A key risk management client asks this question: “What do you think about overnight index swaps compared to traditional Libor-based interest rate swaps?” This blog continues from our blog of June 7, 2010 on the “credibility crisis” in Libor and why sophisticated market participants should prefer OIS-linked swaps to Libor-linked swaps whenever the two markets offer a comparable efficiency in transaction execution.
On June 7, 2010, we posted a study comparing the London interbank offered rate compiled by the British Bankers Association (see www.bbalibor.com) with the same maturity Eurodollar interest rates published by the Board of Governors of the Federal Reserve in its daily H15 statistical release
The title of the blog is given here
van Deventer, Donald R. “Kamakura Blog: Default Probabilities and Libor,” Kamakura blog, www.kamakuraco.com, June 7, 2010. Redistributed on www.riskcenter.com on June 8, 2010.
and a link to the full text can be found here:
In that blog, we showed the huge differences between the two data series from the date of the last change in the Libor panel banks (May 2009) and May 27, 2010. At a one month maturity, the differences in the time series are striking, as we showed in this graph from the June 7 blog:
Although the gaps between the two sources of Eurodollar interest rates narrowed toward the end of the period, on average the H15 Eurodollar 1 month interest rate was 12.4 basis points higher than the “official” Eurodollar quotation with a standard deviation of 10.1 basis points. Given the miniscule spreads in USD interest rate swaps, these differences between two numbers that should be the same can only be described as “ginormous.”
We call the British Bankers Association Libor quotation “official” only because www.isda.org specifically identifies BBA Libor as the basis for US dollar interest rate swaps where the floating rate is a Eurodollar deposit cost. For most other US dollar rate indices, the H15 statistic is the official ISDA source. That is the case with US dollar overnight index swaps, where the H15 statistic for the US domestic Federal Funds rate is the “official” basis for OIS swaps. An OIS swap differs from the floating side of a Libor based swap in that interest is compounded daily at overnight rates that prevail during the floating rate period, rather than the rates that prevailed at the start of the period with payment in arrears like Libor rates. ISDA states the US dollar OIS interest payment calculation as follows:
We now can address the question posed in the first paragraph of this blog:
“What do you think about overnight index swaps compared to traditional Libor-based interest rate swaps?”
There are many thoughtful commentaries on the internet on this topic, many of which can be paraphrased along the lines of “OIS makes a better index because the index does not have the credit risk distortion that is embedded in term Eurodollar deposit costs as surveyed by the British Bankers Association.” We think this explanation is somewhat imprecise and overlooks the many other more important reasons for choosing a swap indexed on OIS rather than Libor. Here are the reasons why, given equivalent quality of execution, one should always prefer the OIS index to Libor:
1. The OIS index reflects real transactions and Libor does not.
The British Bankers Association website www.bbalibor.com makes it very clear that their Libor quotation is simply an indication of funding costs, not the rates at which funding has actually taken place. The H15 federal funds rate, by contrast, represents the weighted average of all overnight borrowing and lending in the US markets on that day. The OIS represents real trades.
2. Libor panel members have an opportunity to falsify their quotations without penalty, but this is not true with the OIS index
Our June 7 blog cites many sources regarding the reasons why Libor panel members have an incentive to mis-state their true funding costs and anecdotal evidence that these mis-statements have taken place. Personally, this author thinks that the graph above is compelling evidence that the mis-statements have been large and persistent for long periods of time, understating true funding costs by 12.4 basis points on average with a 10.1 standard deviation as measured by the Fed’s Eurodollar cost estimates. Market participants who know this is happening can (and presumably do) take advantage of market participants who ignore this phenomenon. Since the OIS is based on actual trades, it is impossible for a similar falsification to take place.
3. The British Bankers Association has been aware of this issue for some time and has not acted to correct the problem.
Given the length of time that these distortions in Libor have persisted and the wide publicity that the problem has gotten, it is quite clear that the BBA has been in no hurry to address this issue. Since the market participants who benefit from these distortions have a major influence as members of the BBA, this should not be surprising to anyone. Again, this issue does not apply to the OIS.
4. The participants in the US Dollar Libor panel are overwhelmingly not US institutions with a US dollar deposit base.
Since May, 2009, the US Dollar Libor panel has consisted of 3 US banks and 13 non US banks. The US banks are
- Bank of America
- Citibank NA
- JP Morgan Chase
and the non-US banks are
- Bank of Tokyo-Mitsubishi UFJ
- Barclays Bank PLC
- Credit Suisse
- Deutsche Bank AG
- Lloyds Banking Group
- Norinchukin Bank
- Royal Bank of Canada
- Royal Bank of Scotland Group
- Societe Generale
- UBS AG
- WestLB AG
The OIS overnight funding cost, by number of participants, is generated by a group of banks whose country of domicile is overwhelmingly the United States. What difference does this make? Since much of the movement in bank costs of funds over the last four years in the dollar market has come from movements in US home prices and related loans and securities values, one would expect US-based bank funding costs to more precisely reflect these movements than a firm with a headquarters in Tokyo or Zurich.
5. The Libor panel cuts off the “high and low” quotations but the OIS includes all transactions.
There are 16 banks on the US dollar Libor panel, and Libor is calculated by averaging the 8 quotes remaining after throwing out the 4 highest and 4 lowest quotations. On the one hand, this helps to reduce the impact of firms understating their true funding costs by throwing out the lowest quotes. On the other hand, it clearly hasn’t been effective enough. The OIS does not suffer from the falsification problem and the inclusion of all fed funds transactions provides the security of the law of large numbers.
6. The OIS reflects a deeper market, the overnight interbank market, than the term funding market that Libor is supposed to reflect.
By number of transactions, the fed funds market has much more depth (many more actual transactions) than the term funding markets. This is a major reason why the OIS reliance on the fed funds rate is a more accurate index of funding costs than Libor
7. Credit risk reflects both markets but in opposite ways. OIS more accurately captures the impact of credit risk on funding costs.
Higher credit risk, all other things being equal, will result in higher fed funds rates unless offset by government intervention. In the Libor market, however, higher credit risk gives panel participants more incentive to understate their funding costs as the graph above clearly shows. This is the opposite effect of the conventional wisdom for advocating the OIS over Libor, but the conclusion is the same: OIS more accurately reflects the true impact of changing credit risk on market conditions.
In short, our view of the Libor market is very much like our view of the credit default swap market. There are a few market participants who control the Libor index and who control credit default swap quotations. As cited in our blogs on CDS market volume, less than 100 of the more than 2,000 corporate names on which CDS are quoted are actively traded.
This means the same thing in both the Libor markets and CDS markets. Some market participants assume that Libor and CDS quotations are “real,” regarded with a reverence like inscriptions on a golden tablet carried by Moses himself. A much smaller number of market participants knows which of these numbers are real and which are not. This is a license to steal, or, um, profit. As one of the first employees of Kamakura, David Kuo, once commented, “Arbitrage is the process by which people who are smart take money from people who are not.”
All other things being equal from an execution quality point of view, the OIS index is much harder to manipulate than Libor. Using Libor as an index, as the numbers above show, puts one in the same position as Mr. Smith in our blog
van Deventer, Donald R. “Mr. Smith goes to Wall Street and Buys a Rolex,” Kamakura blog, www.kamakuraco.com, May 25, 2010. Redistributed on www.riskcenter.com on May 26, 2010.
Using Libor as an index is accepting an invitation to “victimhood.”
Donald R. van Deventer
Honolulu, October 5, 2010