ANALYSIS OF COMMODITY MARKET INFORMATION AND ITS IMPLICATIONS FOR DECISION-MAKING

• Sponsoring Institution: National Institute of Food and Agriculture
• Project Status: COMPLETE
• Funding Source: HATCH
• Reporting Frequency: Annual
• Accession No.: 1013155
• Project Start Date: Oct 1, 2017
• Project End Date: Sep 30, 2022
• Program Code: [(N/A)]- (N/A)

Recipient Organization
VIRGINIA POLYTECHNIC INSTITUTE
(N/A)
BLACKSBURG,VA 24061

Performing Department
Agricultural & Applied Economics

Non Technical Summary
Successful commodity marketing decisions for agricultural market participants critically depend on their ability to analyze the markets and form price expectations. Numerous sources of information are available to market participants that include United States Department of Agriculture (USDA) reports and futures market prices. However, analysis and interpretation of this information is not an easy task and may frequently lead to the difference between profits and losses on the farm. While commodity marketing analysis information is readily available in the main production regions of the Midwest, marginal production areas like Virginia have suffered from the lack of such information since the early 2000s. This lack of commodity marketing analysis information, puts Virginia crop producers at asubstantial disadvantage relative to their more informed counterparts in the Midwestern states with respect to their ability to successfully market their products. In this environment, commodity marketing information including situation and outlook analyses has been highlighted as a number one training need according to the results of the 2016 professional development needs assessment of Virginia Cooperative Extension (VCE) agents. The proposed project will address these needs by providing timely and relevant analysis of commodity market information that can facilitate marketing decisions by producers and agribusiness firms across the state and the region as well as used by VCE agents in their outreach efforts.

Animal Health Component

40%

Research Effort Categories

Basic

30%

Applied

40%

Developmental

30%

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
603	2410	3010	100%

Knowledge Area
603 - Market Economics;

Subject Of Investigation
2410 - Cross-commodity research--multiple crops;

Field Of Science
3010 - Economics;

Keywords

commodity marketing

basis

price forecasting

Goals / Objectives
The goal of this study is to improve commodity marketing efficiency in Virginia and nationwide through improving the quality of commodity market forecasts and demonstrating how they can be used for decision making. This project will address the following problems identified in the previous work:ProblemsObjectivesVirginia producers do not have access to updated information about the difference between cash and futures prices that they can use in their marketing decisions.Collect, analyze, and disseminate basis information and provide guidelines on how it can be used in decision making.Most pricingforecasts are backward looking as most statistical models are based on patterns in an historical price series.Develop forward-looking price forecasts based on futures prices for corn, soybeans, wheat, and feeder cattle.USDA information may no longer be useful in a big data environment, but the USDA is allocating substantial resources to commodity programs.Examine the changing role of USDA information in the big data era.Corn and soybean production forecasts, are inefficient which may cause larger errors and misallocation of resources.Extend forecast evaluation to other commodities as well as focus on implications of "smoothing" on forecast users' welfare.Price interval forecasts have low accuracy and recommendations for improvement have not been adapted.Investigate whether due to the tradeoff between accuracy and informativeness, these forecasts may be preferred to the wider intervals that may be more accurate but less informative.

