3 Statistical Implications of Inflation Targeting in South Africa
- Carol Carson, Claudia Dziobek, and Charles Enoch
- Published Date:
- September 2002
INFLATION TARGETING provides a monetary policy framework within which policy actions are guided by expected future inflation relative to an announced inflation target (Green, 1996). Such a framework was implemented in South Africa when the South African government adopted an inflation target of 3 to 6 percent for the year 2002, as announced by the Minister of Finance in his budget speech on February 23, 2000.
This chapter briefly summarizes the rationale for and statistical implications of inflation targeting, with specific reference to South Africa. It will discuss the consumer price index (CPI) for South Africa, possible variants of the overall CPI to be used for inflation targeting, the quality of the South African overall CPI and subindices, and the index chosen for inflation targeting.
Rationale for Inflation Targeting
In South Africa, the Constitution and the South African Reserve Bank Act instruct the South African Reserve Bank to protect the value of the currency in order to achieve balanced and sustainable economic growth. That requires financial stability: that is, price stability as well as stable conditions in the financial sector as a whole.
The introduction of inflation targeting in South Africa was motivated by the desire for financial stability, which is an important precondition for sustainable high growth and employment creation. By establishing and maintaining financial stability, the monetary authorities make their unique contribution to general economic development in South Africa. If financial institutions and markets are uncertain or unstable, it is difficult to produce, consume, invest, and therefore to increase employment. The emerging market financial crisis in 1997 and 1998 has also clearly illustrated that foreign investment can be withdrawn easily and in large amounts from countries that investors perceive as high-risk destinations. Moreover, it is difficult for a country with a high rate of inflation to remain competitive in a global environment where more and more countries have already successfully reduced inflation to low levels.
The government of South Africa decided to set an inflation target as part of a new approach to monetary policy and price stabilization because such a framework is expected to have certain advantages:
- making the objective of monetary policy clear and thereby improving planning in the private and public sectors;
- forming part of a formalized, coordinated effort to contain inflation in pursuit of the broader economic objective of sustainable high economic growth and employment creation;
- helping to focus monetary policy and enhancing the accountability of the central bank to the public; and
- providing an anchor for expectations of future inflation, which should influence price and wage setting.
Inflation targeting in South Africa is characterized by an announcement of a numerical target for the inflation rate that is intended to be achieved over a specific time period. In this definition it is important to note that inflation targeting is a framework, not a rule; that the numerical rate is made public; and that a definite time horizon is specified.
Although the achievement of the target becomes the overriding objective of monetary policy in an inflation-targeting framework, the adoption of the new framework does not mean that the central bank is forced to apply rigid rules and is left without any discretion.
In the application of inflation targeting in South Africa, allowance is made for serious supply shocks. Some discretion must be exercised in order to avoid costly losses in terms of output and jobs. The South African Reserve Bank has to monitor economic developments closely to determine the origin and likely impact of such supply shocks, including developments affecting the terms of trade of the country or large disruptive international capital flows. Shocks could also arise from natural disasters. If such developments do occur, the public is informed of the likely consequences of attaining the monetary policy objective.
The numerical inflation target is the ultimate objective of inflation targeting. The immediate focus on intermediate targets such as the growth in money supply and bank credit extension falls away, because in inflation targeting, an intermediate target warrants a response only when it is the dominant factor determining inflation within the specified time frame.
In February 2000, the Minister of Finance announced that the South African government would adopt an inflation target, specified as an average rate of increase in the overall CPI excluding mortgage interest cost (called the CPIX) of between 3 and 6 percent for the year 2002.
The inflation target has been specified as a range or band because that affords the central bank some discretion in taking a monetary policy stance, and allows for a degree of the uncertainty and statistical variability inherent in all economic processes. The range indicates that the South African Reserve Bank will have been successful in its pursuit of the target if the average annual rate in the stated measure of inflation lies somewhere within the range of 3 to 6 percent in the calendar year 2002. It does not imply that the measured rate must be at the midpoint of this range. If the inflation rate were required to be at a specific level, a single point target would have been set.
The inflation target will be calculated as the average annual increase in the relevant price index for the year 2002 as a whole. It was decided to use this medium-term target in view of the long lags between monetary policy steps and their impact on inflation. Changes in interest rates in South Africa generally take from 18 to 24 months to have a material influence on the underlying rate of inflation.
