When does prudent bank leverage morph into a form of gambling? And why does it happen?
A new paper by researchers at Stanford Graduate School of Business sheds light on both questions, and comes to an uncomfortable conclusion: If regulators want to substantially reduce the costs of future bank failures they need to force banks to at least double the size of the higher capital cushions that regulators already are imposing because of the 2008 financial crisis.
The research is by Ilya Strebulaev, associate professor of finance, and Will Gornall, a doctoral student, both at Stanford Graduate School of Business.
In a striking new model of how banks approach risk, Strebulaev contends that bank leverage is driven by two sets of forces. The first is what he calls the "CAN" forces, which refers to the ability of banks to safely hold much more debt than nonfinancial firms. The second is what he calls the "WANT" forces, which drives banks to ratchet up debt for competitive reasons — debt that is riskier for the bank but also for the banking system as a whole. U.S. tax policy, which heavily favors debt over equity, helps stoke the "WANT" forces. But government protections, notably deposit insurance and bailouts, push the craving even higher. Bank investors want this high leverage to help them save on taxes. However, society at large does not benefit from banks paying lower taxes and may pay the price for bank defaults.
Almost by definition, banks rely overwhelmingly on borrowed money to generate their profits. About 90% of U.S. bank assets are financed with debt, and that level of leverage has been typical for decades. (It's also typical for banks in most other countries.) By contrast, debt at nonfinancial firms has averaged around 30% of total assets in recent years.
Gornall and Strebulaev find that banks can afford to take on much higher debt, because bank portfolios — the loans they make to borrowers — are far more secure and less volatile than the assets of nonfinancial companies.
Part of the security comes from diversification — banks hold a wide diversity of loans that are very unlikely to go bad in large numbers at the same time. But the much bigger source of security is the bank's seniority as creditor. If a borrower gets in trouble, the bank is at the head of the line for getting paid. The combined effects of diversification and seniority mean that the value of a bank's assets is vastly more stable than that of a nonfinancial company.
The other and more troubling piece of the equation, however, is that banks want higher leverage for competitive reasons. That's because banks derive hefty "debt benefits." Because interest payments on debt are tax deductible but shareholder dividends are not, debt is a cheaper form of financing.
That, in itself, isn't a new idea. But Gornall and Strebulaev take the story further by showing that the tax benefits prompt banks to use leverage as a central competitive tool. The more debt and debt benefits a bank has, the lower it can reduce its interest rates to borrowers and the more business it can generate. The same logic applies not only to tax benefits but also to other benefits generated by debt such as banks providing liquidity to depositors.
"Banks compete with each other through their capital structure choice,'' Strebulaev says. "Banks with low leverage will be out-competed by banks with high leverage." To put it another way: Banks are much less sensitive to debt costs (the risk of losses to their loan portfolios) than to "debt benefits."
The big breakthrough, the researchers contend, is that their model establishes a quantitative relationship between capital requirements, bank leverage, bank risk taking, and the cost of bank defaults.
Based on an analysis of bank leverage and bank default rates, Gornall and Strebulaev conclude that keeping capital requirements at their current level will not help reduce future bank defaults. Under current international banking rules, banks need to maintain a capital buffer equal to about 8% of assets. But the researchers find a sharp drop in default risk only when a bank's equity capital climbs higher. Increasing the capital buffers to 20%, they predict, would reduce bank default rates by 80%.
Federal deposit insurance is another potentially big contributor to high-risk lending and high, future default rates. But the researchers noticed an important distinction: Deposit insurance has very little effect on bank leverage until insured deposits make up about 95% of a bank's total liabilities. However, if banks can game the system by making riskier loans, they will do so. "Most U.S. banks would like to gamble by loading up on loans with high value in good times and low value in recessions," Strebulaev warns, "That way, their shareholders get high profits in good times, and the government comes to the rescue in bad times." As a result, the researchers suggest higher capital requirements on banks that have high levels of insured deposits.
Strebulaev emphasizes that he isn't recommending a particular limit on bank debt or a minimum requirement for bank capital. But he contends that the cost of higher capital requirements, in terms of higher costs to borrowers, would be small relative to the potential gains in bank safety.
Ilya Strebulaev is a tenured associate professor of finance at Stanford Graduate School of Business, Stanford University, and a research associate at the National Bureau of Economic Research. Will Gornall is a PhD student in finance at Stanford Graduate School of Business, Stanford University.