From the point of view of our exercise, this greater variance produces a ‘bias’ in the resulting estimates of the connection between agricultural income and welfare, since we are interested not in the short-run effect of things like weather shocks on expenditures but on the longer-run effects of things like improvements in agricultural productivity. We are also concerned about the related issue of endogeneity; even the simplest general equilibrium models with investment imply simultaneity in the determination of income and expenditures. We address these issues using a simple instrumental variables strategy, using averages of neighboring countries’ sectoral income growth as instruments for own-income growth . Fifth, even after controlling for time, continent, and decile fixed effects in growth, we are concerned that there may be heterogeneity across countries in the way agricultural income growth affects households in different parts of the expenditure distribution. We explore this possible heterogeneity by interacting various fixed or pre-determined country characteristics with income growth from different sectors, reporting those results in Section 5.4. We summarize our main results. First, poorer households’ expenditures grow more in response to growth from agriculture than do the expenditures of wealthier households, and this holds across all deciles. We call this result monotonicity, and it is both very robust and important.
Monotonicity also holds for growth from non-agricultural sources, but in the opposite direction, vertical farming aeroponics with wealthier households’ expenditures responding more than poorer households’. Second, it is not just across deciles that we see an effect: within poorer deciles, households benefit significantly more from growth in agriculture than they do from growth in other sectors. Third and finally, the connection between expenditure and sectoral income growth is importantly and significantly different across different groups of countries. In particular, it is the poorest households in the poorer countries for whom agricultural income growth is most important.From a theoretical standpoint, a long tradition of dual economy models that aggregate the economy into two sectors—agriculture and non-agriculture—has served to identify the transmission mechanisms of an exogenous agricultural productivity increase on welfare . Transmission mechanisms include employment, food prices, real wages, and the demand for non-tradable goods produced in the rural non-farm economy. The tradition in the dual economy literature is to assume that consumption expenditures are equal to real income and that labor income is the source of expenditures while capital income is saved and invested. An increase in growth in one sector would affect the welfare of only the part of the population actually employed in that sector. If expenditures are distributed differently across households in the two sectors, then an increase in employment in one sector will have an effect on the aggregate distribution of expenditures. If, for example, households employed in the agricultural sector tend to be poorer, an increase in agricultural employment will have an equalizing effect on the entire distribution of expenditures .
For a country with a closed economy , an increase in agricultural productivity induces a decrease in food prices. All consumers benefit from lower food prices, but most particularly the poor, who typically spend a larger share of their income on food . If there is surplus labor and wages are tied to the cost of living to secure a fixed real subsistence wage, lower food prices can induce a decrease in the nominal wage, fostering employment and growth in the non-agricultural sector . When workers are mobile and wages are equated across sectors, differences in the rate of growth of different sectors can result in changes in the distribution of expenditures through the employment effect. For example, Loayza and Raddatz formulate a model in which expenditures of the poor are equal to the prevailing wage, while non-poor households can borrow or lend to smooth away the effects of variation in labor income on expenditures . The model shows that the effects of sectoral growth on real wages are larger for sectors with larger employment and a lower elasticity of demand for labor, namely agriculture and services. Another strand of literature is based on a three-sector aggregation of the economy, with a non-tradable sector in addition to the agricultural and, say, manufacturing sectors. A key determinant of the overall effect of an initial growth impetus in agriculture is the linkages created in fostering demand for the non-tradable sector products . To the extent that labor is not fully mobile, then in addition to asymmetric effects on the functional sources of income any growth that originates in the rural economy stands to have a more direct impact on the rural population, where many of the poor live. Much of the empirical support to the claim that agricultural growth is good for aggregate growth, employment, and welfare is based on simulation models that rely on demand and supply elasticities that are not estimated.
Thorbecke and Jung use social accounting with postulated elasticities applied to Indonesia, thus finding that agriculture and services contribute more to poverty reduction that the industrial sectors. Within-country or within-region studies arguably offer the best evidence we have on the connection between aggregate agricultural income growth and household welfare, perhaps because in these contexts one can construct a proper panel dataset. In an important series of papers Datt and Ravallion use panel data for states in India and show a systematic and relatively uniform association between agricultural growth and poverty reduction, but a very heterogeneous relationship between non-agricultural growth and poverty change. With province-level panel data for China over the period 1985–1996, Fan et al. find that agricultural growth is associated with a reduction of rural poverty while non-agricultural growth is associated with an increase in rural poverty. With provincial data for 1983–2001, Montalvo and Ravallion show that the primary sector was the driving force behind the spectacular decrease in poverty in China. Suryahadi et al. conduct an exercise similar to that of Ravallion and Datt but for Indonesia, and are able to distinguish between the rural and urban poor. They find that growth in services is good for both the rural and urban poor, with the effects of agricultural growth focused more specifically on the rural poor. In a similar spirit, Warr uses national data from four Asian countries from the 1960s to 1999 in a panel analysis and finds similar results, in that growth in agriculture and services were associated with a decrease in poverty, with the estimated coefficient on agriculture substantially smaller than the coefficient on services, and the coefficient on manufacturing not significantly different from 0. Looking at the 25 countries with the greatest success at reducing extreme poverty under the period of the Millennium Development Goals, Cervantes-Godoy and Dewbre find that while economic growth was a key determinant, growth in agricultural incomes was especially important. Bresciani and Valdes provide evidence of the role of agricultural growth on poverty reduction through rural labor markets, farm incomes, food prices, and economy-wide multipliers in different country case studies.
