouped from the deposit banking sector, whose reported equity capital currently stands at $1.6 trillion. In short, the deposit insurance system has done more or less what it was designed to do. This experience suggests that a well-designed system of portfolio constraints, capital requirements, risk-based fees, and supervision can have some success in counteracting the effects of moral hazard.
The licensed money approach is incremental; it sees much in the current institutional setup that is worth preserving. It would modernize the current banking system rather than scrapping it, as some narrow banking proponents would do. Still, the similarities here are more important than the differences. Both Cochrane's narrow banking approach and the licensed money approach take panic-proofing to be the central aim of financial stability policy. Once there is agreement on objectives, then it comes down to a question of comparative institutional design.
FINANCIAL REGULATION AS MONETARY SYSTEM DESIGN
The last year has seen two remarkable speeches by senior Federal Reserve officials on the topic of short-term wholesale funding. The first came in February 2013 from William Dudley, president of the New York Fed. "How comfortable should we be," Dudley asked, "with a system in which critical financial activities continue to be financed with short-term wholesale funding without the safeguards necessary to reduce the risk of runs and the fire sales of assets that can threaten the stability of the entire financial system?" His answer: "I don't think we should be comfortable" with such a system.
Yet, he noted, that is the system we have—even after the financial reforms of recent years. "[W]e have not come close to fixing all the institutional flaws in our wholesale funding markets," Dudley remarked. He offered "two broad paths" for addressing the problem:
The first option would be to take steps to curtail the extent of short-term wholesale finance in the system. In principle, regulators across a broader set of institutions and markets could take steps to directly limit the use of short-term wholesale funding to finance longer-term assets…. In other words, regulators could require that a greater proportion of market-based finance be funded by longer-term debt….The other path would be to expand the range of financial intermediation activity that is directly backstopped by the central bank's lender of last resort function.
Three months after Dudley's speech, Tarullo addressed the same topic. "I strongly believe that we would do the American public a fundamental disservice were we to declare victory [over financial instability] without tackling the structural weaknesses of short-term wholesale funding markets, both in general and as they affect the too-big-to-fail problem," he said. "This is the major problem that remains, and I would suggest that additional reform measures be evaluated by reference to how effective they could be in solving it." He emphasized the need for policy measures that apply "more or less comprehensively to all uses of short-term wholesale funding, without regard to the form of the transactions or whether the borrower was a prudentially regulated institution."
It is a remarkable fact that today—more than five years after the panics of 2008—so little progress has been made in the area of short-term funding. How can this failure be explained? Surely part of the answer involves the two objections to panic-proofing that were discussed above: first, doubts as to whether panics in and of themselves are the main policy problem; and second, a vague sense that, even if panics are the main problem, they should be addressed indirectly rather than directly. Whatever the reason, fragile short-term funding has not been dealt with in a meaningful way.
Arguably, we have been making financial stability policy much more complicated than it needs to be. Panics are an age-old problem. They are not about cutting-edge developments in modern finance. Short-term debt is primitive, not complex. The upshot is that panic-proofing does not entail the extension of regulatory oversight or control over the outer reaches of modern finance. Nor does it entail taking aim at nebulous enemies like "systemic risk" or "excessive risk-taking." It is not clear that these are even meaningful concepts—much less that they can provide a sound basis for policy.
Legal limits on the issuance of "private money" are nothing new. At the risk of stating the obvious: current law prohibits the issuance of deposit instruments without a special license. In a prior era, similar prohibitions applied to the issuance of circulating bank notes. In other words, the law has long made money creation a privileged activity. It is widely acknowledged that the financial system's short-term wholesale debt obligations are functional substitutes for deposits. Yet they have no legal-institutional status as such. Hence the short-term wholesale funding markets today represent a form of "free banking": money creation without a license and outside the purview of monetary and banking authorities.
It is sometimes said that enforcing short-term debt limits would be difficult. True, but difficult compared to what? Compared to anti-counterfeiting laws? Capital requirements? Securities regulation? Tax laws? All of those legal systems present opportunities for avoidance and evasion. Yes, enforcement is challenging—welcome to the law.
The licensed money system is not exactly a method of "financial regulation." It is better understood as a modernization of the monetary system. Implicit in the approach is a counterintuitive idea: that financial instability is, at bottom, a problem of monetary system design. In fact, it always has been. This recognition brings a great deal of clarity to the task at hand, and it points toward a far simpler approach to reform.