Project Methods
In order to provide Virginia producers with basis information, this project will 1) collect local cash prices, futures prices and calculate basis relationships for corn, soybeans and wheat over 2000-2014; 2) assess trends in basis relationships over time; 3) develop an on-line tool to assist commodity marketing decisions; 4) provide training and advice on how to use the decision tool and other information developed within this project. The main sources of information regarding cash prices will be Livestock, Poultry and Grain Market News Portal from Agricultural Marketing Service and Commodity Market Newsletters published by the Virginia Department of Agriculture and Consumer Services. Futures price data will be obtained from the CME group.To develop futures-based forecasting models of commodity prices, for each commodity's monthly national cash price, the models will include monthly average nearby (to the forecasted cash) futures price, monthly average nearby futures price for crude oil and monthly average nearby futures price for the US Dollar Index to take into account the impacts of the energy prices and exchange rates on commodity markets. Due to nonstationary nature of price series data, Vector Error Correction Model (VECM) or Vector Autoregression (VAR) models will be used for estimation. Individual commodity equations will be estimated in a system to capture the interactions between agricultural commodity markets and cross-correlation structures in the error terms. Publically available cash and futures prices will be required for this analysis.Value of public information is typically analyzed using the event study approach. Within this framework, information is considered valuable to market participants if prices respond to the information released (the event). Event study methodology may vary widely ranging from simple parametric and nonparametric tests to sophisticated GLS and GARCH-type models. While simple tests may be completely appropriate in cases when the new information in the forecast can be quantified using a market 'surprise' measure (e.g. Colling and Irwin, 1990; Baur and Orazem, 1994; McKenzie, 2008), they may not work as well in cases when the change in information cannot be computed due to the complex and comprehensive nature of the reports (e.g. WASDE reports contain a number of various supply, demand and price forecasts for all major US commodities). In these cases, a dummy variable is typically used to describe report releases and price levels, and volatility can be analyzed on report (event) versus no report days. The problem with this approach is that there is a wide variety of other factors that affect commodity prices that may cause overestimation of report impacts. Failure to control for these other information effects may result in an overestimation of report impacts. Furthermore, if reports are analyzed separately, their impact may be overestimated if several reports are released on the same day. Recent studies (e.g. Isengildina, Irwin and Good, 2006b; Karali, 2012; Adjemian, 2012) developed ways to control for seasonality, day of the week and stock levels as well as other report releases while measuring the impact of public information. GARCH-type models (Xie et al, 2016) allow capturing characteristics of high frequency price data necessary for this type of analysis. Investigation of the changing role of USDA information in a big data era will span a variety of commodities including corn, soybeans, wheat, live cattle, lean hogs, cotton, and orange juice and it will cover multiple USDA reports relevant to these markets over the period from 1970-2016. GARCH-type models will be used to examine the changing impact of these reports on the commodity markets. The data will be obtained from USDA reports and from futures market prices.Analysis of accuracy and efficiency of USDA reports should considerthat most USDA forecasts are fixed-event forecasts as they provide a series of forecasts of the same event at different horizons (e.g., 3rd quarter of 2013 inflation, 2013 corn production). A formal framework to determine the efficiency of fixed-event forecasts was developed by Nordhaus (1987) and was later extended by Clements (1997) and Isengildina, Irwin and Good (2006a). Note the difference from a conventional rolling-event framework, where a series of forecasts of different events is examined. Nordhaus argued that the fixed-event approach may be more powerful than the rolling-event approach, especially in detecting tendencies to systematically adjust forecasts. Rolling-event tests may be unable to detect inefficiency associated with systematic adjustments because forecast errors concerning terminal events occurring at different times are uncorrelated. However, these adjustments would be revealed by the fixed-event approach. The PI's previous work has applied these tests to identify inefficiency in the form of "smoothing" in USDA's corn and soybean production forecasts. Future studies will extend this analysis to other commodities as well as focus on implications of "smoothing" on forecast users' welfare. Most of these tests are typically conducted using OLS regressions. The use of robust regression to account for the impact of outliers (e.g. Xie et al, 2016) may also be explored. Publically available data from USDA reports and futures market prices will be required for this analysis.Evaluation of the tradeoff between accuracy and informativeness in price interval forecasts, where accuracy describes the proportion of time the interval contains the true value and informativeness is measured by the width of the interval, and will be based on the premise that the basis for forecast comparison should be the "value" of the forecast to forecast users. Thus the comparison measure should be based on relative values that reflect the tradeoffs between respective forecast characteristics. This goal can be accomplished by eliciting forecast users' relative values (i.e., willingness to pay) for alternative forecast characteristics using choice experiments. Willingness to pay (WTP) values for alternative forecast characteristics, will be estimated using the random parameter/mixed logit model developed by Revelt and Train (1998). The mixed logit model will be used because it allows efficient estimation of repeated choices by the same respondent within choice-based conjoint experiments. Moreover, this model relaxes the restrictive assumptions of the conditional logit model (Revelt and Train, 1998). Furthermore, we will examine demographic and farm characteristics that may affect forecast user willingness to pay values, such as gender, age, education, risk attitude, years of experience in the market, income/wealth and type of commodity. The impact of these factors on the estimated WTP for each characteristic, will be evaluated in the second stage of the estimation using a system of OLS regressions (following Xie et al, 2016). These experiments will generate data on preferences for forecast attributes, which can be used to estimate a tradeoff function. The tradeoff function will allow integration of various comparison criteria into a single measure that can be used to evaluate and compare forecasts. To the best of our knowledge, this would be the first study to use choice experiments for interval forecast evaluation purposes. If successful, this study will pave way for the use of choice experiments for evaluation of other USDA data products that will result in a direct quantitative measure of benefits that may be compared with the costs of generating this information. We will need to conduct a survey of forecast users to collect data for this study. The main categories of forecast users include agricultural producers, agribusiness firms and market advisory companies. An IRB approval will be obtained before the survey is conducted.

