Tests De Séparabilité dans les Decisioins des Menages Agricoles: Cas du Maroc

The modeling and empirical analysis of the behavior of farming households, particularly in a context of imperfect markets, is important in development microeconomics. Whether the various farm household decisions are made separable or non-separable is necessary for the adequate modeling of their reactions to any economic policy. Indeed, a separable household model and a non-separable household model produce different results, in terms of impact simulations. It is therefore important to test each of these two types of models and to construct and adopt prior simulations. On a Moroccan database, extracted from a national survey on living standards of cross type households, we conduct several tests of separability through various approaches and methods. Specifically, we test the null hypothesis of the first separability on the basis of total turnover and as many turnovers generated in various agricultural products. We rely on both ordinary least squares (OLS) and then on the instrumental variable (IV) to correct the possible endogeneity bias regressors and finally the type of model specifications Seemingly Unrelated Regression Equations (SURE ). Our results show that the hypothesis of the separation of decisions is rejected many times. This depends on the specification used in the dependent variables, independent variables and the retaining estimation method.