Let me begin by welcoming all of you to the National Treasury and many of you to South Africa. I am especially pleased that this convocation to discuss data used for poverty measurement has attracted so much interest. We have here today senior World Bank economists, the chief economist, Franзois Bourguignon, members of the Statistics Council, and the chair of the Council, Hilary Southall, luminaries from the academic community abroad, Angus Deaton and Gary Fields, and from South Africa, representatives of some of our bilateral partners, and of course our ubiquitous government officials from the Presidency, Statistics South Africa, the National Treasury and other government departments. I would also like to extend a special word of thanks to Franзois and his colleagues for suggesting that we jointly host this workshop.
As some of you know, my work in South Africa and in the international community, and in particular in Africa, has had data and tools for measuring the economy as a centrepiece. Whether we wish to adjust the size of our fiscal deficit, increase social spending, pursue macroeconomic convergence in the region, or assess progress in achieving the Millennium Development Goals, accurate, timely, useful data lies at the heart of all these efforts.
That is to state the obvious. What is less obvious, and what many of you have spent your lives doing, is to work with the data, assessing its usefulness, applying techniques to uncover the facts and dynamics lying buried beneath the raw numbers, and applying what you have unearthed to the questions of public policy. Fortunately for your careers, and for the careers of many who follow you, this is a limitless task. And it is especially limitless in Africa.
In South Africa, the limits of these activities have been pushed out by the regime that came before our current democratic dispensation. To put it simply, data collection was bad and biased. The apartheid regime knew little and cared not at all about the well being of the majority of the population.
As a result, not only did we have widespread grinding poverty, but also we had few tools for assessing it and aiding us in developing policies to address it.
Obviously, this has made our task since 1994 rather difficult. As a result, the conduct and design of our household surveys have undergone nearly continuous revision, and as we know, require still further work.
In the coming year, Stats South Africa will pilot the income and expenditure survey using a "diary" method, which we hope will provide better data.
The main reason for this workshop is to explore ways of making the income and expenditure data useful. We should also benefit from discussion of the diary method, potential pitfalls in its use, and direction on how to address them.
Our 2001 Census was a great success, and researchers are now coming to grips with the data that it provides. But conducting censuses also represent a major institutional and operational challenge, and we will continue to learn and develop better means of addressing those challenges - with ever better census results.
Our labour force survey and general household survey, which replace the old October Household Survey, are great improvements.
But even with these, more work needs to be done. Informal sector employment, for instance, has proved volatile. Is it from changes in seasonal employment, miscoding of responses, or a flaw in the design of questions? We have not as yet come to a common view of the problem.
And of course, as we will discuss today and tomorrow, our measurement of poverty does not appear to be as good as it should be. Understanding why this is the case, and what we should do about it, is the test for us now.
The broader challenges around the question of statistics also need consideration, but I don't expect you to dwell on them over the next few days. I will, however, take the opportunity to raise some of them now, because they should provide the backdrop for our discussions. They may also lend themselves to further discussions on next steps in the development of our statistics.
The World Bank, the Fund, and bilateral donors should have as part of their focus in developing countries, advanced and streamlined programmes to build capacity in statistics, the collection of statistics, and helping to develop the institutions that produce statistics. I am certain that by exposing us to new methodologies and insights on the use of data this workshop will excite some of our younger guests to further their careers in statistics and help them to build the institutions and capacity we need.
But one workshop alone is clearly insufficient for South Africa's data needs, let alone the needs of our neighbours. So we need to think further ahead, discuss with the Bank, the Fund and our bilateral partners how to progress much more rapidly in this area, and especially how to extend these ideas and potential programmes to our regional neighbours as well. I would hope that one concrete follow-up to this workshop would be another joint effort to put onto paper the outline of such a joint programme for the region.
The Bank, our bilateral partners, and our academic guests here might also consider assisting Statistics South Africa in its design of surveys and operations, as well as the Statistics Council in its responsibilities for overseeing the quality of our official statistics. But there are many other areas of need, not least the teaching of statistics in our universities, and any of them would benefit from your expertise and enthusiasm (as I know some of them already have).
Building confidence and developing policy
Since the development of our statistics is an ongoing task, it also seems important to build credibility in them. Credibility and confidence are critical to ensure the public uses our statistics and that they believe the numbers reported. Accuracy, one aspect of credibility, is critical for the development and adjustment of policy.
Building confidence in our statistics is neither a short-term endeavour nor is it easy. It is, however, necessary and one to which we are committed, and this implies that as researchers and policy makers our critical faculties must be exercised. It is not sufficient to extrapolate from contested data into policy, although the temptation is always present. Using contested data creates the danger that we talk past each other, either because we use different sets of statistics, or because the different sets don't confirm each other.
As policy makers, and especially those of us, in developing countries, we cannot afford to make large mistakes, although smaller mistakes are inevitable and probably necessary to arrive at the right policies.
So let me speak a little about what we are doing as South Africa policy makers to address the poverty challenges that face us. A recent report for South Africa concludes that the proportion of people in poverty in terms of a South African poverty line has declined (from 51,1% in 1995 to 48,5% in 2002) but that the absolute number in poverty has increased from 20,2 million in 1995 to 21,9 million in 2002. The report also argues that the proportion of the population with living less than $1 per day has increased from 9,4% in 1995 to 10,5% in 2002.
In broad terms there seems to be agreement on the groups of people most vulnerable to poverty - inhabitants of rural areas, women-headed households, the disabled, retrenched farm workers, the elderly, and, increasingly, those directly affected by HIV/AIDS. Poverty has a strong racial dimension, with a far greater proportion of Africans being poor.
