This is a work in progress. Not for citation. Can Higher Food Prices Help the Rural Poor? Some Insights on Potential Impacts of Climate Change in Central Asia Alisher Mirzabaev 01 February, 2013 Conceptual framework Empirical steps and Data Results A Brief Literature Review I Hertel et al. (2010): A potential rise in major global staple prices could decrease poverty among agricultural households. I Mendelsohn et al. (2007): Climate has statistically significant effect on agricultural incomes and climate change is likely to increase poverty among agricultural households. I Nelson et al. (2010): Increases in staple food prices due to climate change may have negative effects on the eradication of child malnutrition, including in Central Asia. 2 Conceptual framework Empirical steps and Data Results Research Question and Objectives I How the climate volatility and the related income and price changes may affect the livelihoods of the rural poor in Central Asia? 1. Differentiate the possible income and price impacts of climate shocks according to different income categories of households. 2. Account for households’ potentially interlinked decision making on production and consumption. 3 Conceptual framework Empirical steps and Data Results Theoretical Framework: Agricultural Household Model I Potential endogeneity of consumption, production and labor supply decisions among agricultural households I Addressing the endogeneity bias through use of instrumental variables I Long-term temperature variability as an instrumental variable in this setting I Temperature variability has important effects on agricultural production, however, clearly it is not caused by current levels of consumption of households residing in that location. 4 Conceptual framework Empirical steps and Data Results Empirical Steps 1. Test for separability of the agricultural household model. 2. Assuming separability, quantile regression of food consumption by households without accounting for endogeneity 3. The non-separability is alternatively assumed and quantile regression is run with instrumented agricultural income. 5 Conceptual framework Empirical steps and Data Results Data Sources I Nationally representative agricultural household surveys (4 countries, N = 1600), from ICARDA-led, ADB-funded project on climate change in Central Asia. I Climate variables: temperature and precipitation I Institutional variables: market access, NASA night-time lighting intensity series. In addition, several proxy variables are created to capture some institutional characteristics. 6 Conceptual framework Empirical steps and Data Results NASA Night-time Lighting Intensity Dataset Source: National Geophysical Data Center, NOAA 7 Conceptual framework Empirical steps and Data Results Testing for Separability Variables Labor market Goods Market *** p<0.01, ** p<0.05, * 0.1 Non-separable Separable Non-separable Separable Dependency ratio Females to males ratio Age of household head Age of household head, squared Gender of household head Average family size Average wage Distance to Markets Night-time lights Market fragmentation Farm size Number of crops grown Constant -0.113 -0.0407 -0.0854* 0.000650 -0.137 0.0715** -0.0961 0.191** 0.0304*** 1.167*** -0.00114 1.797*** 3.083** 0.0260 0.0909 0.0415 –0.000280 0.0536 –0.00579 0.0689 –0.127** 0.0117** – 0.0835 1.28e-05 0.792*** -1.021 –74.12 –39.33 55.73** –0.603** –19.31 –47.37*** –833.2*** 98.41* 11.14** – 1,434*** 0.0221 309.6*** 175.1 –4,092 62.66 473.3 –8.357 6,925 1,113 –8,140 1,500 518.4 18,369 358.2*** 6,052*** 33,447 363 1169 1194 335 Number of households Dependant variables: Household farm labor supply and net trade in agricultural goods,respectively I Finite mixture regression is employed to identify latent idiosyncratic non-separability among heterogeneous households. 