Skip to Main Content
 

Global Search Box

 
 
 

ETD Abstract Container

Abstract Header

Evaluating USDA Agricultural Forecasts

Bora, Siddhartha S
http://orcid.org/0000-0002-8791-9989
http://rave.ohiolink.edu/etdc/view?acc_num=osu16584577802286

Abstract Details

2022, Doctor of Philosophy, Ohio State University, Agricultural, Environmental and Developmental Economics.
The timely availability of accurate forecasts plays a vital role in informing decisions by farm sector stakeholders. In this dissertation, I evaluate the rationality, accuracy, and informativeness of a range of agricultural forecasts and projections published by the United States Department of Agriculture (USDA) and other agencies and examine ways to improve them. The findings have implications for future revisions of the forecasting processes and for policymakers, agricultural businesses, and other stakeholders who use these forecasts. In Chapter 1, I show that some of the reported biases and inefficiencies in USDA forecasts may be due to an asymmetric loss of the forecaster. Many previous studies suggest that many USDA forecasts are biased and/or inefficient. These findings, however, may be the result of the assumed loss function of USDA forecasters. I test the rationality of the USDA net cash income forecasts and the World Agricultural Supply and Demand Estimates (WASDE) production and price forecasts between 1988-2018 using a flexible multivariate loss function that allows for asymmetric loss and non-separable forecast errors. My results provide robust evidence that USDA forecasters are rational expected loss minimizers yet demonstrate a tendency to place a greater weight on under- or over-prediction. As a result, this study provides an alternate interpretation of previous findings of forecast irrationality. Agricultural baselines play an important role in shaping agricultural policy by providing information about the farm sector for a ten-year horizon, yet these projections have not been rigorously evaluated. In Chapter 2, I evaluate the accuracy and informativeness of two widely used baselines for the US farm sector published by the USDA and the Food and Agricultural Policy Research Institute (FAPRI) in three steps. First, I examine the average percent errors of the projections and perform tests of bias. Second, I use a novel testing framework based on the encompassing principle to test the predictive content of the projections for each horizon, determining the longest informative projection horizon. Third, I compare the USDA and FAPRI baseline projections using a multi-horizon framework that considers all projection horizons jointly. I find that prediction error and bias increase with the horizon’s length. The predictive content of the baselines projections for most variables diminishes after 4-5 years. The multi-horizon comparison suggests that neither USDA nor FAPRI projections have uniform or average superior predictive ability over the other for most variables. Multi-step forecasts about commodity market indicators play an important role in informing policy and investment decisions by governments and market participants. In Chapter 3, I examine whether the accuracy of long-term forecasts can be improved using deep learning models. I first formulate a supervised learning problem and set benchmarks for forecast accuracy. I train a set of deep neural networks on a training sample and measure their performance against the benchmark model on a test sample using a walk-forward validation strategy. I find that while the USDA baseline projections perform better for the shorter horizon, the performance of the deep neural networks improves for the longer forecast horizons.
Ani Katchova (Advisor)
Wuyang Hu (Committee Member)
Brian Roe (Committee Member)
Todd Kuethe (Committee Member)
149 p.
Agricultural Economics
forecast evaluation, asymmetric loss, fixed-event forecasts, forecast rationality, net cash income, WASDE., agricultural baselines, USDA, deep learning, commodity forecasts

Recommended Citations

Citations

  • Bora, S. S. (2022). Evaluating USDA Agricultural Forecasts [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu16584577802286

    APA Style (7th edition)

  • Bora, Siddhartha. Evaluating USDA Agricultural Forecasts. 2022. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu16584577802286.

    MLA Style (8th edition)

  • Bora, Siddhartha. "Evaluating USDA Agricultural Forecasts." Doctoral dissertation, Ohio State University, 2022. http://rave.ohiolink.edu/etdc/view?acc_num=osu16584577802286

    Chicago Manual of Style (17th edition)

osu16584577802286
187
© 2022, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.