INFLATION AND/OR EXCHANGE-RATE TARGETS FOR MONETARY POLICY?

1 Introduction

Research into inflation targeting (IT) covers two main issues:

- credibility (or the "inflationary bias" problem)
- policy implementation

As to the first issue, Thygesen stresses that IT is yet another sign of the times of "unreserved orientation of monetary policy towards a purely nominal objective" in line with the policy implications of mainstream theory. However, since his paper is mostly concerned with the second issue, I shall not dwell on the first either. Drawing on some of the paper's many insights, I shall simply put forward a few considerations for further discussion on the problems of policy implementation in general, and more specifically of IT implementation in the light of the recent Italian experience. My view is that, among orthodox monetary policies, IT may offer some advantages if framed in suitable operational schemes. Relatedly, I shall also make a methodological point, namely that theoretical and applied research on monetary policy should be re-focused from the search of central-bank (CB) commitment technologies to the more traditional grounds of how monetary policy works.
 

2 Policy implementation

Implementation is the crux of any monetary policy. Whatever the task assigned to monetary policy by society, scholars and practictioners should find out the instruments most appropriateto the given objective. Thygesen compares the relative merits of IT and exchange-rate targeting under various circumstances, and in particular a) for countries that have the option to join a monetary union (but may be willing to stay outside for a while), b) for the future EMU as a whole.

If the CB's aim is the control of inflation, clearly IT is not an instrument, nor is it an intermediate objective, like the exchange rate or monetary aggregates. Rather, IT is a broad style of conduct of monetary policy whose main ingredients are:

• the announcement of an IT
• the CB's commitment to the announced IT
• an implicit or explicit agreement between the CB, the government, and the public that the CB performance should be assessed by comparing the actual inflation rate with the announced IT.

Searching for instruments takes us to the crucial issue in policy implementation: the so-called "transmission mechanism" from policy-controlled monetary variables through intermediate variables up to the final objective.
 

3 The transmission mechanism

Identifying and mastering the transmission mechanism is a critical step for a number of reasons.

First, as Thygesen recognizes, monetary theory may offer some guidance, mainly in the form of a menu of possibly different mechanisms, but then applied economists and officials in CBs should find their most appropriate procedures under the specific circumstances in which they operate.

Second, these circumstances may vary in different times and places. This is particularly important as regards one of the issues raised by the paper: whether IT will be appropriate to the EMU. I would enlarge the scope of the question by also asking whether one single policy strategy will fit many different economies with possibly different transmission mechanisms.

Third, as Norbert Wiener, the great mathematician, is reported to have said, "economics is a science of 1 or 2 digits". The transmission mechanism, when tested operationally, is not mechanical at all. As is well-known "monetary policy affects the economy with long and variable lags". Moreover, the CB's ability to get close to the target may be limited by a number of uncontrollable factors.

Fourth, the sand in the wheels of the transmission mechanism has consequences on policy assessment. I wish to stress just one of them relating to credibility.

Suppose a CB announces a target over a certain time period, and then the result on average overshoots the target. Did the CB cheat the economy, or was it simply unable to reach the target due to external circumstances? Do we usually have enough information to disentangle the CB's genuine responsibility from poor command over the transmission mechanism? Is there consensus on the measurement of the various "natural rates" that the CB should refrain from tampering with (which depends, by the way, on the state of economic science too)? Thus monetary policy-makers may find the credibility issue over-stated. Credibility is a clear-cut concept insofar as:

• there exists one shared model of the economy, and of the transmission mechanism, and such a model is, in some empirical sense, the right one,
• it is always possible for the public to trace a given state of the economy back to the CB positive actions.

4 The Italian experience

Let me try to give some substance to the implementation problems by referring to Italy's recent disinflationary experience.

