Chapter

Chapter 5: Welfare Effects of Monetary Integration: The Common Monetary Area and Beyond

Author(s):
Joannes Mongardini, Tamon Asonuma, Olivier Basdevant, Alfredo Cuevas, Xavier Debrun, Lars Engstrom, Imelda Flores Vazquez, Vitaliy Kramarenko, Lamin Leigh, Paul Masson, and Genevieve Verdier
Published Date:
April 2013
Share
  • ShareShare
Show Summary Details
Author(s)
Tamon Asonuma, Xavier Debrun and Paul R. Masson 

The formation of currency unions has always been accompanied by intense debates on their costs and benefits for potential members. Even if monetary integration has an important political dimension, it rarely transcends national interest. “No nation has friends, only interests,” Charles de Gaulle, the late president of France, once observed. And in fact, as soon as serious tensions emerge within existing monetary unions, such as the euro area, existential questions about the potential merits of monetary sovereignty resurface. This chapter proposes a model-based assessment of this particular question in the context of the existing Southern African Common Monetary Area (CMA)1 and hypothetical expansions of it.

By nature, the simulation exercises developed in this chapter can only be partial and of a positive nature. Thus, they do not constitute (and cannot be interpreted as) normative assessments underpinning specific policy advice. In particular, there is no pretense that the model-based estimation of “optimal” macroeconomic policies precisely describes what governments and central banks should do (or what they would do under counterfactual benchmarks). Also, a model cannot provide a complete picture of the costs and benefits of monetary integration because, to remain tractable, it must ignore critical dimensions, including the combination of political interests and the varying levels of institutional preparedness across potential members.

Pioneered by Mundell (1961), a vast literature on the costs and benefits of monetary unions mushroomed in the wake of the creation of the euro in 1999. Early studies on optimum currency areas (OCA) emphasized the costs for structurally different countries affected by different shocks of opting for a common monetary policy. Hence, absent fluid labor and capital flows across countries or a fiscal union offering some risk sharing through intraregional transfers, it was generally thought that prospective members of a currency union should first reach a reasonable degree of economic convergence.

The debate on the euro in the 1990s remained dominated by these first-generation OCA arguments, in which the benefits are largely assumed and the costs are directly proportional to the magnitude of cross-country divergences along selected dimensions. This line of thinking provided the theoretical underpinning for convergence criteria with which prospective union members should comply. Yet intense controversy developed around the need to have convergence criteria on public debts and deficits and to turn these criteria into permanent constraints once in the euro area. This heavily polarized issue created demand for formal economic models to investigate the implications of monetary unification for the coordination of the monetary-fiscal policy mix. Unlike OCA-based studies, models by Beetsma and Bovenberg (1998, 1999), among others, revealed explicit benefits of monetary unification with regard to credible monetary policy coordination and better insulation of the common central bank (CCB) from inflationary pressures motivated by public finances considerations.

Debrun, Masson, and Pattillo (2005) developed a model—hereafter DMP—along those lines and calibrated it to assess the net gains from monetary unification in various regions of Africa (see Masson and Pattillo, 2005; and Debrun, Masson, and Pattillo, 2008, 2011). The model allows for a simple and intuitive assessment of the costs and benefits of a given currency union. DMP simulations—which essentially compare the costs of losing monetary sovereignty to the gains associated with greater policy credibility—generally suggest that existing monetary unions in Africa are welfare-enhancing for all or most of their members when compared with the hypothetical counterfactual of a fully flexible exchange rate. The main driver of these credibility gains is that the power of individual governments to extract a higher inflation tax from the regional central bank is diluted proportionally to those governments’ influence on the conduct of the regional monetary policy. As a result, larger currency unions are more likely to be beneficial for all members unless newcomers are relatively large, fiscally undisciplined, and without meaningful trade linkages with the rest of the union. In relative terms, the costs attributable to asymmetric shocks emphasized by the traditional OCA literature—which argues for smaller currency unions composed of countries facing similar terms-of-trade shocks—are small.

In this chapter, a fully updated calibration of DMP is used to assess the net gains arising from the CMA and a number of hypothetical variants selected for their value in illustrating the model’s properties rather than their actual plausibility.2 Specifically, the analysis looks at the welfare impact of (1) being a member of the CMA, (2) expanding the CMA, (3) establishing a regional central bank to conduct monetary policy on the basis of union-wide conditions, and (4) combinations of (2) and (3). For (1), the model suggests that there are significant benefits of being a member of the CMA, particularly for Lesotho and Swaziland. For (2), a larger CMA that includes all current members of the Southern African Development Community (SADC) is desirable for all except the fiscally conserative Mauritius and Tanzania, and possibly Angola, whose terms of trade are very volatile and uncorrelated with its neighbors. For (3), the creation of a genuine CMA-wide monetary union with a regional central bank carries some costs in forgone anti-inflationary credibility because fiscally profligate countries could, despite their small size, extract a higher inflation tax. All members would be better off maintaining the current asymmetric CMA regime in which South Africa sets monetary policy. Finally, creating an SADC-wide currency union under the helm of a regional central bank continues to be beneficial for all except Mauritius, but the gains for existing CMA members—compared with the existing arrangement—are likely to be limited and fall well within plausible margins of error.

By design, the model leaves out a number of relevant issues. These include the political factors that may be essential for making currency unions work,3 the intrinsic benefits from exchange rate stability—less uncertainty promoting investment and intraregional trade—and the creation of a common money market. In addition, the smooth operation of a common currency area also hinges on building institutions (a central bank, a regional financial supervisor, common accounting standards) and on broader coordination of fiscal, structural, and regulatory policies. The scope of the model thus needs to be kept in mind when interpreting the results. Even though the gains are expressed in “welfare” terms, the analysis cannot validate normative statements on its sole merits. For instance, the recent euro area crisis has demonstrated the importance of solidarity and risk sharing among member countries, bringing to the fore fundamental questions about the extent of the loss of sovereignty—and the associated political costs—required for the sustainability of regional monetary unions. The simulations in this chapter could not possibly inform that essential debate. Instead, the model provides one useful benchmark that can be used for further evaluation of monetary union proposals along with more traditional OCA-based assessments and microeconomic analyses of a single currency.

The rest of this chapter is structured as follows. The second section reviews the OCA literature and previous analyses of the CMA experience. The chapter turns to the history and institutional arrangement of the CMA and looks at the state of economic convergence in the third section. The fourth section quantifies the welfare benefits of the CMA and hypothetical enlargements, and is followed by a section that deals with possible challenges of a common monetary union for the CMA member countries.

