Conditional Mixed Process Modeling: Applications from the Agriculture Sector in Ghana

Economists are often faced with situations in which the data generating processes are not continuous and unbounded. This is further complicated by the increasing need to model several and mixed (dependent) variables in a system, either as a system seemingly unrelated regressions, or in a more general system of simultaneous equations. However, few econometric models have been developed to handle such empirical questions and many other complex developments in the Stata ecosystem. The Conditional Mixed Process (CMP) Stata package was developed to respond to the increasing complexity of multi-equation, multi-level empirical research. CMP primarily fits a system of seemingly unrelated equations where the dependent variables are conceived as independent but with correlated errors. It is also flexible in fitting considerably larger class of simultaneous equation frameworks where dependent variables are considered endogenous and appear as explanatory variables in the other models, and in recursive settings. The CMP allows mixing of models in a multi-equation system. With the mixed process not being recursive but fully observed, the CMP relies on a maximum likelihood estimation and thereby circumvents the multistage, less efficient, routines of mixed models estimation. This chapter discusses the CMP with Stata implementation and application to data on Ghanaian agriculture.