Technical Efficiency and Complementarity of Agroecological Innovations in French West Indies Banana Production Jean-Joseph Minviel, M’hand Fares, Jean-Marc Blazy, Alban Thomas To cite this version: Jean-Joseph Minviel, M’hand Fares, Jean-Marc Blazy, Alban Thomas. Technical Efficiency and Complementarity of Agroecological Innovations in French West Indies Banana Production. Applied Economics Letters, 2023, 30 (18), pp.2652-2656. ฀10.1080/13504851.2022.2103074฀. ฀hal-03728189฀ HAL Id: hal-03728189 https://hal.inrae.fr/hal-03728189 Submitted on 20 Jul 2022 HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. Distributed under a Creative Commons Attribution - NonCommercial - NoDerivatives 4.0 International License Technical Efficiency and Complementarity of Agroecological Innovations in French West Indies Banana Production Jean-Joseph Minviel∗, M’hand Fares†, Jean-Marc Blazy‡, Alban Thomas§ This Version: July 2022 Abstract In context of global challenges facing agriculture, our paper addresses the extent to which the synergistic nature of agroecological innovations may reconcile environmental and technical efficiency of farms. We develop an empirical model, namely conditional efficiency framework, which explicitly accounts for contextdependent drivers like synergy and complementarity of innovations. Using a sample of 567 banana farms in the French West Indies, our estimates confirm the complementarity effect since the joint adoption of agroecological innovations increases the technical efficiency scores much more than others drivers and each of the innovations taken in isolation. We also show that advice and extension services as well as human capital variables are key adoption levers for public policy since they reduce the variability of production and thus the risk associated with the joint adoption of the agroecological innovations. Keywords: Conditional efficiency; complementarity; agroecological innovations; farm performance. JEL Classification: C14; 033; Q16; Q55. 1 Introduction Given the challenges of climate change, sustainable food and nutritional security, innovation in the agricultural and food system is a major concern and agro-ecological ∗ Université Clermont Auvergne, INRAE, Vetagro Sup, UMR Herbivores, 63122, Saint-GenèsChampanelle, France. E-mail: jean-joseph.minviel@inrae.fr † Corresponding Author : INRAE, UMR SELMET, 2 place Pierre Viala, 34000 Montpellier, France. E-mail: mehand.fares@inrae.fr ‡ INRAE, UR ASTRO, F-97170, Petit-Bourg, Guadeloupe, France. E-mail : jeanmarc.blazy@inrae.fr § INRAE, UMR Paris-Saclay Applied Economics, AgroParisTech, 75005 Paris, France & Observatory of Rural Development, 31326 Castanet-Tolosan, France. E-mail: alban.thomas@inrae.fr. 1 innovations and bio-based transitions are one of the avenues to be explored in order to meet these multiple challenges (OECD, 2021; WIPO et al., 2017; FAO, 2019). This is particularly true in the French West Indies where the intensive use of chemical inputs in banana monoculture, especially pesticides (chlordecone), has led to soil and ecosystem contamination and human health risks in large areas of the islands (Cabidoche et al., 2009). To mitigate these negative externalities and reduce pesticide use, agroecological innovations have been developed (Blazy et al., 2010). Since agroecological systems are the result of complex interactions at a local level between technologies and biological components of the agroecosystems (Duru et al., 2015), agroecological innovations are then based on a bundle of complementary technologies and practices leading to emergent and synergistic sustainable properties (Altieri, 2002). The main issue addressed by our paper is whether the adoption of complementary agroecological innovations that increase the environmental performance of the farm may also increase its economic performance (Edmeades et al., 2008; Lambrecht et al., 2014). The meta-analysis conducted by Rosenbusch et al., (2011) shows that the innovationperformance relationship is context-dependent, and using recent advances in efficiency analysis we suggest the conditional efficiency framework (Daraio and Simar, 2007; Daraio et al., 2021) to measure performance (technical efficiency) while explicitly accounting