Nominal and real effective exchange rates for Southern African Development Community countries over the period 1980–2004: implications for the expansion of the Common Monetary Area

Abstract

This paper computes both the nominal effective exchange rate (NER) and the real effective exchange rate (RER) for Southern African Development Community (SADC) countries for the period 1980–2004 with a view to identifying those countries that could join the Common Monetary Area (CMA) in the future. The NER and RER variations are used to analyse the suitability of non-CMA SADC countries for membership in the CMA. Variation in the NER suggests an expansion of the CMA to include Botswana and Malawi, while variation in the RER suggests an expansion of the CMA to Mauritius, Botswana and Seychelles. As suggested in the theory developed by Melitz, while the RER variation criterion should be more expansionist than variation in the NER, the RER-based expansion – as this study shows – does not necessarily include all countries suggested by the NER-based expansion.

Keywords:

• Nominal effective exchange rate
• real effective exchange rate, covariance
• common monetary area
• Southern African Development Community

1. INTRODUCTION

The advocates of globalisation believe that the world economic integration should take place through a series of intermediate steps of regional economic integration. In line with this philosophy, Masson and Patillo (2004:12) suggest that achieving a single currency for Africa can be promoted through successful selective expansion of existing currency areas within African regional economic communities.

Two groups of main African regional economic communities can be distinguished on the basis of the existence of currency areas. The first group comprises communities with currency areas: the Economic Community of West African States (ECOWAS), the Economic Community of Central Africa States (ECCAS) and the Southern African Development Communities (SADC). The ECOWAS and the ECCAS have as currency areas the West African Economic and Monetary Union and the Central African Economic and Monetary Community, respectively. These two currency areas constitute the Communauté Financière de l'Afrique (CFA) zone, with each area having, however, a separate central bank – the Central Bank of the West African States and the Bank of Central African States, respectively. Each area also has different member states. Some members within the ECOWAS enjoy the CFA zone but the block intends to launch a new currency for all its members – although the launch has been continually postponed since 2005. The SADC, whose member states at the time of this analysis are Angola, Botswana, the Democratic Republic of Congo (DRC), Lesotho, Malawi, Mauritius, Mozambique, Namibia, Seychelles, South Africa, Swaziland, Tanzania and Zimbabwe, has a currency area called the Common Monetary Area (CMA) with Lesotho, Namibia, South Africa and Swaziland as the area's members. In the meantime, SADC membership has recently changed with the entry of Madagascar. Seychelles, which became a SADC member in 1997, pulled out in 2004 but has now rejoined the organisation. The second group of main blocks includes communities without currency areas: the Intergovernmental Authority on Development, the Arab Maghreb Union and the Common Market for Eastern and Southern Africa (COMESA). The COMESA and SADC have, however, formal programmes to launch a currency area in the future. Africa also counts minor communities, called sub-regional communities, such as the East African Economic Community, the Community of Sahel and Sahara States, the Economic Community of the Great Lakes Countries, the Mano River Union and the Southern African Customs Union. Expansion and merging of current and future currency areas could serve as building blocks in achieving the African Union's plan to have a single currency for Africa by 2021.

The success of the CMA expansion depends on many criteria, including mutually beneficial trade, economic development, alleviation of poverty and macroeconomic convergence; that is, narrowing of differences in economic performance over time toward a specified target (The Banker, 2005:5). Moreover, a SADC (2003:11) Technical Report has recommended specifically – for a sustainable community, and therefore room for expansion of the CMA – the convergence of rates of inflation, ratios of budget deficit to Gross Domestic Product, ratios of the net present value of public debt to Gross Domestic Product and the balance and structure of the external account within countries involved.

While the paper acknowledges these important criteria, the variation of the exchange rate has not been used extensively to assess the successful expansion of the CMA. This paper evaluates the possible expansion of the CMA within the SADC using the variations in the nominal effective exchange rate (NER) and the real effective exchange rate (RER) as criteria for expansion. It first examines the variation of the NER between the rest of SADC countries and the CMA countries. This is followed by computing the variation in the RER, with an emphasis on Melitz's (1995) criterion of the trade-weighted covariance of RER for the expansion of a currency area.