The numerical inflation target was announced explicitly to the public to indicate clearly what the South African Reserve Bank should be held accountable for, and to make the application of this framework as transparent as possible (Van den Heever, 2001). If targets are not met, the central bank has to explain what went wrong. Regular reporting on the stance of monetary policy, as is the case internationally, is made to Parliament.
The monetary policy stance is communicated regularly to the public by means of a monetary policy statement after the completion of every meeting of the Monetary Policy Committee. A Monetary Policy Forum also has been established to open an avenue for ongoing discussions on monetary policy and general economic developments and to ensure that the views of interested parties are taken into account. The Monetary Policy Forum meets twice a year in the major centers of South Africa to allow as many stakeholders as possible to participate in these discussions. To increase transparency, the South African Reserve Bank also publishes twice a year a Monetary Policy Review, which describes in more detail the decisions taken by the central bank and analyzes developments in South Africa and the rest of the world that could affect inflation (Casteleijn, 1999).
These reporting mechanisms help the public understand better the basis on which monetary policy decisions are made.
The Consumer Price Index for South Africa
The CPI is a series of numbers showing how the average price level of a basket of goods and services bought by a typical consumer or household changes over time. The inflation rate is the annual percentage change in the CPI.
The Survey of Income and Expenditure of Households
Statistics South Africa conducts a survey of income and expenditure of households every five years. This survey is conducted by enumerators who visit a number of households throughout South Africa. The information obtained through this survey is weighted according to the Population Census figures in order to represent all households in South Africa.
The results of this survey are used to identify the goods and services bought by a typical consumer or household, which are then included in the CPI basket in order to monitor price changes.
Furthermore, the results of this survey are used to determine the weights of the products/groups (indicator products) in the basket. The weight of a specific product/group is calculated by dividing the total amount spent by all households in South Africa on that product/group by the total amount spent on all goods and services by all households. Statistics South Africa revises the weights every five years according to the latest survey of income and expenditure of households. Currently the 1995 weights, based on the 1995 survey of income and expenditure of households, are used to calculate the CPI. That survey covered a sample of 30,000 households.
Information regarding approximately 1,000 different goods and services groups was collected through the survey. Statistics South Africa made a further breakdown of these groups using supplementary sources. Approximately 1,500 groups are used to calculate the CPI.
Statistics South Africa conducted a survey of income and expenditure of households toward the end of 2000, and will introduce a new basket of goods and services based on this survey in 2002.
Sampling Methodology and Design
Sampling is done in different phases in order to collect information, through the survey of retail prices, needed to compile the CPI. First, a sample of goods and services, based on the information collected through the survey of income and expenditure of households, is designed and selected. An indicator product is selected purposively within each product group. To be selected, a product should constitute a large share of total expenditure. Furthermore, it must be possible to obtain price quotations for the relevant product during the whole year. The sample of indicator products is revised every five years when the weights of the indicator products/groups are revised.
Second, a geographical sample for price collection is designed and selected. Currently all 14 major metropolitan areas, covering all nine provinces in South Africa, are included in the geographical sample for price collection. “Other urban areas” are covered by nine provincial samples of four to five urban areas each, making a total of 39 “other urban areas” sampled.
Third, for each of the 53 sample areas, a frame for sampling retail trade and service outlets is constructed, based on available data sources, mainly the business register of Statistics South Africa, telephone directories, and lists obtained from the head offices of chain stores. The sample is purposively distributed within geographic areas. Specific retail trade and service outlets are selected randomly within each area. Prices are collected from supermarkets as well as specialty shops. The sample of outlets is revised every five years when the weights are revised.
Currently, an average of 11,000 price quotations are collected each month from approximately 2,200 outlets by means of 6,700 questionnaires. The CPI and relevant subindices are based on retail trade and service prices.
Basket of Goods and Services and Weighting Basis
The current CPI basket of goods and services covers approximately 1,500 goods and services. Statistics South Africa classifies these goods and services, according to the International Trade Classification, into 17 main groups: food; nonalcoholic beverages; alcoholic beverages; cigarettes, cigars, and tobacco; clothing and footwear; housing; fuel and power; furniture and equipment; household operation; medical care and health expenses; transport; communication; recreation and entertainment; reading matter; education; personal care; and other.
