The U.K.'s Vickers Report and the EU's recently released Liikanen Group report both recommend a fundamental structural change to the banking system to reduce systemic risk and the likelihood of taxpayer bailouts of banks: "ringfencing." The taxpayer guaranteed deposits of a financial institution would be inside the ringfence, and the ringfenced entity that can fund itself with these deposits can engage only in a limited array of activities that are deemed to be of low risk. Riskier activities, such as derivatives trading and securities underwriting, would have to be undertaken through separately capitalized affiliates that cannot be funded by the ringfenced entity.
This is, in essence, a variation on the Glass-Steagall approach of separating commercial and investment banking. The Volcker Rule is another way of achieving a separation of deposit taking and trading and underwriting activities.
These structural approaches are completely misguided, in my opinion. They are unlikely to have the desired effect of reducing systemic risk or insulating taxpayers from the fallout of risky trading activities.
Insured deposits are a potential source of moral hazard, and there is a case to be made for limiting the risks that can be funded with insured deposits. But historically these activity limits have proved to be a very poor defense against risk taking. Banks have gotten into trouble while engaging in traditional lending activities; the Latin American debt crisis and the S&L crisis are two prominent examples. Ditto Northern Rock in the U.K.
Moreover, it is beyond naïve to believe that making insured depositors whole is the only, or even the most important, cause of taxpayer funded bailouts. A failing investment bank-or a bank-owned trading entity outside the ringfence poses a potential systemic risk that would likely result in a bailout even if insured deposits are not directly at risk. Its derivatives counterparties- which are likely other big financial institutions, and which could well be ringfenced banks, would suffer losses if it went other. Its lenders, which could include, for instance, money market funds that buy its short-term debt- are also at risk. This could lead to destabilizing runs on the money markets. In brief, even entities outside the ringfence would be highly interconnected with the broader financial system, and with entities inside the ringfence. The Fed or the ECB would likely deem problems at such interconnected institutions to be systemically threatening, and would find a way to provide support to prevent a failure.
Put differently, an investment bank, like entity outside the ringfence, can be too big to fail.
Indeed, looking at the 2008 experience, the most vulnerable institutions were not funded by deposits. Bear Stearns was an investment bank, it was bailed out. AIG (AIG) was not deposit funded, it was bailed out, in large part (IMO) to prevent the failure of another non-deposit funded entity, Goldman Sachs (GS) (i.e., the AIG bailout was in part, and likely in large part, a backdoor bailout of Goldman). Merrill was foisted on Bank of America (BAC) to prevent its failure. And of course there is Lehman (LEHMQ.PK). It wasn't bailed out, but you can be metaphysically certain that inaction won't be repeated again.
It is quite easy to envision that there will be firms outside the ringfence that are deemed TBTF. If one of these firms looks acutely vulnerable, central bankers and regulators concerned about systemic contagion will find some way to prop it up. Taxpayer money will be at risk.
Moreover, the ringfencing idea creates a particular kind of vulnerability that proved to be an acute problem in 2008, and which is pretty well understood theoretically. The funding of non-deposit funded entities is often quite fragile, and vulnerable to runs. Like investment banks in '08, entities outside the ringfence are likely to depend on short-term wholesale funding that can (and will) run if the entity looks shaky, and there is doubt that a bailout would be forthcoming.
Deposit funding is stickier, and a destabilizing-and inefficient coordination game equilibrium-run is less likely to occur with sticky funding. Thus, putting trading activities outside the ringfence, and funding these risky activities through the wholesale markets exacerbates a source of financial fragility.
Thus ringfencing increases the risk of liquidity/funding crises that trigger a bailout. It exacerbates a form of fragility that was devastating in 2008, when it was financial institutions that were outside the Glass-Steagall ringfence that proved most vulnerable to the fallout from the decline in real estate prices.
The knotty issue here is why the funding of trading, market making, securities underwriting and similar activities is typically quite fragile. The Diamond-Rajan story is that runs are a way of disciplining opportunistic intermediaries. With the potential for externalities from runs (due to contagion effects) one could make a case that funding is excessively fragile. But that is hardly an argument for ringfencing, which precludes the ability to use less run-prone deposits to fund these activities: Indeed, if anything, it is an argument against ringfencing.
The nostalgia for Glass-Steagall-like structural "fixes" is pervasive. But structural separation of certain forms of financial intermediation, and separation of certain forms of activity from deposit taking, don't address a fundamental source of systemic risk: The reliance on short-term, wholesale funding that is vulnerable to runs. Indeed, it likely makes the problem worse because it increases the reliance of the segments outside the ringfence on such funding. It is delusional to think that just because taxpayer-guaranteed deposits are not directly at risk, that taxpayers are not on the hook for bailouts. If a big, interconnected entity that sits outside the ringfence teeters on the brink, regulators and central banks may well deem it too big to fail: No More Lehmans is their motto. Ringfencing does not isolate derivatives trading, underwriting, etc., from the broader financial system. Entities outside the ringfence will still be interconnected.
So even if there is no explicit taxpayer obligation (as with insured deposits), there is an implicit one. We've seen that implicit guarantee invoked before, and we could see it again. Ringfencing does not address the problem of a non-depository SIFI that is highly interconnected. Indeed, by increasing the number and size of such entities, it almost certainly makes this problem worse.