for contextual drivers. The contextual drivers refer to factors which are neither inputs nor outputs, but that form the backdrop of farmer decision-making. Among the relevant contextual drivers, we mainly focus on complementarity between innovations and its impact on technical efficiency since the literature shows that the adoption of complementary innovations improves firm performance (Arora et al., 2010; Fares et al., 2018; Mohnen, 2019). Our estimates on a database of 567 banana farms in the West Indies confirm this complementarity effect since the joint adoption of two synergistic agroecological innovations, i.e. Disease Free seedlings (DFS) and Fallow, increases the technical efficiency score much more than each of the innovations taken in isolation. We also show that advice and extension services and other human capital variables are key adoption lever since they reduce the variability of production and thus the risk associated with the joint adoption of the agroecological innovations. The remaining sections are organized as follows. In section 2, we present the empirical model and the context of our farm survey database as well as the the variables of test. Econometric results are presented in section 3 adjoining to discussion and section 4 concludes. 2 2.1 Materials and Methods Methods Farmers’ production decisions are modeled using the conditional efficiency framework (Daraio and Simar, 2007; Daraio et al., 2021; Belmonte-Martin et al., 2021), where q p given contextual (condiare combined to produce outputs Y ∈ R+ inputs X ∈ R+ r tioning) variables Z ∈ R+ . In this framework, a production process is defined by a 2 production technology ψ Z that describes the set of all technically feasible input-output combinations given the conditioning variables: ψ Z = {(x, y)|z : x can produce y} (1) A farm is technically efficient (i.e., located on the boundary of ψ Z ) if it produces the maximum possible level of outputs for a given level of inputs1 . The production process defined in [1] can be fully characterized by the joint-conditional probability: S(Y |X,Z) (y|x, z)F(X|Z) (x|z), where S(Y |X,Z) (y|x, z) denotes the conditional survival function Y , with SY (y) = P rob(Y ≥ y), and F(X|Z) (x|z) the marginal conditional distribution function of X, with FX (x) = P rob(X ≤ x). An output-oriented conditional efficiency score is defined as follows by the upper-boundary ψ Z of the support of S(Y |X,Z) (y|x, z) : n θ(x, y|z) = sup θ|S(Y |X,Z) (θy|x, z) > 0 o (2) To account for outlying observations, we define an order-m frontier that characterizes the expected maximum level of outputs achievable for a subset of m randomly drawn production units with X ≤ x as a yardstick (Daraio and oSimar, 2007). That is, for n Z Z b (x) such that the conditional any value y, there exists θm (x, y) = sup θ|(x, θy) ∈ ψbm output-oriented order-m efficiency measure is defined as: Z (x, y) |X ≤ x, Z = z) θm (x, y |z) = E(Y |X,Z) (θbm ˆ ∞ [1 − (1 − S(Y |X,Z) (uy|x, z))m ]du = (3) 0 The survivor function S(Y |X,Z) (y|x, z) is estimated using the following kernel function: Sb (Y |X,Z,n) (y|x, z) = Pn i=1 I(Xi ≤ x, Yi ≥ y)Kĥ (z, zi ) i=1 I(Xi ≤ x)Kĥ (z, zi ) Pn (4) where Kbh (.) = h−1 K((z, zi )h−1 ), with ĥ = (ĥ1 , . . . , ĥr ) a vector of r-estimated bandwidth parameters and I(.) is an indicator function which equals to unity if its argument is true and zero otherwise. Then, the conditional efficiency estimator θ̂m (x, y|z) is given by plugging Ŝ(Y |X,Z,n) (y|x, z) into equation [3] (see for more details Minviel and de Witte, 2017). To investigate the influence of the contextual drivers on technical efficiency, we use a location-scale nonparametric regression model (Badin et al., 2012): θi (xi , yi |zi ) = g(zi ) + σ(zi )ξi (5) where ξi is an error term, g(.) = E[θi (xi , yi |zi )] and σ 2 (zi ) = V [θi (xi , yi |zi )]. The nonparametric functions g(.) and σ 2 (.) are estimated using kernel local linear regression methods. To introduce complementarity effect, we decompose the contextual drivers in two components by rewriting zi = δ 0 s00 + δ DF S s10 + δ F s01 + δs11 + ζXi , where ζXi represent the other drivers and δ 0 s00 + δ DF S s10 + δ F s01 + δs11 the four agroecological innovation strategies of the farmer when he can adopt DFS or Fallow innovations. The latter are said to be complements only if the marginal effect of joint adoption is positive (δ > 0) (Fares, 2013). 