Empirical spatial models of trade that are based on a mathematical programming specification often exhibit a large discrepancy between the equilibrium solution and the observed demand, supply and levels of trade flows among countries. This discrepancy may be due to several causes. Assuming, however, that a trade model is not misspecified – in the sense that behavior of the economic agents involved in the specific commodity markets has been included in the study and that the relevant policy instruments have been properly taken into account – the cause of discrepancy may be traced either to imprecision of unit transaction costs or to imprecision in the measurement of the demand and supply functions' parameters, or both. Policy assessments based on this type of imprecise models are distorted. This paper presents a methodology for calibrating mathematical programming spatial trade models of increasing complexity, from the one-commodity case to a multi-commodity model with asymmetric slope matrices of demand and supply functions. The proposed calibration procedure identifies corrections of imperfectly measured parameters. The calibrated models generate solutions that exactly reproduce quantities produced and consumed in all countries, as well as trade flows among all pairs of countries, observed in a given base year. Such models may then serve as a springboard for assessing the impact of various policy changes on economic agents in the countries under study.