Statistics South Africa also classifies these goods and services, according to the Classification of Individual Consumption by Purpose, into nine main groups: food, beverages, and tobacco; clothing and footwear; housing, water, electricity, gas, and other fuels; furnishings, household equipment, and routine maintenance of the house; health; transport; leisure, entertainment, and culture; education; and miscellaneous goods and services.
The CPI is a fixed weights index, which implies that the weight of each product/group stays the same for the five-year period until the results of the next survey of income and expenditure of households become available.
The Survey of Retail Prices
The survey of retail prices is a sample survey of retail trade and service outlets covering prices of selected consumer goods and services sold to consumers in the 14 metropolitan and 39 other urban areas in the nine provinces of South Africa. Statistics South Africa compiles and publishes a CPI and relevant subindices for the 14 main metropolitan areas, which together represent only about 37 percent of the total expenditure of all South African households. Statistics South Africa also compiles and publishes a CPI and relevant subindices for the 14 main metropolitan areas and the 39 other urban areas, which represent approximately 76 percent of the total expenditure of all South African households. Statistics South Africa aims to broaden the calculations to include the rural areas as soon as the results of the 2000 survey of income and expenditure of households become available and the various indices are reweighted.
Extending the geographical coverage of the CPI and its derivatives to rural areas could happen in one of two ways:
- The prices of the items sold in the rural areas could be collected by enumerators on a regular basis. This will be done if the information regarding points of purchase, collected through the 2000 survey of income and expenditure of households, shows that people in rural areas purchase their goods and services in the local rural shops.
- However, if the information indicates that people in the rural areas mainly purchase their goods and services in the nearby urban areas, the prices collected in the urban areas will be used to compile indices, which will in turn be weighted according to the expenditure patterns of the people in the rural areas.
A final decision on the methodology will be made only after the results of the 2000 survey of income and expenditure of households become available.
Prices of goods and services currently included in the CPI and subindices are collected on any day between the first and the seventh of the month. Although most of these prices are collected monthly (see list of items below), the prices of some items are collected quarterly, semiannually, or annually (see Tables 3.1 to 3.3). The frequency of collection depends on the frequency with which these prices tend to change. However, if Statistics South Africa notices that prices are in the process of changing, these prices are collected even if it is not according to the normal scheduled time.
|Items/Products||Months of Survey|
|Garden tools||January, April, July, and October|
|Washing, ironing, and dry cleaning|
|Reading matter and stationery|
|Tariffs of hairdressing services|
|Ironware and crockery||February, May, August, and November|
|New and retread tires|
|Furniture and equipment||March, June, September, and December|
|Electrical appliances and equipment|
|Medical, toilet, and photographic requisites and services|
|Motor vehicle insurance|
Items/products whose prices are collected monthly:
- vegetables and fruit;
- other groceries;
- alcoholic beverages;
- sweets, nonalcoholic beverages, ice cream, and tobacco products;
- clothing and footwear;
- repairs of clothing, footwear, and furniture;
- interest rates on mortgage bonds;
- coal and wood;
- new vehicles, repairs, and services;
- motor spare parts and accessories;
- petrol; and
|Items/Products||Month of Survey|
|Doctors’ and dentists’ fees||January|
|Motor vehicle license and registration fees|
|Toll fees at tollgates||March|
|University boarding and class fees|
|Telephone and postal tariffs|
|Public transport tariffs||June|
|Newspapers and magazines|
|Entrance fees–drive–ins and bioscopes||August|
|Maintenance of graves|
|Rent of dwellings|
|Items/Products||Months of Survey|
|Winter clothing||March, April, May, June, July, and August|
|Medicine||January and June|
|Contributions to medical aids||January, April, and July|
|Property insurance||January and July|
|Water||January, July, and August|
|Air transport fees||January and August|
|Dog licenses||January, July, and October|
Possible Variants of the Overall Consumer Price Index
During 1999, Statistics South Africa did in-depth research and analysis on different ways to measure inflation, with the idea in mind that the South African authorities might adopt inflation targeting. The research revealed that in countries such as Australia, Brazil, Canada, Chile, Finland, Israel, New Zealand, Poland, Spain, Sweden, and the United Kingdom, which have adopted inflation targeting, different measures of the inflation rate have been derived and applied (Bernanke and others, 1999; Jonsson, 1999):
- Australia adopted the overall CPI excluding the impact of interest rates on mortgage and other interest payments, indirect tax changes, and certain other volatile price items (Reserve Bank of Australia, 1994).
