Meeting with the Governor of the South African Reserve Bank

Meeting report

FINANCE PORTFOLIO COMMITTEE
19 September 2000
MEETING WITH THE GOVERNOR OF THE SOUTH AFRICAN RESERVE BANK

Documents handed out:
Slide Presentation on the Suite of Models Developed by the South African Reserve Bank - text outline only [see Appendix 1]

Acting chair: Ms Joemat (ANC)

SUMMARY
The South African Reserve Bank briefed the committee on the Reserve Bank's economic models. The Reserve Bank's structure and its human resource plans were also outlined with an emphasis on its move toward greater diversity and its recruitment initiatives. Finally the key issues that will arise at the World Bank / IMF meeting from an African perspective were discussed. Discussion included technical questions as well as questions about the place of the central bank in connection with other government organisations and the Black Economic Empowerment Commission, the repo rate, how to retain staff and trade as a means of debt reduction.

MINUTES
Governor Mboweni, Governor of the South African Reserve Bank, indicated that the presentations would cover three areas:
(1) an introduction to the models the Reserve Bank uses to forecast inflation, subject to the caveat that these do not substitute for the Bank's exercise of judgement;
(2) institutional issues at the central bank, discussing its organisation and its transformation; and
(3) any questions arising about the upcoming World Bank meeting in Prague.
He noted his reluctance to go into detail on the last point given that the Committee had recently met with the Minister of Finance.

Modelling
Dr Small, the South African Reserve Bank official responsible for modelling and econometrics, provided a background on modelling and explained the Bank's suite of models, and the matter of models and policy. She noted the monetary policy framework of an inflation target of between six and three percent for 2002 and emphasised that this is the objective that is clear, that focuses monetary policy, that provides for accountability, and that determines inflation expectations.

Dr Small's introduction to modelling emphasised:
- The need for forecasts because of time lags in the economy's response to policy initiatives.
- Models are a simplification of a complex economy and that the limitations of models must be remembered.
- As the structure of an economy changes, so must the model be adjusted.

Dr Small admitted that modelling is not perfect. She gave the example of the Swedish Riksbank's incorrect inflation forecast of 1998/9 that foresaw a continuing rise in inflation. The inflation rate instead dropped significantly before returning to a rising trend. Although the Riksbank did not foresee the drop, once the drop was underway, it forecast correctly the turning point and move back up.

Dr Small next presented several charts concerning current forecasts on South Africa's inflation and GDP growth. The spread between the minimum and maximum projections for the CPI measure of inflation is larger for 2001 (range from 4,8 to 7,5%) than for 2000 (4,9 to 6,0%), indicating uncertainty. For 2000, there is what is called an upside risk, and for 2001 there is a downside risk.

In response to a question, Dr Small explained that an "upside risk" means that more than half the forecasters predict a higher rate than the average. Conversely, a "downside risk" means that more than half the forecasters predict a lower rate than the average. Thus, these are mathematical properties of the distribution of forecasts that come across graphically as a skewing of the bell curve constructed from the forecasts. For GDP growth, there is similarly more uncertainty about 2001 (with forecasts ranging from 2,5 to 4,0%) than about 2000 (range from 2,3 to 3,5%). GDP growth forecasts show a downside risk in both years.

Dr Small gave further background on modelling, explaining that models may be based on varying amounts of theory as opposed to empirical data. The approach to modelling is to begin with economic theory if the model is to include it, then to construct a general empirical model, which must be tested against the data. If there are no co-integrating relationships, it is necessary to go back to the theory and try again or else find an appropriate transformation of the data. If there are co-integrating relationships, then the next step is to estimate them and determine if the model is statistically adequate. If it is not, then it is necessary to go back to the empirical model. If it is, then the modellers can begin testing economic hypotheses and using the model for predictions and policy.

In terms of testing the model, Dr Small explained that the modellers will first do in-sample forecasts, then test the model out of the sample but without actually using it (a holdout period). If it works in these tests, then they will begin using it for full-fledged forecasts.

