The Basel III capital regulation reforms introduced a non-risk based leverage ratio for banks. A new ECB paper shows that a leverage ratio requirement may have unintended negative consequences. It encourages banks that specialize on low-risk lending to raise their share of high-risk loans. And it could make overall bank portfolios more similar, increasing the contamination risk from negative surprises to expected default risks.

“Does a Leverage Ratio Requirement Increase Bank Stability?”, Ilkka Kiema and Esa Jokivuolle
ECB Working Paper Series, no. 1676, May 2014
http://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1676.pdf

The below are excerpts from the paper. Cursive text and highlights have been added.

What is a leverage ratio requirement (LRR)?

“In the aftermath of the global financial crisis, the international rules harmonizing banks’ capital requirements have gone through a major overhaul (the so called Basel III accord). To supplement the previous risk-weighted capital requirements, a non-risk-weighted leverage ratio requirement has been added as a new element. The Basel III recommendation is that a bank’s capital to assets ratio (including off balance sheet items) must be at least 3%. A central motivation to the leverage ratio requirement (LRR) is to provide a more robust capital buffer against the model risk that banks and regulators may get the risk-weights wrong, as happened before the crisis.”

“The Basel Committee on Banking Supervision argues that a LRR would help contain the build up of excessive leverage in the banking system, introduce additional safeguards against attempts to game the risk based requirements, and help address model risk”.”

How has the ECB paper analyzed the LRR?

“We analyze a model of a competitive banking sector in which the risk-weighted capital requirements, which are based on the bank’s internal customer credit ratings, have been supplemented with a leverage ratio requirement (LRR)…We assume that there are two kinds of firms, called low-risk and high-risk firms, a competitive banking sector, and a government which regulates banks…The amount of capital that the internal ratings based (IRB) approach requires for each loan…is a function of the default probability of the loan….The bank is also subject to a LRR which states that the bank must have [a fixed specific proportionate] amount of capital per loan.”

How does the LRR affect bank portfolios and related risks?

“As Repullo and Suarez (2004) have shown, when the IRB requirements are the only capital requirements in the model, banks have an incentive to specialize in either low-risk or high-risk lending. This is because – as banks have the obligation to use not just their capital but also their interest income for covering the losses from the defaulting loans – there is a positive probability that one of two specialized financial institutions fails and the other one does not. In this case the owners of a ‘mixed portfolio’ bank would have to use income from high-risk loans for covering losses from low-risk loans or vice versa. Hence, in order to take full advantage of limited liability, banks prefer to specialize”

“Because banks view (equity) capital as the relatively costly form of financing their lending, the LRR is punitive to banks specializing in low-risk loan customers which may require less risk-weighted capital than the 3% required by the LRR….We show that the LRR might induce banks with low-risk lending strategies to diversify their portfolios into high-risk loans until the LRR is no longer the binding capital constraint on them…[As a corollary] banks previously specialized in high-risk loans will adopt some low-risk loans [to the extent that they are not sold to non-banks] so that consequently, low- risk loans will be held by a larger number of banks and there will be fewer banks specializing only on high-risk loans.”

“In the new equilibrium in bank loan market, formerly low-risk lending banks would add some high-risk loans in their portfolio while banks with formerly high-risk lending strategies would assume some of the low-risk loans. As a result, bank portfolios would become more similar with one another, and hence more correlated.”

“If a model risk materializes in some loan category, meaning that loan default rates turn out to be much higher than the expectation according to which loans were priced and regulatory risk-weighted were set, the problem could now affect a larger number of banks. This is a consequence of the more similar portfolios…[Hence] in the presence of model risk, modelled as an unanticipated shock to the default probability of loans, a relatively low like the current 3% LRR might even reduce bank stability, counter to regulatory intentions.”

“If the model risk is associated with low-risk loans, bank stability would be reduced if the model risk were severe. This is because for a sufficiently high model risk the beneficial effect from spreading (the seemingly) low-risk loans to a larger number of banks is dominated by the effect of contaminating a larger number of banks by the assets turned ‘toxic’. If the unanticipated model risk concerns high-risk loans default probability, then the current moderate LRR (almost) always increases bank failures as a result of the contamination effect.”