Use of Micro and Macro Frameworks in Estimating Poverty Implications of Changes in Food Prices
2 likes1,245 views
The document discusses the use of micro and macro frameworks to assess the poverty implications of food price changes, especially in the context of increasing volatility in food prices impacting poor households. It highlights various methodologies for analysis, including micro-level surveys and macro-economic models, as well as past studies estimating significant poverty increases due to food crises in 2008 and 2010. The findings emphasize the importance of agricultural productivity and the need for effective policies to mitigate the adverse effects of rising food prices on poverty levels.
Use of Micro and Macro Frameworks in Estimating Poverty Implications of Changes in Food Prices
- 1. Use of Micro and Macro Frameworks in Estimating Poverty Implications of Changes in Food Prices Presented by: Maros Ivanic The World Bank www.agrodep.org AGRODEP Workshop on Analytical Tools for Food Prices and Price Volatility June 6-7, 2011 • Dakar, Senegal Please check the latest version of this presentation on: http://www.agrodep.org/first-annual-workshop
- 2. Motivation Food prices appear to have become volatile in the recent years Significant spikes are becoming common (2008, 2010) In the long run, food prices may be on the rise as well Driven by increasing demand for biofuels, population growth Many people are concerned about higher food prices Food expenditure represents 60+ percent of poor households’ budgets, possible welfare implications Food producing households might benefit What do higher food prices mean for development? We need good tools to analyze what the past changes mean and what we can do about it through better policy
- 3. 0 50 100 150 200 250 300 350 400 450 1960 1961 1962 1963 1964 1965 1967 1968 1969 1970 1971 1972 1974 1975 1976 1977 1978 1979 1981 1982 1983 1984 1985 1986 1988 1989 1970’s oil crisis 1990 1991 1992 1993 1995 1996 1997 1998 1999 2000 2002 2003 2008 food crisis 2004 2005 2006 WB food price index 1960–2010 2007 Source: Global Economic Monitor 2009 2010 2010 food crisis
- 4. Many possible methodologies to study impacts of food prices—a few examples Ex post micro-level studies Household surveys asking about impact of previously higher food prices Econometric estimates of the impact of food prices on consumption (need panel data) Very slow… may take too long to conduct a survey Ex ante micro-level studies Simulation of food price impacts on households Need to formulate a set of assumptions Using already available household survey data—answers may be ready quickly Ex ante macro-level studies Using macro (global) models to simulate economy wide impacts of food crises Less precise in calculating poverty impacts—extrapolation of macro changes on poverty Combination of both approaches Macro/global model with detailed household surveys Households may be included in the global model (costly to do)
- 5. Food prices have risen… where do we start our analysis? Methodology depends on the questions we ask Are we interested in ex post analysis or would we like to make predictions? Do we need a macro model or will a micro model be sufficient? What are the driving forces of the changes? Change in global demand/supply? Domestic productivity/weather shocks? How much of the economy is likely to be impacted? What types of behavior do we need to model? Is the impact only short-term or is it likely to stay with us? Can farmers adjust production/input use to higher prices? Are wages and other prices likely to be affected? Data requirements depend on the scope of the analysis Do we have enough data? Global/regional analysis requires a sufficient sample of comparable surveys
- 6. Previous uses of combined micro and macro frameworks in informing policymakers Food crises have been extensively studied using combined micro/macro tools The 2008 crisis is estimated to have put 105 million people in poverty (Ivanic and Martin, 2008) The food crisis of 2010 is estimated to have put 44 million in poverty (Ivanic, Martin, Zaman 2011) They have also been used to make policy recommendations A modest annual increase in agricultural productivity is found to offset most of the negative poverty impacts of rising demand (Ivanic and Martin 2010) Productivity-raising policies found more effective than import protection policies
- 7. Example: Assessment of short-run impact of food 2010 food price crisis on global poverty Work in collaboration with Will Martin and Hassan Zaman We evaluated poverty impacts for a representative sample of developing countries We used a set of 28 recent household surveys Information on households income, food purchases, production & sales We evaluated changes in global and domestic food prices Countries may be to a large degree insulated from the observed changes in global prices We extrapolated the sample results to the rest of the developing countries Weighted by population Separate calculations for the groups of low- and middle-income countries
- 8. Large number of household surveys is a necessary basis for global analysis • We have household surveys for 28 developing countries representing 42 percent of middle-income countries (by population) • Survey conducted between 2000–2009 • All surveys contain information on expenditures, production and sales of 38 agricultural and food, and non-food commodities
