The scale of finance has drawn heightened scrutiny in the years after the financial crisis. Yet in spite of this concern, the size of banks has only grown. There's fear that the government may have to intervene again if another financial crisis comes along, and Fisher, the Dallas Fed president, has blasted this trend.
It is imperative that we end TBTF. In my view, downsizing the behemoths over time into institutions that can be prudently managed and regulated across borders is the appropriate policy response. Only then can the process of “creative destruction”— which America has perfected and practiced with such effectiveness that it led our country to unprecedented economic achievement— work its wonders in the financial sector, just as it does elsewhere in our economy. Only then will we have a financial system fit and proper for serving as the lubricant for an economy as dynamic as that of the United States.
Of course, the regulatory confusion that would arise from breaking up the banks would have massive effects, but are there also other theoretical reasons to be suspicious of "just" breaking up banks? According to the traditional narrative, large banks are too vulnerable to unseen risk. If all the models are calibrated, and something outside those models surfaces, the entire machine could break down. This puts the entire market in jeopardy. Large banks, knowing this, are then willing to take on more risk, as they know that the government will intervene to save them. This combination of factors then creates a "too big to fail" phenomenon, and society pays the price while the bankers continue picking up pennies off of the train tracks.
Yet small banks are very vulnerable as well. What sometimes can be forgotten from discussions of banking is that big banks is why they are so problematic. Systemic risk is at the root of the problem. With their high frequency Gaussian models to hedge alphas, betas, deltas, gammas, a single tail even can cause a the largest bank to collapse, sending ripples through the entire financial system. An analysis of the Fed's interbank lending system showed that 75% of the payments involved 0.1% of the nodes and 0.3% of the linkages between nodes in the banking network. From this image, it's very easy to imagine an explosion causing one bank causing a cascade throughout the network.
But is size the only issue? Not necessarily. Some earlier studies about bank resiliency actually indicated that larger banks should actually decrease bank failure! Bank, as a result of their size, are able to diversify more and limit their exposure to sector-specific volatility. Notably, the United States actually has relatively low bank concentration compared to other countries. The three largest banks in the United States only controlled 19% of the industry in 2003, while the corresponding numbers for Finland and New Zealand were 85% and 77% respectively. Research from the NBER found that a one standard deviation increase in bank size resulted in about a 1 percentage point decrease in bank failure proportions. Considering the percentage risk of bank failure was only 4% in the whole sample, the 1 percentage point would have been a significant decrease in bank failure rates.
Of course, this does not suggest that large banks were better; the fact that many of them collapsed in the recent financial crisis suggests that this isn't the case. Additionally, the focus on the probability of bank failure glosses over the issue of magnitude; the smaller number of bank failures most likely had massive effects. But this does suggest that the relationship is not as simple as one might think, and that the true relationship likely has a severe non-linearity.
Moreover, a market populated by small banks is prone to a crisis because there's "too many to fail". If the market is fed on the same monoculture of debt priced with bad models, there's still the risk that a systemic crisis could run through the system. With VaR models that are ill suited for complex environments and hyperspeed algorithmic trading models, it's not unthinkable that traders could feed on themselves and trigger stock market shocks. This system would be difficult, if not impossible to effectively monitor, especially when any given bank can have large systemic effects.
As a result, some have called for an increase in the diversity in financial systems. The voxEU article specifically outlines four positive externalities from diversity:
- Bailout/Moral Hazard Externality - banks tend to pursue the same investment as they are consequently more likely to be all bailed out.
- Systemic Risk Externality - as banks have a hard time taking into account the effect of its actions on other firms, this leads to inefficient levels of systemic risk with homogenity
- Herding/Momentum Externality - as markets tend to herd, whether for psychological or principal-agent reasons, increasing diversity would limit the swings in the market.
- Insurance Externality - higher diversity makes cross-insurance more robust, reducing risk
With the discussion framed in terms of externalities, the natural answer is a pigouvian tax. The authors propose a system of capital requirements based on how much a given bank's profits or share prices correlate with the market as a whole. The government would "tax" banks who "go with the flow". This would take into account both "too big to fail" as well as "too many to fail". Large banks would be required to hold more capital as they, by virtue of their size, are highly correlated with the market. Small banks would be also pushed to try different strategies to avoid higher capital requirements. The simplicity in the rule is also incredibly elegant; a heuristic, and not a model error sensitive parameter. Thus, in an inverse of Taleb's criticism of the current financial system, this kind of macroprudential regulation may promote a certain level of antifragility as individual banks could play for the lottery tickets with undefined payoffs. It may not be enough, but coupled with robust layers of monetary policy, there may yet be hope for complex economies in an unpredictable world.