In judging whether the CMA can be enlarged, the theory of optimal currency areas can be used. The literature defines a currency area as a group of countries where inter-area currencies maintain a fixed relative value, but vary freely in relation to currencies outside the area (Grubel, 1970:318). The area is considered optimal when this relative constancy of the exchange rate simultaneously achieves internal and external balance for each country in the area. This definition implies that a change in the exchange rate of one country within the area with the rest of the world requires a proportional change of the exchange rate for the remaining countries in the area to keep the inter-area exchange rate fixed. It therefore follows that factors determining the choice of the exchange rate regime, fixed or flexible, also determine the size of the optimal currency area.

The variability in the nominal exchange rate can therefore be used to infer which countries could form an optimal currency area. Knowing that the exchange rate is one of the instruments that can be used to restore internal and external balance, if countries do not have the same frequencies and magnitudes of shocks affecting these balances, it follows that being unable to use the exchange rate to address these shocks indicates which countries could consider forming a currency area.

It should be noted that although joining a currency area has some advantages, such as reduced transaction costs and free movement of goods and people, it has also some disadvantages, such as reduced policy autonomy and ability to adjust to shocks at a national level. An implied fact is therefore that the optimal currency area is not the result of a random process. As the variability of the exchange rate is a key variable in determining an optimal currency area, the literature on this topic is reviewed in relation to this variability. Each of the criteria suggested in the literature (see Holden et al., 1979:329), such as the degree of openness of the economy, factor mobility, diversification, inflation differential and capital mobility in establishing a currency area relate either directly or indirectly to the variability in the exchange rate.

Mundell (1961) links the variability of the exchange rate to the low degree of internal factor mobility. He argues that a country in a monetary area, experiencing a shock that disturbs the desired level of internal and external balances, can achieve the desired balance only through a high degree of factor mobility. Where factor mobility is low, balance can only be achieved by varying the exchange rate. In other words, the degree of variability of the exchange rate is inversely related to the degree of factor mobility. In the Mundellian sense, higher observed variability of the exchange rate would be a sign of an unsustainable monetary area and vice versa.

McKinnon's (1963) theory of optimum currency areas, on the other hand, links the variability of the exchange rate to the openness of the economy by showing that a more flexible exchange rate regime is not appropriate for an open economy. An economy is considered to be more open depending on the size of the tradable goods sector relative to the non-tradable goods sector. As the theory assumes that most countries are price-takers on international markets when an economy is more open, a change in the exchange rate, say devaluation in response to a shock, will not restore balance. In fact, world competitive forces will quickly push up the domestic prices of tradables towards the world market prices. Therefore, concurrent devaluation and an increase in prices and wages have no effect on internal and external balances. This inflexibility in real wages results in a dilemma, for if an attempt is made to reduce inflation through contractionary monetary policy, only unemployment results. In an economy where the tradable goods sector is relatively smaller, price increases resulting from the devaluation would create less inflation and a nominal devaluation of the exchange rate would be reflected in a real devaluation. As an optimum currency area requires a fixed exchange rate between members, countries whose economies are closed to each other or less integrated would form an unsustainable monetary area. In the sense of McKinnon (1963), the variability of the exchange rate would indicate an unsustainable currency area for closed economies and a sustainable currency area for open economies.

Kenen (1969:51) suggests that variability in the exchange rate is required with lower levels of product diversification. With greater diversification, a sector-specific shock will not affect internal and external balance, which reduces the need for a change in the exchange rate. Undiversified economies, on the other hand, need to retain the use of the exchange rate as an instrument. Kenen further suggests that well-diversified developed countries should adhere to the fixed exchange rate regime, while the less diversified developing countries should adopt a flexible exchange rate (1969:52). We could conclude that if countries are more diversified vis-à-vis each other and less diversified against the rest of the world, the formation of a currency union would be unsustainable in the long run. However, this introduces the impact of asymmetric versus symmetric shocks on the choice of exchange rate regime. If countries experience asymmetric shocks then they are less likely to form a successful currency union.

Later development in the theory of optimum currency areas associates variability of the exchange rate with low degrees of financial integration, high levels of inflation differentials, and low degrees of policy integration (Ishiyama, 1975:357). Although the literature recognises that there is a feedback effect from the exchange rate to inflation, large inflation differentials between would-be members of a currency union would require the use of the exchange rate as an instrument to restore internal and external balance simultaneously. Therefore we conclude that where greater variability in the exchange rate is observed in a country, it would be an unlikely candidate for a currency union.