- 9. Also necessary: good information on price changes Global price data for key commodities Obtained from the Global Economic Monitor Domestic prices FAO’s Global Information and Early Warning System (GIEWS) Price coverage is far from perfect Global price changes for 23 out of 38 commodities E.g. Pork, dairy, processed food were missing Domestic price changes for 22 out of 28 countries On average only 2-3 main commodities available When no domestic price was available, use an alternative price transmission parameters E.g. GTAP private consumption shares
- 10. Key global price changes observed June 2010-December 2010 Commodity Price change (percent) Sugar 76 Sorghum 69 Wheat 68 Maize 64 Soybeans 34 Groundnut oil 31 Barley 30 Rice 21 Beef 17 Oranges (proxy for fruits) -40
- 11. Domestic price changes often weakly correlated with global price changes Wheat price change (June 2010-December 2010) 80 70 60 50 40 30 20 10 0 -10 Source: GEM, FAO
- 12. Calculation of changes in household welfare We measure how each household’s per capita real income has been impacted by changes in food prices We consider Change in household agricultural profits Δπ Ignoring input costs and output responses in the short run, only focusing on sales Change in household cost of living Δc We measure the cost of initial level of utility We assume CDE preferences Net welfare change Δw= Δπ - Δc We consider 0.6 as the size elasticity of the cost of living Doubling household size requires only 60 percent higher income to maintain the same level of welfare
- 13. Aggregate poverty statistics We establish the poverty level of utility (income) PovCalNet data Count households that move out and into poverty (poverty headcount) Measure the average distance from the poverty line for those who are poor (poverty gap) Poverty line In poverty Δw Out of poverty Δw Poverty gap Δw Δw Poverty gap
- 14. Change in poverty -2.0% -1.0% 0.0% 1.0% 2.0% 3.0% 4.0% Vietnam Cote dIvoire Cambodia Ecuador Panama Niger Peru Timor Leste Nepal Rwanda Zambia Moldova Indonesia Albania Nicaragua Armenia Mongolia Nigeria each country in our sample India Yemen Malawi Belize Uganda We calculate poverty changes for Sri Lanka Guatemala Bangladesh Pakistan Tajikistan
- 15. Change in poverty -4% -3% -2% -1% 0% 1% 2% 3% 4% 5% Vietnam Cote dIvoire Cambodia Ecuador Panama Niger Peru Timor Leste Nepal Rwanda Zambia Net povertychange Moved into poverty Moldova Moved out of poverty Indonesia Albania Nicaragua Armenia Mongolia Nigeria India Yemen Malawi Belize Uganda We may use the richness of our Sri Lanka model to decompose our answers Guatemala Bangladesh Pakistan Tajikistan
- 16. How can we draw conclusions for the rest of the world? We calculate population-weighted average poverty change for two groups of developing countries Low-income countries (+1.1 percent of population) Middle-income countries (+0.7 percent of population) We apply poverty changes to the global population Low-income countries: 828 million Middle-income countries: 4,758 million We obtain global poverty change Low-income countries: +9.5 million Middle-income countries: +34.1 million Total change: +43.7 million
- 17. How do we assess robustness of our poverty change estimates? Our sample may be biased The results may be driven by any one of the countries included in the sample? Our result may not be generally valid Poverty change may be dependent on poverty definition Our assumptions may be wrong With detailed price changes generally unavailable, we risk feeding the model with irrelevant price changes We run a set of sensitivity analyses in order to understand how robust our results are
- 18. Sample sensitivity analysis We verify that our result is not very sensitive to any specific country result We repeat our poverty calculations 28 times, each time excluding one country We find that our results remain stable throughout 0.00 0.05 0.10 0.15 0.20 0.25 0.30 Density 35 40 45 50 Global change in poverty (millions of people)
- 19. Our results are robust with respect to poverty line Poverty line Net change in Gross reductions Gross increases global poverty, in poverty, in poverty, PPP millions of millions of millions of USD/person/day people people people 1.00 46.5 -19.2 65.7 1.13 47.0 -20.9 67.9 1.19 44.4 -23.4 67.8 1.25 43.7 -24.0 67.7 1.31 42.1 -25.2 67.3 1.38 41.5 -20.9 67.9 2.00 26.9 -23.1 50.0 - Poverty changes vary little with small changes in poverty line - Significant changes in poverty line affect poverty outcomes more significantly
- 20. Possible extensions of the analysis to answer long-run questions Use a CGE model to estimate long-run impacts of several key variables Wage rages Output responses Land use Include quantity adjustments among producers Price and productivity driven Integrate closely micro and macro model Assure consistency between the models The models may also be united into one
- 21. Research question: Poverty impacts of global food price projections 2010– 2050 Global population is projected to rise by 34 percent with regional variation Sub-Saharan Africa rise of 120 percent Very low population growth in East Asia: 10 percent Moderate growth in Latin America: 33 percent Decline in Europe and Central Asia of 4 percent Developing country growth also projected strong Capital stock in East Asia to grow by 800 percent Very high factor productivity growth in East and South Asia
- 22. How can global growth affect food prices? Growth in population raises demand for food and food prices in two ways Population growth increases food demand directly More mouths to feed with the same amount of land Income growth increases demand Income elasticity of food is high for low-income countries Especially East & South Asia can afford more food as their per capita incomes grow significantly Growth can affect supply of food too Growth of capital stock “pulls” resources out of agriculture in developing countries Rybczynski effect Improvements in agricultural productivity raise output The same amount of land can produce more food Higher effective prices attract additional resources
