The Perspectives editors handed me the following assignment: “Ten years from now, what will be the status of human development in sub-Saharan Africa? What will be the trajectory of the AIDS crisis; what actors—if any—can we expect to address problems of health, education, and poverty? What political changes can we anticipate, and how will these affect standards of living (and vice versa)?” The choice of language is significant. Most studies of development in Africa have focused on growth. However, here the interest is in the degree to which growth and other things, most especially the quality of government, translate into standards of living and aspects of “human development” or “capability,” typically measured by school completion rates, child malnutrition, under age-five mortality, and related indicators.¹

The significance of the exercise is not in doubt. Expressing a widely shared concern, Elsa Artadi and Xavier Sala-i-Martin call Africa's economic stagnation the worst economic disaster of the past century:²

• “[B]etween 1960 and 1980, per capita GDP increased slightly from US $1,500 to about $2,000. It then stagnated at this very low level ever since.”³

  Ibid.

  [These figures include North Africa, which has performed better than points south.]

• The poor have grown poorer and the rich have remained about the same over four decades.
• Forty-eight percent of the population of sub-Saharan Africa consumes less than a dollar a day (less than 65 cents, actually).
• There are increasing differences in equality across countries. Worldwide, income inequality has been decreasing, but in Africa it has increased, and between Africa and the rest of the world inequality has increased.
• Social indicators improve with income, but even the African countries that realized economic growth during the 1990s have underperformed countries in other regions in translating higher income into better welfare.⁴

  White et al. 2001.

Not all the news is bad, however. For example, although levels of under-age-five mortality are high across the continent, a number of countries experienced slight improvement between 1990 and 2001. Southern African countries—for example, Botswana and South Africa—fared worse than many others, probably in large part because of the higher prevalence of HIV. Primary school completion rates deteriorated in some countries, but increased substantially in others, most especially Uganda.

Perhaps replying on the need to build the political capital necessary to address these issues, or responding to the current penchant for unified theories and silver bullets, several organizations have recently tried to offer sweeping diagnoses and projections. In 2001 the World Bank's Strategic Partnership with Africa issued a report, African Poverty at the Millennium, which sought to identify causes and solutions. It emphasized continued reform of economic policy, development of stronger institutions, and a step back from some past policies the Bank had advocated, on the grounds they had unanticipated distributive effects (for example, user fees). This year, Tony Blair and Jeffrey Sachs challenged developed countries to increase aid spending in Africa dramatically (to $25 billion per year by 2010) in order to rise to the challenge.⁵

Underlying Perspectives' questions is a similar enthusiasm for bold projections, one I normally resist, partly because of the analytic challenges involved and partly because of the African continent's enormous diversity. I want to echo the pleas of several economics colleagues that we step back from the general models that underlie a lot of the “growth regressions” that are often the stock-in-trade in our field these days. The remainder of this note offers some pretty safe bets and a prayer that I am proved wrong.

Learning from cross-national growth regressions? In forecasting, one first tries to identify the kinds of things that explain current patterns before looking at trends in these as well as potential shocks or new patterns of causality that might emerge over time. The forecaster's impulse is to reach for a variation on the standard growth model, suitably modified to help us understand the translation of economic growth into human development and to accommodate the effects of HIV/AIDS prevalence on human development both directly and indirectly, through higher business costs, greater insecurity, and possibly lower government capacity. One could then simulate the consequences of fixing component variables at given levels and offer projections on that basis, later modified to take into account the effects of a variety of shocks. The key is to be able to make some straight-line projections, mere extrapolations, and then to modify these by running simulations and weighing alternative scenarios. With respect to Africa at least, this approach runs into trouble fast.

Several challenges make this approach impracticable. To begin, we don't understand much about growth in Africa. Over the past decade, economics journals have hosted many attempts to parse the “Africa dummy,” the variable that signifies the effect of being located in Africa on differences in economic performance. Welfare is not the same as economic performance, but whatever accounts for observed patterns of national income growth is likely to influence human development, either through its effect on income and expenditure, or through expansion of resources potentially available for provision of health care, education, and other services. The bad news is that there is much speculation about the causal relationships the “Africa dummy” captures and little convergence upon a standard story. The Africa dummy is explained away by “malaria” (Sachs), geography (land-locked, also fragile soils), high ethno-linguistic fragmentation (William Easterly), conflict (Easterly, partly Paul Collier), natural resource dependence, coups, institutions (Sachs and Andrew Warner, Collier, Daniel Kauffman), rule of law (Robert Barro, Stephen Knack, and Philip Keefer), and high investment risk (Collier), among other things. The recent Sachs report posited five main reasons why sub-Saharan Africa has fallen prey to a persistent poverty trap: very high transport costs and small market size, low-productivity agriculture, a very high disease burden, adverse geopolitics, and very slow diffusion of technology from abroad.

The lack of convergence on an understanding of growth (much less human development) may have to do with the character of the subject matter and a mismatch between analytic tools and the character of the phenomenon. For example, it is almost certainly the case that interaction effects, thresholds, duration of exposure, and endogenous relationships are important in the study of growth and in the translation of growth into lower child mortality, school completion, and other outcomes. When several of these characteristics are present at the same time, it becomes very hard to use standard regression techniques meaningfully. Better understanding often comes from within-country analysis over time or natural experiments within national boundaries. Moreover, in most instances what matters to policy makers is not the mere fact that the level of an important ingredient of growth is low, but why it is so—and the why is a function of local conditions.