Recent research has been sceptical about the use of these stand-alone criteria when analysing currency areas (Bayoumi, 1994; Bayoumi & Prasad, 1997). Using a multi-criteria approach, Bayoumi proposes that the size of underlying disturbances, the correlation between these disturbances, the costs of transactions across currencies, factor mobility across regions and the interrelationship between the demands of different goods together determine the desirability of the currency area (1994:539).

Melitz's (1995:283) theory elaborates on this multi-criteria approach by reformulating the theory of optimum currency areas to incorporate real exchange flexibility as opposed to nominal exchange rate flexibility. Asymmetric shocks among candidates to a currency area can then be resolved by adjustments in the terms of trade and not only by changes in the nominal exchange rate. For example, an event that reduces production in the wine industry of a country in a currency area will require that country to alter the terms of trade (relative prices) with all countries with which it trades wine (inside and/or outside the currency area). Sticky wages and prices, however, do not allow instantaneous adjustments that can be caused by changes in the exchange rate. The extent of these costs depends on the number and the magnitude of adjustments needed, which vary from partner to partner depending on the weight of trade. Therefore, equi-proportionate changes in the terms of trade for each union member with its trade partner determine the costs of joining a monetary union. Melitz (1995:285) measures these changes as the trade-weighted covariance of the real exchange rate. Greater fluctuations in the terms of trade imply a higher covariance that would deter countries from forming a currency union.

Hence, the present paper analyses the suitability of each non-CMA SADC country for joining the CMA. To this end it uses variability of the NER (the NER approach) and variability of the RER (Melitz's [1995] approach). Variability of the NER and RER for each member relative to the CMA variability of the same series is compared for a period of 24 years – 1980–2004. The study uses the statistics of variation (range, standard deviation and covariance), graphical analysis and inferential analysis (cointegration analysis) to rank non-CMA SADC countries in order of suitability for CMA expansion. While achieving the paper's objective of assessing the suitability of each candidate for joining the CMA, the methods used allow comparison of the results from the NER and Melitz's (1995) approach. This paper therefore compares the expansion of the CMA within the SADC using the results from traditional variability of the NER (the NER approach) with results from Melitz's (1995) variability of the RER.

The remainder of the paper is organised as follows. Section 2 discusses the data collection and nominal and real effective exchange rate calculations, Section 3 presents and discusses the results, and Section 4 concludes.

2. DATA COLLECTION AND CALCULATION OF NER AND RER

The nominal and real effective exchange rates are given, respectively, by:

where NER _j is the nominal effective exchange rate for country j as the number of domestic currency units per foreign currency, RER _j is the real effective exchange rate for country j, w _i is the trade weight with country i, p _i is the price level in country i, p _j is the price level of the domestic (SADC) country j, w _i is the trade weight with country i, and e _i is the nominal exchange rate of country j against trading partner i. This exchange rate is not reported in the International Monetary Fund's International financial statistics (IMF, 1980–2004a). Each country's bilateral exchange rate is defined against the US dollar. On the basis of parity with the US dollar, each SADC country's bilateral exchange rate index with each trading partner was calculated. This exchange rate index was later used to calculate the NER and the RER indices.

Quarterly data for the period 1980–2004 were used in the calculations. Exchange rates and price levels for each SADC country were collected from the IMF (1980–2004a) International Financial Statistics. For each SADC country, the top four trading partners were obtained by ranking them according to the total value of trade (i.e. the sum of the value of import and exports). These data were collected from the IMF (1980–2004b) Direction of Trade Statistics. The base year for calculations of all indices is 1990.

The logarithms of NER and RER were plotted and compared with the CMA's NER and RER. The NER and RER indices were transformed into logarithms to overcome the scaling problem in the graphical analysis of the paper. Using the Melitz criterion (1995:284), the incremental covariance for each country was computed and countries ranked according to increasing incremental costs. (See Appendix 2 for more information on how the covariance of the RER for countries making up a currency area is calculated.) The criterion suggests that countries with the lowest covariance with the CMA are the prime candidates for joining the CMA. In fact, existing CMA members would accept new members with the lowest costs to keep the area's costs down.

