Is devaluation expansionary or contractionary? Empirical evidence from Fiji

Abstract

Devaluation has been traditionally promoted as an effective tool for increasing exports and improving the external position of the devaluing country if a nominal devaluation results in expenditure switching. In this article, our aim is to model the relationship between currency devaluations and output for Fiji. Following the approach in Bahmani et al. (2002), we extend the traditional model by incorporating other monetary and fiscal policy variables. We achieve our goal by using the recently developed bounds testing approach to cointegration and the autoregressive distributed lag model and find that devaluation is expansionary in the case of Fiji.

Notes

¹ For an analysis of the Fijian fish and garment industries, see Narayan and Prasad (2007).

² For an analysis of the economic importance of Fiji's sugar industry, see Narayan and Prasad (2005).

³ The exception was 1999 when Fiji, with a democratic government after a long period of time, achieved a 9% growth rate.

⁴ While ideally some supply side variables, such as oil prices, should be included in the model, we refrain from doing so for three reasons. First, with 31 observations, including one more variable will create problems related to degrees of freedom; hence, making our elasticities questionable. Second, given that the objective of this article is to examine the impact of devaluation on output and not examine the relevance of demand and supply side factors to output, our model is suffice and is also consistent with the literature. Third, the major supply side factor for Fiji would be sugar prices rather than oil prices given that sugar has been the main economic activity for over a century until been outperformed by tourism over the last decade or so. However, we cannot model sugar prices because the price that Fiji receives is three times higher than the world market price under the European Union trade agreement.

⁵ A number of channels through which devaluation affects aggregate demand and aggregate supply of domestically produced goods are being specified. On the demand front, devaluation can affect output through the relative price effect, real income effects, imported inputs, income distribution effects and investment, etc. Devaluation affects aggregate supply by affecting nominal wages, imported inputs and costs of working capital. See Lizondo and Montiel (1989) for an in depth analysis.

⁶ The model is estimated using the MICROFIT programme which is specifically designed for conducting the bounds test for cointegration and the ARDL estimation of the long-run and short-run elasticities.

⁷ It should be noted that Bahmani-Oskooee and Goswami (2003) have shown that the F-test results are sensitive to the number of lags. Following them, we experiment with lag lengths ranging from 1 to 6. Our main finding was that while the calculated F-statistic differed at each lag length, we were able to reject the null of no cointegration for all lag lengths. The results are available from the authors upon request.

⁸ Over the period 1988 to 2000, one marked by devaluations, total exports for Fiji grew at 6.5% per annum while GDP grew at around 2.4% per annum.

⁹ All variables in our model are integrated of order one. The unit root tests are not reported here in order to conserve space. The tests are, however, available from the authors upon request.