those on housing, sewing machines, land clearing, andthe other small-scale outlets for peasant funds dis-cussed above, let alone so much lower as to oﬀsetthe diﬀerence between savings rates (p. 184).
Thefailureofpoliciesaimedatinducingrapidindustrialization discredited belief in a take-oﬀ and stages of growth.
The importance of theagricultural sector as a supplier of raw materi-als, food and labor, and as the home marketfor local industrial output, noted by Johnstonand Mellor (1961), was increasingly recognized.The low productivity of many governmentinvestments and the outright failure of somebecame better known. Structural adjustmentimplied ‘‘a shift away from inward-orientedimport-substituting development strategies tomore outward-oriented ones’’ (Alexandratos,1995), and debt-ridden countries have beenurged to increase and diversify agricultural ex-ports.
If GDP rises smoothly as a result of long-term eﬀects of capital formation, labor forceexpansion and technological change, it can bemodeled by econometric analysis, using the 30– 40 years of statistical data which are nowavailable for a large number of countries.Kenny and Williams (2001) have shownthe limited explicatory power of econometricmodels, and their often contradictory results.They note several causes, including, impor-tantly, the assumption that the process of eco-nomic growth is the same, not only in allcountries, but in all periods of time. Theysuggest therefore, ‘‘that mathematical modelingtechniques have invaded territory to which theyare ill-suited’’ (p. 14).This paper uses a few generally agreedprinciples of economic growth to construct anonmathematical model that reﬂects histori-cal experience. The model conforms to thebroad thrust of available statistical data, butshows that economies are fundamentallydiﬀerent at diﬀerent points in time, particu-larly in the relationship of their agriculturaland nonagricultural sectors. Policies thereforeneed to be varied according to the stage atwhich the country ﬁnds itself, rather thanapplied universally. The development of theagricultural sector is initially extremely im-portant. At a later stage, however, increasesin the productivity of towns are required, toimprove urban incomes, to provide alterna-tive occupations to the rural poor, and tostimulate agricultural investment throughgrowing demand.Tomich, Kilby, and Johnston (1995) alsoreturned to a stages approach in their book,
Transforming Agrarian Economies
. The modeldiﬀers from them in emphasizing the change inmarkets rather than in labor disposition, andthe acceleration of change. They identiﬁedcountries with abundant rural labor with 50%or more of the labor force in agriculture, butthought that
it will be decades before they reach the
structural transformation turning point
, when the absolute sizeof the agricultural work force begins to decline. Untilthen, poverty can be alleviated only if productivityand employment in the rural economy are increased(pp. 9–10).
In fact, a number of countries even in sub-Saharan Africa (SSA) are at or near this point.Calculating from (World Bank, 2002), urbanpopulation in SSA increased at an average of 5% p.a. and rural at 2% p.a. 1968–2000, andaveraged 34% of the total in 2000. It is nowheading for 50% in some (Figure 1). Ruralpopulation growth had dropped to 0.4% p.a. orless in eight out of the 20 largest countries in1998–2000 (Figure 2), and not all rural is ag-ricultural. While policy debates have focusedon the pros and cons of exports, the swiftlygrowing internal market has become muchmore important to farmers, and is a maingenerator of change in farming systems, par-ticularly, but not only, in the semi-arid areas.The urban market is attracting not only theirproducts, but also their labor.The model is delineated and explained inSection 2. Section 3 illustrates the dubiety of much of the population and national incomedata relating to SSA, which lies behind allega-tions of low or negative growth rates in incomeper capita, and falling agricultural output percapita (World Bank, 2000a). Wiggins (1995,2000) has already discussed the conﬂict betweenthese statistics and the evidence of change andgrowth gathered from a limited number of vil-lage case studies. As these can be unrepresen-tative, Section 4 illustrates the rapidity of change in four semi-arid African districts,1960–2000. Some 30% of the SSA populationlive in semi-arid areas (Jahnke, 1982), whichprovides a diﬃcult environment for agriculture.If there are achievements in these areas it islikely that this is also the case in the better-endowed areas. Section 5 considers the policiesand government services likely to be most im-portant in the near future.