The disinflationary process in Italy can be dated from the early 1980s. The inflation rate peaked at 22% in the third quarter of 1980; the last official release this year recorded a 1.6% on a yearly basis. Disinflation began after the oil counter-shock, with the lira within the EMS and with no explicit IT; it took the 5 years between 1983-88 for Italy's EMS inflation differential to be halved from 7% to 3%, then it remained stuck at that level. The fall of inflation accelerated dramatically when the lira was outside the EMS with the Bank of Italy following an informal IT, namely between 1994-96 when the inflation differential collapsed to zero. So this is a good case in point of comparison between IT and exchange-rate targeting.

As is well known, there is no agreement about whether exchange-rate targeting within the EMS can be credited with disinflation. Some authors have argued that disinflation in the EMS period is largely explained by world-wide factors, and that Italy's inflation gap did not fall far enough. On the latter point, the usual counter-argument is that the EMS commitment was not tight enough, so that inflationary expectations feeding wage and price setting were not pinned down. There is some truth in this argument, yet it remains to be explained why Italy's inflation gap was eventually closed when the lira was freely floating outside the EMS. This does not mean that IT by itself is a panacea for inflation.

A fair reconstruction of Italy's experience suggests that the commitment technology of the CB is only one element in the picture. Two other crucial factors impinge on monetary policy effectiveness:

•   the behaviours of other social actors, namely the Government, entrepreneurs and workers organizations
•   the CB command over the transmission mechanism, in terms of available information and control variables.

Central bank autonomy, reinforcement of budgetary procedures and a code for collective bargaining are prerequisites for the return to monetary stability (...) It is not by applying the brake of tight liquidity or of an unaccommodating exchange rate that equilibrium will be restored

This should not sound as a disclaimer. Increased CB autonomy and its anti-inflationary stance over the 1980s have certainly helped, but the fact remains that the Italian economy met all three of Ciampi's prerequisites only after 1992, and only in the subsequent years has its inflation gap been eliminated.

As to the transmission mechanism, the post-1992 experience is instructive. Ater the Incomes Policy Agreement of 1992 that linked nominal wages to the official IT set by the Treasury, th Bank of Italy has progressively moved towards IT . Let me point out very briefly some of the lessons that we may draw from this experience

Transmission mechanism specificity and the choice of instrument. The main instrument adopted by the Bank of Italy is control of bank lending rates via discount rates. The (theoretical) procedure is to move bank lending rates to a level consistent with minimizing the deviation of the forecasted inflation from the target over the transmission gestation period. This choice of instrument is consistent with the view that the so-called "credit channel" is important in Italy's transmission mechanism. Most firms in all sectors are largely dependent on credit even for current production. Bank interests account for a significant part of operational costs, no less than wages. For given nominal lending rates, expected price inflation reduces real unit costs for firms and hence induces more labour demand and credit expansion. Expected price inflation can be dangerously self-fuelling. To the extent that the CB is capable of altering bank lending rates, it can also exert some control over future inflation.

Non-neutrality. By altering the cost of credit relative to other factors, this transmission mechanism works mainly through the supply side of the economy, not through the demand side, leaving room for possible non-neutrality effects. This is usually not considered in current models of IT. As already stressed, the interaction with wage setting is crucial. If the CB succeeds in raising bank lending rates and curbing expected and actual inflation, the result will be neutral only if nominal wages decelerate as much as bank rates rise, i.e. if nominal wages are fully indexed to expected inflation. When this is not the case, say because nominal wages also embody some current inflation, disinflation may be costly because nominal wages decelerate less than expected prices, and firms' real unit costs rise.

The private sector and inflation. A real cost of disinflation may also arise for another reason. The 1995 Annual Report of the Bank of Italy contained an unusually harsh remark on the private sector, especially firms, for "behaving inconsistently with the official inflation target". Inflation was not falling fast enough. In the presence of a restrictive fiscal stance, the problem, as seen by the CB, was that not the policy authorities, but the public had an inflationary bias. This, too, is a problem that is still largely ignored in the IT literature.