Other studies have resorted more systematically to cross-sectional country-level time series data, thus looking for average effects across a large set of countries and hence economic structures. Using data from 80 countries spanning 1980 to 2002, Christiaensen et al. find a stronger association between overall poverty decrease and growth originating in agriculture than growth originating in either of the other two sectors. With higher participation, slower growth of agriculture may still deliver more poverty reduction than the growth of non-agriculture. In contrast, using a slightly different method, Bravo-Ortega and Lederman find that in Latin America, it is the non-agricultural sector that has the strongest effect in reducing poverty. Focusing on the role of the unskilled labor market, Loayza and Raddatz find evidence that growth in income from sectors with high unskilled labor shares has a disproportionate effect in reducing poverty rates. In a somewhat different specification, Dollar et al. regress growth rates in incomes of the poorest 20 percent on growth in average income and on changes in the share of agriculture in GDP. The significance of the coefficient on the agricultural variable suggests that, even controlling for aggregate growth, faster growth in agriculture is likely to disproportionately benefit the poor. Lanjouw et al. for example argue that it is the non-agricultural sector in the rural areas that is both more dynamic and more pro-poor,vertical indoor hydroponic system and hence the most important contributor to poverty reduction in rural India. Collier and Dercon note that productivity in agriculture, and especially in the smallholder sector, is so low that economic development and poverty alleviation in Africa will have to come from a radical transformation of the agricultural sector and massive exodus from agriculture. They also cite works on the role of migration in the reduction of poverty in rural areas. Most of the literature that cautions against the importance given to agriculture for poverty alleviation however relates to a different argument: while the relatively strong poverty impact of agricultural growth seems to be a fairly robust result, the cost of investing to obtain a given growth is far higher in agriculture than in other sectors, making it an inefficient instrument for growth and welfare . Our paper does not address this issue at all, but aims at contributing to the literature on the sectoral growth-poverty linkage. An issue in almost all of the studies we have discussed is simultaneity between sectoral growth and the welfare indicator used in the analysis. A contribution of this paper is to tackle this issue by using an instrumental variable approach to try to measure the effect of an exogenous increase in sectoral growth on welfare.
We use the same database collected by the World Bank as do other cross country analyses, although we only select the countries for which welfare is measured by consumption expenditures.2 We also use data on all deciles, rather than only on e.g., poverty rates, as in Christiaensen et al. and other studies described above. When using cross-country evidence on changes in the distribution of income or expenditures one has to make an early choice regarding whether it is better to consider the distribution of these welfare measures within countries or across countries. The former choice leads to an empirical strategy that groups together different welfare quantiles across countries, so that for example, one imagines that the poorest 10 percent of households in Tanzania are similarly positioned to the poorest 10 percent of households in China, despite the substantial differences in the level of real expenditures of the quantile across these two countries. The latter choice construes distribution as a global phenomenon, with the result that the poorest 10 percent of all households globally may all be located in a very small number of countries. If what we want to measure is the global distribution of welfare one also logically ought to weight countries by their populations in any cross-country analysis. different researchers have made different choices.3 In this paper we take the country focused approach, and analyze the relationship between welfare and sectoral growth of all deciles of the distribution within countries, rather than on a measure of poverty level or distribution across countries.4 Over the last several decades, the World Bank has accumulated a large number of datasets from a large number of developing countries which are based on household-level surveys, statistically representative of the populations of those countries, and which include data on non-durable goods expenditures. Though the micro-data from these surveys are not generally available, the World Bank provides data on aggregate expenditures by decile for many of these countries. Our sample is restricted to the countries and years for which we have information on expenditures data for at least three points in time . The sample covers 62 countries, with variable numbers of observations over 1978 to 2011, totaling 310 surveys. This sample of countries and years is not a random sample of the countries of the world. Instead, it is a sample of countries where household expenditure surveys have been conducted . It has however a large coverage, including 81% of the population in low and middle-income countries in 2000. In terms of continents, the sample includes 97% of the population of South Asia, 70% of Sub-Saharan Africa, and 20% of Latin America and Caribbean.There is no clear bias in this sampling of developing countries except for the obvious and egregious absence of all but one Latin American countries.