Suite of models used by the Reserve Bank
Dr Small outlined these as:
· Core Model - a model with 47 equations, of which 24 are structural; based on standard macroeconomic theory about the transmission mechanism - assumes that the repo rate affects the interest rate, which in turn affects both aggregate demand and the exchange rate; the exchange rate affects both export demand and import prices; aggregate demand and export demand affect domestic inflationary pressures, which combine with import prices to determine inflation

· Small-Scale Model - a highly aggregated model with five equations; Australia and Israel use it as their main model, as does Sweden which switched to it after finding three equations to be insufficient

· Phillips Curve - relating inflation to a measure of disequilibrium as a cross-check

· Vector Auto-Regressive (VAR) - based on the dynamic interaction of variables with little theory - all equations are estimated at once as opposed to individually

· Auto-Regressive Integrated Moving-Average (ARIMA) Model - a one-variable model based on the history of that variable

· Indicator Model - to identify sources of inflation

· Fan Chart - to capture uncertainty fanning out from all the variables

· New Collaborative Platform - unique to South African Reserve Bank, a database designed to keep track of who contributed what to the model when.

Dr Small stated that models are used in conjunction with other information, including the Stellenbosch inflation expectations questionnaire and technical assistance from other banks. She explained forecasting as an intensive process, with new data, high-frequency data, Reserve Bank experts, and other financial institutions updating data and formulating assumptions. These two aspects then affect the factors used in the model, which, in conjunction with the determinations of the modelling team and other Reserve Bank staff, leads to a review of forecasts and eventually the locking in of a final forecast subject to interaction with the Monetary Policy Committee (MPC).

Dr Small emphasised the mistaken public perception that policy flows immediately from the models, stating that the Reserve Bank also considers certain assumptions and judgements, other models, and other issues and policy judgements.

Discussion
Gov Mboweni indicated that he wanted to pre-empt one question that he knew would be coming. He stated that the Reserve Bank had not brought its detailed equations to this session. He stated that it was not yet confident it could supply these, and it did not want to create unnecessary uncertainties. He stated that there is still a substantial amount of peer review to be done, although he noted that other central bankers had met a surprise insofar as they expected South African modellers to be barefooted but South African modellers actually knew what they were doing.

Prof Turok noted that investors invest in the developing world based on perceptions, and in some ways it seems unfair by contrast that South Africa has to observe the scientific rigour of a model. Gov Mboweni noted that perceptions have an impact but that the real economy may be driving them. For example, he referred to the United States, where real economic developments have led to a massive perception about the American economy. Gov Mboweni stated that the real world demands that statements be based not on what he called "thumb-sucking" but on something real. He stated that he would speak not from a barefooted football field but from a scientific basis.

Prof Turok also noted that the central bank does so much research and gathers so much information that it seems a shame that it has to confine the use of this information to the realm of monetary policy. He asked if there is any way the country can make use of the data collected by the central bank in other ways and thus get a better division of labour.

Gov Mboweni joked that this question felt a bit curved. He replied that he would have more to say later concerning central banks and the division of labour but that it is important to have a clear division of labour or there may be absolute confusion. It is very important to be clear as to who manages what and that it is clear that the central bank looks after monetary policy. The central bank already engages in some discussions with other public authorities whereby it makes them aware of things that it has detected and learns from them about their sector. For example, there had been bilateral meetings recently with Treasury and Trade and Industry and that there would be a similar meeting on October 14 with Agriculture.

Ms Taljaard (DP) asked if the modellers had fed in GDP growth revised downwards or if they were waiting for third quarter statistics. Dr Small did not reply directly but answered that the modellers forecast GDP growth as one of the variables.

Ms Taljaard (DP) asked if the modelling process is as intensive as it is because there is only quarterly data available and whether there is a problem because the data is only quarterly. Dr Small replied that the quarterly compilation of the national accounts is the most frequent available but that the ARIMA and Indicator Models use monthly data and that monthly data provide for interim numbers that can allow forecasters to adjust their estimates of other variables in the quarter.

Ms Taljaard (DP) asked if the Indicator Model, which is supposed to identify the sources of inflation, distinguishes internal and external pressures. Dr Small did not answer directly but simply repeated that the Indicator Model identifies the sources of inflation.

Mr Feinstein (ANC) asked about the forecasting process, indicating that he took from the charts presented that the two key moments are the updating of the data and the formulation of assumptions by Reserve Bank experts. He asked concerning the latter as to what the nature of these assumptions was and how the experts formulating them ensured as broad a view as possible.

Dr Small replied that they incorporate as much information as possible and stated that they are based on contact with the private sector through sectoral experts. She stated that some assumptions correspond to endogenous variables (those determined by the model) and that there would be comparisons as a result.

