Influence of Contextual Variables: An Application to Agricultural Research Evaluation in Brazil

1. Influence of Contextual Variables: An Application to Agricultural Research Evaluation in Brazil Geraldo da Silva e Souza Eliane Gonçalves Gomes Roberta BlassStaub

2. Introduction • Embrapa monitors, since 1996, the production processes of each of its 37 research centers • Nonparametric production model: DEA • Technical and cost efficiency of each research center • Identification of benchmarks • Identification of exogenous factors (contextual variables) potentially associated with efficiency • Problem: Define a proper data generating process allowing two stage statiscal inference

3. Objectives • Explore the probabilistic interpretations of FDH and the FDH conditional to define a proper data generating process for efficiency measurements • Use the ratio of conditional FDH to FDH, and two stage inference, to assess statistical significance of covariates of interest

4. FDH as an alternative to DEA • FDH does not assume convexity for the underlying technology. Consistent under convexity • Free disposability of inputs • FDH : input orientation

5. FDH: Probabilistic Interpretation • Production process described by a joint probability distribution. One is concerned with dominance • Technical measure of efficiency is defined by • Empirical estimate relative to set 

6. Condtional FDH • A vector Z of contextual variables in Rk affects the production process • Interest is in the conditional distribution of (X,Y) given Z=z • Technical efficiency conditional to Z=z • Empirical Estimator (continous covariate)

7. For multivariate Z • Joint kernel: product of univariate Epanechnikov kernels • Bandwidth: Minimizes aproximated integrated mean square error • Technical efficiency depends on the kernel only through the bandwidth h • Influence of Z: Daraio and Simar (2007) suggest nonparametric regression using as response variable

8. Application • Panel Data: (xjt, yjt, zjt), j=1...n, t=1...T • T small relative to n • Arellano and Bond (1991) model on ranks • Dynamic model allowing serial correlation • GMM • Model assumes panel random effects • It is not robust relative to the presence of second order serial correlation

9. Embrapa’s Production Model • Output: Weighted average of 28 variables falling into 4 categories • Scientific production • Technical publications • Development of technologies, products and processes. • Diffusion of technologies and image • Inputs: Vector of dimension 3 : capital, labor and operational costs • 37 research centers • Period: 1999-2006

10. Contextual variables (covariates) • Intensity of Partnerships (negative association would imply unwanted competiton among research centers) • Income generation effort • Processes improvements • Administrative changes • Size (three levels) • Type (three levels)

11. Graphical Analysis • Evolution of FDH and DEA-BCC suggests differences • Evolution of FDH conditional and FDH suggests differences

14. Goodness of Fit : Dynamic Panel • Instruments: Second order differences of rank efficiencies, first order differences of continous contextual variables, time and categorical dummy variables. • No indication of second order serial correlation (p =11%) • Sargan’s specification test is not significant (p =76% )

16. Final Considerations • The effects size and type are not statistically significant with joint p-values of 84% and 86% respectively • Processes improvements, financial resources acquisition and management change have negative signs. But only financial acquisition resources is statistically significant. Case of favorable covariates • The intensity of partnerships is detrimental to the production process but it is not statistically significant • The lag 2 negative and statistically significant component of the response provides indication of an effort for improvement. Two periods are necessary for that to be achieved

17. References Arellano, M., Bond, S., 1991. Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Review of Economic Studies 58(2), 277-297. Arellano, M., Bover O., 1995. Another look at the instrumental variable estimation of the error-components models. Journal of Econometrics 68, 29-51. Blundell, R., Bond, S., 1998. Initial conditions and moment restrictions in dynamic panel-data models. Journal of Econometrics 87, 115-143. Conover, W.J., 1998. Practical nonparametric statistics. Wiley, New York. Cooper, W.W., Seiford, L.M., Tone, K. 2000. Data Envelopment Analysis: a comprehensive text with models, applications, references and DEA-Solver software. Kluwer Academic Publishers. Boston. Daraio, C., Simar, L., 2007. Advanced Robust and Nonparametric Methods in Efficiency Analysis. Springer, New York.

18. Deprins, D., Simar, L., Tulkens, H. 1984. Measuring labor inefficiency in post offices. In: Marchand, M., Pestieau, P., Tulkens, H. (Eds.), The Performance of Public Enterprises: concepts and measurements. North-Holland, Amsterdam, pp. 243-267. Embrapa, 2006. Manual dos indicadores de avaliação de desempenho das unidades descentralizadas da Embrapa: Metas quantitativas - Versão para ano base 2007. Superintendência de Pesquisa e Desenvolvimento, Brasília. Greene, W.H., 2007. Econometric Analysis. Prentice Hall, New Jersey. Kerstens, K., Eeckaut, P.V., 1999. Estimating returns to scale using non-parametric deterministic technologies: A new method based on goodness-of-fit. European Journal of Operational Research 113, 206-214. Kotz, N., Johnson, L., 1989. Thurstone’s theory of comparative judgment. Encyclopedia of Statistical Sciences 9, 237-239. Podinovski, V.V., 2004. On the linearisation of reference technologies for testing returns to scale in FDH models. European Journal of Operational Research 152, 800-802.

19. Saaty, T.L., 1994. The Fundamentals of Decision Making and Priority Theory with the Analytic Hierarchy Process. RWS Publication, Pittsburgh. Seiford, L.M., Thrall, R.M., 1990. Recent developments in DEA, the mathematical programming approach to frontier analysis. Journal of Econometrics 46, 7-38. Silverman, B.W., 1986. Density Estimation for Statistics and Data Analysis. Chapman and Hall, London. Simar, L., Wilson, P.W., 2007. Estimation and inference in two-stage, semi-parametric models of production processes. Journal of Econometrics 136 (1), 31-64. Soleimani-damaneh, M., Jahanshahloo, G.R., Reshadi, M., 2006. On the estimation of returns-to-scale in FDH models. European Journal of Operational Research 174, 1055-1059. Soleimani-damaneh, M., Mostafaee, A., 2009. Stability of the classification of returns to scale in FDH models. European Journal of Operational Research 196, 1223-1228.

20. Souza, G.S., 2006. Significância de efeitos técnicos na eficiência de produção da pesquisa agropecuária brasileira. Revista Brasileira de Economia 60 (1), 91-117. Souza, G.S., Alves, E., Avila, A.F.D., 1999. Technical efficiency in agricultural research. Scientometrics 46, 141-160. Souza, G.S., Gomes, E.G., Magalhães, M.C., Avila, A.F.D., 2007. Economic efficiency of Embrapa’s research centers and the influence of contextual variables. Pesquisa Operacional 27, 15-26. Torgerson, W.S., 1958. Theory and methods of scaling. Wiley, New York. Wilson, P.W., 2008. FEAR: A software package for frontier efficiency analysis with R. Socio-Economic Planning Sciences 42, 247-254.