The relationship between the price of oil and macroeconomic performance: empirical evidence for the UK

Abstract

This study adopts a vector autoregressive (VAR) approach towards investigating the relationship between the price of oil and UK macroeconomic performance over the period of floating exchange rates. Its distinctive feature is the allowance for a systematic variation in the macroeconomic effects of a change in the price of oil, in recognition of fundamental developments which have occurred with respect to the structure of the UK economy. Empirical analysis indicates that the accommodation of this characteristic within a VAR model increases both the prominence and the pervasiveness of the impact of an oil price shock.

Notes

¹ For example, the papers by Gisser and Goodwin (1986), Mork (1989) and Ferderer (1996) were concerned solely with the US.

² See the articles by Burbidge and Harrison (1984), Bjornland (2000) and, more recently, Jimenez-Rodriguez and Sanchez (2005).

³ For example, see the introduction to the article by Jimenez-Rodriguez and Sanchez (2005).

⁴ Mork et al. (1994) provide an exception. For each of seven O.E.C.D. countries, the growth in output is specified as being dependent upon not only increases and decreases in the real price of oil but also the value ratio of energy imports to G.D.P.

⁵ The real oil price is created by dividing the price of Brent crude oil by the US producer price index.

⁶ The simulation results which are produced by Ng and Perron (2001) provide a justification for relying upon the MAIC.

⁷ The formula for the likelihood ratio test statistic features the small-sample modification that was recommended by Sims (1980).

⁸ Scaled oil price increases and decreases are denoted by SOPI and SOPD, respectively.

⁹ In this article, where reference is made to results which are not shown in any of the tables, these may be obtained, on request, from the author.

¹⁰ The earliest date on which data are available on W* Δlog.(ROILP) is 1972Q3. On the basis of the application of either an ADF or a DFGLS test, the inference can be drawn, at the 1% level of significance, that the associated time series is stationary.