Forecasting the Nominal Brent Oil Price with VARs—One Model Fits All?

IMF Working Papers, November 25, 2015

Source details

Canonical URL: Forecasting the Nominal Brent Oil Price with VARs—One Model Fits All?

The paper carries out an ex post assessment of popular models used to forecast oil prices and proposes a host of alternative VAR models based on traditional global macroeconomic and oil market aggregates.
The bias and underprediction in futures and random walk forecasts are larger across all horizons in relation to a large set of VAR specifications.
The VAR forecasts generally have the smallest average forecast errors and the highest accuracy.
Most VAR specifications outperform futures and random walk forecasts for horizons up to two years.
Combining futures, random walk and VAR models for forecasting have merit for medium term horizons.

While VAR models show overall strength, the paper highlights performance instability:
- Small alterations in specifications, subsamples or lag lengths can provide widely different forecasts at times.
The exact specification of VAR models for nominal oil price prediction is still open to debate.
The findings call for caution in reliance on futures or the random walk for forecasting, particularly for near term predictions.

Subject: Commodities, Econometric analysis, Economic forecasting, Financial institutions, Futures, Oil, Oil prices, Prices, Vector autoregression
Keywords: Brent price, factor VAR, forecasting, forecasting model, Futures, Global, intercept term, lag length, medium-term oil price forecasting model, North America, null hypothesis, oil, oil price, Oil prices, RAC price series, random walk model, rival VAR specification, trended oil price model, VAR forecast, VAR model, VAR system, VARs, Vector autoregression, WP, WTI crude

Source: IMF Working Paper — Forecasting the Nominal Brent Oil Price with VARs—One Model Fits All?, November 25, 2015.

_wp15251 - References
- _wp15251 - References (Markdown version)
- _wp15251 - References (PDF)