Literature Review

The traditional optimum currency area (OCA) literature discusses the costs of forfeiting monetary policy autonomy and the corresponding importance of alternative adjustment mechanisms to external imbalances. In his pioneering study, Mundell (1961) emphasized labor mobility as a crucial adjustment mechanism to idiosyncratic shocks and, therefore, a key precondition for forming an OCA. Price and wage flexibility were also seen as important for coping with idiosyncratic demand shocks. Because shocks were more likely to be similar among highly integrated economies, McKinnon (1963) suggested the degree of openness—defined as the ratio of tradable to nontradable goods—as a key indicator in forming an OCA. Completing the trilogy of classic OCA studies, Kenen (1969) introduced product diversification as an element of an OCA, stressing that a region with a highly diversified production base should be better equipped to maintain a currency union than regions with low diversification because the latter were more vulnerable to asymmetric disturbances. In addition, Kenen highlighted fiscal integration among countries as a mitigating factor because of the implied risk sharing.

More recently, the literature extended the basic economic insights from the classical OCA approach to incorporate new dimensions, including the effectiveness and credibility of monetary policy (Beetsma and Bovenberg, 1999), the centrality of shock correlations (Alesina, Barro, and Tenreyro, 2002), and the endogeneity of OCA adequacy.4 Although it is generally understood that a higher correlation of shocks between countries makes monetary union more beneficial, Mélitz (1991) shows that even if countries face identical shocks, they might still need different policy responses given different initial economic positions and country-specific transmission mechanisms.

A number of studies suggest that monetary integration may be self-validating because OCA criteria are endogenous to the creation of a monetary union. Frankel and Rose (1997) argue that openness (degree of integration) and income correlation are linked because the correlation of business cycles across countries depends on trade integration.5Mongelli (2002) qualifies that claim, showing that the endogeneity of OCA criteria depends on the preexisting degree of convergence. In the same vein, De Grauwe and Mongelli (2005) focus on the endogeneity of economic integration, financial integration, symmetry of shocks, and labor market flexibility.6

Debrun, Masson, and Pattillo (2005) integrate traditional arguments against monetary union—the costs of a one-size-fits-all monetary policy in a heterogeneous region deprived of fiscal federalism—with the potential benefits of enhanced policy credibility, by explicitly modeling the substitutability between monetary integration and domestic institutional reforms (see Box 5.1). They establish the relevance of asymmetries in institutional quality and in the credibility of monetary commitments to macroeconomic stability. In contrast to the OCA literature, they emphasize positive “monetary externalities” associated with larger monetary unions attributable to the greater gains from monetary coordination and from a more effective separation of monetary and fiscal powers.

Box 5.1The DMP Model1

The model used in this chapter is a variant of Debrun, Masson, and Pattillo (2005).2 In line with the vast literature on the European Economic and Monetary Union, the model focuses on the impact of institutional changes on the credibility of a commitment to low inflation (see Beetsma and Bovenberg, 1998, 1999; and Martin, 1995). That approach has considerable appeal in countries and groups of countries in which the risks to macroeconomic stability and the need for further credibility-enhancing institutional reforms are deemed to be high.

To explicitly model credibility problems and institutional solutions, the basic architecture of DMP relies on the positive theory of monetary policy proposed by Barro and Gordon (1983) and extended to fiscal policy issues by Alesina and Tabellini (1987). DMP assumes an n-good, n-country economic area that is small compared with the rest of the world. Countries differ along various dimensions: size, economic governance (propensity to wasteful public spending), budget flows, and Phillips-curve shocks, which are interpreted as terms-of-trade disturbances. Because the welfare analysis rests on an explicit characterization of strategic interactions between monetary and fiscal policy-makers, the underlying economic structure is essentially static, including a new-classical Phillips curve augmented with a distortionary tax and a negative externality from competitive devaluations (monetary surprises) in trading partners, and simple period-budget constraints—there is no public debt.

The benchmark case for welfare evaluations is a regime of complete monetary policy autonomy (flexible exchange rates) with politically dependent central banks. Monetary and fiscal policies are determined jointly by minimizing deviations of the effective tax rate, public expenditure, and inflation from specific objectives. Those objectives are nonnegative constants except for inflation, which fluctuates to partly accommodate Phillips-curve shocks. This captures the preferred trade-off between the variability of inflation and that of output. Finally, as in Barro and Gordon (1983), governments also care about the level of output, welcoming expansions and disliking contractions.

As usual in this literature, equilibrium policies systematically deviate from the first best, reflecting the government penchant for using monetary policy to boost activity beyond its potential—instead of raising the potential through politically costly structural reforms—and the inflationary impact of the waste of tax money levied through distortionary instruments. Overall, inflation is too high and productive public spending too low in comparison with the case in which decision makers could credibly precommit.

Monetary unification is modeled as a change of regime by which monetary policy is determined by a regional central bank, whereas fiscal policy remains in national hands. It yields benefits similar to the delegation of monetary policy to an independent central bank. Indeed, a regional monetary policy is less effective for stimulating output in each individual country because there is no expected gain from a depreciation of the national currency with respect to trading partners in the region. Centralized monetary policy thus brings about lower inflation across the board. With regard to welfare, these credibility gains are proportional to the size of the initial bias (reflecting the slope of the Phillips curve, the reluctance to raise distortionary taxes, the appetite for productive public spending, and the amount of wasteful public spending) and to the intensity of intraregional trade linkages. For instance, little trade implies little loss of effectiveness of regional monetary policy in comparison with national policy, and therefore reduced attenuation of the central banks’ incentive to generate excessive inflation. The costs of unification arise from the inadequacy of the regional monetary policy in the face of country-specific shocks, in line with OCA literature.

1 This box was prepared by Xavier Debrun. A more detailed discussion can be found in Debrun, Masson, and Pattillo (2011).2 A description of the model is presented in Appendix 5C.

The CMA has been the subject of an extensive literature. Van Zyl (2003) discusses the history of monetary integration in Southern Africa and prospects for its extension. Wang and others (2007) review recent developments in the CMA, identify main policy challenges for the members, and discuss implications for further economic integration. Some studies focus particularly on individual countries and their involvements in the CMA (Tjirongo, 1995;Lledo, Martijn, and Gons, 2005;Gons, 2006; and Dwight, 2006). Other papers show that the CMA does not meet the traditional criteria for an OCA, particularly given its vulnerability to asymmetric shocks and its degree of labor mobility (Cobham and Robson, 1994; Van der Merwe, 1996; Metzger, 2004; and Masson and Pattillo, 2005).

The CMA: History, Institutional Arrangements, and Economic Convergence

This section reviews the history and institutional features of the CMA. It also looks at the state of economic convergence across member countries.

History and Institutional Arrangements

Even though the CMA arrangement formalizes the regional role of the South African rand as a means of payment, the CMA is not a full-fledged currency union.7 In 1921, after the establishment of the South African Reserve Bank (SARB), the South African currency (initially the pound and since 1961, the rand) effectively became the only medium of exchange and legal tender in South Africa, Bechuanaland (now Botswana), Lesotho, Namibia, and Swaziland. There were no internal restrictions on capital flows within the area, and virtually all external transactions were executed through banks located in South Africa and subject to South African exchange controls. This system was maintained after Botswana, Lesotho, and Swaziland gained independence in the 1960s and was institutionalized on December 5, 1974, with the signing of the Rand Monetary Area (RMA) Agreement. Botswana left the RMA in 1975 in favor of policy independence.8

The RMA was revamped in April 1986 and transformed into the CMA composed of Lesotho, Swaziland, and South Africa. Under the terms of the CMA Agreement, Lesotho and Swaziland would have the right to issue their own national currencies. Swaziland introduced its currency, the lilangeni, in 1974, followed by Lesotho’s loti in 1980. Namibia, which gained independence from South Africa in 1990, formally joined the CMA two years later, and launched the Namibian dollar in 1993. The currencies of Lesotho, Namibia, and Swaziland (the LNS) have been pegged at par to the rand since their introductions, with bilateral agreements governing access to the South African foreign exchange market.