1 In the conventional approach, this set is not conditional to the contextual drivers (Z). 3 2.2 Materials To control parasitism and therefore reduce pesticide use in French West Indies banana production, two agroecological innovations have been developed: (i) introducing a fallow period (FP) in rotation with banana; and (ii) using disease-free seedling (DFS) after the fallow period, where seedlings are produced in-vitro (Chabrier and Quenerve, 2003). Although both innovations can be used separately, they are complementary in managing pest pressure since DFS makes it possible to avoid exogenous parasitism after the fallow period. If this synergistic effect may ensure environmental efficiency since it reduces the social cost of having recourse to chemical pesticide, it does necessarily generates incentives to joint adoption by increasing the performance of the farm. Table 1: Description of the variables Variables Description Mean Std Min Max Banana production in tons 464.19 975.53 1 10300 Output Banana production Inputs Land Area in banana production (ha) 11.31 19.77 0.5 210 Labor Labor used in annual working unit 7.46 13.36 0.14 151.42 Fertilizer Fertilizers used in tons 23.25 45.24 0.31 453.6 pesticides Pesticides used in kg 15.46 33.85 0.05 441 Contextual drivers DFS (Disease Free Seedlings) 1 if farmer has adopted DFS 0.35 0.48 0 1 1 if the system includes fallow 0.39 0.49 0 1 DFS x Fallow 1 if joint adoption of DFS and Fallow 0.27 0.45 0 1 Intercropping 1 if the system includes intercropping 0.61 0.49 0 1 mechanized banana land (%) 4.318 0.483 0 1 Fallow Share of mechanized land Share of irrigated land Irrigated banana land (%) 0.31 0.41 0 1 1 for farms located in Guadeloupe 0.26 0.44 0 1 Total Farm area (ha) 16.77 29.95 0.5 262 Technical Assistance Number of links with researchers & technicians 31.47 23.41 0 194 Agricultural training 1 if farmer has training in Agriculture 0.51 0.50 0 1 1 if farmer has made high study 0.07 0.263 0 1 Social group 1 if Household belongs to a social group 0.37 0.48 0 1 Older farmer 1 if Age >= 60 0.12 0.32 0 1 1 if farmer expects increasing prices 0.20 0.43 0 1 Guadeloupe Farm size Higher education Price Expectation To address this issue a survey questionnaire on innovation adoption was administered in Guadeloupe and Martinique between March and June 2008, through one-time faceto-face interviews, to a random sample of 607 banana planters with a sampling rate of about 80% in each island. After eliminating missing values and zero values for 4 input-output vectors, the final dataset used contains 567 observations. To estimate our conditional efficiency model using this dataset, we selected one output, four inputs, and fourteen contextual drivers. The output is measured as the physical value of the banana production in tons. The four inputs include the agricultural area, the labor used, the quantity of chemical fertilizers used, and the quantity of chemical pesticides. Among the relevant contextual drivers, we mainly focus on complementarity between innovations and its impact on technical efficiency (see Table 1). 3 Results and discussion The estimates for the mean effects obtained from the order-m conditional efficiency model are reported in table 2. The average conditional technical efficiency score amount to 0.67, while the unconditional score is only 0.6. This suggests first that farmers in our sample generate 33% less outputs than it is technically feasible. That is, farmers could increase their output by 33% without increasing their input use. Second, in average contextual drivers induces a 7% increase of the efficiency score. Regarding the average marginal effects of these contextual drivers, our results show that the potential of increase mainly comes from our variable of interest, i.e., complementarity between DFS and Fallow agroecological innovations since marginal effect of joint adoption induces an increase in farmer efficiency of about 20%, while the adoption of DFS alone generates only 15% of increase and Fallow alone has no significant effect. These results highlight