Skip to main content

Objective-based forecast evaluations for crude oil volatility.

Periodic Reporting for period 1 - FOROIL (Objective-based forecast evaluations for crude oil volatility.)

Reporting period: 2017-08-01 to 2019-07-31

Oil price volatility forecasting is of major importance due to the financialisation of the oil market and the fact that the oil market participants’ decisions are based on such forecasts (e.g. oil-intensive industries, policy makers, portfolio traders). Currently, forecasters mainly predict oil price conditional and realized volatility using primarily GARCH and HAR models and evaluate the forecasts’ performance using statistical loss functions. Nevertheless, oil price volatility users are faced with (i) multiple volatility measures apart from conditional and realized, (ii) multiple forecasting models and (iii) different applications for which they use oil price volatility forecasts (e.g. policy making, portfolio allocation, risk management). Hence, the evaluation of the different forecasts using statistical loss functions is not adequate. Thus, in order to make informed decisions, oil volatility users need to know the most appropriate volatility measure in combination with the most accurate forecasting model. Thus, this project provides a framework which considers a range of volatility measures and models and allows oil volatility users to choose the most appropriate volatility measure combined with the best forecasting model, according to the economic decision for which the forecast will be used. To achieve this we develop loss functions that reflect the purpose of the oil price volatility forecasts, i.e. objective-based loss functions rather than stand-alone statistical ones. The development of such framework allows participants to make informed decisions which lead to better policy mix by policy makers, portfolio allocation by investors, or well-performed risk management by oil-intensive industries or regulators. By contrast, should end-users evaluate the forecasts based on statistical loss functions, then their economic decisions are sub-optimal. Hence, this innovative project has laid the foundations for an advanced econometric model framework to be used as a policy and practice suite of tools for the evaluation of the most appropriate oil volatility measures combined with the most accurate forecasting models, based on objective-based loss functions.
The extant literature supports that oil price forecasting is important for a number of stakeholders, including policy makers, investors, firms and households, given the role that oil prices play on several aspects of economic activity. International institutions and global media also link the macroeconomic stability with oil price fluctuations.
Thus, the overarching aim was to lay the foundations for an advanced econometric model framework for the evaluation of the most appropriate oil volatility measures combined with the most accurate forecasting models, using objective-based loss functions. The first research objective was to construct the variations of volatility measures that had been proposed in the econometric theory. The second research objective was to estimate the most appropriate volatility measures combined with the most accurate forecasting models, suitable for oil traders, portfolio managers and risk managers. Finally, the third research objective was to estimate the most appropriate volatility measures combined with the most accurate forecasting models, suitable for policy makers and regulators. In this project we confine our interest to the usefulness of oil price volatility forecasts for investors and policy makers.
We first show that the construction of the different intraday volatility measures do contain different information for the future path of oil price volatility. This is important as the different information could assist in the improvement of oil price volatility forecasts.
Having constructed the different intraday volatility measures, we show that, contrary to the current practice that mainly considers stand-alone statistical loss functions, oil price volatility forecasts should be assessed based on objective-based evaluation criteria, given that different forecasting models may exhibit superior performance at different applications.
Our results convincingly show that our forecasting framework is economically useful, since different models and volatility measures provide; (1) superior after-cost profits depending on the financial use of the volatility forecasts and (2) improved conditional forecasts of macroeconomic variables depending on the indicator of interest.
The conclusions of the project were widely disseminated using various means, as suggested in the project’s proposal. We presented the findings of the project in five staff seminar series in the UK and international academic institutions, as well as, in public talks at the Festival of Learning (Bournemouth University) and Panteion University of Social and Political Sciences. In addition, we presented our results in three short videos at website. Furthermore, we presented the findings in five international conferences and two workshops. The Marie Sklodowska-Curie researcher, Stavros Degiannakis, has successfully accomplished the project, delivered the four working packages, and submitted papers in international journals that have been produced during the fellowship. The working papers of the projects were also made available to the wider public via the project’s website. The Community support of the Marie Sklodowska-Curie Action is being acknowledged in publications and presentations.
FOROIL has made a significant contribution beyond the current state-of-the-art in forecasting oil price volatility. In particular, we have added to the literature of oil price volatility forecasting in two very important ways. First, we evaluated for the first time the forecasting power of models that have not been yet considered in the area of oil price volatility. Second, we used for the first time several objective-based loss functions in order to evaluate oil price volatility forecasts of different oil volatility measures. These objective-based loss functions reflected the different purposes for which oil volatility forecasts are required. There are substantial benefits to various stakeholders from developing our framework, including the ability to: i) assess predictive content; ii) make recommendations which are conditioned on a broader set of information; iii) pass the internal consistency and accuracy tests of the forecasts and iv) adopt the proposed framework even when the market conditions have changed.
So far the project has managed to achieve the intended short-run impact. The fellow gained expertise in state-of-the-art oil price forecasting techniques and in the development of objective-based loss functions, as well as, in the energy market’s structure and operations. Furthermore, the fellow gained expertise in outreach activities, consultancy and policy formulation skills, and wider dissemination strategies. The fellow has also strengthened his affiliations with the UK and international academic and non-academic sector, which allowed him to expand his research network and obtain consulting experience. The fellow was also trained in methods for effecting counselling, managing expectations and effective communication of the practical implication of the research output. Furthermore, the fellow has acted as Marie Skłodowska-Curie Ambassador and communicated effectively his Marie Skłodowska-Curie Actions experience and thus raised awareness of the ERA efforts for research excellence.
First page of the FOROIL technical report