RDP 9811: Effective Real Exchange Rates and Irrelevant Nominal Exchange-Rate Regimes 5. Concluding Remarks

In this paper we have re-examined the relationship between nominal exchange-rate regimes and the short-term volatility of the real exchange rate (RER). Our paper has incorporated two innovations. First, we focused our attention on the effective RER (as opposed to the bilateral RER used in previous studies of this type). We presented results for a set of industrialised countries, using the Bretton Woods and post-Bretton Woods eras to distinguish between fixed and floating exchange-rate regimes respectively. Like earlier studies, the variance of monthly percentage changes in the effective RER was significantly higher under the more flexible exchange-rate regime of the post-Bretton Woods period. However, whereas the short-term volatility of bilateral RERs increased about 12 fold post-Bretton Woods, for effective RERs the increase was just less than three fold (based on results pooled across all countries). Using within country analysis, for bilateral RERs this increase was of the same order of magnitude and statistically significant for all countries, however, in the case of effective RERs, there was no significant increase in volatility for three of the 14 countries in the sample.

The second innovation of our paper was to examine results based on a larger sample of countries using more recent monthly data on nominal exchange-rate regimes compiled from IMF records. Three broad regime classifications were identified – fixed, managed and floating. This data set had the advantage of significant variation in the timing of regime switches across different countries.

We focused our efforts on a group of 27 countries which experienced low and stable inflation and stable growth over the sample period (1978 to 1994). The results of pooling observations across countries, within regimes, implied that the short-term volatility of the effective RER under floating exchange-rate regimes was almost double the volatility under fixed exchange-rate regimes. This difference was statistically significant. However, the results from the within country analysis suggest that the relationship, if any, between the volatility of the effective RER and the nominal exchange-rate regime is quite weak. The within-country results are more appropriate than the pooled analysis if the behaviour of the RER is influenced by characteristics which vary across countries (although we attempted to control for some of these differences by restricting our sample to countries with common inflation and growth experiences).

There were 17 countries with low and stable inflation and stable growth rates, which had experienced more than one regime since 1978. Of these, only five displayed significantly greater short-term volatility of the effective RER under more flexible exchange-rate regimes. Twelve countries displayed no significant increase in the short-term volatility of the effective RER under more flexible regimes. Two of these countries actually displayed significantly lower volatility under more flexible regimes.

Our finding is not inconsistent with earlier findings which show that bilateral RERs are more volatile under floating exchange-rate regimes. First, a country's effective RER typically has lower variance than most of its component bilateral RERs. Second, it seems plausible that the nature of the fixed exchange-rate regime of the Bretton Woods system may differ from that of a fixed exchange-rate regime post-Bretton Woods. The variance of the effective RER under fixed exchange rate arrangements may depend on the extent to which major trading partners are also in a fixed exchange-rate regime. In other words, the fixed exchange-rate regime during Bretton Woods is quite likely to result in lower variance of the effective RER than the fixed exchange-rate regime post-Bretton Woods.[20] However, as our data set stands, there are almost no countries which are in the fixed exchange-rate regime both during and after Bretton Woods and are fixed to a single currency. The few that we can clearly identify as remaining fixed post-Bretton Woods also simultaneously went from fixed to the US dollar to fixed to a basket of currencies – and experienced a fall in their effective RER volatility post-Bretton Woods.[21]

One potential bias in our estimates of RER volatility comes about because we have excluded observations near to regime transitions. If anything, we feel that we have underestimated the volatility of the RER under the fixed exchange-rate regime because most of the changes in RERs at the time of regime transitions are probably due to fixed exchange rates (we leave this to further research).

Higher volatility of bilateral RERs under more flexible exchange-rate regimes has been explained in the past as being due to higher volatility of nominal exchange rates, combined with sluggish adjustment of prices. Our result does not discount this explanation – it may be that volatility across a range of bilateral nominal exchange rates is averaged out considerably when constructing effective real exchange rates. However, there is an alternative interpretation of our findings. It may be that countries choose nominal exchange-rate regimes in order to minimise RER volatility; that is, the choice of nominal exchange-rate regimes may be endogenous.[22] If this is the case, it may be difficult to find evidence of a systematic relationship between the nominal exchange-rate regime and effective RER volatility across all countries, even though the nominal exchange-rate regime may influence the volatility of the effective RER.


Our results are consistent with this idea, although the comparison must necessarily be across a different set of countries over the two periods. The variance of monthly changes in the effective RER was 0.98 for the pooled sample of 16 industrialised countries during Bretton Woods. The pooled result for fixed exchange-rate regimes after 1978, based on a set of countries with low and stable inflation and stable growth rates, was 1.71. [20]

To investigate this issue more carefully would require an extension of our regime classification back to at least 1973. [21]

For a discussion of this topic see Edwards (1996). [22]