that by jointly adopting complementary innovations, farmer becomes more productive (Miravette and Pernias, 2006). For the other contextual variables, we find that technical assistance and agricultural training significantly increases the conditional efficiency scores since both provide precise information and know-how on agroecological innovations use (Xayavong et al., 2016). Likewise, expectation of increasing selling prices may play as an insurance and incentive to improve farm productivity and this can explain its positive impact on the conditional efficiency score (Karian et al., 2014). For the dispersion effects, the results indicate a positive effect of the DFS/Fallow joint adoption on the variance of the conditional efficiency scores. That is, while synergistic agroecological systems have the advantage of increasing marginal efficiency scores, they also have the disadvantage of increasing their variability, which seems to be a general rule of coupled innovations since intercropping systems (intercropping) also increase the variability. These innovative systems are indeed complex to manage and therefore require high human capital. Our results show that some key variables, such as the existence of technical assistance (Xayavong et al, 2016), higher education (El-Osta, 2011), more experienced farmer (older farmer ; Ainembabazi and Mugisha, 2014; Karki et al., 2020) and some insurance against risks (expectation prices; Karian et al., 2014), reduce the risk associated with agroecological innovation by decreasing the output (and thus the efficiency scores) variability. 5 Table 2: Conditional Efficiency Estimates Marginal effects DFS 4 Dispersion effects Estimate Bootstrap S.E Estimate Bootstrap S.E 0.1547** 0.0683 -0.0142 0.0099 Fallow -0.0384 0.0564 -0.0129 0.0101 DFS x Fallow 0.1980** 0.0820 0.0295** 0.0148 Intercropping -0.0416 0.0266 0.0122** 0.0059 Share of mechanized land 0.0051 0.0479 -0.0025 0.0120 Share of irrigated land 0.1248*** 0.0532 -0.0019 0.0104 Guadeloupe 0.0422 0.0485 -0.0137 0.0087 Technical assistance 0.0013* 0.0008 -0.0007*** 0.0001 Agricultural training 0.0530* 0.0282 0.0026 0.0066 Higher education 0.1152 0.0828 -0.0263*** 0.0099 Social group 0.0232 0.0347 -0.0072 0.0064 Older farmer 0.0364 0.0344 -0.0249*** 0.0056 Price expectations 0.0658** 0.0328 -0.0158** 0.0077 Mean conditional TE 0.67 Mean unconditional TE 0.60 Conclusion This paper uses a context-dependent framework, the nonparametric conditional efficiency model, to examine the innovation-performance nexus in agriculture. Our estimates on a database of 567 banana farms in the West Indies confirm this complementarity effect since the joint adoption of two synergistic agroecological innovations, i.e. Disease Free seedlings (DFS) and Fallow, increases the technical efficiency score much more than each of the innovations taken in isolation. We also show that advice and extension services, as well as well as farmer human capital increase, are key adoption levers for public policy since they reduce the variability of production and thus the risk associated with the joint adoption of the agroecological innovations. The main limitation of this work is that we analyze the complementarity effect using only two agroecological innovations. It would be interesting in a future research to study the impact and interactions between a large number of agroecological innovations to test the robustness of our complementary result. References [1] Ainembabazi, J.H. and Mugisha, J. (2014). The Role of Farming Experience on the Adoption of Agricultural Technologies: Evidence from Smallholder Farmers in Uganda. The Journal of Development Studies 50(5):666–679. 