Breaking up large banks risks increasing rather than reducing systemic risk and the cost to taxpayers of that risk.  It is, admittedly, an unpopular view as more politicians and regulators take up the cause of breaking up the banks (most recently, Neel Kashkari https://www.minneapolisfed.org/news-and-events/presidents-speeches/lessons-from-the-crisis-ending-too-big-to-fail, but also Democratic hopeful Bernie Sanders and many more). The syllogism upon which the argument for breakup rests is that the banks need bailing out because they are so large that they would inflict damage on the economy if they failed; therefore, the answer is simple: we make them smaller, and voila, they wouldn’t ever need bailing out.  I would argue that there is a benefit to an economy of having appropriately sized banks that surpasses any costs of supporting banks when necessary, and that the diversification that comes with size is a critical factor in minimizing the chances of bank failures. That begs the question of what is an optimal size, and how does one really calculate the cost of bailing banks out.

Bank failures rarely occur in isolation.  Rather, they occur in waves as a particular factor, such as loans to a particular sector of the economy, causes distress.  That distress tends to be widely dispersed throughout the banking industry.  Lending is basically what economists call a commodity business, a standard product for which the only basis for competition is price.  It engenders what is often called lemming like behavior, where banks stampede into particular types of lending.  A race to the bottom ensues until too many loans are made a too low a price.  Borrowers become overleveraged, so a small turn in industry conditions triggers a wave of defaults.  The resulting losses affect not just one imprudent bank but a large chunk of the banking industry.  When a particular problem affects many banks, and suspicion spreads about the effect on other banks, eroding confidence in the entire system, we get systemic risk.

Some conclusions can be drawn from this.  Breaking up the banks will merely mean that the losses will be divided among a greater number of institutions, so the number of institutions failing will be greater.  The proportion of the industry as measured by assets will not necessarily change.  The economic impact of that proportion of the industry failing could be just as bad, only more difficult to manage because of the greater numbers of institutions.  The pain of financing being withdrawn from a particular sector of the economy generally creates pressure to support the institutions that finance it.  One has only to think back of the farm loan crisis of the mid-1980’s, and the rescue of the Farm Credit System, with its myriad of small, cooperatively owned banks. Similarly, the captive auto finance companies in 2008-9, which were relatively small institutions, were supported in order to help the auto sector.

More difficult to tackle is the notion that size actually provides benefits.  Academics have not found proof of economies of scale. That may be because of the ways in which they set up their studies.  Intuitively, it makes sense that at least certain businesses within banks must have economies of scale.  For example, the credit card and mortgage banking industries have evolved seemingly naturally to become highly consolidated so that the vast majority of all lending in those sectors now is done by only a handful of institutions.  The high cost of building the infrastructure of call and processing centers, risk management systems and tricky funding mechanisms (securitization) present barriers to entry that are insurmountable for very small institutions.

The credit card and mortgage banking businesses are not the only ones requiring infrastructure.  Other types of banking activities also require a very large scale to be viable.   Broadly speaking they are those that service the large corporate and institutional clients.  Large institutions want large loans—too large to be prudent holdings by even the largest global banks.  Exposures of $10-60 billion to one borrower, for example, would wipe out a large portion of the $170 billion of common equity of Wells Fargo bank if the borrower failed.  Banks typically syndicate the loans to share the risks with others, but there is a limit to how many can be in the syndicate, and there is still an initial holding period until the loans are sold.  Large, multinational corporations also require a host of other services from cash management, to currencies to bond and equity flotations, loans to foreign subsidiaries, trade finance, and hedges, not to mention pension management for their employees. Each of these require large scale infrastructrure, from maintaining a global network of offices, to massive, multi-hundred million dollar software systems.  For some of the sophisticated services, reputation and superlative expertise are required, so that only those institutions that are the biggest and the best receive serious consideration from clients.  A small institution simply could not provide these services.  We must give serious thought to whether it would benefit the US economy if our large corporations would have to travel to other continents to receive the services they require.  The loss of jobs for highly skilled professionals, the loss of tax revenues from a profitable sector of the economy could put a dent in GDP and US Treasury coffers.

