*University of Tunis and North Africa Bureau of Economic Studies, Tunis, Tunisia.
 

Let me first say at the outset that I will make my comments on this paper as an outsider to the subject. I don’t have any special expertise and most of what I know about it I learned from this paper. So my comments will be more general and relate to what someone who has not been well aware of the details of this subject might think of some of the issues raised and the conclusions reached.
 

The concept of CBI

I will start with some remarks about the concept of central bank independence (or CBI). I have found myself somewhat confused about the concept. For instance the various solutions or mechanisms which are considered to constitute "independence" for a central bank such as "a conservative central banker" or optimal contracts do not sound to represent "independence" in any meaningful way to me. A central bank (CB) working under such mechanisms sounds very much like a dependent institution with respect to government through for instance the setting of goals and objectives, or incentive schemes. In fact as Blinder has observed such solutions would be equivalent to directing or instructing the central bankers to behave as if the loss function has specific values of the parameters: for instance k=1 or w higher than that of society as a whole. The concept of "independence" seems to involve so much confusion which led to the introduction, discussed or alluded to in the paper, of at least 6 concepts (personal independence, policy independence, goal independence, financial independence, instrument independence, autonomy).

I do not want to go into the details of these various concepts and their relevance. But what I would like to observe is that what in fact seems to be the object or the essence of most of the discussion is how to "insulate the conduct of monetary policy from opportunistic short-term interference of government or even from the CB own temptations for short-termism". That in the long-run some significant dependence of the CB remains with respect to government is clear from the various arguments about accountability, need for overriding of central banker under some conditions, and so on. It seems to me that using the term "independence" does not help in clarifying or illuminating the discussion of the issues. But I am not sure what better term one might use. Some have suggested in this conference the term "commitment technology for a central bank"
 

Normative analysis

Let me turn now to the normative or theoretical analysis of CBI. A useful way to look at the arguments in favor of CBI is to interpret it as a solution to a government failure or to set it in the framework of transaction cost politics used by Dixit. The time inconsistency argument is essentially an argument about the failure of government or the political institutions to adequalely weight short term gains from opportunistic use of inflationary monetary policy against long term gains from of a commitment to low inflation. The inflation variability argument is a government failure argument also since it is changes in governments with differing preferences which result in higher inflation and higher variability. But one may ask whether it is not a revealed preference of voters through their choice of changing government rather than a failure of governemnt. The problem however is that voters do not make a choice about inflation or monetary policy when they choose a government. They choose a bundle or package of policies, and it not clear that the outcome in terms of any one policy is a preferred outcome for the voters. In such a sense the inflation variability argument may be seen also as a government failure argument.

The other premise of the analysis is that it is possible to design appropriate institutional mechanisms to deal with such government failures. The aim is principally to make the central bank behave with a long term horizon and shy away from opportunistic short termism.

From this perspective, I find the discussion of CBI (with whatever meaning you give to this term) in terms of a government and an agency which is the CB is too much couched in terms of the situation of any given government and CB. I think that setting up of CBI through whatever mechanism of "conservative central banker" or "contract" is not done by any government. It is at some point in the history of a country and under some special circumstances that a government works to institutionalize CBI. It is usually a singular episode. This happens in the context of a political process which strives to correct the government failure when it appears that it is becoming too costly to society.

What this means is that the government-CB game is embedded in a larger game involving the various actors of the political process, or it is part of a multi-level principal-agent problem. What does this imply ? It implies that the arguments for central bank inependence based on the time inconsistency analysis are not wholly convincing. If the analysis is to be limited to a government and a CB, and it is the government which is the principal at the source of failure it remains to be explained why it is not the government which does change its behavior or to device other mechanisms than that of CBI. For instance, does this minimize transaction costs (à la Dixit) in any sense? We do in fact see many instances of governments committing themselves to low inflation without CBI. On the other hand if by government it is meant the political process in general it is not clear why solutions to the problem would be limited to those one can deduce from a framework of analysis based on the sub-game between government and CB.

