RDP 9811: Effective Real Exchange Rates and Irrelevant Nominal Exchange-Rate Regimes 4. New Evidence

Christopher Kent and Rafic Naja

October 1998

We examined the variance of monthly percentage changes in the effective RER across three possible nominal exchange-rate regimes for the period November 1978 to December 1994. We arrived at our preferred set of results in a number of stages. In the first stage we derived the pooled results for the full sample of countries. In the second stage we dealt with the problem associated with observations close to regime transitions. We elected to remove three monthly observations on either side of regime switches and maintained this refinement thereafter. In further stages we eliminated countries from the sample until we were left with a preferred sub-sample of countries with low and stable inflation and stable growth experiences; at this final stage we derived both pooled and within-country results.

Stage 1: Full Sample

The results for the full sample of countries using the pooled analysis are shown in row (1) of Table 2. The pooled variance of effective RER changes was substantially higher under the floating exchange-rate regime than the fixed and managed exchange-rate regimes (389 versus 113 and 86 respectively). However, the variance under all regimes appears to be high, in part because of large and infrequent changes in nominal exchange rates which were associated with switches in nominal exchange-rate regimes.

Table 2: Variances of Effective RERs Across Nominal Exchange-rate Regimes
Stage   Fixed Regime Managed Regime Floating Regime
Variance Variance Variance
1 Full sample of 90 countries
– no correction for regime switching 		113
(7,874)		 86
(6,121)		 389
(2,477)
2 Full sample of 90 countries
– with correction for regime switching 		46
(7,634)		 76
(5,838)		 32
(2,272)
3a Sample of 45 high inflation countries
– inflation more than 10 per cent per annum
– with correction for regime switching 		77
(3,521)		 159
(2,752)		 50
(1,426)
3b Sample of 45 low inflation countries
– inflation less than 10 per cent per annum
– with correction for regime switching 		20
(4,106)		 2.4
(3,083)		 3.4
(828)
4 Sample of 27 low and stable inflation and stable growth rate countries
– inflation less than 10 per cent per annum
– standard deviation of annual growth less than 4.5 per cent
– standard deviation of monthly inflation less than 1.5 per cent
– with correction for regime switching 		1.7
(1,207)
 		 2.2
(2,840)
 		 3.3
(815)
 
 
 

Notes: Figures in parentheses are the number of observations that pertain to each pooled variance.

Stage 2: Regime Switching

Regime switches often involve large movements in both the nominal exchange rate and the RER. These transitions were problematic because it was not clear whether these movements were associated with the new or the old regime (or simply because the regime was in transition). Indeed, a large depreciation may take place at the onset of a flexible regime but may in fact be a correction of a significant overvaluation that occurred during the fixed exchange-rate regime. We controlled for this problem by deleting the last three months at the end of an old regime and the first three months at the beginning of a new regime. Row (2) of Table 2 shows the pooled results with this correction. Compared with the results in the first row, the short-term volatility of the effective RER was substantially lower for all regimes, especially for the floating exchange-rate regime (which is consistent with the example of regime switching mentioned above). All of the results that follow in the paper are based on this correction for regime switching.[16]

Stage 3: High and Low Inflation Countries

We divided the sample of countries between those with high inflation and those with low inflation; we used a cut-off rate of 10 per cent per annum averaged over the sample.[17] This helped to account for systematic differences across country types that might be relevant to the behaviour of the RER but not necessarily related to the nominal exchange-rate regime.[18] The results shown in rows (3a) and (3b) of Table 2 confirm our suspicion that high inflation countries had much greater RER volatility than low inflation countries. We concentrated the rest of our analysis on the set of low inflation countries.