Progress 10/01/19 to 09/30/20

Outputs
Target Audience:Agricultural market participants that use USDA forecasts for decision making, scientists that examine the impact of USDA forecasts on the markets, USDA forecast providers. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?A Ph.D. student, Xiang Cao has been involved in one of the papers that have been published. He has made a considerable contribution to this research and presented it as one of the chapters of his dissertation. Another Ph.D. student, Armine Poghosyan is actively involved in a new study associated with goal 2. She has made a considerable contribution to the review of literature and model development. How have the results been disseminated to communities of interest?Because of the COVID pandemic, no conference presentations have been made. However, I have been very active in disseminating the findings of these studies and related issues in workshops and popular media interviews: May 5, NBC Today Kayla Ruble interview https://www.today.com/food/why-there-meat-shortage-if-farmers-have-plenty-animals-t180781 May 4. Impact of COVID-19 on Virginia Grain and Oilseed Markets. https://aaec.vt.edu/extension/resources.html April 28. Interview with Brett Molina, USA Today https://www.usatoday.com/story/money/2020/05/01/meat-shortages-how-worried-should-americans-be/3067983001/ April 27, email interview with Dave Ress, Daily press https://www.dailypress.com/business/consumer/dp-nw-pork-prices-20200427-whv6lq4ptfhy3b5qsrrytpow2u-story.html April 21, interview with Niv Elis, the Hill. https://thehill.com/homenews/news/494014-five-threats-to-us-food-supply-chains April 14, NBC Today Kayla Ruble interview - Farmers are throwing away fresh food and dairy. Food banks want to change that April 13, WBDJ interview https://www.wdbj7.com/content/news/Prices-could-increase-as-the-coronavirus-closes-meat-processing-plants-569607801.html April 6, Richmond times interview https://www.richmond.com/business/pandemic-is-presenting-challenges-for-local-farmers-but-they-are-adjusting/article_8eba710f-5769-5a4a-a57d-95e71531a54b.html April 2, Ag Policy and Market Outlook forum presentation https://video.vt.edu/media/t/1_dhmn6n6l March 26, VTnews interview (https://vtnews.vt.edu/articles/2020/04/cals-food-supply-chain.html) March 23, Sarah Vogelsong interview What do you plan to do during the next reporting period to accomplish the goals?My activities in the next reporting period will be focused on objectives 1 and 2. While considerable progress has been made on objective 1 last year, additional tasks associated with developing training tools for using Virginia basis information still remain and have not yet been completed Considerable progress has been made on objective 2 with a lot of the research completed. The next step for this study would be to refine several research methods and prepare the paper for a conference presentation and journal publication. For goal 5, we have initiated a new project aimed at collecting and analyzing feedback on how to improve USDA's outlook programs. Within this project, we will conduct 3 workshops, 8 personal interviews, and 2 focus groups with the goal of outlining a roadmap to increase sustainability and enhance accuracy and competitiveness of USDA outlook products.

Impacts
What was accomplished under these goals? Goal 1: Basis information was updated and used to analyze the Impact of COVID-19 on Virginia Grain and Oilseed Markets. https://aaec.vt.edu/extension/resources.html Goal 2: I have been working with a graduate student, Armine Poghosyan, on developing forward-looking price forecasts based on futures prices for corn, soybeans, wheat. Her proposal for a selected paper was accepted for 2020 AAEA meetings but not presented due to the COVID pandemic. We are continuing to make progress on this study and plan to present it at AAEA meetings next year. Goal 3: Considerable progress has been made toward this goal: Two manuscripts have been accepted and published: Isengildina-Massa, O., Cao*, X., B. Karali, S.H. Irwin, M. Adjemian, and R. Johansson. 2020. "When Does USDA Information Have the Most Impact on Crop and Livestock Markets?" Journal of Commodity Markets, https://doi.org/10.1016/j.jcomm.2020.100137. Isengildina-Massa, O., B. Karali, S.H. Irwin. 2020. "Can Private Analysts Beat USDA? Analysis of Relative Accuracy of Crop Acreage and Production Forecasts." Journal of Agricultural and Applied Economics, https://doi.org/10.1017/aae.2020.18. For objective 4, we examined an interesting puzzle of how supply fundamentals contained in USDA forecasts affect grain futures price movements and published the following article: Karali, B., S.H. Irwin, and Olga Isengildina-Massa. 2020. "Supply Fundamentals and Grain Futures Price Movements." American Journal of Agricultural Economics 102(2):548-568.

