A dynamic semi-nonparametric demand system: An application to U.S. pork import demand
Price and substitution elasticities are important measures when conducting import demand analysis. However, measurement of these elasticitites is confounded by several factors. First, importing firms typically serve as the demanders of agricultural imports. Yet, agricultural import demand is often modeled as a final consumer demand. Second, use of flexible functional forms often results in violation of not just global, but also local regularity conditions of demand (either input or final good). This thesis addresses both of these issues using data on U.S pork imports from Canada, Denmark and the rest of the world. A dynamic Fourier cost function trade model based on producer theory is developed and a dynamic globally flexible AIDS model is also developed based on consumer theory. Dynamics are incorporated into the Fourier series expansion components in terms of habit persistent formation for the consumer model and the adjustment cots for the producer model. An empirical illustration of the difference in estimates between consumer model and producer model is provided. Estimates from the producer model satisfy the required regularity conditions, such as curvature condition of cost function and negative own price effects, whereas the consumer model does not satisfy the curvature condition. Results support the proposed dynamics and point to the proposed firm-based model as a potentially useful model in future applications. This study is the first empirical analysis of US pork import demand using the dynamic globally flexible function.