Saturday, January 25, 2014

Designing the Monetary Policy Committee

The establishment of a proper Monetary Policy Committee is central to a well functioning monetary policy process. There is a lot of interest in alternative designs of the Monetary Policy Committee. In this post I analyse a few designs. Before we get to that, let's think of the foundations.

One judge or a bench?

You are accused of a crime. The prosecution is going to make its case at a court and you're going to argue that you are innocent. What would make you more comfortable? One judge or a bench of judges?

Almost instinctively, our answer is: A bench of judges. We would feel more unsafe if there was just one judge. Why?

  1. We might be unlucky and that one judge might be a bad one.
  2. A well intentioned and sensible judge might just make a mistake.
  3. The independent thinking of one judge might be compromised by other considerations. It is harder to contaminate the thinking of an entire bench.

For these three reasons, a bench always does better than a single judge.

Institutional capacity for monetary policy as opposed to individualistic monetary policy

In everything that we do in public life in India, we aspire for State structures to work in an impersonal way. Individuals come and go, but the institutional apparatus of the Indian State must exhibit consistency and predictability.

If monetary policy is controlled by the Governor, then it becomes personalised. Our job is to establish a framework through which the monetary policy strategy is consistent and predictable across 50 years.

When monetary policy is controlled by the Governor, every action is identified with him. This makes it  harder to do unpopular things. The Governor will have too much of a temptation to play to the gallery, or score political capital by doing the bidding of politically powerful people. If we want independent thinking, the rate setting should be done by the committee where the Governor is more important than other members, but not much more.

Combination of forecasts, or portfolio diversification

Ever since Bates & Granger, 1969, we have known this formally in the field of forecasting. A combination of forecasts is, in general, better than one forecast. The optimal combination gives a greater weight to a better forecaster. But even naive combinations are better than any one forecast.

The intuition is much like portfolio diversification. The diversified portfolio is always superior to an individual security. Each forecaster makes mistakes. As long as the forecasts are uncorrelated, the mistakes tend to cancel out, and the portfolio outperforms the individual security.

Design 1: The Governor makes monetary policy

This is a solution with one judge. This is a bad design for so many reasons:

  • This is one forecaster making predictions.
  • This is a portfolio with one security.
  • We might be unlucky and have a bad Governor.
  • A well intentioned and sensible Governor might be wrong.
  • Under this arrangement, monetary policy is personal. Every time a new Governor shows up, there is a significant change in the nature of monetary policy. This generates policy risk. In contrast, a good MPC arrangement gives greater consistency and predictability.
  • Politicians know this one person can deliver the desired outcome, so when elections are approaching, there will be extreme pressure on this one person to deliver a rate cut.
  • The Governor is tempted to accede to these pressures, earn IOUs, and pursue his personal interests.

It's a bit worse than this in one setting:  the last 1 year before the date of retirement of the Governor. Monetary policy acts slowly. On a horizon like 1 year, the impact upon the economy, of a change in the policy rate, is small. Hence, when a Governor has a year to go before his term ends, accountability goes down sharply. A Governor can cut deals on the policy rate and not get caught out.

This arrangement is good for the Governor and bad for the country. This is today's RBI.

Design 2: An MPC with the Governor and two of his juniors

In this design, the MPC has three persons -- the Governor and two of his juniors.

Does this buy us a lot? The juniors are unlikely to challenge their boss.

For all practical purposes, this collapses into Design 1. We have not improved on Design 1.

This is the MPC you show the world, in order to claim we are on international best practice of having an MPC, but in truth, nothing has been done.

Design 3: The Governor appoints three independent experts to the MPC

In this design, the MPC is six people: two juniors of the Governor, and three independent experts appointed by the Governor.

The persons who are chosen by the Governor will feel they owe him something and will avoid disagreeing with him.

The persons who are chosen by the Governor are likely to have a shared view of the world with him. This will give correlated forecasts. The gains from combination of forecasts will be limited as all the six forecasts are much like each other.

The Governor has 3 votes in his pocket. For him to have his way, he has to only persuade one out of the three external members. All three are similar to him in worldview and all three owe him. In almost any situation, with almost any proposal, the Governor will be able to go down on bended knee and persuade one of the three external MPC members to go with him.

