1. Introduction

The debate on whether South Africa should formally introduce an inflation target as an anchor for monetary policy, recently revived again. Such targets were introduced by a number of other countries over the past twenty years, in most cases with relatively good results. Amongst these countries can be mentioned New Zealand, Canada, Australia, Israel, Sweden, the Czech Republic, Turkey and the United Kingdom. The new European Central Bank recently decided to follow a more eclectic approach in which both the money supply and inflation will be targeted, with due recognition of an important role for a number of other monetary/financial aggregates in the process of inflation creation.

In the case of South Africa, formal targeting of the M3 money supply was introduced in the mid-1980's to serve as an anchor for monetary policy. Like in many other countries, this base for monetary policy served South Africa well for some years and made an important contribution towards the gradual reduction in the level of the rate of inflation from a well-embedded double-digit figure of between 12 and 20 per cent over a twenty-year period from 1972 to 1993, to an average of well below 10 per cent over the past five years.

The use of the money supply as an anchor for monetary policy is based on the assumption that there is some stable relationship over time between changes in the money supply and total spending on goods and services and prices, that is inflation. Monetary policy is therefore directed towards controlling the rate of expansion in the total money supply as an intermediate objective with the ultimate goal of protecting the value of the currency. To achieve the intermediate objective, the central bank uses various operational instruments such as open market operations, variable minimum cash reserve requirements and changes in the conditions of discount window facilities to influence the amount of liquidity in the banking sector (supply of money) and the level of short-term interest rates (demand for money).

In this monetary policy model, the central bank therefore has an ultimate objective (to protect the value of the currency), an intermediate objective (to control the money supply), a supportive objective (to influence the amount of bank credit extension) and a number of operational instruments that can be used to achieve the goals of monetary policy. It must be pointed out that, apart from the fairly generally accepted ultimate goal of protecting the value of the currency, central banks can choose any one of the more important elements of the model as an intermediate target for guiding its shorter-term decisions on monetary policy. Some central banks may prefer a target for total domestic bank credit extension (DCE); others may prefer more direct controls over the amount of liquidity in the banking sector; others may set a more direct objective for the level of interest rates. Be that as it may, the monetary policy model as described above is a consistent model in which fairly stable relationships exist between the various components of the model. Whatever element may be chosen as an anchor for monetary policy, the other components of the model cannot and should not be ignored.

2. Moving from an intermediate objective to the ultimate objective of monetary policy

In South Africa, like in many other countries with a comparable experience of economic development, changes in the money supply in recent years lost some of its usefulness as an anchor for monetary policy, and as an intermediate objective for achieving the ultimate goal of protecting the value of the currency. This has happened, as in many other countries, mainly because of a major liberalisation of the financial markets, a huge increase in the volume of transactions in the money and capital markets, and an opening-up of the country for international participation, not only in respect of trade, but also for the inward and the outward movement of capital across international borders.

In the process, huge increases in the money supply occurred that were not related directly (or indirectly) to total spending on real goods and services. In technical terms, the additional money supply remained in financial circulation and was reflected in a large decline in the income velocity of circulation of the money supply. More money became available for every one unit of goods and services produced in the country. And yet, these increases in the money supply did not lead to increases in the rate of inflation. On the contrary, inflation in South Africa remained on a steady downward path, interrupted only by the adverse effects of the exchange rate depreciations of 1996 and 1998.

Reserve Bank guidelines for an acceptable rate of increase in the money supply, based on the objective of a gradual reduction in the rate of inflation and providing for a desirable but realistic rate of growth in real economic activity, were overshot by a substantial margin during each of the past four years. Taking account of developments in the financial markets, and the acknowledgement of a growing need for more money to service the growing volumes of financial market transactions, the Reserve Bank remained indulgent about these "excessive" increases in the money supply, as long as inflation continued to decline gradually.

These developments, however, impinged on the credibility of the continued use of M3 as the intermediate target of monetary policy in South Africa, and opened up the way for unreasonable criticism of monetary policy decisions taken by the Reserve Bank. After more than four years of a persistent "excessive" growth in M3, with a clear downward trend in inflation at the same time, the question has arisen whether the time has not come for South Africa to follow the example set by many other countries and to switch from a money supply anchor for monetary policy to a more direct targeting of the ultimate goal, that is inflation in its own right.

In many statements made since 1997, the Reserve Bank has expressed its support for gradually moving towards this new approach. In the annual monetary policy statements issued by the Bank in March 1998 and again in March 1999, the Bank indeed distinctly moved towards what it referred to as "informal" inflation targets. The view was expressed that this new approach was indeed more clearly linked to the directive contained in the Constitution of the Republic of South Africa in terms of which the Reserve Bank has been tasked with the responsibility of protecting the South African rand in the interest of economic development in the country.