3. RESULTS AND DISCUSSION

The multi-criteria modern approach to expansion of the monetary union using the RER needs to be compared with the traditional approach of a relatively fixed NER.

3.1 The nominal exchange rate

According to the traditional view, the expansion of the CMA should include SADC countries that have maintained their nominal exchange rates relatively fixed in relation to those of the CMA over the period 1980–2004. We determine this relationship by graphically examining the variation of the series and computing statistics of variation. Figure 1 presents the logarithm of the NER index for each SADC country against the CMA's NER calculated as the weighted average of NER of member states. In this figure, the logarithm of the NER of each country is denoted log(NER). For example, log(NER) Ang denotes the logarithm of the NER index for Angola. Figure 1a–c indicates that the NER movements for Angola, DRC, Malawi, Mozambique, Tanzania, Zambia and Zimbabwe depart substantially from the CMA's NER. The variation in the NER for Botswana, Mauritius and Seychelles, on the other hand, are relatively closer to the variation in the NER for the CMA (Figure 1d). The more closely exchange rates move together, the greater the likelihood that imbalances between countries are similar between them. Therefore countries where the NER variation is very different from the CMA's would not succeed in an expanded monetary area.

Figure 1: Comparison of the NER between CMA countries and each non-CMA SADC country

Statistics of variation, specifically the covariance, have been central to recent theoretical developments in determining the order in which countries could join a currency area. The covariance is defined as the sum of the product of deviations around the mean of two series. The relative range and standard deviations for each country vis-à-vis the CMA are also computed. The covariance is likely to be high when the percentage of variation (standard deviation) of a given country relative to that of the CMA is high. Statistics of variation, namely the relative range and standard deviation, are presented in Table 1.

Table 1: Comparison of relative range and standard deviations

NER range	%	NER standard deviation	%
CMA = 0.418	100	CMA = 0.073	100.0
Seychelles	53	Botswana	98.0
Botswana	83	Seychelles	137.7
Mauritius	133	Zambia	267.0
Malawi	460	Mauritius	349.0
Tanzania	485	Angola	382.0
Mozambique	584	Malawi	383.0
Zambia	904	Zimbabwe	384.9
Zimbabwe	958	Tanzania	418.9
Angola	2238	DRC	544.0
DRC	3342	Mozambique	815.0
Source: Authors' calculations on basis of the NER indices.

As can be seen in Table 1, there is a certain consistency in the rank order that the range and standard deviations of the NER relative to the CMA suggest. Using the variation in the NER for the CMA as the basis of comparison, Seychelles, Botswana and Mauritius are shown to be the better candidates as variation in their NERs is not greatly different from the variation in the CMA's NER, even though the relative standard deviation for Mauritius is still high (349 per cent). These results tend to be consistent with the graphical analysis suggesting that Botswana and Seychelles would be the best candidates for union with the CMA.

As the statistics of variation cannot provide inferential evidence of the extent to which the NER for each non-CMA SADC country and the NER for CMA have relatively fixed value over time, cointegration analysis was performed. All of the NER series were found to be trended and integrated of order one, as can be seen in Appendix 1.

Table 2 shows at the 5 per cent level of significance that Botswana and Malawi would be the more likely candidates for union with the CMA. Using various tests, Botswana consistently emerges as the best candidate for union. This finding is not surprising since Botswana's policy regime is to peg their currency, the pula, to a basket of currencies that reflects the country's trading patterns in order to stabilise the trade weighted real exchange rate. Since South Africa is a major trading partner for Botswana and the South African rand trades one-to-one with the CMA currencies, the variation of Botswana's exchange rate is expected to follow that of the CMA. Therefore, the emergence of Botswana as the most suitable for joining the CMA might rather reflect the effectiveness of its currency policy regime. Until an analysis considering other criteria for a successful currency area formation is carried out, the variation of the exchange rate is not conclusive on Botswana's suitability for the CMA.