Mr Feinstein (ANC) indicated that he would like to request more information as to how the experts assess various sectors. The Acting Chair noted this request. Gov Mboweni noted that he had tried to make a preventive strike at the outset and indicated that the Reserve Bank is providing and will provide as much data as possible. He stated that the models had just been completed in July. He also mentioned the story of his attendance yesterday at the COSATU conference, where he had received some criticism. He stated that this was on account of both policy and his use of "technocrats" who are not part of the mass democratic movement. Gov Mboweni interpreted this as underlining the importance of considering all relevant considerations. He indicated again that the Reserve Bank would try to provide information.

Dr Woods (IFP) noted that Reserve Bank statements seem to describe modelling as merely one tool, yet it seems quite dominant. He asked for reflection on what else the Reserve Bank uses. Gov Mboweni indicated that the Monetary Policy Committee (MPC) takes time to consider everything and that it considers along with the models such considerations as the international macroeconomy, general discussion of the domestic economic situation, international financial developments, domestic capital market developments, and the views of private sector economists. He added that he enjoyed hearing the views of private sector economists but that he would invite more participation, especially from the NGO movement. He indicated, for instance, that he would like to hear from the research wing of the trade union movement. The Reserve Bank staff and the MPC both spend a lot of time on the process.

Dr Woods (IFP) sought to clarify whether the core model is an equilibrium model based on aggregate demand and aggregate supply curves with a range of variables. Dr Small indicated that it is a basic standard macroeconomic model based on supply, demand, a vertical Phillips Curve, and standard macroeconomic theory.

Dr Koornhof (UDM) noted that the Black Economic Empowerment Commission is coming out with its final report by the end of September and that this report could have far-reaching implications. He asked if the Commission had consulted with Gov Mboweni and how monetary policy could assist with the Commission's proposals. Gov Mboweni stated that the Commission had not engaged in any discussion with him, although it likely knew his private views. He stated that it had invited him to comment on September 29 and that he would do so. In terms of assisting the Commission's proposals, he stated that the best assistance would be price stability. He also noted the Reserve Bank's affirmative procurement policy on tenders. He described how after a black accounting firm had pointed out that Form D1006 (which every bank must submit to the Reserve Bank for its approval of the bank's appointment of auditors) had an in-built structural discrimination against black accounting firms, this form would be changed (as its Annexure asked for number of partners, with the form looking for hundreds, and the size of the firm's library) . The result was no chance that a black accounting firm would ever audit a bank. They have also implemented a policy of encouraging big firms to take black firms along as partners so they learn about these audits.


Governor Mboweni, Governor of the South African Reserve Bank, presented on the bank as an institution and on the upcoming World Bank meeting in Prague. He stated that he would introduce the bank as an institution very briefly, leaving written materials with the members, and then talk about how it sees its transformation in the coming period.

South African Reserve Bank Structure
Gov Mboweni explained that half the directors are appointed by banks and half by government. There are also four governors, himself and three deputy governors. He proceeded through the structure of their management. Deputy Gov G Marcus oversees bank supervision, financial services, and exchange control, some of which could possibly fit outside the Reserve Bank structure but which the government had put there. Dep Gov T Thahane oversees business systems and technology, legal services, corporate services, and international relations. Dep Gov J Cross oversees currency and protective services, international banking (concerned largely with exchange rates, as well as gold and foreign reserves), money and capital markets (concerned with the repo rate), internal audits, the mint, and the South African Bank Note Company.

Gov Mboweni explained that his adviser, Dr X P Guma, advises but also oversees human resources and the college where the Reserve Bank does training. He noted that this college is important to all of southern Africa and that it also liaises with the IMF and the Bank of England. It has both a two-week diploma in central banking and a longer two-year program. Dr Guma also oversees the Vuindiela Unit, which opens any blockages. Finally, Dr E J van de Merwe, Chief Economist and Head of Research, oversees research and the national payment system. The latter had been the site of the Reserve Bank's one major human resource crisis last year when twenty key technocrats walked out and formed their own consulting company. The Reserve Bank had nobody left who could do their jobs and had to contract with this consulting company, so they all rapidly became millionaires.

Gov Mboweni explained that the Board of Directors has very limited powers, such as setting the salaries of bank officials and is composed of outside directors. The Board of Governors and the MPC are the key Reserve Bank decision-makers. The Board of Governors meets every two weeks.