- Canada adopted the overall CPI excluding the prices of food products and energy, as well as the effect of indirect tax changes (Bank of Canada, 1991a and 1991b; Freedman, 1995).
- Finland adopted the overall CPI excluding indirect taxes, government subsidies, house prices, and mortgage interest payments (Åkerholm and Brunila, 1995).
- New Zealand adopted the overall CPI excluding interest cost components, indirect taxes, subsidies, government charges, and significant price effects due to changes in the terms of trade (Roger, 1998).
- The United Kingdom adopted the retail price index excluding mortgage interest payments (Haldane, 1995; Yates, 1995).
- Brazil, Chile, Israel, Poland, Spain, and Sweden (Andersson and Berg, 1995) adopted the overall CPI.
On the basis of the above-mentioned research, Statistics South Africa, in consultation with the South African Reserve Bank and the Department of Finance, decided to compile, in addition to the existing overall CPI and the core CPI (that is, the overall CPI excluding fresh and frozen meat and fish, fresh and frozen vegetables, fresh fruits, interest rates on mortgage bonds and personal loans, assessment rates, and changes in value-added tax), two other variants of the overall CPI. The additional variants would furnish policymakers with more options from which to choose the best index to adopt for inflation targeting. The two additional variants are the core index excluding petrol and the CPIX.
In order to ensure an accurate derivation of different measures of inflation, Statistics South Africa recompiled the CPI from January 1997, based on the originally collected prices of items contained in the CPI. From this, the above-mentioned variants of the overall CPI were compiled.
The tables in Appendices 3.1 and 3.2 show the above-mentioned four indices (Statistics South Africa, 2000 and 2001) and the monthly and annual percentage changes in each index, and Figures 3.1 and 3.2 are graphical presentations. From Appendix 3.1, column 5, and Figure 3.1, it is clear that the highest and lowest annual percentage changes in the CPI for the metropolitan areas were 11.1 percent at April 1995 and 1.7 percent at October 1999, reflecting a variation of 9.4 percentage points. However, during the period January 1995 to July 2001, the highest and lowest annual percentage changes in the core index for the metropolitan areas were 9.9 percent at April 1997 and 6.2 percent at April 1996, reflecting a variation of 3.7 percentage points (see Appendix 3.1, column 8; and Figure 3.1). Similar to the core index, the highest and lowest annual percentage changes in the core index excluding petrol for the metropolitan areas during the above-mentioned period were 9.3 percent at April 1997 and 6.3 percent at March and April 1996, reflecting a variation of 3 percentage points (see Appendix 3.1, column 11; and Figure 3.1). Furthermore, during the above-mentioned period, the highest and lowest annual percentage changes in the CPIX for the metropolitan areas were 10 percent at April 1997 and 5.3 percent at April 1996, reflecting a variation of 4.7 percentage points (Appendix 3.1, column 14; and Figure 3.1).
Figure 3.1.South Africa: Annual Percentage Change in the Derivatives of the CPI for the Metropolitan Areas
Source: Statistics South Africa. CPI, consumer price index.
Figure 3.2.South Africa: Annual Percentage Change in the Derivatives of the CPI for the Metropolitan and Other Urban Areas
Source: Statistics South Africa. CPI, consumer price index.
During the period January 1998 to July 2001, the annual percentage changes in the CPI for the metropolitan and other urban areas varied between 9.0 percent at November 1998 and 1.9 percent at October 1999, reflecting a variation of 7.1 percentage points (Appendix 3.2, column 5; and Figure 3.2). However, during this period, the annual percentage change in the core index for the metropolitan and other urban areas varied between 8.1 percent at January 2001 and 6.5 percent at February 1998, reflecting a variation of only 1.6 percentage points (Appendix 3.2, column 8; and Figure 3.2). Furthermore, during this period the annual percentage change in the overall CPIX for the metropolitan and other urban areas varied between 8.1 percent at September 2000 and 6.4 percent at July 2001, reflecting a variation of only 1.7 percentage points (Appendix 3.2, column 14; and Figure 3.2).
