Are there public good justifications for worrying about rural incomes?Agriculture in Pro-Poor Growth:A South American Perspective William Foster Can new products and new international and domestic marketing channels increase rural incomes? Ideas, models, and evidence. Sonoma, June 19-20, 2005
The importance of overall growth to poverty reduction • Dollar and Kraay (“Growth is good for the poor,” 2002) • The growth elasticity of poverty is one. • Policy-related factors usually considered important to reduce poverty (health and education spending, farm labor productivity) were found to have little marginal effect on the average income of the poorest. • Controversy: Ravallion and others. • The lesson we learn from the literature is that economic growth can be more pro-poor in some circumstance and less in others, but just by itself, growth is pro-poor.
What can agriculture do to promote overall growth and poverty reduction? • Recent World Bank effort for Latin America and the Caribbean. • How big is agriculture? • Econometric evidence of the contribution of agriculture to growth. • Econometric evidence of the contribution of agriculture to poverty reduction. • Policy implications: assisting agriculture in promoting overall growth. • Beyond agricultural production to the importance of the rural economy as a whole.
Basic results: • Agriculture can promote growth directly through its own expansion, and indirectly through its spillovers to the rest of the economy. • Agricultural growth in contrast to non-agricultural growth can contribute differentially to the increase in the income of the poorest, • and sometimes its contribution to income growth of the poorest exceeds its relative size in the overall economy.
How big is agriculture? • Development lit. in 1950s generally pessimistic with respect to the sector’s potential for productivity and export growth (Hirschman, Prebisch, Lewis). • A reassessment of the efficiency of producers and the sector’s growth potential, following the work of Schultz. • Large lit. on farm sector’s significant multiplier effects (Hazell and Bell, Haggblade, Mellor, Delgado, Adelman) • Consumption effects dominate linkages in low income countries. • Probably production links dominate in middle income.
An expanded definition of agriculture using input-output linkages: Source: Authors’ calculations in Beyond the City (2005), based on methodology in Anríquez, Foster and Valdés (2003).
The contribution of agriculture to growth:“old and honorable question, dating back to the Physiocrats.” An equation of interest • With no spillovers, the elasticity of national growth (GDP p.c.) with respect to sectoral growth is share of GDP (SA) • With spillovers, we must estimate the cross-sectoral growth elasticities. • How can one estimate externalitities and multipliers? • Timmer (2002), Bravo-Lederman for World Bank, Beyond the City (2005)
Timmer, using 65 developing countries (1960-1985): • A contemporaneous increase of one percent in the growth rate of agriculture would contribute to about a 0.2 percent increase in the non-agricultural growth rate. • But this does not show causality (both sectors could have grown in response to other factors, such as macroeconomic policies). • More interestingly for inferring causality, a one percent increase in the lagged agricultural growth rate (five years) would contribute to about a 0.14 percent increase in the non-agricultural growth rate.
Bravo-Lederman (World Bank), using 120 countries (1960-2000): • Their approach controls for non-agricultural GDP level (a way to control for the level of development). • Granger “causality” tests. • Does agricultural growth lead non-agricultural growth, or vice versa, or are there reverse effects, each sector to the other?
Which sector leads which? • In developing countries historically a 1% increase in ag. growth leads to between 0.12 % (LAC) and 0.15% (other developing) increase in non-ag. growth. • High income: ag. growth has been associated with a subsequent fall (–0.09) in non-ag. growth (resource pull?). • Also a reverse effect: a 1% increase in non-ag. growth rate leads to a decrease in ag. growth in non-LAC developing countries. • In LAC and developed, non-ag. growth appears not to be related one way or the other to subsequent ag. growth. • But much heterogeneity within country groupings.
Cross-sector growth elasticities: A 1% increase in one sector’s GDP on the other sector’s GDP. Granger-causality tests from Bravo-Ortega and Lederman (2005), also Figures 3.3 and 3.4 and Table 3.16 in de Farranti et al. (2005).
Econometric evidence is consistent with the I-O linkage results • Chile has the strongest linkages (particularly forward) and the strongest cross-sector growth elasticity. • Mexico has weaker linkages than Chile, but stronger than Colombia, and has a cross-sector growth elasticity double that of the regional average. • Colombia has the weak linkages and a cross-sector growth elasticity equal to the regional average.
