Finally RBI seems to have figured out the banks and market players shouldn’t allowed to set benchmark rates based on “polls”. They’ve a draft report out on how benchmark setting should be changed to use actually traded data rather than asking people what the rate should be.
The LIBOR scam was on the principle that the LIBOR rate was set by asking a few banks on the phone on what they would demand and offer as interest rates, and those responses were collated and the rate determined. This rate was manipulated downwards by those very banks, because those banks had positions on a direction that was benefited by their giving a different rate from what the market demanded.
In India, many benchmarks continue to have “polled” inputs. The RBI ref rate, which is what RBI publishes everyday around 1 pm, is a rate they get by calling banks and asking for their rates.Overnight MIBID/MIBOR (Mumbai Interbank Bid and Offer interest rates) are set by polling. Yield curves are set through polling of quotes of various tenures of securities.
These rates are not just for benchmarking. They are used to settle transactions. The MIBOR is used to settle interest rates swaps (called OIS) which currently have a notional outstanding of 16 trillion rupees (16 lakh crore).
But don’t gape just yet, swaps have a large notional but a very tiny net impact – I can borrow 1 billion rupees (100 cr) at 10% and lend at (floating) 9.5% and the difference is just 0.5%, or 5 million (50 lakh). The notional is very high in comparison.
But the point is that the rate for settlement for any OIS is currently obtained by polling banks. Instead, it should be based on what participants actually pay for overnight borrowing. The RBI is now agreeing that should be the case.
RBI recommends that:
- FIMMDA takes over Fixed income benchmarks, and FEDAI the forex benchmarks.
- MIBID/MIBOR rates to use actual trade data (volume weighted trades on NDS’s CCIL platform ) instead of polling.
- RBI Ref rates for the dollar/yen etc. to use actual trade data instead of polling.
- Don’t take “last traded yield” to set benchmarks, use volume weighted prices or yields instead.
- All other rates must move to use trade data where available instead of doing polls.
- Benchmarks should have codes of conduct for participants including integrity requirements, whistleblower protection, oversight, complaints etc.
Capital Mind View: While this is necessarily bad for certain powerful banks in the short term, such regulation is good for the market. Actual traded data is better than polls, and polls can only be used if enough trades are not taking place. Using polls is like asking me “What price would you buy Reliance at?” and then publishing the answer as the price of the Reliance stock. I could say anything I wanted, especially if a rupee here or there makes a difference to my profits. And in the face of a well traded Reliance share, it is downright stupid to ask me – just take the average prices of the market.