Three Essays Evaluating Long-Term Agricultural Projections

Hari Prasad Regmi, Purdue University

Abstract

This dissertation consists of three essays that evaluate long-term agricultural projections. The first essay focus on evaluating Congressional Budget Office’s (CBO) baseline projection of United States Department of Agriculture (USDA) mandatory farm and nutrition programs. The second essay examine USDA soybean ending stock projections, and the third essay investigate impact of macroeconomic assumptions on USDA’s baseline farm income projections. The three essays are summarized as follows: Essay1: AnEvaluation of Congressional Budget Office’s Baseline Projections of USDA Mandatory Farm and Nutrition Programs.The CBO projections of USDA’s mandatory farm and nutrition program outlays play a vital role in shaping agricultural policy and in agricultural policy debates. Using CBO projections and observed outcomes from 1985 through 2020, we examine the degree to which projections of farm, supplemental nutrition assistance, and child nutrition program outlays are unbiased, efficient, and informative. We find that projections for farm program and child nutrition program outlays are unbiased. Supplemental nutrition assistance program outlays are unbiased at short horizons but are downward bias beyond a three-year horizon. We find that all three series of projections are inefficient. The projections for supplemental nutrition assistance program and child nutrition program outlays are informative up to a five-year projection horizon, but the farm program outlay projections are informative for only a one-year horizon. Disaggregated farm program outlay projections since 2008 further suggest that the uninformativeness principally stems from conservation program projections. The findings may provide valuable insights for CBO to improve future projections and for projection users, including policymakers, to adjust expectations and future Farm Bill discussions. Essay 2: Evaluation of USDA’s soybean ending stock baseline projection.The carryover inventory of a specific commodity – ending stock, which summarizes a commodity market’s supply and demand components – is an important measure of level of scarcity in a market. Using USDA baseline projections and realized values from 1997 through 2020, we examine bias and informativeness of soybean ending stock projection. We decompose soybean ending stock projection errors using a machine learning algorithm, Extreme Gradient Boosting (XGBoost). Our results indicate that soybean ending stock projections are unbiased (except for nowcast), however ending stock projections are informative for only one-year horizon. The decomposition of ending stock projection indicates that demand components (crushing, seed and residual, and exports) are primary sources of ending stock projection error for a majority of projection horizons. US soybean market is directly linked with global soybean import and exports, thus, we further investigate USDA projection of soybean export and import country-by-country. Results indicate that USDA’s foreign soybean import projections are barely informative for most of countries including China whereas export projection from Argentina, Brazil, and Other South American countries are informative for four years, two years, and not informative at all, respectively, under our conservative estimates. Finally, an analysis of US soybean export projection error to foreign export/import destinations indicates that errors on export projection from Brazil and other South American Countries (except Argentina) cost USDA the most. Results may help market participants form expectation when making plan and business decisions.

Degree

Ph.D.

Advisors

Kuethe, Purdue University.

Subject Area

Agriculture|Nutrition

COinS