Price jumps and volatility in U.S. agricultural futures markets

Live cattle;futures;microstructure;noise;realized volatility;integrated variance;corn futures;intraday;price jumps;jump risk;information shocks;nonparametric test;agricultural commodity;heterogeneous autoregressive model;artificial neural network;forecasting;long-memory;nonlinearities.

Couleau, Anabelle

关键词: Live cattle;    futures;    microstructure;    noise;    realized volatility;    integrated variance;    corn futures;    intraday;    price jumps;    jump risk;    information shocks;    nonparametric test;    agricultural commodity;    heterogeneous autoregressive model;    artificial neural network;    forecasting;    long-memory;    nonlinearities.;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/105638/COULEAU-DISSERTATION-2019.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Agricultural commodity futures markets have changed with the arrival of electronic trading. Electronic trading platforms have facilitated the emergence of automated systems in these markets which are now experiencing a race among traders to gain speed in implementing transactions. This new trading environment raises questions about the increasing role of high-speed traders and their effect in agricultural futures markets. In the first two essays in this dissertation, I examine how recent structural changes associated with increased speed of trading and release of public information experienced by these markets affect prices and volatility dynamics. In the third essay, I investigate whether more flexible research approaches should be employed to provide market participants and policy markets more accurate volatility forecasts within the context of the new more heterogeneous trading environment.The first essay identifies both the magnitude and the duration of the bias caused by market microstructure noise in measuring efficient price variance in the live cattle futures market from 2011 to 2016, with emphasis on price variance behavior in recent years. The U.S. live cattle futures prices have experienced high levels of intraday price variance, starting in 2015, which have raised concerns about the possible impact of microstructure noise from high frequency trading on market instability. Market microstructure noise increases observed price variance, but its effects are not large and do not last more than three to four minutes in response to changing information. Intraday price variance has increased in recent years, but the findings provide little evidence that high frequency traders were responsible for economically meaningful market noise. Informatively, steps taken by the CME and cattle producers to mitigate noise have not been fruitful to date, and signal that the magnitude of noise will likely vary with the magnitude of changes in demand and cyclical supply.The second essay demonstrates that jumps in corn futures prices have increased with electronic trading and the shift to real-time announcement of USDA reports. Using intraday prices from 2008 to 2015, we employ a nonparametric test to detect jumps and variance analysis to estimate jump or execution risk. Real-time trading of major USDA reports has substantially increased the frequency and clustering of price jumps, and results in higher market liquidity costs. In contrast, while the presence of jumps on non-announcement days has doubled recently, their magnitude has declined as have transactions costs during their occurrence. The largest jump risk or execution risk is experienced by high frequency traders due to heightened microstructure noise during price jumps.The third essay investigates the ability of artificial neural network (ANN) to forecast realized volatility in the corn futures market. Forecasting volatility is complicated by heterogeneous expectations from a diversity of traders and by nonlinearities such as seasonality or public information shocks (USDA announcements). Recent applications of artificial neural networks in econometrics suggest this model is particularly suited in capturing unknown nonlinearities forms. Using corn futures prices observed between 2009 and 2017, this paper compares the volatility forecasting performance of nonlinear autoregressive ANN models against other alternative linear specifications, such as the heterogeneous autoregressive (HAR) model which account for heterogeneity in volatility expectations. Our findings indicate that the nonlinear ANN model works better at all horizons (1-day, 1-week, and 1-month) than the standard HAR model even when the HAR model is augmented for seasonality or public information shocks, pointing to the importance of accounting for unknown forms of nonlinearities through more flexible approaches.