6 [2] Altieri, M. (2002). Agroecology: the science of natural resource management for poor farmers in marginal environments. Agriculture, Ecosystems & Environment 93:1–24. [3] Arora, A., Forman, C. and Yoon, J.W. (2010). Complementarity and information technology adoption: Local area networks and the Internet. Information Economics and Policy 22(3):228–242. [4] Badin, L., Daraio, C. and Simar, L. (2012). How to measure the impact of environmental factors in a nonparametric production model? European Journal of Operational Research 223:818–833. [5] Belmonte-Martin, I., Ortiz, L. and Polo, C. (2021). Local tax management in Spain: A study of the conditional efficiency of provincial tax agencies. SocioEconomic Planning Sciences 78:101057. [6] Blazy, J.M., Tixier, P., Thomas, A., Ozier-Lafontaine, H., Salmon, F. and Wery, J. (2010). BANAD: a farm model for exante assessment of agro-ecological innovations and its applications to banana farms in Guadeloupe. Agricultural Systems 103:221– 232. [7] Cabidoche, Y.M., Achard, R., Cattan, P., Clermont-Dauphin, C., Massat, F. and Sansoulet, J. (2009). Long-term pollution by chlordecone of tropical volcanic soils in the French West Indies: A simple leaching model accounts for current residue. Environmental Pollution 157:1697–1705. [8] Chabrier, C. and Queneherve, P. (2003). Control of the burrowing nematode (Radopholus similis Cobb) on banana: impact of the banana field destruction method on the efficiency of the following fallow. Crop Protection 22:121–127. [9] Daraio, C. and Simar, L. (2007). Advanced robust and nonparametric methods in efficiency analysis: Methodology and applications. New York: Springer. [10] Daraio, C., Simar, L. and Wilson, P.W. (2021). Quality as a latent heterogeneity factor in the efficiency of universities. Economic Modelling 99:105485. [11] Duru, M., Fares, M. and Therond, O. (2015). Designing agroecological transitions: A review. Agronomy for Sustainable Development 35:1237–1257. [12] Edmeades, S., Phaneuf, D., Smale, M. and Renkow, M. (2008). Modelling the Crop Variety Demand of Semi-Subsistence Households: Bananas in Uganda. Journal of Agricultural Economics 59:329 – 349. [13] El-Osta, H. (2011). The Impact of Human Capital on Farm Operator Household Income. Agricultural and Resource Economics Review 40(1):95–115. Publisher: Cambridge University Press. [14] FAO (2019). HLPE Report #14. Agroecological and other innovative approaches. For sustainable agriculture and food systems that enhance food security and nutrition. 7 [15] Fares, M. (2013). Using a multinomial probit to test for complementarity. Applied Economics Letters 21(3):180–184. [16] Fares, M., Raza, S. and Thomas, A. (2018). Is There Complementarity Between Certified Labels and Brands? Evidence from Small French Cooperatives. Review of Industrial Organization 53(2):367–395. [17] Karki, S., Burton, P. and Mackey, B. (2020). The experiences and perceptions of farmers about the impacts of climate change and variability on crop production: a review. Climate and Development 12(1):80–95. [18] Karlan, D., Osei, R., Osei-Akoto, I. and Udry, C. (2014). Agricultural Decisions after Relaxing Credit and Risk Constraints. The Quarterly Journal of Economics 129(2):597–652. Publisher: Oxford University Press. [19] Lambrecht, I., Vanlauwe, B. and Maertens, M. (2014). Integrated soil fertility manage- ment: From concept to practice in eastern DR Congo. In: Bioeconomics Working Paper Series. [20] Minviel, J.J. and de Witte, K. (2017). The influence of public subsidies on farm tech- nical efficiency: A robust conditional nonparametric approach. European Journal of Operational Research 259(3):1112–1120. Publisher: Elsevier. [21] Miravete, E.J. and Pernias, J.C. (2006). Innovation, Complementarity and Scale of Production. Journal of Industrial Economics 54(1):1–29. [22] Mohnen, P. (2019). R&D, Innovation and Productivity. In: W. Green and T. Ten Raa (eds.) The Palgrave Handbook of Economic Performance, pp. 97–122. Cham, Switzerland: Palgrave MacMillan. [23] OECD (2021). Agricultural Policy Monitoring and Evaluation 2021: Addressing the Challenges Facing Food Systems. Paris: Organisation for Economic Cooperation and Development. [24] Rosenbusch, N., Brinckmann, J. and Bausch, A. (2011). Is Innovation Always Beneficial? A Meta-Analysis of the Relationship Between Innovation and Performance in SMEs. Journal of Business Venturing 26:441–457. [25] WIPO, Cornell-University and INSEAD (2017). The Global Innovation Index 2017: Innovation Feeding the World. Tech. rep., World Intellectual Property Organization, Geneva. OCLC: 1129168119. [26] Xayavong, V., Kingwell, R. and Islam, N. (2016). How training and innovation link to farm performance: a structural equation analysis. Australian Journal of Agricultural and Resource Economics 60(2):227–242. 8