Publications

• Type: Journal Articles Status: Published Year Published: 2020 Citation: Isengildina-Massa, O., Cao*, X., B. Karali, S.H. Irwin, M. Adjemian and R. Johansson. 2020. When Does USDA Information Have the Most Impact on Crop and Livestock Markets? Journal of Commodity Markets, https://doi.org/10.1016/j.jcomm.2020.100137.
• Type: Journal Articles Status: Published Year Published: 2020 Citation: Isengildina-Massa, O., B. Karali, S.H. Irwin. 2020. Can Private Analysts Beat USDA? Analysis of Relative Accuracy of Crop Acreage and Production Forecasts. Journal of Agricultural and Applied Economics, https://doi.org/10.1017/aae.2020.18.
• Type: Journal Articles Status: Published Year Published: 2020 Citation: Karali, B., S.H. Irwin, and Olga Isengildina-Massa. 2020. Supply Fundamentals and Grain Futures Price Movements. American Journal of Agricultural Economics 102(2):548-568.
• Type: Other Status: Published Year Published: 2020 Citation: Impact of COVID-19 on Virginia Grain and Oilseed Markets. https://aaec.vt.edu/extension/resources.html

Progress 10/01/18 to 09/30/19

Outputs
Target Audience:Agricultural market participants that use USDA forecasts for decision making, scientists that examine the impact of USDA forecacts on the markets, USDA forecast providers. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?A PhD student, Xiang Cao has been involved in one of the papers that is currently under review. He has made a considerable contribution in this research. How have the results been disseminated to communities of interest?Three of these studies have been presented at the NCCC-134 Conference on Applied Commodity Price Analysis, Forecasting and market risk management and published in the conference proceedings. Two of these studies have also been presented at the American Agricultural Economics meetings. Food Policy and the Journal of Agricultural and Resource Economics are top tier outlets with extensive readership. What do you plan to do during the next reporting period to accomplish the goals?My activities in the next reporting period will be focused on concluding work on objectives 1 and 2 and starting work on objective 3. While considerable progress has been made on objective 1 last year, additional tasks associated with developing training tools for using Virginia basis information still remain and have not yet been completed due to personnel issues. Interested students have been identified and the work on this objective will resume and hopefully be competed next year. Considerable progress has been made on objective 2 with a lot of the research completed and presented at a conference last year. The next step for this study would be to refine several research methods and prepare the paper for a journal publication. For objective 4, we will examine an interesting puzzle of how supply fundamentals contained in USDA forecasts affect grain futures price movements.

Impacts
What was accomplished under these goals? Considerable progress has been made for objective 3: Examine the changing role of USDA information in the big data era. Two papers have been published and two papers are under review.

Publications

• Type: Journal Articles Status: Published Year Published: 2019 Citation: Karali, B., O. Isengildina-Massa, S.H. Irwin, M.K. Adjemian, and R. Johansson. 2019. Are USDA Reports Still News to Changing Crop Markets? Food Policy 84:66-76.
• Type: Journal Articles Status: Published Year Published: 2019 Citation: Karali, B., O. Isengildina-Massa and Scott H. Irwin. The Changing Role of USDA Inventory Reports in Livestock Markets. Journal of Agricultural and Resource Economics 44(3):591-604).
• Type: Journal Articles Status: Under Review Year Published: 2019 Citation: O. Isengildina-Massa, X. Cao, B. Karali, S.H. Irwin, M. Adjemian and R. Johansson. When Does USDA Information Have the Most Impact on Crop and Livestock Markets? Journal of Commodity Markets.
• Type: Journal Articles Status: Under Review Year Published: 2019 Citation: Isengildina-Massa, O., B. Karali, S.H. Irwin. Can Private Analysts Beat USDA? Analysis of Relative Accuracy of Crop Acreage and Production Forecasts. Journal of Agricultural and Applied Economics.
• Type: Conference Papers and Presentations Status: Accepted Year Published: 2019 Citation: Isengildina Massa, O., B. Karali and S.H. Irwin. "Relative Accuracy of Crop Acreage and Production Forecasts." Invited paper presented at the Agricultural and Applied Economics Association Annual Meeting, Atlanta, GA, July 21-23, 2019.