- 23. How global growth affects poverty Rising food demand and food prices Harm consumers who have to pay more for food Help net-selling farm households whose incomes rise Higher productivity and larger endowments Raise real incomes and lower poverty Higher agricultural productivity Raises agricultural output and effective producer prices Higher supply lowers actual prices Technology adoption rate A high rate of adoption of higher agricultural productivity by smallholder farmers may help the poor who are often smallholders as well
- 24. What analytical framework? Use a global CGE model to examine the impact of projections on agricultural outcomes Exogenous variables: population growth, factor accumulation, productivity changes Outputs: price changes To examine the impact of price changes on poverty we use a detailed household model Exogenous variables: price changes & productivity gains Outputs: individual households’ changes in welfare and poverty
- 25. Global general equilibrium model Standard GTAP model Six World Bank defined regions Latest GTAP database Expanded agricultural commodity detail (FAO data) 22 original agricultural and food commodities split in 38 E.g. split Soybeans and Groundnuts from “Oil seeds” Focus on “orphan” crops such as sorghum, cassava etc. Parameter values Doubled the trade elasticities for long run analysis Raised value-added substitution in “Forestry” and “Fishing” to allow for growth in these industries
- 26. Household model Calculate welfare changes for each household Change in the cost of living Change in business income Change in wage income Take into account the second-order impacts Households adjust their consumption and production with respect to changing prices Demand (CDE) Supply (CRETH) Supply and demand parameters consistent with CGE model
- 27. Scenarios Four broad scenarios Baseline (projected growth) Higher agricultural TFP worldwide (additional 1pct p.a.) Higher agricultural TFP in developing countries (additional 1pct p.a.) Protection of primary agriculture in developing countries Raise self-sufficiency—halve import shares in 2050 Additional analyses Higher productivity in individual crops Type of technical change Factor productivity or total productivity Assessing the importance of the adoption rate
- 28. Global aggregate changes Commodity Baseline Ag TFP+1 Output Price (CPI) Output Price (CPI) Primary agriculture 126.9 115.6 179.1 4.8 Processed food 124.6 14.0 154.6 -1.8 All food 126.0 48.3 164.7 1.4 Energy 206.7 -6.0 212.4 0.8 Nondurables 204.7 -4.7 209.7 -0.6 Durables 203.5 -6.4 205.5 -0.2 Services 177.1 -5.9 179.6 -0.4 - Food prices are significantly lowered by higher agricultural TFP
- 29. Poverty impacts by country of higher agricultural TFP & protection (relative to baseline) 25 PAK 20 15 Poverty change (pct) 10 BGD TLS 5 NGA KHM VNM ALB ALB NGA GTM MWI ALB 0 NGA -5 KHM VNM KHM GTM VNM GTM -10 TLS TLS MWI MWI PAK -15 PAK BGD BGD -20 Higher agricultural productivity Higher agricultural productivity Protection developing countries
- 30. Decomposition of global poverty impacts (relative to baseline) 4 3 Change in poverty (pct) 2 1 0 -1 -2 TFP + 1 -3 -4 TFP + 1 developing -5 Protection -6 Raw agric Raw agric Food prices Food prices Other prices Other prices Wages Total prices on prices and on consumers on producers on consumers on producers consumers productivity on producers - Higher agricultural productivity lowers poverty mainly through favorable agricultural price and wage impacts - Protection works in opposite direction, mainly hurting poor through lower wages
- 31. Comparison of regional poverty impacts 4 Change in poverty (pct) 2 0 -2 SSA LAC -4 SAR+EAP ECA -6 -8 Raw agric Raw agric Food prices Food prices Other prices Other prices Wages Total prices on prices and on consumers on producers on consumers on producers consumers productivity on producers - Simulation: 1 pct additional agric growth by region - Significant regional differences appear: - Africa benefits mainly from direct impacts on the price of agriculture - Asia benefits manly from wages - Latin America and Eastern Europe benefit more equally
- 32. Impact of higher productivity on poverty by commodity (1pct TFP) 1 Change in poverty (pct) 0.5 0 -0.5 -1 -1.5 -2 -2.5 -3
- 33. Adoption rates & poverty impacts of total factor productivity (1pct p.a.) 0 100 80 60 40 20 0 -0.2 Change in poverty (pct) -0.4 -0.6 -0.8 -1 -1.2 Rice-total -1.4 Maize-total -1.6 Vegetables-total Adoption rate
- 34. Adoption rates & total output productivity impacts on poverty (1pct p.a) 0 100 80 60 40 20 0 -0.5 Change in poverty (pct) -1 -1.5 Rice-TFP Rice-Output Maize-TFP -2 Maize-Output Vegetables-TFP -2.5 Adoption rate Vegetables-Output
- 35. Conclusion Micro and micro tools are extremely useful in analyzing the likely poverty impacts of significant food price shocks Simple scenarios can often be answer satisfactorily and quickly using simple methodology (household survey studies) Complex and ex ante research questions may require the use of a CGE model Long-run projections Impact of unobservable variables (wages) Merging CGE with micro-simulations is not necessary as long as we assure broad consistency Data remains the greatest stumbling block of the analysis Not enough household surveys Not enough timely and comprehesive price information