Oil Prices and Real Exchange Rates in Oil-Exporting Countries: A Bounds Testing Approach
Jahan-Parvar, Mohammad R., Mohammadi, Hassan, The Journal of Developing Areas
We test the validity of the Dutch disease hypothesis by examining the relationship between real oil prices and real exchange rates in a sample of fourteen oil exporting countries. Autoregressive distributed lag (ARDL) bounds tests of cointegration support the existence of a stable relationship between real exchange rates and real oil prices in all countries, suggesting a strong support for the Dutch disease hypothesis. As for short-run, there are evidence of causality from oil prices to exchange rates in four countries; from exchange rates to oil prices in two countries; and bidirectional relations in another four countries. There is no evidence of causality in the remaining four countries.
JEL Classification: C32, C52, F31, F37, F47.
Keywords: Oil prices, Real exchange rates, Dutch disease, Cointegration, Autoregressive distributed lags.
(ProQuest: ... denotes formulae omitted.)
The sharp increase in oil prices over the past decade has renewed interest in the "Dutch disease" hypothesis. According to the hypothesis, the inflow of oil windfalls into an oil exporting country may cause appreciation of the real exchange rate, reduce its competitiveness in the non-oil exporting sector, and limit its ability to build a diversified exports base. The culprit for the disease is the "spending effect". More specifically, higher oil income may increase the demand for non-traded goods, and increase their prices relative to those of traded goods. This appreciation of the real exchange rate will reallocate resources from the non-oil traded sector into the non-traded sector, contracting the former to the extent that it is exposed to international competition. Early literature on the subject includes Dornbusch (1973), Gregory (1976), Forsyth and Kay (1980), Corden (1984), Corden and Neary (1982), Buiter and Purvis (1982), Bruno and Sachs (1982), Eastwood and Venables (1982), Enders and Herberg (1983), Edwards and Aoki (1983), Edwards (1986), van Wijnbergen (1984). More recent studies include Gylfason (2001), Torvik (2001), and Stevens (2003).1
Despite a great deal of theoretical work on mechanisms of Dutch disease, formal empirical work on the subject has received limited attention. Furthermore, available evidence is not conclusive. Edwards (1984) finds that exogenous shocks to world price of coffee have monetary and inflationary effects on the Colombian economy. Taylor et al. (1986) finds a negative relation between Nigerian agricultural exports and its oil export revenues. Warr (1986) concludes that higher oil revenues enabled the Indonesian government to defer the much-needed currency devaluation in the 1970s, and were the primary source of subsequent financial problems. Brunstad and Dyrstad (1997) find significant demand and cost-of-living effects following the intensive build up period of the Norwegian petroleum sector, suggesting that Norwegian petroleum sector has been the culprit for the country's weak manufacturing performance. In contrast, Bjorland (1998) finds evidence of weak response in UK's manufacturing but positive and significant response in Norwegian manufacturing in response to oil and gas sector shocks. Hutchinson (1994) finds that developments in oil and natural gas sectors of the Netherlands, UK, and Norway had no significant effects on manufacturing sectors of these economies, and thus there is no support for the Dutch disease hypothesis. Jahan- Parvar and Mohammadi (2009) study the potential loss of competitiveness due to higher oil prices in a sample of six oil producing countries using a dynamic simultaneous equations method, and find weak evidence for the monetary channel of Dutch disease.
This paper provides a formal test of the Dutch disease hypothesis by examining the possibility of a long-run relationship between real oil prices and real exchange rates in monthly data for a sample of fourteen oil exporting countries. Our empirical results using the "autoregressive distributed lag" (ARDL) model of Pesaran, et al. …