Three essays on quantitative analysis in commodity markets

Li, Jiarui

Permalink

https://hdl.handle.net/2142/115329 Copy

Description

Title

Three essays on quantitative analysis in commodity markets

Author(s)

Li, Jiarui

Issue Date

2022-03-29

Director of Research (if dissertation) or Advisor (if thesis)

Irwin, Scott H.

Doctoral Committee Chair(s)

Irwin, Scott H.

Committee Member(s)

• Schnitkey, Gary Donald
• Serra, Teresa
• Etienne, Xiaoli
• Hubbs, Todd

Department of Study

Agr & Consumer Economics

Discipline

Agricultural & Applied Econ

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Energy prices
• Fuels markets
• Wholesale fuel spreads
• Carbon policy
• Low carbon fuel standard (LCFS)
• USDA
• Crop condition survey
• Crop condition index
• Early yield prediction
• Public information
• Commodity, Futures markets, Financialization, Index investment, Directional predictability, Granger causality, Quantile

Abstract

This dissertation consists of three essays that apply quantitative analysis in commodity markets. The first paper essay studies the pass-through impacts from the Low Carbon Fuel Standard (LCFS) program to wholesale fuels provided in California. The goal of LCFS is to reduce carbon intensity of transportation fuels provided in California. LCFS mandates an overall carbon intensity reduction for fuels in gasoline pool and diesel pool by 10% by 2020 through the tradable LCFS credit system. To evaluate if LCFS effectively discourages the consumption of traditional fuels, we estimate the long-run equilibrium and short-run dynamics of the pass-through from the LCFS credit prices to wholesale gasoline and wholesale diesel prices in California from 2016 to early 2020. Our pass-through models control for fundamental time-constant fixed effects and time-varying seasonal patterns in wholesale fuel prices. Wholesale gasoline fuels have quick and complete pass-through over the full sample period, suggesting fuel suppliers can pass the full LCFS credit costs to downstream buyers about 4 business days; for wholesale diesel fuels, they have incomplete long-run pass-through, and over 15 business days, they can only recoup 64% of the LCFS credit costs. The second essay examines the forecasting accuracy of a batch of yield forecasting models that directly transform the ordinal crop condition ratings to the numeric condition index along with a recently developed model introduced by Begueria and Maneta in 2020 that applies the cumulative link mixed model to transform the condition ratings to the continuous condition index. We conduct the out-of-sample yield forecasts recursively for corn and soybean from 2000 through 2020 for all models. We measure the forecasting errors of this group of models and find throughout the growing season, the average root-mean-square-percentage-error (RMSPE) is about 5% for corn and 6% for soybean. Our findings suggest this group of models that use crop conditions data provide accurate yield forecasts. Next, we compare the model developed by Begueria and Maneta (BM model) with its four competing yield forecasting models that have already been widely applied by industry practitioners. Single-horizon model forecasting comparison tests like modified Diebold Mariano test and Model Confidence Set test fail to show that BM model significantly outperforms its competitors for each week throughout the growing season. We also conduct the multi-horizon forecasting comparison test. The results from the average Superior Predictive Ability test show that despite BM model has more complex model specifications, throughout the growing season, it does not provide superior out-of-sample yield forecasts than its competitors. The third essay applies a recently developed cross-quantilogram (CQ) test to examine the impact of Commodity Index Traders (CIT) positions on returns in four agricultural futures markets from 2004 – 2019. Most previous studies reject the basic tenet of the Masters Hypothesis that financial index investments have pressured agricultural futures prices upward. However, the impact of this investment may be more complicated and nuanced than can be detected by the relatively simple linear Granger causality tests used in many previous studies. We conduct three linear causality tests to provide the baseline about the relationship between CIT positions and futures returns. Test results fail to reject the null of no causality in most of the cases across the different tests, measures of position pressure, or the sample period considered. Next, we apply the CQ test of directional predictability in the tails of the distributions of the CIT positions and price movements. Consistent with the standard linear causality tests, we find no evidence that supports the Master Hypothesis.

Graduation Semester

2022-05

Type of Resource

Thesis

Copyright and License Information

Copyright 2022 Jiarui Li

Owning Collections