Box 5.2 summarizes key features of the CMA. The long history of monetary association and strong financial and trade linkages explain why the CMA coincides with a customs union enjoying fairly unrestricted capital mobility. This unique characteristic contrasts with two long-standing monetary unions in Africa, the Central African Economic and Monetary Community and the West African Economic and Monetary Union, where some trade restrictions persist and intraregional capital mobility, though in principle free, remains low (Masson and Pattillo, 2001).

Economic Convergence

A prominent characteristic of the CMA is the economic and financial weight of South Africa, which accounts for more than 90 percent of the region’s GDP and trade. Prior to the impact of the global financial crisis on the region, fiscal positions remained fairly benign, on average, although Swaziland posted fairly sizable and increasing deficits (5.6 percent of GDP on average during the period 2008–10, with 13.4 percent, on a commitment basis, during fiscal year 2010/11), while Lesotho’s public debt, at 41.5 percent of GDP, stayed well above the regional average. See Table 5.1.

Table 5.1Countries in the Common Monetary Area (CMA): Selected Economic Indicators, 2008–10(Average, unless otherwise indicated)
CMAMemorandum
LesothoNamibiaSouth AfricaSwazilandCMA TotalBotswana
Nominal GDP (millions of U.S. dollars)1,9399,947307,7433,167322,79513,323
Real GDP growth rate (percent)3.72.81.62.1n.a.1.7
Inflation (percent, period average)6.67.97.67.3n.a.9.2
Fiscal balance (including grants; percent of GDP)−0.8−1.9−3.8−5.6n.a.−8.9
Total government debt (percent of GDP)41.517.631.216.1n.a.12.3
International reserves (months of imports)5.13.85.03.8n.a.19.3
Current account balance (percent of GDP)−0.81.1−4.7−13.6n.a.−1.3
Total exports (millions of U.S. dollars)8704,14991,8401,88898,7464,757
Sources: National authorities; and IMF staff estimates.Note: n.a. = not available.
Sources: National authorities; and IMF staff estimates.Note: n.a. = not available.

Pegged exchange rates and almost free intraregional capital mobility suggest that monetary policy rates are bound to move in parallel among CMA countries (Figure 5.1), with SARB’s monetary stance being the area’s anchor. Monetary hegemony is the only possible equilibrium in a fixed exchange rate system without formal cooperation procedures to set interest rates in line with CMA-wide conditions. Although discount rates in Namibia and Swaziland have been closely aligned with the SARB repo rate, Lesotho implements monetary policy through the Treasury bill market, hence the spread observed with respect to the policy rates of the other members. As a result of monetary policy convergence, LNS benefited from South Africa’s largely successful adoption of a formal inflation-targeting framework in 2001. The convergence of inflation rates across the CMA supports the view that the area effectively functions as a currency union under the leadership of the SARB.

Figure 5.1Common Monetary Area Countries and Botswana: Central Bank, Treasury, and Inflation Rates

Sources: IMF, International Financial Statistics (panels a and b), and World Economic Outlook database (panel c).

Note: Data for Botswana on Treasury bill rates (panel b) are not available.

Box 5.2The Common Monetary Area1

The CMA is an arrangement in which Lesotho, Namibia, and Swaziland (the LNS) have pegged their domestic currencies at par to the South African rand. Within the CMA, each country issues its own currency, and bilateral agreements define the states in which these currencies are legal tender. Although the South African rand is legal tender in all member countries of the CMA, the three other currencies are only legal tender in their own countries. The South African Reserve Bank (SARB) has adopted an inflation-targeting framework.

In each LNS country, the local currency and the rand are perfect substitutes, with no conversion cost, and no restrictions on funds transfers, whether for current or capital transactions. All four members of the CMA (together with Botswana) belong to the Southern African Customs Union. As a consequence, capital and goods are highly mobile across the CMA region,2 although further progress is needed in removing non-tariff barriers to trade. In normal times, LNS benefit greatly from goods and capital mobility because mobility gives them access to South African investments and markets. However, in times of crisis, sharp reversals in net capital flows (moving to South Africa) can strain these smaller economies. This is exacerbated by the absence of a conversion cost between the local currency and the rand.

Although all CMA members effectively share the same monetary policy, the CMA is not a full currency union. There is no common central bank conducting monetary policy for the region as a whole and pooling external reserves, and there is no formal regional surveillance of domestic policies to ensure that they remain consistent with the smooth working of the area. The exchange rate arrangements of the smaller countries under the CMA have certain characteristics of a currency board—domestic currency issues are required to be fully backed by foreign reserves (except for Swaziland). However, unlike a typical currency board, there is no restriction that the central bank of a small member country must hold domestic assets. Importantly, the one-to-one parities with the rand are not backed by irrevocable commitments such as the promise of mutual assistance in case the peg comes under pressure. The SARB may, however, make foreign exchange available to other members of the CMA. Finally, there are no fiscal transfers aimed at cushioning the impact of asymmetric shocks.

Another peculiarity of the CMA is that unlike other members, Swaziland has the option to adjust its exchange rate unilaterally. Such an adjustment would not require formal consultations with South African authorities.3 Additionally, Swaziland is not required to hold foreign exchange at the SARB to cover its currency in circulation, although Lesotho and Namibia are required to do so. Swaziland’s reintroduction of the rand as legal tender was done with the concurrence of the rest of CMA in 2003.

1 This box was prepared by Olivier Basdevant and Borislava Mircheva. See details of the CMA’s institutional framework in Appendix 5B.

2 The only exceptions result from the member countries’ investment or prudential liquidity requirements prescribed for financial institutions.

3 Should Swaziland decide to adjust its exchange rate, it would have to provide six months’ notice to the SARB.

In the absence of CMA-wide benchmarks for external reserves, the deterioration of fiscal balances in the aftermath of the global financial crisis may raise concerns about reserve adequacy. However, conventional reserve adequacy indicators are not alarming. International reserves in LNS countries covered close to three months of import.9 However, because the credibility of fixed parities can always be tested, reserve levels can usefully be compared to the size of base money and broad money aggregates, as is routinely done when assessing reserve adequacy in currency boards. Table 5.2 shows that Lesotho has maintained a relatively high level of reserves to cover broad money since 2008, but the ratio declined sharply in 2011. The situation in Swaziland is less benign, with reserves covering slightly less than half of broad money in 2011, which represents a dramatic deterioration since 2008. Although Namibia exhibited much lower reserve coverage ratios of broad money—a reflection of its more developed financial sector—its reserves largely exceed its short-term external debt (an indicator not available for Lesotho and Swaziland), suggesting that reserves would not come under immediate pressure if there were to be a sudden stop in capital inflows.