Hence, while this looks like the show of an MPC, it's actually just Design 1.

Similarly, giving the Governor a veto takes you to Design 1. There is no MPC in the world where the Governor has a veto.

Design 4: The Ministry of Finance appoints all the six members of the MPC

The Ministry of Finance, in any case, appoints the Governor and his juniors. Suppose we have an MPC where the three external members are also appointed by the Ministry of Finance.

The three external members will not owe their appointments to the Governor. This will generate greater independence in thinking and voting.

The three external members are likely to be less ingrained in the Governor's way of thinking. There will be less group-think; their forecasts will be less correlated with those of the Governor.

For any idea, the Governor still needs just one of the three external MPC members to side with him, and he's won the vote. But a few truly bad ideas will get shot down.

Design 5: Give the Governor only two votes out of six

In this design, only one of the juniors of the Governor is on the MPC. There are four external MPC members. All six are appointed by the Ministry of Finance.

The four external members do not owe their position to the Governor. And, they are not part of the group-think of the central bank.

In order to win a vote, the Governor needs 4 votes. He's got two in his pocket. He has to persuade at least 2 out of the 4 external members.

He will not be able to push through ideas where even half of the externals aren't persuaded. This sounds right.

The combination of forecasts will be best when the four external MPC members bring diverse thoughts to the table which are less correlated with those of the governor. One way to assist this is to have the four persons from the four regions of the country -- as is done in the US and in the ECB. Each one will bring a distinctive perspective about economic conditions all across India.

Odd numbers and ties

Assuming all members are in a meeting, it's simplest if the size of a bench is always an odd number so as to eliminate ties. When I use examples like six member committees above, this is faulty on this respect.

Even when an MPC has an odd number of members, if an odd number of persons is absent, we're left with an even number of persons present. This can be solved by giving the chairman a casting vote. But the same kind of reasoning (about power) should be conducted in those settings. E.g. consider an MPC with Governor + 1 internal + 2 externals + casting vote for the Governor. In this, the Governor can never lose, so it's just Design 1.

Feedback loops and the quality of external members

For external members, a considerable investment in time and trouble is required in being a good MPC member. This is related to the role that they play on the committee.

If the Governor has overwhelming power in the MPC meeting, this takes away the incentive for an MPC member to invest in figuring out stuff. Why bother, when your vote is irrelevant? For all members to go into a meeting with careful preparation, meetings have to have consequences, and every vote has to matter.

A well structured MPC creates feedback loops where external members build the knowledge and get better as they go. A badly structured MPC creates a low level equilibrium where external members tend to not acquire the requisite knowledge.


Many different designs of the MPC can be envisioned. This article shows how to think about the alternatives. The key questions to ask are:

  1. How many of the externals does the Governor need to win the day?
  2. How hard will it be for him to go down on bended knee and woo the externals? Do they owe him?
  3. How uncorrelated is the thinking of the persons on the MPC? Are we successfully bringing in diverse perspectives?
  4. How will monetary policy work when the Governor has only a year to go before he leaves, a period in which there is a sharp breakdown in accountability owing to the lags in monetary policy? 
  5. How to create an arrangement where there is stability in the thinking of the MPC, across personnel changes?

1 comment:

  1. Very well argued. An alternative is to neither have one person nor have a committee of people, but to just shift to a rules-based monetary policy regime whereby monetary policy instruments are exercised predictably to meet clearly established policy goals, based on pre-defined triggers. An NGDP or nominal income target or even inflation target coupled with a standard policy rule, such as the Taylor rule or the modified Evans rule, would be a huge improvement to the non-transparent way in which monetary policy is set currently in India. At least that way, we would know the target, and the instrments that will be ued, will build better accountability of the central bank.
    In summary, while I agree it would be an advance to go from the decision of one individual to a committee, I don't think it would go far enough


Please note: Comments are moderated. Only civilised conversation is permitted on this blog. Criticising me is perfectly okay; uncivilised language is not. I delete any comment which is spam, has personal attacks against anyone, or uses foul language. I delete any comment which does not contribute to the intellectual discussion about the blog article in question.

Please note: LaTeX mathematics works. This means that if you want to say $10 you have to say \$10.