Table 2: Cointegration of the CMA NER and non-CMA countries' NER

Countries	Test statistics	Critical values
		1%	5%	10%	Cointegration
Angola	–1.949	–0.2600	–1.950	–1.610	*
Botswana	–3.363	–3.511	–2.891	–2.580	**
DRC	–1.757	–0.2600	–1.950	–1.610	*
Malawi	–3.155	–3.511	–2.891	–2.580	**
Mauritius	–1.414	–0.2600	–1.950	–1.610	****
Mozambique	–1.488	–0.2600	–1.950	–1.610	****
Seychelles	–1.044	–0.2600	–1.950	–1.610	****
Tanzania	–1.275	–0.2600	–1.950	–1.610	****
Zambia	–1.635	–0.2600	–1.950	–1.610	*
Zimbabwe	–0.649	–0.2600	–1.950	–1.610	****
*Cointegration at 10% significance level only. **Cointegration at 5% and 10% significance levels. ***Cointegration at 1%, 5% and 10% significance levels. ****No cointegration at all.
Source: Authors' calculation from NER indices.

3.2 The real exchange rate

Melitz (1995) has shown that movements in the NER are no longer the crucial criteria for fitness for a currency area. Hence Melitz's theory (1995) incorporates changes in both the NER and relative prices (terms of trade) or, in other words, changes in the RER as criteria for a currency area. Greater fluctuations in the RER make it difficult to agree on a common NER trend as they may be costly for joining and existing members. The extent of differences in these fluctuations between countries is measured by the covariance of RER as per Melitz's theory. This section analyses the expansion of the CMA on the basis of this covariance.

For the sake of consistency with previous NER analysis, the paper first presents fluctuations of non-CMA SADC members' RER around the CMA's RER, using graphical analysis and statistics of variation. Because relative prices – components of the RER – may reflect asymmetric shocks, the paper presents fluctuations of the RER and relative prices separately in order to see whether asymmetric shocks are reflected in the NER.

Figure 2 shows fluctuations of the RER and relative prices for each non-CMA country relative to those of the CMA. As can be seen in Figure 2a,b, Botswana's RER and relative price series seem to move closer to the CMA's. Mauritius's and Malawi's RERs (Figure 2c) are the next closest. Generally, it is shown that countries diverge more from the CMA on the basis of their relative price movements than on RER movements, which indicates that non-CMA countries have experienced severe asymmetric shocks that have not been reflected in changes in their NER.

Figure 2: Real exchange rate and relative price movements over the period 1980–2004

Using the range and standard deviation, Table 3 presents movements in the RER and relative prices of each non-CMA country against the variation of the RER and relative prices for the CMA. The range or standard deviation of non-CMA countries' RER and relative prices is measured as a percentage of the range or standard deviation of the CMA's RER and relative price. In terms of range, Botswana, Seychelles, Zambia and Mauritius are found to be closest to the CMA. However, there are marked differences in the direction of price movements for Mauritius and Seychelles, with relative ranges for these countries of –66 per cent and –117 per cent, respectively.

Table 3: Comparison of the range and standard deviations of RER and relative prices for the CMA and each non-CMA country

Range				Standard deviation
RER	%	Relative prices	%	RER	%	Relative prices	%
CMA = 0.036	100	CMA = 0.380	100	CMA = 0.073	100	CMA = 0.119	100
Botswana	542.7	Botswana	35.3	Botswana	98.2	Botswana	34.6
Seychelles	614.9	Mauritius	–66.0	Seychelles	138.7	Mauritius	81.7
Zambia	1019.2	Seychelles	–117.9	Zambia	267.0	Seychelles	135.4
Mozambique	1106*	Mozambique	–149.3	Mauritius	348.9	Malawi	339.5
Tanzania	1319	Malawi	242.1	Angola	382	Mozambique	412.3
Mauritius	2137	Tanzania	409.1	Malawi	383.2	Tanzania	445.3
Malawi	2739	Zimbabwe	788.6	Zimbabwe	384.9	Zimbabwe	642.1
Zimbabwe	2746	Zambia	896.8	Tanzania	418.9	Zambia	1050.8
DRC	3143.8	Angola	2196.3	DRC	544.7	Angola	2833.5
Angola	1103.2	DRC	3374.7	Mozambique	815.2	DRC	4232.4
*Mozambique series available only for the period 1990–2004.
Source: Authors' calculations on the basis of the indices.

In terms of standard deviations, Botswana, Seychelles and Mauritius are closest in terms of RER and relative price movements: relative prices in Botswana are 65.4 percentage points lower than the CMA, while Seychelles is 35.4 percentage points higher. As can be seen, RER-based suitability differs from the NER-based one.