Table 5.2Small Countries in the Common Monetary Area: Reserve Adequacy Ratios
2008200920102011 estimate
(Months of imports)
Gross reserves/imports
Lesotho6.05.63.82.6
Namibia3.84.72.82.5
Swaziland4.63.92.82.3
(Percent)
Gross reserves/short-term external debt
Lesothon.a.n.a.n.a.n.a.
Namibia403908599591
Swazilandn.a.n.a.n.a.n.a.
Gross reserves/base money1
Lesotho939844670510
Namibia542636312274
Swaziland749496384237
Gross reserves/broad money
Lesotho15513210467
Namibia39532925
Swaziland116845448
Source: National authorities; and IMF staff estimates.Note: n.a. = Not available.

Rand in circulation are not included as part of base money.

Source: National authorities; and IMF staff estimates.Note: n.a. = Not available.

Rand in circulation are not included as part of base money.

Although the global crisis led to a synchronized deterioration of fiscal balances in the CMA and to some convergence, LNS fiscal positions are more volatile than South Africa’s mostly because of these countries’ smaller size and lack of diversification. The revenue-sharing formula in the Southern African Customs Union (SACU) also contributed to the volatility of fiscal positions, as evidenced by considerable differences during cyclical upturns, as in 2004–07, when LNS recorded much larger surpluses than did South Africa (Figure 5.2). This allowed LNS to maintain low and relatively stable public debt levels, and for Lesotho to gradually converge to the 20–40 percent of GDP range.

Figure 5.2Countries in the Common Monetary Area and Botswana: Fiscal Balance and Total Government Debt

Sources: National authorities; and IMF staff estimates.

Model-Based Assessments of the Common Monetary Area and Variants of it

Is CMA participation in the best interests of its members? The DMP model can provide some elements of an answer, comparing the costs of sharing a single monetary policy—forgone stabilization—with the benefits of policy coordination brought about by fixed exchange rates. As indicated earlier, welfare effects are calculated as deviations from a hypothetical counterfactual in which each country could set its own monetary policy to maximize a weighted average of national policy objectives. The counterfactual implicitly assumes that nominal exchange rates would be allowed to fluctuate and to be a key transmission channel of monetary policy. However, the specific costs related to exchange rate volatility and risks are not captured by the model. The analysis first looks at the CMA as it is and then explores the implications of other SADC member states joining the CMA. These exercises were selected for their illustrative virtues, deliberately abstracting from other relevant dimensions of integration. Finally, the analysis estimates the impact of establishing a regional central bank in which each member state would influence the common monetary policy proportionally to its economic size.

The CMA

On the cost side, what matters are the correlations of LNS terms-of-trade shocks with South Africa’s: the lower the correlation for a given country, the greater the cost entailed in the absence of monetary sovereignty. On the benefit side, the model emphasizes two key elements. The first is the intensity of intraregional trade flows. The greater they are, the greater is the externality related to uncoordinated monetary policies (for instance, leading to competitive devaluations). A monetary union shuts off this transmission channel of monetary policy, which, in turn, prevents a common central bank (CCB) from exploiting monetary policy to boost output in one country at the expense of the others.10 Therefore, the CCB enjoys greater anti-inflationary credibility than does a purely national institution. In addition, the CCB can more easily deflect pressures from an individual member state to provide monetary financing for that state’s budget. Note that model simulations assume an allocation of seigniorage and inflation tax revenues according to GDP shares. This allocation of seigniorage differs from the rules-based allocation under the current CMA framework. In practice, however, the induced discrepancies between the formula-based and the GDP-share revenues are too small (about ½ percentage point of GDP for Lesotho, for instance) to be significant for these welfare comparisons.11

To account for the leadership role of the SARB in setting the area-wide monetary stance, this analysis calibrates the CCB such that LNS have no influence. In this scenario, the hypothetical SARB has no incentive to seek competitive devaluations vis-à-vis other CMA countries, which lowers the credible inflation rate. However, SARB is not better insulated from eventual pressures to raise the inflation tax than if it sets monetary policy for South Africa only. Likewise, LNS central banks boost their anti-inflationary credibility because they cannot use devaluations in a strategic fashion, but in addition, the CMA arrangement—assuming that it is in itself fully credible—perfectly insulates local monetary conditions from LNS budget financing considerations. For LNS, CMA membership thus represents a powerful tool to cement their commitment to price stability.

Key inputs for the model simulations are summarized in Table 5.3. Low terms-of trade correlations of LNS with South Africa as well as generally more volatile terms of trade point to nonnegligible costs related to the absence of monetary sovereignty for these countries. At the same time, LNS are generally more open to trade12—particularly with South Africa—and have larger “financing needs” from the budget, suggesting that gains in anti-inflationary credibility are likely to be large.

Table 5.3Countries in the Common Monetary Area: Key Model Inputs, 1994–2010
TOT correlations with South AfricaFN1Standard deviation of TOT shocksOpenness:

0.5(X + M)/GDP
Adjusted standard deviation of TOT
Lesotho28.1766.552.4276.551.85
Namibia20.5241.553.8345.281.73
South
Africa100.0033.341.4327.340.39
Swaziland9.3244.432.7881.732.27
Memorandum
Botswana36.0642.173.9542.831.69
Source: Authors’ calculations.Note: FN = financing need; M = imports; TOT = terms of trade; X = exports.

Financing need estimates the resources required to finance the socially optimal level of productive (welfare-generating) public spending. FN therefore combines measures of the desired level of productive public spending and waste. The methodology to capture these unobservable variables is discussed in Debrun, Masson, and Pattillo (2011). Underlying data reflect 1994–2005 averages to exclude the large changes in fiscal variables observed around the global financial crisis.

Source: Authors’ calculations.Note: FN = financing need; M = imports; TOT = terms of trade; X = exports.

Financing need estimates the resources required to finance the socially optimal level of productive (welfare-generating) public spending. FN therefore combines measures of the desired level of productive public spending and waste. The methodology to capture these unobservable variables is discussed in Debrun, Masson, and Pattillo (2011). Underlying data reflect 1994–2005 averages to exclude the large changes in fiscal variables observed around the global financial crisis.