Progress 10/01/17 to 09/30/18

Outputs
Target Audience:Virginia agricultural producers, agribusiness firms, as well as VCE agents. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?A number of undergraduate students were involved in research associated with objective 1, including Morgan McCoy, Jordan Pulley, Lamie Carrol, and Meghan McLaughlin. These students received credit for undergraduate research work and are currently successfully employed by various agribusiness firms and agricultural lending companies. Furthermore, Hunter Watkins and Meleah Shadler received credit for Kohl Center class for their work on assessing training needs associated with commodity marketing information. How have the results been disseminated to communities of interest?A website was built to disseminate commodity basis information to Virginia agricultural market participants and extension agents: https://aaec.vt.edu/extension/va-commodity-market.html The paper describing the futures based forecasting models for U.S. crops was presented at the NCCC-134 Conference on Applied Commodity Price Analysis, Forecasting and market risk management and published in the conference proceedings: http://www.farmdoc.illinois.edu/nccc134/conf_2017/pdf/Zhu_Isengildina_NCCC-134_2017.pdf What do you plan to do during the next reporting period to accomplish the goals?My activities in the next reporting period will be focused on objective 3: examine the changing role of USDA information in the big data era. External funding for this project was obtained through a cooperative agreement with the office of Chief Economist of USDA. This study will examine the theoretical models and empirical evidence of the value of public information in agricultural markets in order to determine how big data and access to information affects the role and impact of USDA situation and outlook programs. Furthermore, additional progress will be made on objective 1 associated with developing training tools for using Virginia basis information.

Impacts
What was accomplished under these goals? Considerable progress has been made for objective 1: collect, analyze, and disseminate basis information and provide guidelines on how it can be used in decision making. Several undergraduate students (Morgan McCoy, Jordan Pulley, Lamie Carrol, and Meghan McLaughlin) were hired to collect crop cash price data from Agricultural Marketing Service database (https://www.ams.usda.gov/market-news/custom-reports), feeder cattle price data from VDACS and futures price data from Quandl. The resulting data sets were organized to calculate basis values for main Virginia locations. These data sets were made available at: https://aaec.vt.edu/extension/va-commodity-market.html The next steps should be devoted to developing training materials for using these data. Several undergraduate students (Hunter Watkins and Meleah Shadler) participated in the Kohl Center project in the Fall of 2018 with the goal of better identifying the training needs for using commodity marketing information. Their project summary provides important guidelines for the next steps that should be taken to accomplish this objective. Due to funding availability and specific interests of a Master's student participating in this project, objective 2 was accomplished ahead of schedule. This study proposed two futures-based models for forecasting cash prices of corn, soybeans, wheat and cotton over the period 2000-2016. The difference model predicted changes in cash prices as a function of changes in futures prices. The regime model specified different market regimes and modeled cash price levels based on observed futures prices in various regimes. The out-of-sample performance of both models was compared to the benchmark of a 5-year moving average over the 2013-2016 sub-period. The findings indicated that the regime model performed best for corn and soybeans. While, neither model beat the benchmark for wheat at the longer forecasted horizons, the difference model performed well at short forecast horizons (up to 5- months ahead). Both models performed better than the benchmark for cotton price forecasts, but they were not significantly different from each other. This paper was presented at the NCCC-134 Conference on Applied Commodity Price Analysis, Forecasting and market risk management and published in the conference proceedings: http://www.farmdoc.illinois.edu/nccc134/conf_2017/pdf/Zhu_Isengildina_NCCC-134_2017.pdf

Publications

• Type: Conference Papers and Presentations Status: Published Year Published: 2017 Citation: Zhu, J., and O. Isengildina-Massa. 2017. Futures-Based Forecasts of U.S. Crop Prices. Proceedings of the NCCC-134 Conference on Applied Commodity Price Analysis, Forecasting, and Market Risk Management. St. Louis, MO. [http://www.farmdoc.illinois.edu/nccc134].