Now, using Melitz's (1995) covariance of the RER, the paper analyses the covariance (cost) of joining the CMA. The covariance of the RER for a particular country is calculated between that country's RER and the average of its trading partners' RER. It is calculated using the formula , where X is the NER of a particular country over the period 1980–2004 and is its average, and Y is the weighted average of its trading partners' NER over the period 1980–2004 and is its average. Despite Melitz's (1995) distinction between intra-area and external covariance of the RER, Table 4 presents the overall covariance for each country because of the very low level of intra-SADC trade. In fact, only a few SADC countries – Botswana, Malawi and Seychelles – have South Africa as their main trading partner with insignificant respective intra-trade covariance. Considering the CMA as one country, Table 4 shows that Mauritius, Seychelles and Botswana have lower covariance of the RER than the CMA. Their respective covariances of the RER are 0.003, 0.005 and 0.083, while the CMA's covariance of the RER is 0.085. If each country individually joined the CMA, incremental covariance (i.e. additional covariance due to new entry to the CMA) would be the joining country's own covariance. It can be understood that a country with an incremental covariance lower than that of the CMA would be the most welcome by existing members. The entrant country would, however, lose, as it would be subjected to a higher covariance of the RER (the average covariance of the RER for the area, i.e. total covariance of members states divided by their number) than its own. The opposite would happen for countries with RER covariance greater than the CMA, such as Zimbabwe, Tanzania and others in Table 4.

Table 4: Incremental and average covariance (cost) to joining the CMA

Real exchange rate			Relative prices
Country	Incremental	Average	Country	Incremental	Average
Lesotho	0.096	0.096	Lesotho	0.096	0.096
Namibia	0.114	0.105	Namibia	0.098	0.097
South Africa	0.025	0.078	South Africa	0.025	0.073
Swaziland	0.108	0.085	Swaziland	0.102	0.080
CMA	0.343	0.085	CMA	0.401	0.080
Mauritius	0.003	0.069	Mauritius	0.003	0.067
Seychelles	0.005	0.058	Seychelles	0.006	0.059
Botswana	0.083	0.062	Botswana	0.083	0.062
Zimbabwe	0.087	0.065	Zimbabwe	0.087	0.064
Tanzania	0.091	0.068	Tanzania	0.091	0.067
Malawi	0.135	0.074	Malawi	0.114	0.071
Zambia	0.137	0.080	Zambia	0.137	0.077
Mozambique	0.314	0.099	Mozambique	0.283	0.093
Angola	0.314	0.115	Angola	0.314	0.109
DRC	0.414	0.136	DRC	0.414	0.129
Source: Authors' calculations from the RER and relative price indices.

A point that arises from the above discussion is that countries that lower the CMA's covariance would be reluctant to join, as they would incur relatively greater costs, and those that raise the CMA's covariance (costs) would be unacceptable to existing members if they joined. This analysis leads to a dilemma; namely, those countries acceptable to the CMA in the event of enlargement will be unwilling partners, and those countries wishing to join will be undesirable to the CMA. This dilemma could be resolved through a multi-criteria consideration. In fact, there are many other issues under consideration when joining a monetary area, and the costs and benefits of a monetary union are weighed against each other both for existing and entrant countries. The costs of joining the area, mainly the inability to use an exchange rate policy in case of shocks, should be weighed against other factors such as beneficial mutual trade, development policies and long-term macroeconomic convergence. Table 4 shows that countries with lower covariance of the RER also have lower covariance of the relative prices. Although previous graphical analysis of diverging relative prices and RER suggested that asymmetric shock was in general not reflected in the NER, the covariance of the RER and relative prices seem to suggest greater suitability with less divergence of these covariances in general. This finding suggests that applying Melitz's (1995:284) criterion of RER to currency areas is more expansionist than the NER criteria.