For the set of low inflation countries, the managed and floating exchange-rate regimes displayed significantly lower RER volatility than the fixed exchange-rate regime (2.4 and 3.4 compared with 20). This apparently perverse result may reflect the fact that many countries that experienced the fixed exchange-rate regime had low average inflation but highly variable inflation and highly variable growth rates, which should have increased the volatility of the RER for these countries.[19]

Stage 4: Countries with Low and Stable Inflation and Stable Growth Rates

Figure 3 shows the standard deviation of inflation and growth rates for the set of low inflation countries. We chose to exclude from our sample of low inflation countries those countries with either a standard deviation of growth greater than 4.5 per cent or a standard deviation of inflation greater than 1.5 per cent (that is, those countries shown in the upper and right quadrants of Figure 3).

Figure 3: Variability of Inflation and Growth for Low Inflation Countries
Figure 3: Variability of Inflation and Growth for Low Inflation Countries

Notes: These calculations were based on monthly inflation and annual growth of real GDP. Real GDP data was from the Summers and Heston database in the Mark 5.6 version of the Penn World Tables.

The results of the effective RER volatility for 27 low and stable inflation and stable growth countries are shown in row (4) of Table 2. The ratio of the pooled variance of percentage changes in the effective RER during the floating exchange-rate regime compared with the fixed exchange-rate regime was 1.9 to 1. This difference was significant at the 5 per cent level. The short-term volatility of the effective RER during the managed float regime lay in between the other regimes (and was significantly different from both of these regimes).

The pooled results suggest that the effective RER displays greater volatility under more flexible nominal exchange-rate regimes, consistent with earlier findings based on bilateral RERs. However, the within country results in Table 3 show that there was no systematic difference in the short-term volatility of the effective RER across nominal exchange-rate regimes. Within the sample of 27 countries with low and stable inflation and stable growth rates, there were 17 countries which experienced more than one regime over the sample period. Of these countries, only five had significantly more volatile effective RERs under more flexible exchange-rate regimes. Two countries displayed significantly lower RER volatility under more flexible regimes. Ten countries displayed no significant difference in the short-term volatility of the effective RER.

Table 3: F-Tests for Differences in Variances of Effective RERs Across
Nominal Exchange-rate Regimes(a)
  Country Fixed regime Managed regime Floating regime F-Test(b) Direction(c)
Variance Variance Variance
1 Australia   1.60 6.48 4.04* +
2 Canada   1.14 0.95 1.19 0
3 Finland 1.26   3.42 2.72* +
4 India   4.94 1.08 4.58* –
5 Italy   0.66 2.72 4.15* +
6 Japan   6.74 4.95 1.36 0
7 Korea 1.15 1.61   1.40 0
8 Morocco 1.54 2.38   1.55* +
9 New Zealand(d)   1.46 1.10 1.10 0
10 Norway 0.92   0.51 1.80* –
11 Pakistan 2.94 2.91   1.01 0
12 Singapore 0.66 0.68   1.03 0
13 Spain   1.71 1.01 1.69 0
14 Sweden 1.89   3.65 1.94* +
15 Thailand 1.85 2.42   1.31 0
16 United Kingdom   3.68 3.68 1.00 0
17 United States   2.37 1.79 1.33 0

Notes: (a) Variances were based on monthly percentage changes in the effective RER.
(b) * indicates significance at the 5 per cent level.
(c) A positive (negative) sign indicates that the variance is significantly higher (lower) for the Bretton Woods regime. A zero indicates no significant difference in the variance across regimes.
(d) For comparative purposes, the quarterly percentage changes for NZ were converted to the equivalent monthly changes.

Footnotes

Although we do not consider it in this paper, the behaviour of the RER at the time of regime switches is often dramatic and is one way in which the nominal exchange-rate regime may in fact have important implications for the RER and the economy in general. [16]

Inflation rates were calculated as monthly changes in the CPI (or the WPI when the CPI was unavailable) from the IFS. [17]

This is also a useful distinction because the regime classification scheme is likely to be more accurate for low inflation and more developed economies (as mentioned in Section 3). [18]

This volatility must reflect greater instability both in terms of external shocks, such as the terms of trade, and internal shocks such as changes in domestic policy settings. Most of these countries are members of the CFA African Franc Zone, which experienced a very substantial real and nominal devaluation in January 1994 (Savvides 1996). [19]