I would like also to suggest that basing essentially the argument for CBI on the time inconsistency analysis may actually weaken the argument. It makes it look too much dependent on some pre-conditions such as: 1) the existence of a trade-off between inflation and unemployment, and 2)the attempts by governments to exploit this trade-off in the short run. Would the non existence of those conditions render the argument for CBI irrelevant? I think that making the argument in terms of government failure is more general and powerful. This failure may come from the time inconsistency analysis using a Phillips curve or from the inflation tax temptation, or any other argument such as those from the public choice view or the Sargent and Wallace argument discussed, but just in passing, by the paper.
 

Determinants of CBI

Let me turn now to the discussion about the determinants of CBI and take my point of departure at a puzzling last sentence in the paper which states that "current research leads us to the conclusion that every society gets the central bank it deserves". It sounds like similar common statements to the effect that countries get the leaders they deserve or the political system they deserve. This statement about central banks is puzzling for at least two reasons. The first is that the empirical evidence from the study of the determinants of CBI is almost totally negative with none of the determinants suggested found relevant empirically. This means that we know little about the empirical determinants of CBI. So one wonders what is the empirical basis for the statement. It may be the somewhat limited evidence about the role of the public opposition for inflation, including financial opposition, and the implication that CBI is associated with society’s fundamental support of the objective of price stability. But this explanation leads to the second reason which is how to interpret the statement. Should this be interpreted as just a recognition of ignorance. Or should it mean that the type of CB a country gets is the most efficient institution given the country’s conditions? This would be a strong statement about the existence of no mismatch between institutions and socio-economic conditions or the non existence of inefficient institutions. This is clearly contradictoty with the accumulating evidence about the existence of such inefficiencies, and about path-dependence in institutional analysis. As I just mentioned the institutional framework of a CB does not change frequently. The establishment of CB "independence" is usually a singular episode. So the existence in a given country at a given time of an "institutional setting" for monetary policy which is not "efficient" or appropriate is often a likely situation.

This leads me to a remark about the empirical analysis described in the paper. Most of the analysis uses data for less than a 10-year period for a cross-section analysis, this assumes practically immediate adjustment of the institutional features of a central bank to some determinants. Since institutional change of the magnitude of CBI would be a much longer process it is not surprising that empirical results are so weak and not significant. Some empirical analysis is based on much longer time periods, but it is not clear whether this is a combined historical analyis and cross section or just cross section using averages for long time periods.
 

The developing countries

The paper deals almost exclusively with developed countries. Some reference is made for developing countries in the context of the empirical analysis of the determinants of CBI when some of these countries were included in the data sets. Therefore, it is wothwhile to ask some questions about the relevance of this literature to developing countries.

If the arguments for CBI are based on how to correct government failures then the issues of CBI are probably of greater relevance for LDCs where such failures are likely to be greater. In recent history the inflation bias was certainly mch greater in some countries, most notably Latin America, than what was observed in OECD countries. But is the analysis appropriate ? Two aspects are worth mentioning in this respect.

First, most of it is in the context of democratic political institutions, with elections, parties, changes in government which are the outcome of political competition and so on. While such settings exist in some developing countries and increasingly so, the most typical situation in such countries is that of autocratic or/and single party systems governing for long time periods, and changes in government in non-democratic ways.

Second, it is very unrealistic to assume that any inflationary bias of monetary policy in developing countries would come from attempts by governments to exploit short-term trade-offs between inflation and unemployment. The main reason is the temptation of using the inflation tax, very often due to the weakness of the tax systems.

What this means is that the normative analysis presented in the paper about CBI is of little relevance for developing countries. But the issue of CBI is probably more important in those countries than in developed countries. In practice mechanisms for insulating CB from short term government interference have included not subjecting CB to Ministry of Finance control, putting limits on CB finance of the budget, pegging exchange rates and even establishing currency boards, and so on. These issues are not discussed in the paper and there is little theoretical analysis of them.