Sources: Columns 1-4 are based on World Bank (1997e, Tables 1-5 and 1-8). Column 4 is an approximation, based on the assumption that export growth in 1991-2 was equal to the 1991-5 average for all regions. Column 5 is taken from World Bank (1997d).
However, all of these conclusions depend critically on the reliability of the projections, which may be questionable. Firstly, as discussed in Annex 1, the export growth figures for Sub-Saharan Africa appear inconsistent with the assumptions for commodity prices. This would suggest either that the region's export growth will be slower and/or that commodity prices (and thus the prices of the region's exports) will be lower. In either case, the growth rate of Sub-Saharan Africa's export revenues will be slower, and the sustainable interest rate will be still lower, at least until 2006.
Secondly, there are widespread doubts about the realism and reliability of the Bank's forecasts, which are often seen as being over-optimistic. According to Caufield (1996), for example,
"[The World Bank's] record as a forecaster leaves much to be desired. It failed, for example to anticipate the oil price rises of 1973, 1979 or 1985. The Bank's researchers tend toward overoptimism. Its predictions have frequently exceeded not only reality, but even the estimates of other agencies.... The Bank has...consistently produced overly rosy projections of economic growth.... In 1991, the Bank acknowledged that its predictions of growth were 'too hopeful'. It had, it explained, overestimated how fast world trade would grow, underestimated interest rates, overestimated the flow of capital to developing countries, underestimated how far oil prices would decline, and overestimated global savings." (Caufield, 1996 pp 300-1)
Has the Bank improved its projections since this admission? Some assessment can be made by comparing its projections for the 1990s in the first (1991) edition of Global Economic Prospects and the Developing Countries (World Bank, 1991) with actual outcomes for 1991-5, as recorded in the latest (1997) edition (World Bank, 1997e). At first sight, this suggests that export volume growth projections at that time were, if anything, under-estimated. For developing countries as a whole, export growth in 1991-5 was slightly above the "high scenario" projection (8.5% compared with 8.3%). The results for East Asia, South Asia and Latin America were all above the high scenario by substantial margins (between 3% and 7% PA), while that for Europe, the Middle East and North Africa was slightly above the "downside scenario" (the second lowest of the four provided).
However, there are two critically important caveats.
The conspicuous exception to the pattern of under-estimation is Sub-Saharan Africa, where export growth was in any case the lowest of any region. Actual export growth in 1991-5 was just 0.7% per year, compared with a projection for the 1990s of between 1.3% per year and 4.1% per year (3.0% per year in the baseline scenario).
While export outcomes were much better than expected, this did not feed through into an equivalent over-performance on economic growth - even though other circumstances (eg., FDI and real interest rates) were also more favourable than predicted12.
In fact, developing countries as a whole actually grew more slowly in 1991-5 than the Bank projected even in its worst-case scenario (2.3% PA, compared with 2.9% PA), as did both Sub-Saharan Africa, and Europe, the Middle East and North Africa. Of the regions which exceeded the "high" scenario export growth projection, South Asia's overall growth was in line with the baseline scenario, while Latin America's was in line with the "downside" scenario. While East Asia's growth was again better than the "high" scenario, the margin of over-performance was much less than that for exports (1.7% PA, compared with 7.0% PA); and extending the period to 1998 or beyond would almost certainly eliminate the over-performance altogether.
In other words, while the Bank found their projections for developing countries in the 1980s to have been too positive because of over-optimistic assumptions about global economic conditions, in the 1990s the underlying assumptions were much more cautious and, in the event, seriously under-estimated the reality; but the growth projections were still much too optimistic. This spectacular failure to predict overall economic performance, coupled with the considerable over-optimism of the export projections for Sub-Saharan Africa, raise serious concerns about the Bank's ability to model the linkages from global developments to export performance, or from exports to economic growth.
The Bank's projections of financial developments seem to be particularly unreliable. The 1991 projections predicted average FDI in developing countries as a whole of $27bn PA; even if there were no further growth from the 1996 level, the actual figure would be in the order of $78bn. If the 1990-6 growth rate (28.3% PA) were sustained, it would be $128bn, five times the projected level. The Bank also failed quite spectacularly to predict the East Asian crisis - "the most severe regional financial disruption since the Latin American debt crisis of 1982 and perhaps since the Creditanstalt default of 1931" (Chote, 1998, p2). As late as September 1997, the Bank dismissed the prospect of a "Mexican-style crisis" in East Asia, observing that "while near-term growth will likely slow [in East Asia], a full-blown Mexican-style crisis appears unlikely because of better economic fundamentals in several respects than in Mexico in 1994" (World Bank, 1997e, p21).