As anticipated, the costs arising from shock asymmetry are meaningful—a permanent decline of per capita income equal to 0.3–0.5 percent. Because the CCB/SARB now sees monetary policy as less effective than if exchange rates with LNS were floating, the inflationary bias inherent to its incentive to stimulate production beyond potential is smaller. The overall welfare effect from the monetary externality is by construction the same (0.46 percent) for all countries; and with the exception of Swaziland, it is large enough to offset the costs of a one-size-fits-all monetary policy. For South Africa, the anchor of the CMA, this credibility effect is, in fact, the only determinant of net gain because the equilibrium inflation tax and response to shocks are the same as under monetary sovereignty. LNS can count on a third, and quantitatively much larger, benefit: a credible CMA arrangement insulates their central banks from the pressure to finance strained budgets by raising the inflation tax. The magnitude of the gains—in the range of 2 to 6 percent in permanent per capita income—reflects the very high equilibrium inflation tax rates found by the model under the assumption of monetary sovereignty. This is rooted in a greater estimated appetite for public spending as well as weaker economic governance (and heightened likelihood of waste). For instance, the model estimates that Lesotho’s financing need is twice as large as South Africa’s (last column of Table 5.4). Overall, the model’s estimates of the permanent per capita income gains due to the CMA are significant: 6.1 percent for Lesotho, 2.1 percent for Swaziland, 1.8 percent for Namibia, and 0.5 percent for South Africa.

Table 5.4Common Monetary Area: Welfare Gains and Losses(Percent of GDP)
Due to:
Welfare gainMonetary externalityFiscal asymmetryShock asymmetryGDP share1Shock correlation2FNA/FN2
Lesotho6.070.466.03−0.310.5728.1750.10
Namibia1.790.461.64−0.282.6220.5280.25
South Africa0.460.460.000.0095.75100.00100.00
Swaziland2.110.462.19−0.511.059.3275.04
Source: Authors’ calculations.Note: FN = financing need; FNA = average financing need. It is assumed that South Africa sets monetary policy for the Common Monetary Area.

In percent.

Compared with South Africa’s terms-of-trade shock and financing need, respectively, in percent.

Source: Authors’ calculations.Note: FN = financing need; FNA = average financing need. It is assumed that South Africa sets monetary policy for the Common Monetary Area.

In percent.

Compared with South Africa’s terms-of-trade shock and financing need, respectively, in percent.

Hypothetical Expansions of the Common Monetary Area

A natural question at this stage is whether expanding the membership of a mutually beneficial arrangement such as the CMA would make sense both for existing members and for potential newcomers. For illustrative purposes, the universe of potential newcomers among countries of the SADC member states is used—with the exceptions of Seychelles and Madagascar because of gaps in data. The main appeal of SADC as the focus group for these simulations is that it is a large and heterogeneous club to which all CMA countries belong. Although SADC has plans to put in place a monetary union, the horizon is distant and subject to considerable uncertainty.13 Moreover, one of its members, Tanzania, has already committed to participating in a monetary union with other members of the East African Community, whereas Botswana has shown no interest in joining a monetary union. To shed light on the strategic dimension of integration, the analysis proceeds in two steps: first, it expands the existing CMA one country at a time and assesses the welfare impact; second, it introduces potential members all at once.

Adding a Single Country to the Existing CMA: Who Gains, Who Loses?

SADC comprises a large and diverse group of countries. However, even in this broader area, South Africa’s economy remains by far larger than all the others combined (it makes up more than 70 percent of SADC’s GDP). As Table 5.5 illustrates, countries differ widely in the key dimensions of this model: terms-of-trade correlations and volatilities, government financing needs, and intensities of intra-regional trade linkages. Given the results above, one could again expect that financing needs will ultimately shape a given country’s willingness to join and the existing CMA members’ incentive to welcome it. The very large dispersion of financing needs reflects highly heterogeneous institutional environments. Tanzania and all CMA members except Lesotho have relatively low financing needs, in sharp contrast with Angola, Lesotho, and Zimbabwe. Correlations of terms-of-trade shocks are generally low or negative, again pointing to significant costs of monetary unification, when compared with the hypothetical benchmark case of monetary sovereignty. That is, relying exclusively on OCA criteria to analyze the desirability of a larger CMA would likely cause the idea to be dismissed out of hand.

Table 5.5Countries in the Southern African Development Community (SADC): Selected Indicators, 1994-2010
TOT correlations with South AfricaFN1Standard deviation of TOT shocksOpenness: 0.5(X+M)/GDPAdjusted standard deviation of TOT
CMA members
Lesotho28.1766.552.4276.551.85
Namibia20.5241.553.8345.281.73
South Africa100.0033.341.4327.340.39
Swaziland9.3244.432.7881.732.27
Other SADC members
Angola−0.8457.7910.0872.517.31
Botswana36.0642.173.9542.831.69
Congo, Dem. Rep. of24.2537.565.8635.382.07
Malawi−25.5546.719.5434.833.32
Mauritius−18.4927.587.3158.814.30
Mozambique28.0342.582.4232.570.79
Seychelles14.4810.2985.358.78
Tanzania−32.5231.065.2923.081.22
Zambia228.0946.587.5536.432.75
Zimbabwe3−25.3052.142.3739.500.94
Source: Authors’ estimates.Note: CMA = Common Monetary Area; FN = financing need; M = imports; TOT = terms of trade; X = exports. Seychelles is omitted from the calculations because its high per capita income gives implausible figures for its FN variable.

Data from 1994-2005

Data from 1999-2010.

Data from 1999-2007.

Source: Authors’ estimates.Note: CMA = Common Monetary Area; FN = financing need; M = imports; TOT = terms of trade; X = exports. Seychelles is omitted from the calculations because its high per capita income gives implausible figures for its FN variable.

Data from 1994-2005

Data from 1999-2010.

Data from 1999-2007.

A number of interesting results emerge from the exercise (Table 5.6). First, all CMA members appear to benefit—albeit marginally in many cases—from the membership of any other individual SADC country provided the SARB continues to set monetary policy for the enlarged CMA. Second, three potential candidates for CMA membership would lose out from joining on their own initiative: Angola, Mauritius, and Tanzania. These three countries would clearly suffer from having to adopt South Africa’s monetary policy because their terms-of-trade shocks are orthogonal to or negatively correlated with South Africa’s. In Angola and Mauritius, terms of trade are also extremely volatile, making a national monetary policy desirable. These are prominent cases in which OCA arguments dominate credibility considerations. For Mauritius—and to a lesser extent, Tanzania—the loss would be aggravated by the likelihood that the CMA-wide inflation tax would exceed what the model describes as optimally required to cover their relatively low financing needs.

Table 5.6Welfare Gains or Losses from Adding a Single SADC Country to the CMA(Percent of GDP)
AngolaBotswanaCongo, Dem. Rep. ofMalawiMauritiusMozambiqueTanzaniaZambiaZimbabwe1
Lesotho0.020.250.020.050.050.090.010.080.20
Namibia0.030.310.030.060.060.110.010.100.25
South Africa0.030.330.030.060.070.120.020.100.26
Swaziland0.030.310.030.060.060.110.010.100.24
Angola−0.29
Botswana2.25
Congo, Dem. Rep. of0.93
Malawi1.92
Mauritius−2.56
Mozambique2.32
Tanzania−0.18
Zambia2.40
Zimbabwe4.11
Source: Authors’calculations.Note: CMA = Common Monetary Area; SADC = Southern African Development Community. Monetary policy is assumed to be set by South Africa. Welfare is relative to monetary autonomy for new entrants, and to CMA for existing members.

