After the first allegations of LIBOR manipulation in May 2012 – which eventually resulted in investigations into banks and individuals in various countries as of June 2012 – the reliability and credibility of unsecured reference rates in various currencies (the LIBOR in pounds, dollars, euros, and yen, and also the EURIBOR) have been severely questioned. With a view to restoring the credibility of these important reference interest rates, financial regulators have launched a broad consultation to study possible options to avoid similar manipulation in the future. There has been an intense debate between defenders of a low-key approach to reforming money-market references and the supporters of a more drastic overhaul. Several questions arise: Do we need reference interest rates? How/why did manipulation occur and what can we learn from it? And which short- to longer-term adjustments could be envisaged?
Do we need reference interest rates?
The answer to this question is emphatically yes, for at least three reasons.
First, given the traditional interaction between the secured and unsecured segments of the money market through hedging activities, it is important for both market participants and central bankers to be able to observe changes in market conditions following certain types of events. Since most cash and derivatives segments of the money market are over the counter – hence reducing considerably the quantitative information available on transaction volumes – it is useful to have a reliable reference rate which summarises market conditions for these transactions.
Second, although various indicators suggest that the importance of unsecured funding for banks has changed substantially over time (see Figure 1), the benchmarking role of the LIBOR and EURIBOR for the derivatives segment has strongly increased. Figure 1 shows that wholesale interbank unsecured operations (all maturities included) represented at most 15–20% (in Austria and Belgium) of banks’ balance sheets in the second quarter of 2012, with less than 10% for Germany, France, and the UK (for both deposits and loans, on average). At the same time, if we cumulate over the same period short-term borrowings by banks through both secured and unsecured instruments, they appear more important in various countries, including in Asia, Europe, and North America – reaching a peak of 40% of banks’ balance sheets in France.
Figure 1. Wholesale interbank unsecured operations as a % of total liabilities in Q2 2012
Source: SNL Financial.
However, the key issue is not with cash, but with derivatives instruments. The daily turnover in over-the-counter derivatives as calculated for the Bank for International Settlements’ triennial central bank survey on derivatives (and composed for the most part of LIBOR- and EURIBOR-indexed interest-rate swaps) has almost tripled in ten years, with a daily turnover of around $2 trillion in 2010. We therefore face a paradoxical and potentially threatening situation, in which immense positions are taken through very large volumes of derivatives instruments that rely on benchmarks, the basis of which has gradually been eroded.
Last but not least, existing reference interest rates for the unsecured market are still used as a standard to set financial asset prices, despite a decreasing turnover in the underlying market. In some countries, a substantial proportion of total loans to non-financial corporations and to households are floating, with their interest rates referenced mostly to benchmark interest rates like LIBOR or EURIBOR. The importance of reference rates thus goes well beyond the very short end of the yield curve.
How/why did manipulation occur and what can we learn from it?
As nicely explained by Persaud (2012) on Vox, any manipulation of these reference interest rates necessarily requires the coordination (or collusion) of a significant number of panel member banks sharing the same characteristics (i.e. having similar net positions). The names of the banks having reached settlements with the US and UK financial regulators and the US Department of Justice confirm this view.1
The motivation behind the manipulation could be at least twofold. On the one hand, attempts to make abnormal profits from trading positions – including those in non-cash segments (especially derivatives) – are certainly one important consideration. On the other hand, reputational considerations – like the reluctance to trade with counterparties – cannot be totally ruled out. The latter consideration has surely been strengthened by the market distortions implied by the 2007–2013 financial crisis, which led to a substantial reduction of market activity in the unsecured segments of money markets. Empirical evidence shows that market activity in term unsecured interbank segments significantly declined as of August 2007 (see Figures 2 and 3). Kuo et al. (2013) provide evidence of a declining trend in daily turnover of interbank transactions in the US money market – especially for transactions beyond the one-month maturity, which further decreased on account of the intensification of the crisis with the collapse of Lehman Brothers on 15 September 2008. A similar picture for the Eurozone is provided by Figure 3, based on a survey conducted by the ECB, which shows that market activity in that segment of interbank unsecured loans/deposits vanished from 2009 on.
Figure 2. Daily turnover in the US money market
Source: Kuo et al. (2013).