8 Conceptual framework Empirical steps and Data Results The Separable Agricultural Household Model Dependent variable Daily per capita food expenses, log Farming profits (log) Age of household head Age of household head, squared Education of household head Education of household head, squared Gender of household head, (0–female, 1–male) Family non-farm workers Family on-farm workers Livestock value (USD) Value of total assets (log) Market fragmentation Cooperation Rural development Night-time lighting intensity Distance to markets (log) Number of crops grown Subsistence farmer (0–yes, 1–no) Vegetable price (log) Potato price (log) Wheat price (log) Maize price (log) Net agri trade position Share of privately owned land Farm size (log) Constant Quantiles 10th 30th 50th 70th 90th 0.0438* 0.0215 -0.000262 0.132 -0.0272 -0.0169 0.0286 -0.0216 4.84e-07 0.0405 0.172 0.00511 0.000112 -0.00950 -0.0933 0.0449 0.295 -0.0338 0.708*** 0.0329 -0.0797 -0.310*** -0.298** 0.0198 0.0318* 0.00143 -3.53e-05 -0.137 0.0279 -0.0851 0.0910*** -0.0593*** 2.65e-07 0.0404*** –0.0229 0.0248 2.95e-05 -0.00365 -0.0683 0.00136 0.416 -0.0365 1.063*** -0.0709 -0.0439 -0.348*** -0.263*** 0.0198 0.0427*** 0.00218 -3.03e-05 -0.230*** 0.0499*** -0.0452 0.0773*** -0.0627*** 1.34e-08 0.0407** 0.0318 0.0353* 2.12e-05 -0.00525 -0.0492 -0.0237 0.413** -0.0334 1.238*** 0.0181 -0.0303 -0.292*** -0.207*** 0.0218* 0.0370** -0.00509 5.17e-05 -0.146 0.0319* -0.0587 0.0818*** -0.0578*** -3.26e-07 0.0487** 0.0259 0.0456** -1.49e-06 -0.00167 0.00240 -0.0147 0.424 -0.0492* 0.951*** -0.213 0.0649 -0.247*** -0.0913* 0.0356** 0.0291 -0.00756 9.80e-05 -0.0903 0.0147 -0.0223 0.0620** -0.0631*** -7.33e-07 0.0519** 0.00709 0.0567 1.15e-05 -0.00389 -0.0463 -0.00621 0.253 -0.0550 0.831*** -0.120 0.0705 -0.250*** -0.165*** 0.0362*** -0.698 0.718 1.301** 0.531 1.389* 9 Conceptual framework Empirical steps and Data Results The Non-Separable Agricultural Household Model Dependent variable Daily per capita food expenses, log Quantiles 10th 30th 50th 70th 90th Farming profits (log), instrumented Age of household head Age of household head, squared Education of household head Education of household head, squared Gender of household head, (0–female, 1–male) Family non-farm workers Family on-farm workers Livestock value (USD) Value of total assets (log) Market fragmentation Cooperation Rural development Night-time lighting intensity Distance to markets (log) Number of crops grown Subsistence farmer (0–yes, 1–no) Vegetable price (log) Potato price (log) Wheat price (log) Maize price (log) Net agri trade position Share of privately owned land Farm size (log) 0.614*** 0.0280 -0.000326 0.00494 -0.0272 0.0489 0.0528* -0.0373** -1.27e-06 0.0672*** –0.129 0.00798 0.000116* -0.0133*** -0.175*** 0.0376* 0.302*** -0.0318 0.753*** 0.154 -0.0654 -0.307*** -0.181* 0.0131 0.413*** 0.0104 -0.000127 0.0353** 0.0279 -0.0986** 0.0720*** -0.0581*** 3.39e-07 0.0611*** –0.0795 0.0156* 6.87e-05* -0.00720** -0.113*** 0.0185 