For these reasons I join other voices in suggesting that if the aim is to develop explanations that are sufficiently accurate and specific to generate policy prescriptions or the kinds of projections that take into account the enormous diversity within Africa, one might wish to back away from the quest for a single model that holds across countries.⁶

There are also big gaps in data and knowledge about important parts of the relationship between key variables. For example, the impact of HIV/AIDS on income and welfare remains difficult to capture, in part because of the relative newness of the disease, the slow pace at which the effects unfold, and the still poorly understood influence on household economies, business costs, and institutional capacity. Of the 35–42 million people living with HIV/AIDS globally, around 25.5 million live in Africa. According to UNAIDS, in 2003 HIV prevalence in southern Africa was 16 percent; in east Africa, 6 percent; and in west and central Africa, 4.5 percent.⁷

Thirteen million have already died from the disease. Twelve million children have lost at least one parent. Because population growth will continue on the continent, the number living with HIV will increase through 2025 if current policies continue.

We have fragmentary information about the consequences of the HIV/AIDS epidemic for income and welfare. UNAIDS points to a variety of influences, most discerned only in small scale studies.

• AIDS-affected households suffer a collapse in income. Findings from Zambia suggest that most households experienced an 80 percent reduction of monthly income. In Cote d'Ivoire, incomes fell by half. Effects were smaller in Botswana, but the Botswana studies revealed that each income earner could expect to take on four more dependents.⁸

  Ibid., 31.

• The disease has strongly affected public service. Nearly a third of all teachers are HIV positive, and shortages of teachers are expected in most parts of Africa. Illness and mortality among health care workers has increased five to six times.

• Businesses in southern Africa, including the South African mining sector, will experience sharply higher costs from absences, lower productivity, lower benefits from training, and higher medical costs. About one third of mineworkers in South Africa will be HIV-positive this year. Losses to the labor force in other parts of southern Africa will gradually rise to about 25 percent over the next 15 years.⁹

  Ibid., 32.

The existing growth models for Africa do not incorporate these effects.

Although “institutions” are undoubtedly important in the translation of growth into human development, most studies of institutional performance lack sufficient detail to motivate policy choices among alternative forms. Using International Country Risk Guide and World Bank data for countries for which information is available, there are statistically significant correlations between “bureaucratic quality” in 1995 and 2000 and key human development indicators in 2000/2001 (not controlling for other factors). Unfortunately, we don't really know which aspects make a difference—or even whether, as seems likely, there is something that causes both better bureaucracy and higher levels of human development. Although the theory behind the promotion of democracy as a way of building accountable government is elegant, it is not clear that new democracies perform as anticipated. In policy, as in academic research, there is undoubtedly an overemphasis on voting as a key to better government and a neglect of other aspects of institution building. Again, this is an area where detailed but highly systematic, theory-based field research would be helpful.

Blunt projections. In the absence of a refined understanding of the kinds of things that influence human development in African settings, it is impossible to offer precise predictions. That said, here are some blunt projections:

• Although there are variations in human development across countries in Africa, conflict and HIV/AIDS are likely to depress most human development indicators in most places for the next ten years. Many of the costs of the HIV pandemic materialize only after a delay (for example, the need to care for and train orphaned children, or bankruptcies in the face of high training costs and lower competitiveness).

• Although some countries have experienced impressive levels of economic growth in recent years and have seen an improvement in human development, in most cases rapid improvement has come only after total or near-total economic collapse. The recipes that helped them recover from extended civil conflict or bad macroeconomic policy may not prove useful in moving to the next level.

• Several countries in Africa grew at a fast pace in the 1960s and early 1970s, then experienced a decline after the 1973 and 1978 OPEC oil price hikes. The high cost of oil now will affect the continent adversely. Similarly, if it materializes, as seems likely, an international economic crisis triggered by high U.S. deficits and trade surpluses would undermine growth in Africa.

• The relationship between economic growth and levels of human development is variable. Certainly, higher tax revenues from healthier economies enable governments to enhance public health systems and expand education, as well as build roads. But good governments can help even where growth is sluggish. High HIV-induced turnover in the civil service will complicate efforts to enhance public sector capacity, but innovative solutions could help a government have an impact even amid lackluster economic performance.

• Competition for oil may undercut pressure for better government performance. Already we see China and other countries attenuate demands for accountable government in the Sudan in the quest for oil. China is willing to offer cash and high prices for oil, with the promise that it won't comment on a government's performance. As new oil comes on line in several parts of Africa and demand escalates, international pressure for performance may dissipate unless the international community, including private companies, decide that better government will eventually lead to less conflict and lower costs—and then coordinate their actions.

• Persistently low human development indicators coupled with the effects of HIV and conflict could shorten time horizons and willingness to invest, aggravating the continent's governance problems and increasing the likelihood that armed gangs will move from country to country to loot for survival. Devising ways to draw youth into constructive work and education, as well as buffering neighbors from conflicts next door, will be of utmost importance during the next decade.

• An infusion of aid on the scale Tony Blair has proposed might also be counted as an exogenous shock. Whether it would prove helpful or not is unclear. To date, most studies of aid have found relatively little benefit. Much would depend on how and where the money is spent.

The seeds of improvement after 2015 will be sown in the next few years. A decade with little visible improvement in human development may be inevitable, but the focus of attention ought to be on creating the base for a big rebound thereafter.