VoxEU Column International Finance

Implementation of Basel III in the US will bring back the regulatory arbitrage problems under Basel I

Rejigging financial regulation is in vogue. But, in the world of international finance, how well do different regulatory systems join up? This column argues that the US Dodd Frank Act and Basel III are, in part, incompatible and that harmonising them may lead to unintended consequences. The US ought to tread carefully here but should also try hard to maintain the spirit of better financial regulation.

This column is a lead commentary in the VoxEU Debate "Banking reform: Do we know what has to be done?"

In the aftermath of the global financial crisis, many countries have been redesigning their financial regulatory frameworks. In the US, the Dodd Frank Act of 2010 specified the directions for new financial regulations. The US financial regulatory agencies, including those that were newly created by Dodd-Frank, have been busy writing and rewriting the rules.

At the international level, the Basel Committee on Banking Supervision has come up with the third implementation of the international standard for minimum capital regulation, ‘Basel III’. However, in several areas, the requirements of Dodd-Frank are apparently inconsistent with those of Basel III and thus US regulators face a difficult task of reconciling the two regulatory initiatives.

Harmonising regulation, unintended consequences

On 7 June this year, the US Office of the Comptroller of the Currency, the Federal Reserve System, and the Federal Deposit Insurance Corporation came up with three ‘Notices of Proposed Rulemaking’ on regulatory capital rules. These proposals entail new capital regulations for the US banks, reconciling apparent discrepancies between Dodd-Frank and Basel III.

I focus on one aspect of the Notices of Proposed Rulemaking from which there is likely to be a serious unintended consequence, if implemented. First, it is worth noting that the implementation of Section 171 of Dodd-Frank includes the following requirement:

“The appropriate Federal banking agencies shall establish minimum risk-based capital requirements on a consolidated basis for insured depository institutions, depository institutions holding companies, and nonbank financial companies supervised by the Board of Governors. The minimum risk-based capital requirements established under this paragraph shall not be less than the generally applicable risk-based capital requirements, which shall serve as a floor for any capital requirements that the agency may require, nor quantitatively lower than the generally applicable risk-based capital requirements that were in effect for insured depository institutions as of the date of enactment of this Act”.

Dodd-Frank sets “the generally applicable risk-based capital requirements” as a floor that regulated financial institutions must satisfy in addition to any other minimum risk-based capital requirements that the regulators may impose. “The generally applicable risk-based capital requirements” are defined as follows:

“For advanced approaches banking organisations, the regulatory capital requirements proposed in this [Notice of Proposed Rulemaking] and the Standardized Approach [Notice of Proposed Rulemaking] would be ‘generally applicable’ capital requirements for purposes of section 171 of the Dodd-Frank Act (August 2012 NPR).

Even those banks that calculate the risk-weighted assets using advanced approaches are required to hold enough capital required by the standardised approach. The problem is that the standardised approach is based on the old methodology of classifying assets into several risk buckets, which was originally used in Basel I regulation.

A flawed approach

The flaws of this approach have been noted by many researchers and practitioners1. Since the risk weights classification in the Basel I regulation was coarse, the same ‘bucket’ included the assets with very different risk levels. This led some banks to shift their portfolios to hold more risky (and hence higher return) assets within the same risk assets category, thereby increasing their risk without increasing regulatory capital. Because all the sovereign bonds of investment grade had the same risk weights (zero), banks were able to increase the return by increasing the holding of the most risky ones.

Because highly rated tranches of securitised loan products carried lower risk weights than individual loans, banks were able to economise on regulatory capital by selling the loans that they originated and by buying (highly rated) securitised loan products.

It is now well understood that the incentive for these regulatory arbitrages created by the Basel regulation increased risk in the banking system without a corresponding increase in risk-weighted assets – and hence regulatory capital2. The standardised approach in the Basel III Notices of Proposed Rulemaking include some improvements over the Basel I approach. For example, the risk weights classification is now finer. The classification is, however, still insufficient to make the risk-weighted assets sensitive enough to risks calculated by more advanced approaches.

Assigning risk is too ad hoc

Another problem is caused by the inevitably ad hoc nature of assignment of a risk weight to each category of assets. Because of this problem, creation of finer “buckets” can actually distort bank behaviors even more if the allocated risk weights differ from the risk differentials that banks perceive. For example, according to the Notice of Proposed Rulemaking on the Standardised Approach, residential mortgages are classified into eight buckets depending on the loan-to-value ratio and the type of loan and given different risk weights. Thus, the classification is finer than that in Basel I, which assigned 50% risk weight to all mortgage loans. Now a 30-year amortising mortgage with the loan-to-value ratio between 60% and 80% gets risk weight of 50%, while an interest-only loan with the same loan-to-value ratio receives increase the 100% risk weight.

This would be fine if the bank sees the interest-only loan as twice as risky as the 30-year amortising loan and requires twice as much capital. If that is not the case, the bank will have an incentive to reduce one type of loan and increase the other. For example, if the bank sees that the risk of the interest-only loans they originate are not quite twice as high as the risk of the 30-year amortising loans, the new capital regulation will discourage the bank from originating such interest-only loans. At the same time, some other banks may find the type of interest-only loans that they generate are actually more than twice as risky as their 30-year amortising loans. In that case, these banks will actually amount of interest-only loans.


The international Basel III allows banks to use the advanced approach to calculate the regulatory capital, so the banks in Europe and Japan that are qualified to use the advanced approach do not have the problem faced by US banks. The problem is that Dodd-Frank requirements for US banks set a floor of the “generally applicable risk-based capital requirements”. To avoid reviving the problems we know from Basel I regulation, US regulators should find a way around imposing the standard approach to advanced approaches banks, while respecting the spirit of Dodd-Frank at the same time.

Author’s note: The views expressed here are those of the author and do not necessarily represent those of the institutions with which they are affiliated.


Acharya, V V(2012), “The Dodd-Frank Act and Basel III: Intentions, Unintended Consequences, and Lessons for Emerging Markets”, ADBI Working Paper Series, 392.
Bruno, V, and H S Shin (2012), “Capital Flows, Cross-Border Banking and Global Liquidity”, AFA 2013 San Diego Meetings Paper, 1 December.
Dewatripont, M, J-C Rochet and J Tirole (2010), Balancing the Banks: Global Lessons from the Financial Crisis, Princeton, NJ, Princeton University Press.
Jones, D (2000), “Emerging Problems with the Basel Capital Accord: Regulatory Capital Arbitrage and Related Issues,” Journal of Banking and Finance, 24, 35-58.

1 See Jones (2000) and Dewatripont, Rochet, and Tirole (2010, Chapter 3) for example.
2 This is clearly shown in Figure 1 of Acharya (2012). Similarly, Figure 4 of Bruno and Shin (2012) shows this was the case for Barclays.

5,039 Reads