Land rental, farm investment, and efficiency in Burkina Faso

Transcription

1

2 Land rental, farm investment, and efficiency in Burkina Faso December 1, 2016 AEC Conference, Abuja, Nigeria December 1, / 16

3 Outline 1 Context and Background 2 Research Questions 3 Conceptual Framework 4 Data 5 Econometric Analysis 6 Conclusion AEC Conference, Abuja, Nigeria December 1, / 16

4 Context and Background Land and agricultural productivity Land is likely the most important factor in African agricultural production; However its distribution is very skewed in most countries. In Burkina Faso, land rights have been historically weak and land markets are underdeveloped. In recent years, the Burkinabe government has increasingly promoted reforms. The underlying assumption is that clear property rights and reduced transaction costs will generate efficiency because more able farmers will obtain more land. AEC Conference, Abuja, Nigeria December 1, / 16

5 Research Questions Implications of land rental for agriculture Are we seeing efficiency gains? Who is participating in land rental markets? Are renters using more modern inputs? Do renters use these inputs more efficiently? AEC Conference, Abuja, Nigeria December 1, / 16

6 Research Questions Our contribution Earlier studies focus on Asia (Deininger & Jin, 2005) and Latin America (Deininger, Zegarra, & Lavadenz, 2003). But recently, there is increasing interest in land markets in Africa. Most studies are limited to analyzing the determinants of participation in land rental (e.g. Deininger et al., 2003). Recently, the literature has shifted attention to the impact of land rental on productivity, income, and poverty. Our study fits in this recent development and extends the discussion to the impacts of land rental of farm investment and technical efficiency. AEC Conference, Abuja, Nigeria December 1, / 16

7 Conceptual Framework Theoretical model General setting: agricultural household model Farmers have fixed amount land Ā h and labor L h as endowments. Production function depends on farmers skills s h, and on-farm labor l a h and amount of cultivated land A h is q = F (s h, l a h, A h). Households can rent in land or rent out land we can have A h Ā h. Renting land incurs transaction costs γ. AEC Conference, Abuja, Nigeria December 1, / 16

8 Conceptual Framework Theoretical model... Comparative Statics Probability to rent in land and the amount of land rented in are with farmer skills. Probability to use modern inputs with land rental. AEC Conference, Abuja, Nigeria December 1, / 16

9 Data Data Three waves of the Continuous agricultural survey (EPA) , , and Sample size consists of 4130 households per year. Focus on cereals (maize, rice, sorghum, millet) producers representing 99 percent of the sample). About 35 percent of farmers are in land rental markets with an average of 0.5 ha of land rented. AEC Conference, Abuja, Nigeria December 1, / 16

10 Econometric Analysis Question 1: Participation in land rental markets Models We model the decision to rent in land as both binary and continuous. Probit Tobit Double Hurdle Test which model better fits the data. AEC Conference, Abuja, Nigeria December 1, / 16

11 Econometric Analysis Question 1: Participation in land rental markets... Control variables households ability, demographic and economic characteristics village and region fixed effects, crop fixed effects, and time fixed effects How to obtain ability? Following Deininger et al. (2003) we estimate a modified Cobb-Douglas production function to elicit households ability as the predicted household fixed effect. AEC Conference, Abuja, Nigeria December 1, / 16

12 Econometric Analysis Participation in land rental markets... Table 1: Double hurdle model results Tier1 Tier2 Coeff Coeff Marg Effect Only significant coefficients (1) (2) (3) Estimated household farming ability *** *** *** (0.0367) (1.6622) (0.0195) Female-headed household *** ** *** (0.0769) (0.3527) (0.0185) Household size *** *** ** (0.0031) (0.0054) (0.0036) Log total cereal land *** *** *** (0.0074) (0.2508) (0.0006) Log of total own land *** *** *** (0.0475) (0.1196) (0.0028) Number of tropical livestock units *** *** *** (0.0540) (1.9801) (0.2595) Access to credit ** *** (0.1120) (0.8371) (0.0134) Observations 11,950 11,950 11,950 Standard errors in parentheses are obtained by 100 bootstrap replications*** p<0.01, ** p<0.05, * p<0.1 AEC Conference, Abuja, Nigeria December 1, / 16

13 Econometric Analysis Question 2: Land rental markets and farm investment Models Decision to use mineral fertilizer, hybrid seeds, organic manure, plant protectants, and hired labor as functions of land rental status and other control variables We use multivariate probit to account for the simultaneous nature of the demand for inputs. Our identification relies on : Variation in the data; Household fixed effects and correlated random effects to account for heterogeneity and attrition bias; Remaining endogeneity is likely but is less severe given that rental is predetermined. AEC Conference, Abuja, Nigeria December 1, / 16

14 Econometric Analysis Question 2: Land rental markets and farm investment... Table 2: Alternative models for input demand and land rental in Burkina Faso Only coefficient of land rental Mineral Fertilizer Hybrid seeds Organic manure Protection Chemicals Hired labor Treatment: Has rented land Multivariate Probit *** (0.0274) (0.0382) (0.0264) (0.0269) (0.0263) Probit *** (0.0335) (0.0475) (0.0394) (0.0301) (0.0203) Linear probability model *** (0.0094) (0.0037) (0.0082) (0.0112) (0.0091) AEC Conference, Abuja, Nigeria December 1, / 16

15 Econometric Analysis Question 3: Land rental and efficiency Models Estimate a stochastic frontier production function q ht = β 0 + β i lnz iht + v ht u ht q ht is value of production Z vector of input cost v ht error terms u ht inefficiencies Following the literature, we specify a model for the inefficiency (u ht ) with land rental as an explanatory variable. AEC Conference, Abuja, Nigeria December 1, / 16

16 Econometric Analysis Land rental and efficiency... Table 3: Determinants of technical inefficiencies Only significant coefficients (1) (2) Participate in land rental market *** (0.0227) Amount of land rented * (0.0308) Age of the head of household * ** (0.0011) (0.0011) Female-headed household ** * (0.0377) (0.0393) Household size *** *** (0.0056) (0.0047) Household head is alphabetized *** *** (0.0364) (0.0339) Log total farm land *** *** (0.0473) (0.0403) Number of tropical livestock units ** *** (0.5051) (0.3660) Log of off-farm income * (0.0021) (0.0017) Observations 12,502 12,510 AEC Conference, Abuja, Nigeria December 1, / 16

17 Conclusion Summary and Implications Land rental is increasingly prevalent in Burkina Faso. Farmers with high ability are more likely to expand farm operations through land rental. Mixed effects of land rental markets on input intensification : Farmer participation in land rental markets has a positive effect on the likelihood of using crop protection chemicals. But there is no effect on mineral fertilizer, organic manure, and hybrid seeds. However, land renters are more efficient per unit input. Much of the gains from land policies might be in term of efficiency and not increased adoption modern inputs. AEC Conference, Abuja, Nigeria December 1, / 16