It also seems to be the case that poverty is more likely if people don't have jobs, a finding that I suspect is common to many countries. But it suggests that our large number of unemployed would materially benefit - move out of poverty - if they were able to take employment. And clearly Government needs to design policy so that demand for our products is maximised, as is our ability to produce them at a reasonable cost that uses as much labour as possible.
At the risk of being trite, we need to create more jobs. But this also should be done in the context in a shift in the attitude of many South Africans - and better statistics has a role to play here. We tend to take a dim view of informal sector economic activity, but given the insufficiency of our data, we know next to nothing about it. What sort of data should we have to explore the informal sector, to understand its dynamics, to help us understand the incentives to enter the labour market, to tell us whether ensuring the enforcement of basic labour regulations would be more employment-creating and better for welfare than extending to small business the regulations designed for Anglo-American and Iscor. Achieving the right policies requires the management of trade-offs informed by good statistics. This prerequisite for improvements to government policies holds most strongly when it comes to fiscal policy.
Managing trade-offs in fiscal policy
From 1994, Government policies and programmes have been guided by an ambitious programme for social reconstruction, the Reconstruction and Development Programme, focused in four broad areas - meeting basic needs, building the economy, democratising the state and society, and developing human resources and nation-building. Key underlying objectives were to deal with social legacies of apartheid, particularly our entrenched poverty.
To engineer a decisive shift in the focus of the state apparatus, so-called "lead programmes" were identified and many started through earmarked funding (the separate RDP Fund). Special and separate poverty relief allocations were made and specific institutions were also established (National Development Agency, Umsobomvu).
In recent years, however, Government has in some sense moved away from a broad strategic statement on poverty reduction (one could call it the "Poverty Reduction Strategy Paper approach") and earmarked funding for poverty relief. Instead, Government has opted to adjust its overall orientation within a broad fiscal stance that seeks to maximise available resources for social and economic expenditure without compromising the sustainability of the growth of those resources.
The composition of expenditure has been adjusted to focus resources at direct and indirect mechanisms of poverty relief, while maintaining a balance between expenditure that supports people's ability to engage in productive economic activity (education, housing, transport, skills development, economic services and infrastructure) and providing direct, welfare-type, support (cash grants, school feeding, food relief). Strengthening the assets of the poor through land reform is another major commitment.
Given our improving macro-economic and fiscal balances, we have been in a position in recent years to strengthen spending in a range of areas, including in economic infrastructure, income support through increasing access to cash grants (especially the means-tested child support grant), and also extending school feeding and emergency relief (drought, food prices).
For all services, real allocations have been maintained or increased although in some areas such as health the demands are large in the context of the HIV and Aids epidemic (1). Our system of social grants (means-tested cash grants to the elderly, disabled and families with children) is widely regarded as the country's primary tool for poverty relief. Currently 2,1 million elderly, 1,3 million disabled and 4,3 million poor children between ages 0 and 9 years receive grants.
Total beneficiary numbers increased from about 2,5 million in April 1997 to 7,9 million in April 2004 - mainly driven by the implementation of the child support grant from April 1998 onward. At this stage, about 17% of South Africans receive cash grants and this proportion will continue to increase as the child support grant is extended up to the age of 13 over the next two years.
This has resulted in an increase in the cost of grants as a proportion of government spending and GDP, rising to 13 and 3 percent, respectively.
Two data-related issues stand out from this brief review of policy. The first is the need for any serious measurement of poverty to take into account the impact of government spending, or the "social wage." The second is the need for data that helps us to understand how our system of grants and other social spending is biased against some people and over-favours others.
Other questions arise. We need to know why some people may not receive income support that they are entitled to, and why some people are better able to take advantage of economic opportunities that do arise. Answering those sorts of questions, and many others, are critical for the policy development process and making further inroads into poverty.
Within the National Treasury, we have engaged for some time in a continuous process of Budget Reform, which to my mind is an important support-type function, which does not directly alleviate poverty - but is essential for ensuring (and monitoring) that the right information flows into prioritisation and funding processes.
Among other things, this entails ensuring that allocation processes are sufficiently open and stable, and politically driven, so that the real priorities around poverty and development are served by allocations.
We therefore see our Budget Reform programme as a critical part of a poverty alleviation strategy. Budget reform has helped to increase the transparency of the budget process and enhance decision-making by political principals. The three-year budgets allow for greater certainty and increased focus on reprioritisation, and - an increased focus on performance assessment and value for money.
Eradicating poverty is complex and takes time. Seemingly simple solutions that rely on weak and generalised assumptions about who the poor are, where they live, what they need and what they want, are destined to fail.
We need to answer these questions. The data must be reliable, verifiable, corroborated. The data must be collected and produced by an efficient, state of the art, statistics institution with the systematic help of the academic community. And, of course, this data must be used effectively... by applying the right techniques that help us to answer questions that are both of an academic interest in their own right, but also satisfy the needs of policy makers.
In terms of our poverty challenge, currently we have too little disaggregated information on the characteristics of the poor and why they are poor, and too little information about the dynamics of poverty. The data needs to help us to measure the impact of our poverty alleviation programmes, to identify areas of success, weakness, and failure in the delivery of income support and the delivery of economic and social services.
Obviously, as South Africans, we have much to do. And I am conscious of having not made your deliberations any simpler. I have, however, tried to provide some sense of the importance of this workshop to me, and to the South African Government more generally.
(1) Most concrete progress has been with regard to access to some basic household services: 1,6 million new houses, 700 primary health care clinics, improved access to water and sanitation, million new electricity connections, increased access to telephones.