Human Resource Plan 2005
Gov Mboweni stated that for historical reasons, the Bank has been managed by Afrikaner men and that it is now time to engage the skills of all South Africans. He indicated that the current overall staff profile is 57% white, 33% African, 6% coloured, and 4% Asian. It is 58% male and 42% female. However, if one looks only at managerial positions, they are 93% male and 81% white. He stated that, under the present situation, blacks and women fix coffee and the air conditioning. The Reserve Bank has a target of achieving 50% black representation and 33% female representation across all occupational categories at the Bank by 2005. Thus, departments have gone through a review and made plans. And the Bank has engaged in an important recruitment drive to hire anyone from anywhere whose skills are not being utilised and to identify capable young people. Capable young people will go to the Reserve Bank's college, or it might take them to New York or Europe for education and practical business learning. There was some worry regarding a project with Wharton Business School that some who went there would not come back, but they all have. Gov Mboweni noted the great talents of these young people.

He mentioned that the Reserve Bank is developing a program with the University of Pretoria to deal with the potential impact of HIV/AIDS on its staff.

Discussion
Mr Louw (ANC) asked how to prevent the skills of talented people from disappearing without them coming back to the Bank. Gov Mboweni stated that the trick is not to send disgruntled employees to Europe or America and that if happy employees go, they will come back.

Dr Woods (IFP) asked about the recent commentary in Business Day to the effect that the Reserve Bank merely pretends that the repo rate is a market-determined rate. Gov Mboweni replied that the repo rate is a floating rate but that it is true that the Reserve Bank's actions have a major role in the mechanism, depending on if the Reserve Bank supplies all the funds required.

World Bank / IMF Meeting in Prague
Gov Mboweni identified the key issues so far as Africa is concerned:
- With regard to the position of poor countries in the IMF system, he referred to the Highly Indebted Poor Countries (HIPC) Initiative, which looks for debt reduction for such countries as Mozambique, Ethiopia, and Zambia that are highly indebted. He referred to the imposition of new forms of conditionality on poor countries trying to deal with the World Bank and IMF. These countries are required to produce complicated Poverty Reduction Strategy Papers (PRSPs) for both institutions, which wastes a great deal of time that could be spent actually implementing the strategies. He explained that countries become so burdened with meetings and paperwork that they have less time to tackle the problems.
- There is the issue of the place of African technocrats within the IMF.
- The proposals to reform the quota system that would mean many African countries would lose their voting power. He stated that the two that would remain significant are South Africa and Nigeria. He stated that other countries are asking what is fair about this when the organisation is supposed to be a 182-member institution.
He concluded by emphasising that poverty reduction is certainly the most important issue.

Discussion
The Acting Chair asked about the oil boycotts and expressed her concern that there might be protests in Prague. Gov Mboweni stated that if this happened, it would be great fun. He stated that there was a need to find mechanisms for grievances.

Dr Rabie (NNP) referred to the matters of debt reduction and conditionality and asked the Governor's views on trade as a way to reduce debt. He noted that the odds seemed stacked against the Third World.

Gov Mboweni stated that better trade flows would help all kinds of processes. Debt piles up because of interest costs. Initiatives like the American African Opportunities Bill might help, but he noted his suspicions that they would contain new conditionalities. He added, however, that the world has moved a lot. Discussion is no longer about the comparative advantage of Africa in mineral production, but about the competitive advantage of Africa, which is yet to be determined. He hinted at a hope that this would be about human resource development.

In closing he stated that this Committee is an important forum for accountability.

Appendix 1:
THE SUITE OF MODELS DEVELOPED BY THE SOUTH AFRICAN RESERVE BANK
19 September 2000

Contents

• Background
• New suite of models
• Models, forecasts & policy

Monetary Policy Framework

• Formal inflation targeting framework announced on 23 February 2000
• Target range set at 6 to 3 per cent for the year 2002
• Achievement of target becomes objective of monetary policy

Advantages of new framework

• Makes objectives of monetary policy clear
• Helps to focus monetary policy
• Enhances the accountability of the central bank to the public
• Provides an anchor for inflation expectations

The use of models

• "Economics is a science of thinking in terms of models joined to the art of choosing models that are relevant to the contemporary world" - John Maynard Keynes, 4 July 1938

The use of models in monetary policy

• "Implicit in any monetary policy action or inaction is an expectation of how the future will unfold, that is, a forecast. There is no way to avoid making a forecast, explicitly or implicitly." - Alan Greenspan, 1994

Why are policy makers concerned with the future?

