Abstract: This article compares the theoretical and functional specification of production in partial equilibrium (PE) and computable general equilibrium (CGE) models of the global agricultural and food system included in the AgMIP model comparison study. The two model families differ in their scopepartial versus economy-wideand in how they represent technology and the behavior of supply and demand in markets. The CGE models are deep structural models in that they explicitly solve the maximization problem of consumers and producers, assuming utility maximization and profit maximization with production/cost functions that include all factor inputs. The PE models divide into two groups on the supply side: (1) shallow structural models, which essentially specify area/yield supply functions with no explicit maximization behavior, and (2) deep structural models that provide a detailed activity-analysis specification of technology and explicit optimizing behavior by producers. While the models vary in their specifications of technology, both within and between the PE and CGE families, we consider two stylized theoretical models to compare the behavior of crop yields and supply functions in CGE models with their behavior in shallow structural PE models. We find that the theoretical responsiveness of supply to changes in prices can be similar, depending on parameter choices that define the behavior of implicit supply functions over the domain of applicability defined by the common scenarios used in the AgMIP comparisons. In practice, however, the applied models are more complex and differ in their empirical sensitivity to variations in specificationcomparability of results given parameter choices is an empirical question. To illustrate the issues, sensitivity analysis is done with one global CGE model, MAGNET, to indicate how the results vary with different specification of technical change, and how they compare with the results from PE models.

Abstract: Data quality is a key factor in determining the quality of model estimates and hence a models’ overall utility. Good models run with poor quality explanatory variables and parameters will produce meaningless estimates. Many models are now well developed and have been shown to perform well where and when good quality data is available. Hence a major limitation now to further use of models in new locations and applications is likely to be the availability of good quality data. Improvements in the quality of data may be seen as the starting point of further model improvement, in that better data itself will lead to more accurate model estimates (i.e. through better calibration), and it will facilitate reduction of model residual error by enabling refinements to model equations. This report sets out why data quality is important as well as the basis for additional investment in improving data quality. No Label