Extending the definition of agriculture to include the food processing: • LAC average cross-sector growth elasticity from ag. to non-ag. increases to 0.18. This suggests that positive spillovers of ag. are stronger when the downstream industries are included in the “rural” economy. • By contrast, with food processing in non-LAC developing countries’ ag., average cross-sector growth elasticity falls. • Interpretation: in non-LAC developing, much of the subsequent growth in non-ag. related to current primary ag. growth is in processing industries related to ag. A large part of what is measured as the non-ag. growth correlated with ag. is in the food processing sector. • Ideally, we should disaggregate non-LAC.
Considering ag’s 2 contributions, direct (GDP share) and indirect ( ): • For non-LAC developing ag. contributes 1.5 x the size of the sector to growth. For LAC ag. about 1.8 x size. • Non-LAC developing : non-ag. contributes slightly less its share to GDP growth. (Resource pull?) • In LAC and developed the non-ag. contribution is approximately equal to its share in GDP. • Significant spillovers of ag. to non-ag. in developing countries. (And comparable to Timmer’s long run.) • Ag. growth would lead over time to a lower share of agriculture in total GDP, which corresponds to the historical trends we have seen.
Agriculture’s GDP share falls with development per capita GDP in 1995 US$, annual data from 1960 to 2002
What are the policy implications of a strong indirect effect of ag. on non-ag. (and a lesser reverse indirect effect)? • Not an argument for subsidizing agricultural production! But an argument for not taxing the sector. • Granger tests suggest but do not really prove causality in a mechanical sense: it shows predictive links. • Mechanisms must be understood. • There is a cost of underinvestment in rural public goods relative to urban. (More on waste later.) • Social project evaluation: meant to understand the mechanisms. • Important to screen public projects, promote high and reject low returns.
The contribution of agriculture to poverty reduction: Differing sectoral impacts on poverty • With spillovers, EoC is augmented by the indirect effect of cross-sectoral growth impacts. Relate average income in each quintile (j = 1,…,5) to the sectoral labor productivities (gi = Gi÷ Li : Li labor force in Ag. and Non-Ag. We would like to know the Elasticities of Connection:
Income disparities matter for ag.’s impact on the income of the poorest. • Timmer: where disparity between richest and poorest small, growth in Ag. labor productivity is “slightly but consistently” more important in generating p.c. income in every quintile. • When income gap large, EoCs of both sectors for poorest quintile small, but rise sharply by income class. • High income gap countries: poorest quintile is “nearly left out of the growth process altogether.” • Implication: Ag. growth less successful than Non-Ag. growth at raising the incomes of the poorest for countries with high income disparity.
Bravo-Lederman (World Bank), using 84 countries (1960-2002): • EoCs (direct effects on poverty) are higher for Non-Ag. than for Ag. growth across quintile groups. • For non-LAC developing, the EoCs for the poorest quintile are 0.36 for ag. and 0.64 for non-ag. • Absolute impact: Non-Ag. growth more important than Ag. growth, in both LAC and non-LAC developing. • Relative impact of Ag. growth least for the lowest quintile compared to higher quintiles (Timmer’s inequality scenario). • EoCs for Ag. compared to non-ag. even less for LAC, where the Ag. EoCs fall relative to non-LAC developing countries and the Non-Ag. EoCs increase.
But there are indirect effects of Ag. on poverty via influence of Ag. growth on Non-Ag. growth, stimulates poverty reduction as well: • For LAC total elasticity is 0.28 Ag. and 0.77 Non-Ag. • Other developing countries: 0.48 Ag. and 0.58 Non-Ag. • Note: indirect effect of Ag. growth on poverty is a notable proportion of total effect both in LAC (≈0.33) and non-LAC developing (≈0.20). • Note: relative to LAC, in non-LAC developing Ag. growth has slightly higher impact on Non-Ag. growth, but Non-Ag. growth has smaller impact on poverty. • So: for non-LAC developing, the direct effect of Ag. growth is relatively more important than in LAC. • Nevertheless: in non-LAC developing, growth of the Non-Ag. sector still more important (in absolute terms).
Relative to its GDP share agriculture has a greater impact on poverty reduction than non-agriculture: • Agriculture’s GDP share averages 0.12 for LAC and 0.22 for non-LAC developing countries. • Relative to their shares in GDP, on average, agriculture’s contribution to raising the incomes of the poorest is at least 2.5 times that of non-agriculture • 2.5 for LAC, 2.9 for non-LAC developing countries.