The public may not endorse the official IT for sheer statistical reasons, becuase of the low credibility of the CB, or because a heavily indebted non-bank sector may be willing to inflate more than the official IT. In these cases, reaching the IT first requires the convergence of the economy's expected inflation to the IT itself. This is not only a matter of the CB's willingness and determination. There are also serious informational problems.

If the CB ignores the economy's inflationary bias, then it will set the discount rate too low and will miss the target. This will confirm the public's belief that the official IT is wrong or untenable in a vicious circle. As long as the expected inflation by the private sector dwells above the official IT, the real cost of disinflation grows, for the discount rate will have to be raised above the level required by the official IT alone. The CB may find its action unduly penalized in spite of its genuine commitment. Thygesen also recognizes this problem of the over-reaction of markets as a weakness of exchange-rate targeting, though it may become a general problem of monetary policy of any kind.

In Italy the discount rate peaked at 9% in March 1995 and was stuck there for 16 months. The real cost of credit to firms actually rose in a general context of falling rates in the rest of the world and of appreciation of the lira by some 15%. Over the same period, the inflation rate only fell by 1% and overshot the target. It was only after September 1996 that the inflation rate fell substantially and the discount rate was reduced. The preliminary results of the estimation of a model of the Italian transmission mechanism that we are running suggest that the anti-inflationary use of bank lending rates is effective, but the results come at some real costs in terms of long-run equilibrium output.

The resulting picture is strikingly the reverse of the familiar Barro-Gordon one. The real costs of disinflaton persist as long as the CB finds it necessary to punish the private sector for its inflationary bias. The incentive of the private sector to cheat the CB by embodying a higher inflation rate into contracts (perhaps "black" contracts at the firm level) persists as long as the CB is unable to anchor the inflation rate firmly to the IT. As already remarked, here the problem is not one of credibility but of the CB's effectiveness. On the other hand, to the extent that the CB succeeds in keeping the inflation rate close to the target, it may act as guarantor of the commitment of private parties to keep nominal contracts and prices in line with the target itself. It is perhaps not by chance that after subscribing the Incomes Policy Agreement of July 1992, the unions in Italy became strong supporters of the CB's effort to drive the economy towards the IT.
 

5 Conclusions

Thygesen's paper provides us with a broad and thoughtful discussion of pros and cons of IT as compared with other orthodox monetary policy techniques. One of the paper's merits is to focus on policy implementation problems. Implementation is a key factor in monetary policy design, and its effectiveness is no less important than the CB's commitment technology. To some critics, IT is more difficult to implement and assess, and hence is less accountable, and hence is less desirable than other techniques of orthodox monetary policy. In my view, these alleged drawbacks of IT can be turned into advantages. I have in mind more recent schemes which suggest that IT should be designed as an inflation forecast target, over a suitable period of time, and within a suitable range of inflation values. The points of strenght of these schemes are mainly two.

1) The closer interplay between the CB and the public at each of these steps make the whole process more, not less, transparent.

2) These schemes recognize the effectiveness problem, which may explain their growing popularity among CBs. They offer a framework for monetary policy that balances accountability of the CB's commitment, on the one hand, with autonomy in its technical choices and the necessary degree of tolerance and time-to-build in assessment of its performance, on the other.

There is also a theoretical feedback from this paper. One can only agree with the author's claim that, at least from Europe, it is hard to see any dramatic gap between mainstream monetary theory and policy as far as the first principles of the role of monetary policy is concerned. However, the gap seems substantial at the policy implementation level. The current large-scale production of game-theoretic models where almighty policy-makers play over-sophisticated strategies in over-simplified worlds is probably well within its range of diminishing returns. There are very good reasons to believe that economic agents play rather simple strategies to survive in a very complicated world. Personally, I would subscribe for independent, competent and effective monetary policy-makers who take care of monetary stability at large. To this end, continuous groundwork on the many complications and pitfalls of transmission mechanisms is needed.
 

References