1. The Welfare Impact of Land Redistribution: Evidence from a Quasi-Experimental Initiative in Malawi Mariapia Mendola (Università di Milano Bicocca) Franklin Simtowe (AGRA-Kenya) Washington, D.C., March, 2014

2. 2 Aim Investigate the economic consequences of land redistribution in a poor developing context Assess the causal impact of a market-based land program in Malawi using a quasi- experimental approach

3. 3 Motivation- 1 Land is a key asset in Africa (where rural households are overwhelmingly the majority) Land-concentration is an endemic problem in many developing countries- There is evidence that –Land inequality is detrimental for growth and development growth (Aghion et al., 1999; Alesina and Rodrik, 1994). –Small-holder farmers are more efficient and productive (e.g. Binswanger et al, 1995).

4. 4 Motivation-2 Marginal (and one-time) land redistribution are costly and politically unfeasible (Aghion et al., 1999; Piketty, 2000) Many African countries have either initiated or are contemplating major land reforms-- intense policy debate. Little evidence on the causal effects of (a market-based) land- reform on household well-being.

5. 5 Land inequality issues in Malawi dates back as the 19 th century. From 2008 estimates –55% of smallholder farmers in Malawi cultivate less than one hectare –there are about 30,000 estates cultivating between 10 and 500 hectares. –28% of the country’s cultivable arable land (about 2.6 million hectares) lies underutilized or unutilized in large estates In 2002, a new National Land Policy was adopted by the Government of Malawi to correct some of the historical wrongs on land tenure and inequality. In 2004, with financial assistance from the WB, the Government of Malawi, started implementing a Community Based Rural Land Development Project (CBRLDP). Background

6. 6 Decentralized land reform carried out in 6 pilot districts in the South of Malawi (pilot test for community demand-driven land redistribution programmes) Final goal is to increase the incomes of about 15,000 poor rural families through easing land pressure and improving land access to needy rural communities. The project provides conditional cash transfer (about $1,000) to land-poor hhs to relocate, purchase, develop and cultivate (larger) plots of farm land. The CBRLDP is a voluntary land acquisition program implemented via a quasi-experimental design, i.e. involving the use of before- and-after matched control and project (beneficiary) groups. The CBRLDP

7. 7 4-year household panel data collected among 1194 households in 6 districts (Mulanje, Thyolo, Mangochi, Machinga, Balaka and Ntcheu) in Malawi between 2006 and 2009. –“treatment group”: 391 beneficiary households –“control group”: 397 households –“indirectly treated group”: 190 households left behind in the vacated areas; 214 households in receiving areas. Although baseline data collection was conducted right after households’ relocation in 2007, a number of variables are retrospective and can therefore be interpreted as “true” baseline values (WB, 2009). The Data

9. 9 Difference-in-difference (DID) regression model Y it =α + βT it *t + ρT it +γt + ε it where β is the post-program treatment effect –Unobserved heterogeneity in participation (or “missing data for the counterfactual”)  Propensity Score Matching (PSM) at the baseline Final regression model is a combination of DID and PSM –Parallel trend assumption remains an issue (i.e. time varying factors do not affect program participation). Empirical strategy- 1

10. 10 Our outcome variables (Y) are - land holding size (either agricultural, residential or both); - crop and livestock portfolio; - agricultural productivity for major crops (i.e. maize and tobacco); - household food security (measured as meals taken in the lean season and months of reserve food); - household incomes (levels and main sources); - household expenditure (levels and expenditure items); - use of improved farm inputs such as fertilizer and seed; - access to social amenities such as schools, health. Standard household- and farm-level controls. Short- and long-term effects Empirical strategy- 2

11. Table 2. Baseline sample statistics by group Non-weighted sampleWeighted sample treatedcontroldiff.treatedcontroldiff. Tot.agricultural land size (hectares)0.690.570.12***0.550.57-0.02 HH size4.264.61-0.35**4.54.57-0.07 HH head female0.280.240.040.23 0.00 HH head age43.9238.745.18***37.7238.83-1.11 HH head married0.740.77-0.03**0.780.770.01 HH head ever attended school0.70.75-0.050.790.750.04 HH head attended primary schooling0.25 0.000.240.25-0.01 Muslim0.480.53-0.050.550.520.03 Matrilinear inheritance0.850.790.06**0.820.80.02 Housing conditions (index)-0.130.41-0.54**0.370.4-0.03 Area cultivated with maize0.50.450.05**0.470.450.02 Access to agriculture extension services0.610.480.13***0.50.490.01 Value of assets17750.111619.396130.71*10835.4811738.25-902.77 (Log) average monthly income3140.32535.4604.90*2718.962537.38181.58 Obs.399391192385 Note: This table compares the characteristics of treated and control households. The first three columns report non-weighted averages for each group, as well as the between-group difference for each variable. In the last three columns observations are weighted by the inverse propensity score matching. Standard errors are reported in parentheses. Between-group difference are: * significant at 10%; ** significant at 5%; *** significant at 1%.

12. 12 The project significantly increases land holdings (on average by 0.72 ha), agricultural output and crop-specific land productivity (i.e. maize and tobacco) of beneficiary households It also improves households’ food security (e.g.expect.duration of staple food increases by 3.5. months), income and durable asset ownership In general, these impacts are higher in the short term while they slightly decrease over time, when they remain significant. Negative or no impact on beneficiary households is found with respect to access to social services such as schools, health centres and water facilities, both in the short and medium term. Results

13. 13 Heterogeneous treatment effects by the gender of the household head and across the marriage/inheritance system followed by the household. –On average, we find smaller impact on female-headed hhs- BUT we find higher impact on female-headed hhs in patrilineal households. Spillover effects on neighbouring hhs in both vacated and surrounding areas (by comparing ‘neighbouring’ and non- beneficiary hhs) –Overall no major spillover effects Other results

14. 14 Overall, the findings suggest that land redistribution may be good in terms of productivity and income enhancing process There is scope for reducing poverty and inequality in developing countries by implementing a market and community-based land reform Though, the provision of land to the landless and poor households must be complementary to other assets, such as social services. Conclusions