Partial and total sectoral poverty elasticities: Impact of a 1% increase of each sector’s GDP on average income of the poorest quintile.
Did the expansion of Chilean agriculture after the reforms contribute to the alleviation of national poverty? • Lopez and Anriquez analyze sector’s influence on poverty alleviation via impact of ag. growth. • Skilled and unskilled labor the focus. • Important finding: asymmetric response of the two types of labor to expansion of agriculture and non-agriculture. • Demand for unskilled workers more sensitive to expansion of agriculture than skilled workers • Critically, expansion in ag. Leads to more unskilled labor demand compared to expansion in non-ag. output. • Increasing ag. GDP by 1% (with 0.17% decline in non-ag. output – a zero-growth scenario) would lead to a 0.51% expansion of the employment of unskilled.
The importance of a growing agriculture sector for poverty alleviation in Chile: Estimated Labor Demand Elasticities (evaluated at sample means) Note: Standard errors in parentheses. *** Significant at the 1% level, ** Significant at the 5% level, *Significant at the 10% level. Source: Lopez and Anriquez (2004).
Consolidated effects of Chilean agricultural growth on decreases in poverty headcount. *Elasticity estimates based on price, wage and employment effects taken together. Other elasticities extimates for the effect of aggregate growth on poverty in Chile are in the range of 0.8 to 1.2. Source: Lopez and Anriquez (2004).
What are the implications for government expenditures? • For developing countries a 1% increase in Non-Ag. growth has stronger impact on both growth and poverty reduction than a 1% increase in Ag. • If the decision is to choose between a 1% growth in Non-Ag. versus a 1% growth in Ag., one should select the growth in Non-Ag. • But, of course, the policy choice is not trading off a 1% increase in growth in one sector foregoing a 1% increase in growth in another sector. • One cannot trade-off growth rate percentage points one-for-one as if they summed to a budget constraint.
From a public policy point of view, what are the restrictions? • Human resources: • The marginal impact of human expertise in promoting non-agricultural growth directly is very likely significantly larger than in promoting agricultural growth directly. • Improvements in “non-agricultural” policy would certainly benefit agriculture. • Budgets: question of public expenditures is different. • Should it be allocated according to the relative size of the sector in national GDP? • Yes, if no cross-sector growth effects, and if additional monies are equally efficient in promoting sectoral growth across sectors, and if the only objective is increasing national growth.
The answer is, no, however, if relative to their sectoral shares agriculture contributes more than non-agriculture to national growth. • Ag. contributes about 1.8 times its size the contribution of Non-Ag. to national growth in LAC countries. • For non-LAC developing, Ag. contributes about 1.6 times its size the contribution of Non-Ag. to national growth. • If Ag.’s GDP share is 12% (LAC average), using estimated cross-sector growth effects (and only interested in promoting growth), one should allocate to Ag. on average slightly less than twice its GDP share (24% of budget). • If Ag.’s GDP share is 22% (average non-LAC), should allocate to Ag. 1.8 times its GDP share (40% of budget). • For Non-Ag., the government should allocate less of a windfall than its sectoral share. • Very important caveat: assuming that expenditures are equally efficient in promoting sectoral growth.
The importance of the composition of rural public expenditures: • LAC FAO data set on public expenditures in rural areas used by Lopez: while public spending levels can promote slightly Ag. GDP per rural person, mix of spending much more important. • Reallocation of 10 percentage points of total rural public expenditures (e.g., from 40% spending on public goods to 50%) raises Ag. GDP per rural person by 2.3% – and this without spending a tuppence more in total. • A dollar added to total rural expenditures would be shared by public and private goods. An intra-marginal shift of a dollar is claimed entirely by public goods and is lost to private subsidies, and the latter might in fact have a negative contribution to agricultural growth
Beyond agricultural production to the importance of the rural economy as a whole: • Gardner concludes that where growth in rural household income has been achieved, 5 factors present: • Macroeconomic and political stability, • Property rights and incentives, • Productivity enhance new technology, • Access to competitive input and product markets, • And real income growth in the Non-Ag. economy. • “To remedy rural poverty, what is most needed is improvement in the labor market generally more than, say, improved crop varieties. This is not to say that agricultural research and rural infrastructure investment are not valuable, or that the net effect on poverty is in the right direction…. Agricultural economics is the discipline that can analyze the possibilities of these and other profitable investments… Yet it is becoming evident that rural growth and poverty alleviation are not sub-fields of agricultural economics.”
