stuartschitsulo/fisp-malawi-econometrics

# Econometric Analysis of Agricultural Productivity and the Farm Input Subsidy Programme (FISP) in Malawi Advanced Econometrics coursework (LUANAR, Applied Data Science) estimating the causal effect of inorganic fertilizer use on smallholder maize yield, using the 2019–2020 Malawi Fifth Integrated Household Survey (IHS5 / LSMS). The analysis works through four complementary estimators on the same plot-household panel: | Estimator | Purpose | |---|---| | OLS (baseline → controls → district FE) | Descriptive association, likely biased by farmer-ability selection into fertilizer use | | 2SLS (IV: distance to fertilizer seller, community FISP availability) | Causal LATE, correcting for endogenous fertilizer choice | | Control function | Endogeneity test via the first-stage residual; numerically equivalent to 2SLS, extensible to nonlinear models | | SUR (yield + sales system) | Joint estimation exploiting correlated errors between productivity and commercialisation | Full write-up with derivations, interpretation, and policy discussion is in `paper/FISP_Malawi_Report.md`. ## Repository structure ``` do/ 01_build_analysis_data.do builds the plot-level analysis dataset from raw IHS5 modules 02_descriptives_and_ols.do Table 1 (descriptives) and Table 2 (OLS) 03_iv_2sls.do Table 3 (2SLS, first stage, overidentification test) 04_control_function.do Table 4 (control function, bootstrap SEs) 05_sur_system.do Table 5 (SUR system) 06_synthesis_table.do Table 6 (all estimators side by side) paper/ FISP_Malawi_Report.md full write-up output/ esttab/graph output lands here (git-ignored; kept out of version control) ``` ## Data Raw microdata (IHS5 / LSMS 2019–20) is **not included** in this repository — it is distributed by the National Statistical Office of Malawi and the World Bank and is subject to the World Bank Microdata Library terms of use. To reproduce: 1. Download the IHS5 modules from the catalog ab …