4. CONCLUSION

This paper first calculated the NER and RER for all SADC countries. Although the ratio of the prices of tradables to non-tradables as the definition of the RER is the most appropriate approach for developing countries, data deficiencies prompted the use of the purchasing power parity approach instead. This approach allowed us to compute the NER by summing the weighted bilateral exchange rate indices, based on parity with the US dollar, and the RER by the product of the NER with the ratio of weighted price indices of the main trading partners and the domestic price index for the country. The study ranked countries according to their declining degree of suitability for union with the CMA on the basis of both the NER and the RER. According to the variation in the NER, the order of suitability was: Botswana, Seychelles, Zambia, Mauritius, Angola, Malawi, Zimbabwe, Tanzania, DRC and Mozambique. However, only for Botswana and Malawi did the NER move together with the CMA's NER. Countries were also ranked according to the fluctuations in the RER; in other words, according to the trade-weighted covariance of the RER. Since the RER can change as a result of the nominal exchange rate and the relative prices, the ranking procedure on the basis of this criterion considered two options that did not give different results: holding constant the NER and allowing both the NER and relative prices to vary. The order of suitability was as follows: Mauritius, Seychelles, Botswana, Zimbabwe, Tanzania, Malawi, Zambia, Mozambique, Angola and DRC. However, only for Malawi, Mauritius, Zambia and Seychelles did the RER or relative prices move together with those of the CMA. The research shows that, applying Melitz's (1995:284) covariance of the RER criterion, the real exchange rate appears to be more expansionist than the NER when selecting countries into the CMA.

A word of caution is in order, however, regarding the conclusion of the study. It should be noted that the results obtained are suggestive but not conclusive, given that the various criteria work simultaneously when determining the sustainability of currency area expansion. Analysis of covariance of RER for joining the CMA did lead to a dilemma that may need to be resolved by applying other criteria. The criterion cited by the literature as crucial to the sustainability of a currency area and economic block is beneficial mutual trade – but as there is very low intra-SADC trade, other criteria may need to be taken into account when deciding suitability, such as similar economic interests, policy priority, economic structures, macroeconomic shocks and convergence of the joining countries.

Notes

APPENDIX 1

Table A1: Order of integration of series used in the analysis

Countries	NER	Relative prices	RER
CMA	I(1)	I(1)	I(1)
Angola	I(1)	I(2)	I(1)
Botswana	I(1)	I(1)	I(1)
DRC	I(1)	I(2)	I(2)
Malawi	I(1)	I(1)	I(1)
Mauritius	I(1)	I(1)	I(1)
Mozambique	I(1)	I(2)	I(2)
Seychelles	I(1)	I(1)	I(1)
Tanzania	I(1)	I(1)	I(1)
Zambia	I(1)	I(1)	I(1)
Zimbabwe	I(1)	I(2)	I(1)
Source: Authors' calculations.

APPENDIX 2

To define the costs of the monetary area, Melitz (1995) determines the optimum currency area as the area that minimises the costs. He considers these costs as depending both on size u and openness x of the country. The size u is a variable that takes values between zero and one inclusively. The value zero corresponds to a situation where no enlargement of the area beyond existing borders is possible, while one corresponds to a situation where a country can be part of the universal monetary area. Melitz (1995) holds openness (value-added of trade in total output) constant and large enough (25 per cent) to formulate the costs of the monetary union as follows:

where cov(1) is the weighted average of expected individual covariance of the relative prices with foreign countries. If a country has n partners, cov(1) will be the weighted average of expected individual covariance of the relative prices.

The members of the union that a country wishes to join are themselves related to the rest of the world so their covariance of relative prices with this world needs to be taken into account. So, the costs c(u, x) depend on the proportion of the existing covariance cov(1) that pertains to fellow members of the union. According to the Melitz's theory (1995), a set of countries form a successful monetary area if they minimise the ratio below:

where cov(u) stands for trade adjustment costs with the union members, cov(1) stands for trade adjustment costs both with union members and outside members, is the weighted average prices in the rest of the union, e is the effective exchange rate, p is the price at home, is the price outside the union, p _u is the price inside the union ( combined with p), p* is the price abroad, and combined, and e _u is the multilateral effective exchange rate of the union currency.

To see whether or not the ratio in Equation (2) above is at a minimum, one must first calculate the covariance of the existing members and see whether adding a new member increases the costs significantly. A member increases the costs of existing members when its own covariance of relative prices is greater than the average covariance of relative prices of the countries it is joining. Joining will increase the ratio in Equation (2) by increasing cov(u), while joining will decrease the ratio in case of an opposite scenario. Of course expansion of the monetary union depends both on entrants and existing members' additional costs.