The Zimbabwe exercise is based on data up to 2007 and, as a result, the estimated gains owing to low inflation from participating in the CMA are measured relative to the monetary policy pursued before dollarization in 2009. Benefits from low inflation would be smaller with unchanged moderate gains due to monetary externality if Zimbabwe were to join the CMA.

Source: Authors’calculations.Note: CMA = Common Monetary Area; SADC = Southern African Development Community. Monetary policy is assumed to be set by South Africa. Welfare is relative to monetary autonomy for new entrants, and to CMA for existing members.

The Zimbabwe exercise is based on data up to 2007 and, as a result, the estimated gains owing to low inflation from participating in the CMA are measured relative to the monetary policy pursued before dollarization in 2009. Benefits from low inflation would be smaller with unchanged moderate gains due to monetary externality if Zimbabwe were to join the CMA.

The greatest winners among potential new entrants would be Botswana,14 Zambia, and Zimbabwe.15 All three countries would benefit significantly from a lower inflation rate because fiscal pressures on their monetary policies would be lower. In Botswana and Zambia, positive terms-of-trade correlations also help contain the costs from no longer having their own monetary policy to stabilize output in the face of shocks. These hypothetical gains and losses also depend importantly on estimated financing needs, and can change over time as countries adopt fiscal reforms relieving pressures on their monetary policies. Conversely, South Africa, the anchor for the CMA, could see its finances deteriorate to the point of nullifying the hypothetical gains calculated here for other countries.

Expanding the CMA Arrangement to All SADC Countries

The logic of the DMP model suggests that a block expansion could be more desirable than piecemeal monetary integration. To illustrate this property of the model, the results in Table 5.7 indicate that only Mauritius would remain a net loser from CMA membership, essentially for the same reasons as above. For Angola and Tanzania, notably, joining the CMA with all other SADC countries at the same time would yield net benefits, whereas joining the CMA alone would not. As the decomposition of welfare effects shows, the determining factor between the two integration strategies is that the credibility effect associated with a single monetary policy is quite sizable (in excess of 1.3 percent of permanent per capita consumption for SADC candidates and about 0.9 percent for existing CMA members).

Table 5.7A Greater CMA/SADC: Welfare Gains and Losses(Percent of GDP, unless otherwise specified)
Cause:
Welfare gainMonetary externalityFiscal asymmetryShock asymmetryGDP share (Percent)Shock correlation1FNA/FN1
Angola0.421.354.59−5.295.07−0.8457.70
Botswana2.771.351.76−0.252.9936.0679.07
Congo, Dem. Rep. of1.771.350.85−0.402.6824.2588.76
Lesotho0.680.890.000.000.4328.1750.10
Malawi2.671.352.62−1.170.97−25.5571.38
Mauritius−1.701.35−1.21−1.900.62−18.49120.91
Mozambique3.041.351.84−0.061.9528.0378.30
Namibia0.840.890.000.001.9620.5280.25
South Africa0.890.890.000.0071.61100.00100.00
Swaziland0.920.890.000.000.799.3275.04
Tanzania0.711.35−0.47−0.194.54−32.52107.33
Zambia3.121.352.59−0.701.8828.0971.58
Zimbabwe4.661.353.61−0.124.52−25.3063.95
Source: Authors’calculations.Note: CMA = Common Monetary Area; FN = financing need; FNA = average financing need; SADC = Southern African Development Community. South Africa is assumed to set monetary policy. Welfare is relative to monetary autonomy for new entrants, and to CMA for existing members.

With respect to the union’s average shocks and financing need (FNA), respectively, in percent.

Source: Authors’calculations.Note: CMA = Common Monetary Area; FN = financing need; FNA = average financing need; SADC = Southern African Development Community. South Africa is assumed to set monetary policy. Welfare is relative to monetary autonomy for new entrants, and to CMA for existing members.

With respect to the union’s average shocks and financing need (FNA), respectively, in percent.

Hegemony Versus A Regional Central Bank

So far, the analysis has assumed that CMA monetary policy would continue to be determined by the current rules of the game, giving the SARB explicit monetary hegemony. However, as integration proceeds, the commitment to a regional monetary union could be further cemented by the establishment of a regional central bank in which each member would have a voice. Specifically, scenarios are simulated in which a regional central bank sets the common monetary policy to maximize a weighted average of individual welfare functions, using country shares in regional GDP as weights. The analysis first quantifies the welfare impact on existing CMA members to move toward such a model, and then revisits the effect of a regional CCB in the context of the greater SADC currency union.

A Full Common Monetary Union with Current Members

The simulations reported in Table 5.8 suggest that under the current calibration of the model—which reflects past data—no CMA member would benefit from a CCB replacing the SARB. The gains associated with a more stabilizing monetary policy for LNS would be more than offset by LNS governments’ pressures on the CCB to raise the inflation tax. By definition, there could not be any additional gain resulting from policy coordination because exchange rates are already fixed. Admittedly, these effects are very small, and fall well within reasonable margins of error related to the uncertainty surrounding the model’s calibration. The overwhelming influence of South Africa’s economic conditions on the region is such that there would arguably be little difference between the current, explicitly hegemonic model and a regional CCB. Moreover, that particular comparison can be sensitive to certain limitations of the model. As mentioned earlier, the DMP framework does not capture some potential benefits of establishing a monetary union, including the presumed elimination of currency risk and the greater induced convergence of nominal interest rates. Also, specific guarantees on the political independence of the CCB could better insulate it from the influence of fiscally profligate countries and could conceivably be a precondition imposed by the anchor of a fixed exchange rate system before moving to a full-fledged monetary union (Debrun, 2001).

Table 5.8Welfare Effect of a CMA Monetary Union Versus Existing Arrangement(Percent of GDP, unless otherwise specified)
Due to:
Welfare gainMonetary externalityFiscal asymmetryShock asymmetryGDP share (Percent)Shock correlation1FNA/FN1
Lesotho−0.080.00−0.080.000.5727.1950.89
Namibia−0.080.00−0.100.012.6231.5881.50
South Africa−0.110.00−0.110.0095.7599.19101.56
Swaziland−0.090.00−0.100.011.0514.7876.21
Source: Authors’ calculations.Note: CMA = Common Monetary Area; FN = financing need; FNA = average financing need. South Africa sets monetary policy for the existing CMA.

With respect to the union’s average shock and financing need (FNA), respectively, in percent.

Source: Authors’ calculations.Note: CMA = Common Monetary Area; FN = financing need; FNA = average financing need. South Africa sets monetary policy for the existing CMA.

With respect to the union’s average shock and financing need (FNA), respectively, in percent.