The Bank has always held the view that, in the case of South Africa, it is essential that the Government through the Minister of Finance should endorse Reserve Bank targets or alternatively set quantitative inflation targets together with the Reserve Bank for the Bank to pursue. Inflation expectations play a major part in the ongoing inflation process, and a commitment by not only the Reserve Bank but also Government to lower inflation will add credibility to a resolute and collective objective of bringing inflation in South Africa in line with the relatively low levels that now apply in most other countries of the world.

A recent view expressed by the Minister of Finance that South Africa should consider the possibility of introducing inflation targeting, fuelled renewed speculation on the possibility of such a move in the near future. The ignorance revealed about inflation targeting in the subsequent public debate exposed once again the need for a better understanding of the intricate process of inflation in a modern economy. It is often overlooked, for example, that inflationary pressures can be created in many sectors of the economy and, if not depressed at source, can require draconian monetary policy measures to avoid monetary accommodation, with a continuing process of vicious inflationary circles that can harm the country for a long time.

It is also a fallacy to believe that a switch from money supply to inflation targeting will enable South Africa to have lower interest rates almost instan-taneously. Taking account once again of the intimate relationships that exist between the various components of the monetary policy model, there is but little difference between inflation targeting and money supply guidelines (or for that matter a DCE, liquidity or interest rate anchor for monetary policy). Interest rates will only decline to a lower level once inflation has been reduced on a sustainable basis to a lower level, and expectations about higher inflation in future have been eliminated with success from the minds of the majority of economic agents.

3. Introducing inflation targets

The introduction of inflation targeting is no easy task. Apart from the question of who should be responsible for the establishment of the target, the following questions must also be answered:

How should inflation be measured?

Although it is widely accepted that some form of a cost-of-living index or a consumer price index should be used for this purpose, there is no consensus on how exactly inflation should be measured ideally for the purpose of implementing monetary policy. One of the major deficiencies of any kind of price index for this purpose is that the statistics are based on past consumption expenditure, and do not take account of possible future purchases -- a deficiency that does not apply to money supply targeting where accumulated money balances form part of the base aggregate.

At what level or range should the target be set?

"The purpose of an inflation target is to provide the central bank with a rule for making monetary policy decisions. For that purpose, a target band is not necessary: a single point target is sufficient" (William A. Allen: Inflation Targeting: The British Experience, a Bank of England publication). A fixed point, however, is much more difficult to hit than staying within a range. Moving within a band also leaves some discretion to the central bank and can also provide more flexibility in the case of unforeseeable future price shocks.

What models should be used for forecasting inflation?

It must be understood that inflation targeting is about the future. Any monetary policy decision implemented today, for example a decision to raise interest rates, will normally only affect actual measured inflation with a time lag of, say, twelve to eighteen months (or even longer in some countries). Inflation targets can therefore only be set in a forward looking model that takes account of long-delayed effects. A reliable system for forecasting future inflation therefore becomes an essential precondition for any monetary policy model based on inflation targeting.

To what extent should inflation targeting be concerned with asset price inflation?

It is well-known that the Board of Governors of the Federal Reserve System in the United States of America is always concerned about the dangers of financial and property asset price inflation, and of the risk of a "bursting bubble". The experience of Japan in the late 1980's can be quoted where measured consumer price inflation never exceeded 4 per cent per annum, and yet excessive increases in bank credit extension and the money supply created unrealistic asset prices. Eventually, when the bubble burst, the total Japanese economy collapsed.

4. Conclusion

It is mainly for these reasons that many countries still prefer to stick to a more controllable intermediate target such as M3 or DCE, rather than the more complex and difficult inflation targets. South Africa's expressed intentions to move towards inflation targeting therefore need careful preparation. The Reserve Bank is in the process of improving its techniques of forecasting inflation, and continues to analyse the changing relationships that are now emerging in the new situation between the various components of the monetary policy model.

Reference should, in conclusion, be made to a third alternative in addition to intermediate targets, such as the money supply and inflation targeting, that can also be considered by countries as an anchor for monetary policy, and that is explicit targeting of the exchange rate of the currency. Countries such as Argentina, Hong Kong and the Democratic Republic of China, have opted for this alternative with a reasonable measure of success. There are, however, many reasons why this option is not regarded as appropriate for the present South African situation. Suffice it to say that, in the macroeconomic monetary policy model used by the Reserve Bank, the exchange rate is included as an important element in the comprehensive model, but it comes out as a result of the policy process, and not as an independent objective.