• Monetary policy works with a considerable time lag
• It differs across countries and over time
• Policy makers wish to know what will happen if they change the central bank's interest rate
• Careful use of models can reduce the uncertainty inherent in this type of decision making

Benefits of models

• Provide a simple framework with which to analyse and quantify short to medium-term macroeconomic developments
• Assists in the construction of forecasts in a systematic manner
• Useful for alternative policy simulations

Requirements for models

• To be responsive to structural changes in the economy
• Not necessarily to supply precise forecasts, but at least a good feel for directional effects of policy options
• To correspond to linkages in the economy and to track the dynamics in the model easily
• To follow a pluralistic approach: not one model for all occasions, but a suite of models
• To follow a pragmatic approach: supplement model output with surveys, value judgements, etc.
• To organise thoughts about policy and the economy, but their limitations must always be kept in mind.

Experiences of other policy makers

• Models are dynamic, changing entities
• Improvements are constantly being made without changing the theoretical structure of the steady-state model
• Over the medium-term, judgmental forecasting becomes more difficult and model properties more important
• With a forward-looking target, a mechanism is needed to set instruments today in order to achieve the future target
• Staff projections that are incorporated into the policy process are most useful

Riksbank: CPI outcome and forecast for 1999

[PMG Ed. Note: graph not included]

Sweden: CPI outcome and forecast for 1999

[PMG Ed. Note: graph not included]

Consensus forecast: CPI
[PMG Ed. Note: graph not included]

Headline inflation rate : Risk

[PMG Ed. Note: graph not included]

Consensus forecast: Real gross domestic product

[PMG Ed. Note: graph not included]

Real gross domestic product : Risk

[PMG Ed. Note: graph not included]

Types of models

• No theory, simply empirically best fit of data
• Derived from theoretical principles
• Spectrum between two extremes

Steps in model building

• Literature study
• Gather data and estimate individual equations
• Model validation
• Tests for system dynamics & stability
• Test forecasting ability: A lengthy process

A modern approach to modelling

[PMG Ed. Note: diagram not included]

Model Evaluation

[PMG Ed. Note: diagram not included]

New suite of models

Incorporates latest techniques and thinking in modelling

• Core model
• Small-scale model
• Phillips-curve model
• Vector auto-regressive (VAR) model
• Auto-regressive integrated moving-average (ARIMA) model
• Indicator models
• Fan chart
• Collaborative platform

Core model

• Keep small: 47 equations of which 24 are structural equations
• Focus on key economic variables
• Long-run equilibruim for real variables independent of price levels
• No long-run trade-off between inflation and output
• Takes time to restore equilibruim - short-run trade-off

Transmission mechanism
[PMG Ed. Note: diagram not included]

Small-scale model

• Highly aggregated: 5 equations
• Easier to experiment with different behavioral equations
• Reserve Bank of Australia and Bank of Israel use primarily this type of model

Phillips-curve model

• Directly relates price inflation to a measure of real disequilibrium
• Cross-check forecasts derived from core model

Vector auto-regressive (VAR) model

• Dynamic interaction between a set of variables
• Do not require strong theoretical assumptions
• Useful over shorter time horizons

Auto-regressive integrated moving-average (ARIMA) model

• Single equation model
• Model a single variable in terms of an auto-regressive component and a moving-average component
• No strong theoretical assumptions
• Useful over shorter time horizons

Indicator models

• Identify source of inflationary pressure
• Not used to forecast inflation
• Useful over short time horizon

Fan chart

• Highlights the risks around the forecasts
• Mathematical equation
• Require mode and median

New collaborative platform

• Built around a centralised database
• Communicate results easily
• Effective management reporting
• Econometric functionality

Survey of inflation expectations

• Contracted an independent institution to conduct surveys 4 times a year
• Questionnaires to 4 sectors: Households, business, labour unions, economists/analysts
• BER to publish the results

Technical assistance and advice from other central banks

• England
• Canada
• Australia
• Sweden
• Israel

Forecasting process in the Bank

• Intensive process spanning 6 weeks 4 times a year : from data collection to final discussion at MPC meeting
• Initial assumptions and forecast by technical staff and sectoral experts in the Bank
• Comments and suggestions by senior staff
• Comments and suggestions by members of the MPC
• Discuss forecast, risks and uncertainties at MPC meeting
• Interim periods: "direction of revision forecast"

Schematic illustration of the forecasting process in the Bank

[PMG Ed. Note: diagram not included]

Role of forecasts

• " In an ever-changing economy, no single model can possibly assimilate in a comprehensive way all the factors that matter for policy. Forming judgements about those factors, and their implications for policy, is the job of the Committee, not something that can be abdicated to models or even modellers. But economic models are indispensible tools in that process." - Bank of England

Popular perception about models, forecasts and policy

[PMG Ed. Note: diagram not included]

Correct perception about models, forecasts and policy

[PMG Ed. Note: diagram not included]