Figure 3. Maturity breakdown for average daily turnover in Eurozone unsecured lending (index: cash lending volume in 2003 = 100)
Source: European Central Bank (2013).
Evidence of collusion
In Brousseau et al. (2009), we already provided statistical evidence of such manipulation or collusion. The starting point of the analysis was the unlikely lower dispersion observed in the daily return (i.e. daily change) of the three-month EURIBOR fixings against that of the three-month Euro OverNight Index Average (EONIA) swaps as reported in Figure 4. Figure 4 shows that – as may be expected in a volatile environment – the three-month EONIA swaps kept a quite disperse distribution of daily returns over the sample under review, but the distribution of the daily returns for the EURIBOR was much more concentrated. Since the pricing of both instruments should be very close in practice through hedging activities, this obvious disconnection of their daily returns could not happen by chance – suggesting that pricing did not reflecting market conditions. The resulting suspicion was then confirmed by the econometric analysis, which yielded the following, more detailed, results:
- First, the usual relationship (Granger causality in econometric terms) between the three-month deposit interest rate and the three-month swap index is not observed anymore during the financial crisis – especially after the collapse of Lehman Brothers.
- Second, the breakdown observed in the relationship between the three-month deposit rate and the three-month overnight index swap seems to be explained by an observable linear pattern (characterised by the presence of a deterministic trend). More specifically, it was shown that the three-month deposit interest rate decreased systematically on a daily basis, which could be captured by a trend variable.
- Third, this different profile in the distribution of daily returns of both the deposit interest rate and the overnight index swap rate would only appear on the basis of a random stochastic process with an extremely low probability.
Figure 4. Histogram of the daily returns on three-month EURIBOR and EONIA swap indices (October 2008–July 2009)
Source: Figure 13 in Brousseau et al. (2009).
Which adjustments could be envisaged from a short- to longer-term perspective?
The ongoing legal investigations that started in May 2012 confirmed our suspicion of price manipulation (or more broadly coordinated pricing) based on our previous results obtained in 2009. Today, the crucial issue is to rehabilitate credible pricing benchmarks in money markets, since it is commonly acknowledged that they are useful to anchor the pricing of financial instruments. Some participants in the public debate appear quite embarrassed (or reluctant) to acknowledge the limited representativeness (and hence value) of the benchmark fixings based on the wholesale unsecured segments of the money market. This conclusion would force them to recognise the need for a more substantial overhaul – potentially implying legal complications and financial redenomination risks hanging over the trillions in notional amounts of financial contracts referenced on the legacy indices. Our view is that a substantial redefinition (or even change) of the current reference rates is needed today – not only because they are subject to manipulation, but also because the funding of financial institutions has changed over time. The problem is therefore twofold:
1. Finding a suitable successor to the LIBOR/EURIBOR indices;
2. Handling cautiously the phasing-out of legacy indices in such a way that the huge market risks for the trillions in exposures are avoided.
Among the possible candidates, one possible approach would consist in transforming the current definition of the LIBOR/EURIBOR into a flexible trade-weighted average one. In this approach, fixings would not simply reflect quotes at one point in time in a day, but would reflect the banks’ wholesale funding cost based on real transactions over one trading session. While the merits of this proposal are obvious, it assumes that market transactions take place at sufficiently regular frequency (at least daily) in order to ensure continuity in the time series of the reference rates. The assumption that the underlying level of trading in the short-term unsecured wholesale market will again be sufficient to support the production of reliable and unequivocal money market fixings in future can be easily challenged in case of market fragmentation, as the recent financial crisis has reminded us. To be a reliable, suitable, and resilient reference rate, the successor to current fixings must have at least five key characteristics. As discussed in the Bank for International Settlements’ report, these are: reliability, robustness, frequency, availability, and representativeness – all of which presuppose that the new candidate will be based on an active market segment (see Bank for International Settlements 2013).