The probability of systemic crises should also increase if institutions are less diversified.  Diversification is perhaps the most important single tenant of risk management for banks.  Diversification comes in the form of product offerings, geographic dispersion among different markets, or just sheer number of small transactions.  The idea is that all banking activities are not subject to the same cycles and pitfalls at the same time; at any given time, some activities will be doing well while others will be under stress, but the diversified institution will sail on a stable course.  Large size is no guarantee of diversification; the blend of activities must not be completely correlated. But small size always involves concentrations.  If we break up banks along geographic lines, we will get geographic concentrations; if along product lines, we will get product concentrations.  If the economy of a narrow region or particular product areas, such as let us say commercial real estate lending, spiral down, the concentrated banks will fail.  It is helpful to remember the dark days of the financial crisis.  In 2008, the specialized investment banks failed (Merrill Lynch and Lehman) while diversified banks reaped earnings from lending operations to offset trading losses.  In 2009 the spike in loan losses hurt traditional commercial banking.  Those commercial banks that had investment banking operations in 2009 (like Bank of America, Citigroup and JPMorgan) were able to offset some of the losses on loans with the rebound in values of their trading portfolios.  Banks that were merely lenders to commercial real estate, like Colonial Bancorp, failed.

If we must have large banks for efficient delivery of financial services, we can still minimize taxpayer exposure with appropriate capital cushions and regulation that is aimed at preventing systemic crises.  The latter task is complex and can range from regulation to attempt to prevent lending bubbles from occurring, to ensuring adequate market liquidity in the case of a freeze-up like the one that occurred in 2008.  Nevertheless, there will be times when systemic crises occur.  In that case, support for solvent but illiquid banks is a cheaper way to go than liquidation of an entire banking system.  The costs to taxpayers  in this country have always been temporary, essentially in the form of loans to be repaid by the industry in the ensuing years, with interest.  While arguably taxpayers are at risk until the repayment, history says that banks always return to profitability after their bad assets are cured because their basic franchise, that of taking deposits and making loans is a stable, annuity-like business as long as borrowers pay back their loans.

Of course insolvent banks need to be dealt with as well, in a way that prevents moral hazard.  The industry has always been asked to take on the resolution of large insolvent banks. Equity holders and managements were wiped out.  For example, in the last crisis, Bank of America was asked to take over Countrywide and Merrill Lynch.  These two banks would have been insolvent but the larger Bank of America never was not, although the costs it bore to make that acquisition probably totaled over $100 billion.  This brings up the issue that occurred in 2008, that there were too few large healthy banks to take over the unhealthy.  Bank of America could take on two of them but not the next one, Lehman Bros.  No others were healthy enough to step up to the plate. It may be that we need to ensure that there are a sufficient number of large banks to act as a pool of acquirers.  What that number should be would merit some study.

Another factor that minimizes costs to taxpayers is banks’ access to capital, or their ability to attract private capital through share sales in case of need.  US banks currently have strong capital thanks in part to their ability to issue shares in 2009 as investors looked forward to a return to healthy profitability when the crisis blew over.  One could question whether investors would be so eager again, now that increased capital requirements and costs of regulation have halved the returns of equity the industry can earn.  Capital standard setters would do well to consider whether there is such a thing as too much capital, so that banks cannot cover their cost of capital.

In conclusion, considering the costs to taxpayers only in terms of the financial advances made in the immediate aftermath of a crisis is arguably too narrow.  The discourse would better be aimed at the broader question of costs to the economy, that affects all taxpayers.  The traditional regime of capitalization, regulation and using all means possible to maintain confidence in the system, while ensuring that the banking industry can take care of its really sick brethren and repay any taxpayers funds employed is the way to go.