Testing the Separation Hypothesis with Farmer Heterogeneity: A New Use of the Structural Break Test

Agricultural households respond differently to market imperfections: production decisions are not separated from preferences or utility. Therefore, in order to correctly model household decisions it is important to test for this separation. Many innovative tests have been developed in the literature but most of them assume no heterogeneity across the farmers (i.e. all farmers are assumed to behave in the same way) and so these tests identify only one model for every farmer in the sample. By using a structural break test from time-series econometrics and by developing a search dimension in cross-section data that can act like a time dimension in time-series data, the test in this paper can allow for heterogeneity across the farmers. Thus, it is able to identify the correct model for each farmer in the sample. * I would like to thank John McLaren, John Pepper and Sanjay Jain for their helpful advice and comment. All remaining errors are my own.