A Larger Currency Union with SADC Members

The set of net beneficiaries of a larger SADC currency union under a regional CCB is the same as if SARB maintained its leadership position: only Mauritius would have no interest in joining the union. Quantitatively, however, an SADC union with a regional central bank would spread the losses from a one-size-fits-all monetary policy across virtually all countries, confirming that SADC is not an OCA in the traditional sense. Also, existing CMA members would now lose out from the larger inflation tax imposed by the participation of the more profligate members of the SADC. The net result is that current CMA members would gain little, if anything, from a larger SADC currency union under a regional central bank.

Conclusion

Model simulations shed new light on incentives to form currency unions in a context in which central banks internalize the government budget constraint, pointing to three broad policy implications: 16

  • In line with the traditional OCA literature, DMP simulations suggest that the costs of a one-size-fits-all monetary policy can be significant if external shocks affecting individual economies are large and uncorrelated with the rest of the region (e.g., Bayoumi and Ostry, 1997). The model suggests that CMA countries thus benefit from their monetary association because it provides large offsetting gains in policy credibility and macroeconomic stability.

  • Although the model captures the value of a rand anchor for the CMA, a regional CCB conducting policy on the basis of area-wide averages would still appear to be more beneficial than full monetary autonomy. This shows that regardless of specific institutional guarantees on the independence of the CCB, monetary unification per se can deliver major credibility gains. By the same token, moving toward a regional CCB would likely require that all parties prefer such guarantees to the existing arrangement.

  • Mechanisms alleviating the costs of currency unions with regard to inefficient shock stabilization could help, particularly in the context of an expansion of the CMA. In scenarios explored above, shock asymmetry—which makes unionwide monetary policy inadequate—can be quite costly in some cases, pointing to the importance of more-countercyclical fiscal policies and ultimately, risk-sharing mechanisms. Agreeing to and implementing an effective transfer system would pose important challenges for a heterogeneous grouping of countries like SADC; however, these issues are beyond the scope of this chapter.

The simulations further suggest that the current monetary arrangement is beneficial for all CMA members, including South Africa. Lesotho and Swaziland gain the most because the CMA insulates their monetary policies from fiscal pressures. The potential gains of expanding the CMA to SADC countries depend on the strategy. If current SADC countries were to join the CMA together, all of them except Mauritius could be better off. However, Angola, Mauritius, and Tanzania would lose out from individual membership because Angola is subject to large idiosyncratic disturbances and Mauritius and Tanzania already have strong fiscal policies.

The creation of a genuine CMA-wide monetary union with a regional central bank carries some costs in forgone anti-inflationary credibility because the model assumes that fiscally profligate countries could have some influence over monetary policy and extract a marginally higher inflation tax. All members would therefore be better off maintaining the current CMA regime. Furthermore, if this monetary union were enlarged to be SADC-wide with a regional central bank, all members except Mauritius would receive benefits, but the gains for existing CMA members—compared with the existing arrangement—are likely to be negligible, falling well within plausible margins of error.

Appendix 5A. History of the Common Monetary Area
Table 5A.1Major Events in the History of the Common Monetary Area (CMA)
Year
1974Lesotho, South Africa, and Swaziland signed the Rand Monetary Area (RMA) treaty.
Swaziland established a monetary authority and issued its own national currency, the lilangeni, pegged at par to the rand. Botswana did not sign the RMA agreement: it had withdrawn from the negotiations in September.
1975–76Botswana established a central bank, and replaced the rand at par with its own national currency, the pula.
1980Lesotho established a central bank and issued its own national currency, the loti, pegged at par with the rand.
1986Lesotho, South Africa, and Swaziland signed the CMA Trilateral Agreement to replace the RMA, making additional provisions regarding the capital account, intra-CMA fund transfers, and seigniorage compensation.
Swaziland discontinued the use of the rand as legal tender alongside the lilangeni.
1989The CMA was amended, removing exchange restrictions resulting from limitations on conversion of balances upon termination of the monetary agreement or the withdrawal of one party.
1992Following its independence from South Africa, Namibia formally joined the CMA. The Multilateral Agreement replaced the Trilateral Agreement.
1993Namibia initiated issuing its own national currency, the dollar, pegged at par to the rand.
2003Swaziland reauthorized use of the rand as legal tender.
Appendix 5B. Institutional Framework of the Common Monetary Area

Wang and others (2007) provide a comprehensive summary of the institutional framework of the CMA as follows.

Currency Arrangement

Article 2 of the CMA (Multilateral) Agreement gives the three small member countries the right to issue national currencies, and their bilateral agreements with South Africa define the areas where their currencies are legal tender. The local currencies issued by the three members are legal tender only in their own countries. The South African rand, however, is legal tender throughout the CMA.17 The bilateral agreements also require LNS to permit authorized dealers within their territories to convert, at par, notes issued by their central banks or the South African Reserve Bank (SARB) without restriction and subject only to normal handling charges.

Under the Lesotho–South Africa and Namibia–South Africa bilateral agreements, the central banks of Lesotho and Namibia are required to maintain foreign reserves at least equivalent to the total amount of local currencies they issue.18 Such reserves may comprise the central bank’s holdings of rand balances, the rand currency the central bank holds in a Special Rand Deposit Account with the SARB, South African government stock (up to a certain proportion of total reserves), and investments in the Corporation for Public Deposit in South Africa.

Movements of Funds within the CMA

Under the terms of the CMA Agreement (Article 3), no restrictions can be imposed on the transfer of funds, whether for current or capital transactions, to or from any member country. The only exceptions result from the member countries’ investment or liquidity requirements prescribed for financial institutions. The small member countries view investment and liquidity requirements as a measure of savings mobilization for development purposes. The regulations requiring the investment of funds by financial institutions in domestic securities or credits to local businesses or individuals are, in effect, minimum local asset requirements. These regulations are meant to address the concern of the three small, less developed, CMA members that funds generated in their territories and deposited with local financial institutions tended to flow to the more developed capital markets of South Africa.

Access to South African Financial Markets

The CMA Agreement provides for the three small member countries to have access to the South African capital and money markets, but only through prescribed investments or approved securities that can be held by financial institutions in South Africa, in accordance with prudential regulations in Lesotho, Namibia, and Swaziland (LNS). The terms and timing of such issues are subject to consultation and agreement with the South African government, and the issues have the same rating as South African municipal bonds. As for the short-term money market, there are no regular arrangements for the taking up in South Africa of treasury bills issued by LNS. However, the CMA Agreement recognizes the right of the other member countries, in special circumstances, to enter into bilateral negotiations with South Africa to obtain temporary central bank credit.

Gold and Foreign Exchange Transactions

Although LNS have the right to authorize foreign transactions of local origin, and are responsible for doing so, the CMA Agreement (Article 5) requires their exchange control regulations to be—in all material aspects—similar to those in effect in South Africa. Gold and foreign exchange receipts of residents are subject to a surrender requirement. There are no exchange restrictions on current international transactions and for nonresidents.

Compensation Payments

Since the rand is legal tender in all CMA countries (but the currencies of the three small CMA members are not legal tender in South Africa), South Africa compensates them for forgone seigniorage. Compensation is based on a formula equal to the products of (1) two-thirds of the annual yield on the most recently issued long-term South African government stock, and (2) the volume of rand estimated to be in circulation in the member country concerned. The ratio of two-thirds was established on the assumption that it approximated the yield of a portfolio of reserve assets comprising both long-term and short-term maturities, assuming that the average yield would be less than the full long-term yield.

