Determining Time-Varying Drivers of Spot Oil Price in a Dynamic Model Averaging Framework

Abstract

This article presents results from modelling spot oil prices by Dynamic Model Averaging (DMA). First, based on a literature review and availability of data, the following oil price drivers have been selected: stock prices indices, stock prices volatility index, exchange rates, global economic activity, interest rates, supply and demand indicators and inventories level. Next, they have been included as explanatory variables in various DMA models with different initial parameters. Monthly data between January 1986 and December 2015 has been analyzed. Several variations of DMA models have been constructed, because DMA requires the initial setting of certain parameters. Interestingly, DMA has occurred to be robust to setting different values to these parameters. It has also occurred that the quality of prediction is the highest for the model with the drivers solely connected with the stock markets behavior. Drivers connected with macroeconomic fundamental indicators have not been found so important. This observation can serve as an argument favoring the hypothesis of the increasing financialization of the oil market, at least in the short-term period. The predictions from other, slightly different modelling variations based on DMA methodology, have happened to be consistent with each other in general. Many constructed models have outperformed alternative forecasting methods. It has also been found that normalization of the initial data, although not necessary for DMA from the theoretical point of view, significantly improves the quality of prediction.

Forecasting the Crude Oil Spot Price with Bayesian Symbolic Regression

“…Further, the dividend-to-price ratio was taken, constructed as the difference between the logarithm of 12-month moving sums of U.S. stock dividends [81,82] and the logarithm of the S&P 500 index. Similar variables were found to be important commodities price predictors by, for example, [7,83,84], or for financial markets in general, by [85].…”

Forecasting Selected Commodities’ Prices with the Bayesian Symbolic Regression

“…However, some benchmark models require, for example, stationary time-series. Secondly, even if not necessary, transformed data can often result in a better forecast accuracy of the final models (Coulombe et al 2021;Medeiros et al 2019;Drachal 2018a). The widest basket of commodities was attempted to be collected.…”

Predicting House Prices Using DMA Method: Evidence from Turkey

“…However, it was decided not to transform i_rate further, because of its economic interpretation and properties. Secondly, in the dynamic model averaging scheme, stationarity is not a significant obstacle towards inserting a variable into the modeling scheme (Drachal 2019a). Notes: RPPI denotes the logarithmic differences of Residential Property Price Index, mortgage-logarithmic differences of the level of mortgages in thousands of TRY, dwellings-the number of two or more dwelling residential buildings, cpi-logarithmic differences of Consumer Price Index, fdi-foreign direct investment in real estate activities, i_rate-interest rate for housing, ipi-Industrial Production Index, u_rate-unemployment rate, stocks-logarithmic differences of Borsa Istanbul 100 Index, usd-logarithmic differences of the USD/TRY exchange rate, eur-logarithmic differences of the EUR/TRY exchange rate, gt-the Google Trends index for the search query, ird-the difference between the 10-year Turkish bond yield and 2-year bond yield.…”