Agric and Resource Economics
Non Technical Summary
This research project aims at measuring the impact of trade disputes on agriculture in the United States. Because U.S. farmers and food processors sell a significant share of their production abroad, the growing number of trade disputes is a primary concern for the future viability of agriculture. However, so far we know little about the implications of these policies because reliable counterfactuals for causal inference are generally unavailable. Therefore, this project aims at providing the necessary tools to precisely measure the impact of international trade disputes on U.S. agriculture. We accomplish this task by developing novel and innovative analytical methods based on machine learning techniques. These methods allow us to construct credible counterfactual trade flows, obtain a more precise identification of trade effects and evaluate the impact according to dispute characteristics, product specificities, and timing of trade measures. The analysis will enable us to not only assess the importance of trade disputes for agriculture in the United States but also to provide the necessary means to determine the impact of future trade disputes. Therefore, our project will enhance the understanding of a highly relevant foreign trade issue that has been so far largely neglected in the empirical trade literature while being of vital importance for the future of agriculture in the United States. Such understanding provides the foundation for a public discussion that is based on facts and allows for informed decisions. Consequently, our research will enhance market efficiency and performance by providing essential knowledge on the functioning of markets in light of trade disputes.
Animal Health Component
Research Effort Categories
Goals / Objectives
The long-term goal of this project is to precisely measure the impact of international trade disputes on U.S. agricultural exports. We will accomplish this task by developing novel and innovative methods based on machine learning techniques to construct credible counterfactual trade flows for causal inference. These benchmarks allow us to obtain a more precise identification of the trade destruction effect caused by trade disputes involving the United States.The research objectives are as follows:Assemble a complete dataset of international trade disputes targeting U.S. agricultural exports over the period 1990 to 2018.Prepare a time-consistent monthly dataset of U.S. foreign trade carefully disaggregated by specific product at the customs-district level.Develop a novel statistical approach based on machine learning techniques to create credible trade flowcounterfactuals and measure the impact of agricultural trade disputes more precisely.Evaluate the impact of trade disputes on U.S. agricultural trade and explore differences in the trade impact according to dispute characteristics, product specificities, and timing of trade measures.
The primary challenge for accuratelymeasuring the impact of trade disputes on U.S. agricultural exports relates to the specification of a plausible benchmark for the statistical evaluation of policy effects. Such a counterfactual can be either based on a prior probability distribution or derived from a data generating process. An assessment based on priorsis prone to error because reliable information on market characteristics and trend similarities are often not available, particularly at the disaggregated time and product levels. Therefore, we will use additional information on trade flows to improve our benchmark. For this purpose, we will exploit temporal variation in foreign trade data at the product level with machine learning techniques. We build on the assumption that alternative products show similar trade patterns as the products targeted in a trade dispute. These products are neither substitutes nor complements of the affected products. By excluding substitutes and complements, we can ensure that spillover effects in product space do not affect the causal inference strategy. We will implement five modules to create valid counterfactuals and infer the causal relationship between trade duties and U.S. agricultural export trade flows.