The policy point: not exclusively sectoral but national and territorial. How to facilitate the transition from a rural economy based on small farms to one diversified in income sources, competitive in internat’l markets, dynamic. • This might mean migration, public policies to facilitate the Non-ag. rural economy, education, R&D, public infrastructure, etc. • But there is an analytical problem: we know more about how to promote agricultural growth than we do about how to promote non-agricultural rural activities. This is because we know so little about what drives the rural economy and territorial development. • As Hewings notes, there have been many fads in territorial development, but few satisfying economic evaluations.
The importance of non-farm employment and income: • The evidence suggests that wages, self-employment outside agriculture, and other earnings from commercial activities, manufacturing and other services are significant sources of income for rural households. • Rural nonfarm income tends to be positively correlated with national development and case studies indicate positive growth over time of nonfarm income as a share of total household income in rural areas. • As a proportion of total employment in rural areas, nonfarm employment averages approximately 25% in Latin America and 44% in Asia, usually representing a lower share than nonfarm income relative to total income.
A portfolio of livelihoods for generation of rural income • Evidence not only for Latin America. Foster and Rosenzweig (2004) underline importance of labor intensive small scale manufactures in India. for the generation of rural income • Frank Ellis and Freeman (2002) (Uganda, Kenya, Tanzania and Malawi) conclude: • “Better off households are distinguished by virtuous spirals of accumulation typically involving diverse livestock ownership, engagement in non-farm self-employment, and diversity of on-farm and non-farm income sources. Lessons for …the creation of a facilitating rather than blocking public sector environment for the multiplication of non-farm enterprises….”
Productivity drivers found in the rest of the economy have implications for scale in farming • Requiring investments and fixed costs that influence the “structure of farming” – the distribution of farm sizes and returns. • The rest of the economy and trade is changing, producing a flow of new opportunities for farmers. • A major issue for this conference is to what extent small farmers can participate both as producers and workers in taking advantage of this flow of opportunities. • What can be done to encourage their participation?
But is there anything inherently pro-poor about maintaining the existence of a large number of small farms? • No, the point is not to maintain millions of small farmers, but to eliminate poverty. • BUT: we have to deal with what we have. • Best case: dynamic overall economy spurred by Ag. growth and in turn spurring the Ag. growth and the structure of farming, while at the same time offering better non-farm income opportunities. • Beyond having a stagnant economy, most discouraging case: dynamic economy, both domestic and global, driving changes in the structure of Ag. production but not providing opportunities to the poor, who then remain as small farmers without being able to take advantage of non-farm employment opportunities • for geographic, demographic and cultural reasons, and labor market restrictions.
This leads to a discussion of safety nets • More R&D, better infrastructure and rural finance, and other initiatives in poorer regions will improve Ag. productivity and the rural economy generally. • But some households in some regions will lag behind, because they do not have the assets that complement the positive developments elsewhere. • Age, illiteracy, poor soils and climate, low population densities all will conspire against finding productive employment opportunities, whether on or off the farm. • How should policies be designed to aid these households and communities?
Many governments have made use of programs to provide income support and cash transfers • Some compensatory response to trade liberalization, but more broadly applicable to lagging regions and families. • Focus on income based rather than farm related support. • 3 common features for the conditional cash transfer payments: • targeted to rural areas and to poor households, • based on number of children, • continuation contingent on final impact on human capital. • Perhaps the most important point is to support the long-term human capital investments in children. • Objective : not to keep people on the farm or even rural, but to promote opportunities and mobility of future generations (economically and geographically). • For the immobile, such as the elderly, other safety net aspects are being adopted in several developing countries.
Back to the importance of growth • Middle income countries have at least the potential to finance such schemes, and in many countries they do. • To what extent can least-developed countries finance and implement such safety nets? • This returns us to the theme of rapid growth and agriculture’s role in securing that growth. • Rapid growth allows at least the potential for a rapid growth of fiscal revenues. • This connects us again to the importance of the composition of government spending. • Foreign assistance can help, but it cannot in a sustained manner replace domestic resources wisely spent.