From the above it is clear that the annual percentage change in the CPI for the metropolitan areas as well as for the metropolitan and other urban areas fluctuated more during the period January 1995 to July 2001 than the annual percentage change in the other variants of the CPI. Thus it became clear to South Africa that one of the variants of the overall CPI—the core index, core index excluding petrol, or CPIX—had to be adopted as the measure for inflation targeting, as each was more stable over time than the overall CPI.
Statistics South Africa introduced the different variants of the CPI during a workshop with stakeholders on February 15, 2000 (Statistics South Africa, 2000).
Quality of the South African Overall CPI and Subindices
Statistics South Africa received an independent international review of the production of the monthly CPI. With the South African authorities contemplating inflation targeting, it was desirable to benchmark the quality of the index from which an inflation-targeting measure might be derived.
The review was carried out by Mats Haglund, former head of the prices division at Statistics Sweden, and funded by the Swedish Independent Development Agency. The key finding of Haglund (2000) was that “the samples are the result of well-thought-out sampling procedures, the index is compiled by means of internationally recommended index formulas and methods for dealing with substitutions, and quality changes are expected to be appropriate for most of the different product areas covered. The index is computed from an extensive number of price quotations received for samples of indicator products from samples of retail trade and service outlets. The index, or derivatives such as the Core Index of Statistics South Africa or the CPIX, that is, the consumer price index excluding interest rates on mortgage bonds, which are currently published, is likely to result in reliable results regarding the rate of change of South African consumer prices.” This report and findings were also discussed at the workshop with stakeholders February 2000 (Haglund, 2000; Statistics South Africa, 2000).
Index Chosen by South Africa for Inflation Targeting
If the inflation targets are indeed to influence inflation expectations, it is crucial that the measurement used enjoy widespread acceptance and be easily understood. This cautions against excluding too many goods and services or items that feature significantly in household spending, such as food and transport. For this reason, after in-depth research and discussions, South Africa chose the CPIX, which is the overall CPI less only interest rates on mortgage bonds, as the inflation measurement. This measurement had the advantage of being uncomplicated, covering all those items to which poorer households (who often do not have mortgages) are most sensitive, and avoiding the perverse impact that changing interest rates can have on the inflation measure.
The South African authorities decided against using the core inflation rate compiled by Statistics South Africa for inflation-targeting purposes, despite certain advantages that the use of this index could have had. The measurement of core inflation has the advantage of excluding prices directly affected by policy measures as well as some prices over which policy has no direct control and which could lead to misleading signals when they are affected by economic shocks. However, it does not exclude the impact of some kinds of prices, such as changes in oil prices. The measurement of core inflation has further disadvantages in that it does not comprehensively reflect the cost of living, is difficult for the public to understand, and is less credible than headline inflation.
The South African Reserve Bank, in agreement with the Department of Finance, therefore opted to target the overall CPI excluding mortgage interest cost as measured for metropolitan and other urban areas, CPIX(mu). Using the CPI for only the metropolitan areas would have restricted the measurement of inflation to 14 metropolitan areas. By opting for this more comprehensive measure of inflation, price changes in 53 metropolitan and other urban areas, covering approximately 76 percent of all consumption expenditure, are taken into consideration in the inflation target. This expansion enhances its credibility. If Statistics South Africa were to broaden the calculations to include the rural areas in measuring CPIX(mu), this measure of inflation could possibly be used for inflation-targeting purposes.
In a wider context, inflation targeting is part of a process in which economic policymaking is becoming more transparent and subject to more accountability and technical rules, and less susceptible to discretionary actions. Discarding the use of subtle intermediate variables and nuances is a great virtue, and the new framework may well prove to be more than a temporary fad. It is no panacea: central banks will still need to earn their credibility and work hard at maintaining it, through tight policies where necessary. However, to the extent that the inflation target focuses policies and expectations, it can reduce the frictional costs associated with curbing inflation.
The key parameters of the inflation-targeting framework implemented in South Africa from February 2000 were determined on the basis of thorough research and best international practice and advice.
|Consumer Price Index (CPI)||Core Index||Core Index Excluding Petrol||CPI Excluding Interest Rates|
on Mortgage Bonds (CPIX)
|Consumer Price Index (CPI)||Core Index||Core Index Excluding Petrol||CPI Excluding Interest Rates|
on Mortgage Bonds (CPIX)