Consultation and Other Provisions

To facilitate implementation of the CMA Agreement, the member countries have established a commission in which each of them has one representative (along with advisors as needed). The commission holds regular consultations—at least once a year—with the aim of reconciling the interests of member countries on common issues pertaining to monetary and foreign exchange policies. It also convenes at other times at the request of a member country. Article 9 of the CMA Agreement provides for the establishment of a tribunal to arbitrate disputes that might arise between member countries regarding the interpretation or application of the agreement.

Appendix 5C. Description of the DMP Model
National policymaking
Phillips curve with regional spilloversyi=yN+c(πiπieτi)Σki,k=1nθi,kc(πkπke)+εi.i=1,,n(5.1)
Government budget constraint (no debt)gi=ρ¯i+μπi+τiδi(5.2)
Government utility functionUiG=12{a(πiπ˜i)2i2γ(gig˜i)2}+yi(5.3)
Trade-off between output and inflation variabilityπ˜i=ηεiwithη>0(5.4)
Supranational monetary policy
Phillips curve faced by the common central bank for each member of Myi=yN+c(1θiM)(πMπMe)iΣkMθi,kc(πkπke)+εi,iM,withθiM=ΣkMθi,k(5.1)
Key variables and parameters
πiInflation rate in country i. A superscript e designates a rationally expected value.
yiLogarithm of output in country i.
yNLogarithm of the natural level of output at zero taxation. Without loss of generality, the analysis assumes yN = 0.
τiCorporate income tax rate (also tax revenues in percentage of output).
θi,kMarginal effect of monetary policy in country k on output in country i.
εiTerms-of-trade shock (zero-mean, transitory, and with finite variance).
giSocially beneficial government expenditure in percentage of output.
μInflation tax base in percentage of output.
ρ¯iPermanent nontax revenue from natural resource endowment in percentage of output.
δiFunds diverted from socially beneficial government expenditure in percentage of output.
ηRelative preference for output stability against inflation stability.
Note: Complete solutions are available from the authors upon request.
Note: Complete solutions are available from the authors upon request.
Inflation Rates under Alternative Monetary Regimes
Monetary regimeEquilibrium inflation (country i)
Autonomy
The equilibrium (time-consistent) inflation is…πi*=πi**+γμbΛδi+(b+γ)Λc(5.5)
…while the socially optimal rate is…πi**=γμbΛ[g˜iρ¯i]Sizeoffinancingrequirement+γμΛcOutputcostoftaxationa(b+γ)ηΛεiOutputstabilization,(5.6)
with Λ = a(b+γ) + γμ2b > 0
…so that the inflation bias is…πi*πi**=(b+γ)Λc"AugmentedBarroGordoninflationbias+γμbΛδiPublicsectorinefficiency(5.7)
Monetary union M (utilitarian common central bank)πiM*=γμbΛ(FNAM)+(1θAM)(b+γ)+γμΛca(b+γ)ηΛεAM,(5.8)foralliM,withxAM=ΣiMωiMxi,forx{FN,θ,ε}

(cross-country, output-weighted averages within M), and FNi=gi˜+δiρ¯i. Hence, πAM*=πA*AverageinflationundernationalpoliciesθAM(b+γ)ΛcAveragereductionintheBarroGordonbias.
Legally independent national central banksπi*=πi**+λi(b+γ)Λc+γμ(b(1λi)γ)Λδi(5.9)
with 0 ≤ λi ≤ 1, the extent of political interference. If λi the government has no influence on the central bank’s decisions; and if 1, λi = 1 the government effectively sets monetary policy (see equation 5.5).

The CMA comprises Lesotho, South Africa, and Swaziland.

These estimates are not meant to make precise comparisons among the countries but are intended to provide broad estimates of welfare gains and losses from CMA participation. More precise estimation for each country would require capturing more country-specific factors (policymakers’ preferences and institutional quality) and detailed measures of financial needs based on recent data.

Some broader considerations are surveyed in Hawkins and Masson (2003).

Beetsma and Giuliodori (2010) provide a detailed survey on OCA theory and the Economic and Monetary Union in Europe.

Frankel (1999) notes that the endogeneity of OCA criteria means that some parameters, such as openness and income correlation, are not irrevocably fixed, but instead can change over time in response to the countries’ fundamental policies and to exogenous factors.

Blanchard and Wolfers (2000) point out the endogeneity of labor market institutions, and Issing (2001) stresses the endogeneity of political integration.

Appendix 5A summarizes major events in the development of the CMA.

Although Botswana left the RMA in 1975, it continued to use the rand until August 1976.

IMF (2011a) cautions against a uniform metric for reserve adequacy—including the traditional rules of thumb—across all low-income countries. In particular, the quality of the overall policy framework should play a role in the determination of optimal reserve levels.

This chapter considers two cases: (1) the SARB chooses monetary policy for the whole area, and (2) a CCB weighs the welfare of South Africa and LNS countries proportionally to their economic size.

Pastor and Ramirez (2012) find that in international comparisons, the revenues from seigniorage in the CMA countries and Botswana are less than one-fourth of what generally accrues to governments in countries with similar rates of inflation.

The model simulations reflect trade intensities from a bilateral trade matrix using the IMF Direction of Trade Statistics.

The plan is outlined in the SADC Regional Indicative Strategic Development Plan. It envisages a common market by 2015 and a common currency by 2018.

For Botswana, the largest component of welfare gains is due to the fiscal asymmetry (1.76 percent of GDP) with moderate gains due to monetary externalities (0.79 percent) offset only partially by the welfare loss associated with asymmetry in terms-of-trade shocks (0.25 percent).

Kramarenko and others (2010) review pros and cons of alternative monetary regimes for Zimbabwe, including the possibility of joining the CMA. The paper also provides a welfare analysis of Zimbabwe’s participation in CMA and predicts that existing members’ welfare would fall marginally by about ⅓ percent of CMA GDP. Zimbabwe, however, would gain more than 24 percent of GDP, with most of the gain stemming from the fiscal externality. Differences between the current results and previous ones are the result of a significant reduction in Zimbabwe’s estimated financing need, which reduces the necessary inflation tax.

De Cecco and Giovannini (1989) and Kenen (1995) discuss necessary policy suggestions for the monetary unification in Europe.

Swaziland suspended the use of the rand as legal tender in 1986 despite the fact that the rand continued to be widely accepted in the country. In the fall of 2003, the Swazi authorities re-authorized the use of the rand as legal tender alongside the lilangeni.

This provision was not included in the Swaziland–South Africa bilateral agreement of April 1986, in part for reasons detailed in note 17. However, the Central Bank of Swaziland has maintained foreign reserves larger than the total amount of local currencies it issued throughout the past two decades.

    Other Resources Citing This Publication