Seven Principles for Reforming Financial Benchmarks

A number of regulators have been looking at how to dig financial markets out of the Libor mess. There isn’t much disagreement that bank-run benchmarks based on made-up numbers that can be readily manipulated have to be very significantly reformed. The problem is figuring out how to reform and what to replace them with, particularly given the large variety of benchmarks potentially at stake. It isn’t possible to simply abandon these benchmarks and parties to contracts have continued to use the existing ones, well-known warts and all, because they need something.

The International Organization of Securities Commissioners (IOSCO) is the latest body to weigh in on what to do. We were invited to comment on it. You can read what we said here. We used this as an opportunity to elaborate on our earlier analysis of the Wheatley Review findings. We make seven key points:

1. First, whenever possible benchmark indices should be based on actual transactions as opposed to quotes. This offers a number of advantages, not the least of which being its robustness (though not immunity) to fraud and manipulation. Where a transactions basis is not viable, the second-best alternative is to use committed quotes. The final option, uncommitted quotes of the type used for LIBOR, should not be adopted.
2. Second, some care should be taken in the manner of calculating the benchmark given the input data (either transactions or committed quotes). While averages are a common and certainly reasonable choice, there are alternatives – taking medians or modes for example – which may be more robust to manipulation.
3. Third, it must be understood that following the LIBOR scandal and associated investigations and liabilities, voluntary participation in a quote-based benchmark is unlikely to be attractive. Compulsory participation may be required, as tied to market participation.
4. Fourth, calls for “full transparency” – either of the calculating methodology or of the input data themselves – may be misplaced. Transparency facilitates cheating, and while some disclosures will certainly be necessary to inspire market confidence in a benchmark, full disclosures should be limited to the maximum extent possible.
5. Fifth, robust governance and supervision should be adopted to minimize the possibility of fraud and manipulation. The Administrator of the benchmark should be demonstrably disinterested in the benchmark itself. This may preclude public agencies from acting as Administrator, as well as trade associations, among others. Also, it must be recognized and accepted that submitters will almost always have material interests in the final benchmark result. Those interests must either be offset, as tends to happen naturally with transactions data, or proper incentives must be put in place to dissuade fraud, as can happen with an appropriate commitment mechanism for quotes.
6. Sixth, there is little to no role for regulatory direction of the content of the benchmark. The benchmark calculation methodology should be the proprietary interest of the private Administrator with an objective of reliability, continuity and accuracy. Regulation on how inputs must be formed and how the benchmark must be calculated risks introducing a wedge between the benchmark and the market it is meant to service. And it shouldn’t be strictly necessary if care is taken to base the index on proper inputs and if robust governance controls are adopted.
7. Finally, despite all the best efforts mentioned above, any benchmark should be routinely screened for attempts at manipulation. There will always be parties with very deep, material interests in the outcome of a benchmark index. No type of input data is immune from manipulation. No calculation methodology can be totally robust against fraud. No governance structure is perfect. Whatever structures are adopted, it is critical that ongoing reviews – so-called screens – take place to enhance the integrity of the benchmark.