An Efficient Forecasting Procedure for Kuhn-Tucker Consumer Demand Model Systems

摘要: This paper proposes an efficient and accurate forecasting algorithm for the multiple discrete-continuous nested extreme value (MDCEV) model. The builds on simple, yet insightful, analytical explorations with Kuhn-Tucker conditions of optimality that shed new light properties MDCEV For specific, but reasonably general, functional forms consumption utility specification, circumvents need to carry out any iterative constrained optimization procedures have hitherto been used (KT) demand model systems. non-iterative nature contributes significantly its efficiency accuracy. Further, although developed in context model, proposed can be easily modified other maximization-based consumer systems literature. Simulation experiments highlight compared a traditional procedure. example, forecast expenditures 4000 households 7 expenditure alternatives, 500 sets error term draws each household, takes less than 2 minutes. On hand, routine would take around days do so.