In a recently published study (Brousseau et al. 2013), we therefore discuss the secular trends at play in the functioning of money markets and banks’ funding over the last 20 years, and highlight the increasing challenges posed by these trends to the production of a reliable unsecured money market index. Following the evidence reported in Brousseau et al. (2009), the practical feasibility of a narrowly trade-weighted approach for new reference rates is clearly questionable. Our main argument comes from the evidence of a decreasing trend in market activity in the term unsecured segments of money markets, as already shown by Figures 2 (for the US money market) and 3 (for the Eurozone). Based on this observation, we review in Brousseau et al. (2013) other possible candidates to replace LIBOR/EURIBOR benchmarks in the future. These range from a gradual migration to the overnight index swap indices (OIS, Option 1) to what we call a total cash pool index (TCPI, Option 2) benchmark. We see merits in both options, but they are mutually exclusive. Option 1 (the overnight index swap migration) is self-explanatory. By contrast, Option 2 (the total cash pool index migration) deserves further clarification.
In practice, Option 2 would consist of a systematic calculated pooling of all short-term wholesale funding operations of banks. According to these authors, such a benchmark reference would be holistic and would not be centred on the exclusive specific nature (unsecured vs. secured) of the underlying funding instrument, but would focus instead on the cost of the overall funding mix for wholesale operations below one year. For instance, it would, for a given maturity, aggregate all the short-term borrowing operations of money-centre banks and calculate one single trade-weighted average rate encompassing: unsecured interbank borrowing, issuance of certificates of deposit, general collateral repo operations involving different types of collateral, securities lending, and cross-currency funding operations through foreign exchange swaps. In a nutshell, any transaction involving a cash leg bearing interest would be pooled within a corresponding maturity bucket to calculate this synthetic index. This collection of information could be required for instance by the statistical officer and/or regulator in each country.
The overnight index swap migration proposal would have technical merits. It would facilitate the pricing, hedging, quotation, and computation of margin calls for the core over-the-counter instruments, both in conceptual terms and by leading to more tractable calculations.
The total cash pool index migration proposal would have many advantages. First, its large underlying base would make it an instant benchmark, devoid of any risks of manipulation. Second, its use for indexation could facilitate banks’ asset-liability management, naturally reducing risk mismatches. Third, the natural diversification of the sources of wholesale funding that it would bring would support, on both sides of the balance sheet, the convergence towards the new Basel III prudential liquidity regulations.
The LIBOR scandal that emerged in May 2012 demands appropriate answers. This necessarily involves finding a successor which can be used in a reliable manner in all circumstances and thus not lead to controversy amid reduced market activity. With the benefit of hindsight, the recent crisis showed that the overnight index swap market continued to work, while banks continued to make secured short-term transactions. Therefore, we believe that a migration to the overnight index swap indices is possible. At the same time, establishing a more ambitious benchmark like the total cash pool index would be useful in order to have a precise view of banks’ funding costs. Of course, the latter option – however desirable – could not be implemented overnight. But the legacy aspect of the debate should not undermine the chances of a successful and rapid transition towards an effective solution. This apparent difficulty must not prevent policymakers from addressing this possibility in an open manner to increase transparency in the financial industry. A serious discussion on the options at hand is more necessary than ever, but we should recognise that the success of the new candidate will also depend on its adoption (or not) by the financial industry.
Disclaimer: The views expressed in this column are solely those of the authors and do not necessarily reflect those of the European Central Bank, the Eurosystem, or the International Monetary Fund. This column should thus not be reported as representing the views of these institutions.
Bank for International Settlements (2013), “Towards better reference rate practices: a central bank perspective”, March.
Brousseau, V, A Chailloux, and A Durré (2009), “Interbank offered rate: Effects of the financial crisis on the information content of the fixing”, Lille Economie and Management Working Paper 2009-17.
Brousseau, V, A Chailloux, and A Durré (2013), “Fixing the Fixings: What Road to a More Representative Money Market Benchmark?”, IMF Working Paper 13/131, May.
European Central Bank (2013), “Euro Money Market Survey”.
Kuo, D, J Vickery, D Skeie, and T Youle (2013), “Identifying Term Interbank Loans from Fedwire Payments Data”, Federal Reserve Bank of New York Staff Report 603, March.
Persaud, A (2012), “Notes on a scandal: Libor”, VoxEU.org, 21 July.
1 To date, Barclays has paid a combined fine of GBP 290 million in June 2012, UBS a combined fine of USD 1.5 billion (including the fine to Swiss authorities) in December 2012 and Royal Bank of Scotland a combined fine of GBP 390 million in February 2013. More recently (October 2013), Rabobank agreed to pay a combined fine of USD 1 billion to US, UK and Dutch authorities, the second largest one after that paid by UBS.