0.381*** -0.0492** 0.989*** 0.117 -0.0380 -0.313*** -0.201*** 0.00222 0.458*** -0.00812 7.45e-05 0.0420*** 0.0499*** -0.0607 0.0708*** -0.0613*** 7.76e-08 0.0593*** –0.0421 0.0235 3.75e-05 -0.00642 -0.0810** 0.00539 0.394*** -0.0260 0.991*** 0.0552 0.0786 -0.235*** -0.0323 0.0219* 0.553*** -0.00394 4.29e-05 0.0462** 0.0319* -0.130 0.0756*** -0.0610*** -2.14e-07 0.0565*** – 0.215** 0.0302 2.24e-05 -0.00491 -0.0710* 0.00757 0.296** -0.0299 0.957*** -0.0626 0.0857 -0.166*** 0.0153 0.0334*** 0.448*** -0.0150 0.000133 0.0275 0.0147 -0.114 0.0624** -0.0658*** -5.57e-07 0.0690*** 0.0661 0.0266 -2.37e-05 -0.00142 -0.0508 -0.0218 0.147 -0.0647** 0.747*** 0.0948 0.101* -0.178*** 0.0845 0.0305** Constant -4.549*** -1.915*** -1.692** -2.543** -1.074 10 Conceptual framework Empirical steps and Data Results The Non-Separable Agricultural Household Model Dependent variable Daily per capita food expenses, log Quantiles 10th 30th 50th 70th 90th Farming profits (log), instrumented Age of household head Age of household head, squared Education of household head Education of household head, squared Gender of household head, (0–female, 1–male) Family non-farm workers Family on-farm workers Livestock value (USD) Value of total assets (log) Market fragmentation Cooperation Rural development Night-time lighting intensity Distance to markets (log) Number of crops grown Subsistence farmer (0–yes, 1–no) Vegetable price (log) Potato price (log) Wheat price (log) Maize price (log) Net agri trade position Share of privately owned land Farm size (log) 0.614*** 0.0280 -0.000326 0.00494 -0.0272 0.0489 0.0528* -0.0373** -1.27e-06 0.0672*** –0.129 0.00798 0.000116* -0.0133*** -0.175*** 0.0376* 0.302*** -0.0318 0.753*** 0.154 -0.0654 -0.307*** -0.181* 0.0131 0.413*** 0.0104 -0.000127 0.0353** 0.0279 -0.0986** 0.0720*** -0.0581*** 3.39e-07 0.0611*** –0.0795 0.0156* 6.87e-05* -0.00720** -0.113*** 0.0185 0.381*** -0.0492** 0.989*** 0.117 -0.0380 -0.313*** -0.201*** 0.00222 0.458*** -0.00812 7.45e-05 0.0420*** 0.0499*** -0.0607 0.0708*** -0.0613*** 7.76e-08 0.0593*** –0.0421 0.0235 3.75e-05 -0.00642 -0.0810** 0.00539 0.394*** -0.0260 0.991*** 0.0552 0.0786 -0.235*** -0.0323 0.0219* 0.553*** -0.00394 4.29e-05 0.0462** 0.0319* -0.130 0.0756*** -0.0610*** -2.14e-07 0.0565*** – 0.215** 0.0302 2.24e-05 -0.00491 -0.0710* 0.00757 0.296** -0.0299 0.957*** -0.0626 0.0857 -0.166*** 0.0153 0.0334*** 0.448*** -0.0150 0.000133 0.0275 0.0147 -0.114 0.0624** -0.0658*** -5.57e-07 0.0690*** 0.0661 0.0266 -2.37e-05 -0.00142 -0.0508 -0.0218 0.147 -0.0647** 0.747*** 0.0948 0.101* -0.178*** 0.0845 0.0305** Constant -4.549*** -1.915*** -1.692** -2.543** -1.074 10 Conceptual framework Empirical steps and Data Results Engel curve: potato share I Following Deaton (1989) and using non-parametric kernel regressions. 11 Conceptual framework Empirical steps and Data Results Engel curve: potato share 12 Conceptual framework Empirical steps and Data Results Proportion of Potato Producers and Sellers 13 Conceptual framework Empirical steps and Data Results Welfare effects of potato price changes 14 Conceptual framework Empirical steps and Data Results Welfare effects of potato price changes 15 Conceptual framework Empirical steps and Data Results Key Conclusions I The poorest households are more vulnerable to the impacts of weather and climate shocks since their food security more strongly depends on their agricultural incomes. 16