Journal Issue
Back Matter

Back Matter

Thomson Fontaine, Dalmacio Benicio, Joannes Mongardini, Genevieve Verdier, and Gonzalo Pastor
Published Date:
January 2011
    • ShareShare
    Show Summary Details
    APPENDIX 1 A Brief History of the SACU Agreement and the Revenue-Sharing Formula

    The SACU dates back to 1910, when Basutoland, Bechuanaland, South Africa, and Swaziland signed an arrangement in Potchesftroom, South Africa’s old capital.16 This agreement lasted until the British protectorates gained independence in the mid-1960s. It was then renegotiated, culminating in the 1969 agreement, which preserved an important element of import dependence by the smaller member countries on South Africa, although it also included a revenue-sharing formula for the division of customs and excise revenue collected in the union.

    Namibia formally joined SACU in 1992 after attaining independence from South Africa in 1990. However, political conditions in South Africa were already changing at that time, auguring for an eventual revision of the agreement at the earliest possible time. Negotiations for a new SACU agreement began in late 1994, months after the formation of the first South African government of national unity.

    Cementing a new SACU agreement took almost eight years of “stop-and-go” negotiations. Reportedly, discussions among country representatives covered the new institutional arrangement (including accountability and voice in policy decision making in the customs union) and a new revenue-sharing formula. The new SACU agreement was signed in October 2002. It became effective in July 2004 following the completion of technical annexes and ratification by all member countries.

    The new SACU agreement, containing a total of 51 articles, covers three main areas: governance and administration, economic policy and regulatory issues, and revenue sharing. The thrust of the new SACU agreement is to facilitate the development of common policies and strategies for member states in a global economy, while ensuring an equitable sharing of revenue from customs, excise, and additional duties.

    The new SACU revenue-sharing formula deals with customs and excise revenues separately through two distinct components, and establishes a third “development” component. Total customs revenues collected are to be paid into a common revenue pool (administered through the South African Revenue Fund) and distributed according to each country’s share of total intra-SACU imports (i.e., each country’s total imports from the other SACU members).17 The customs component for the current financial year (t) is distributed to all member states in relation to each member state’s intra-SACU imports for the most recent financial year (t—2). No adjustments are made owing to revisions of import data, but adjustments are made if actual customs duties collected are different from the initial estimates up to two years (t + 1 and t + 2). Countries that import most from within the union receive the largest share of the customs pool, thereby providing implicit compensation for the presumed “cost-raising” effects of the customs union. A fixed share (initially 15 percent) of total SACU excise collections (also paid into the common revenue pool and distributed across member countries net of administration costs) amounts to a development component. This component is allocated using the per capita GDP of each country.18 The remaining SACU excise revenues are distributed on the basis of each country’s share of total SACU GDP for the most recent calendar year (t—2)—a proxy of the value of excisable goods consumed.

    As noted, the new revenue-sharing formula includes significant changes to the manner in which revenue shares are calculated, managed, and distributed. The shares of each component for each member are estimated from the most recent and actual trade, GDP, and GDP per capita data. These shares are applied to agreed customs and excise forecasts, with adjustments necessary over the next two years to reflect revised/actual tax collections.

    The SACU Revenue-Sharing Formula

    The formula for sharing customs revenue is described in the 2002 SACU Agreement. The total share (S) of SACU revenue to each member country (i) from the common revenue pool is calculated as follows:

    • Si = Ci + Ei + Di


    • Ci = (Mi / M) * C

    • Ei = (GDPi / SACUGDP) * (E-D)

    • Di = 20*(1-(GDPPi/SACUGDPP-1)/10))/100*E


    • Ci = the share to each member of total customs duties collected in SACU C, where Mi is the CIF value of the goods imported by each member from other SACU members as a share of total intra-SACU imports M (The Customs Component)

    • Ei = the share to each member of total excise duties E collected in SACU less the development component D (currently 15 percent of total excises) (The Excise Component)

    • Di = the share of the development component to each member of the total excise duties collected in SACU distributed as a share of the member country’s per capita GDP as a share of per capita SACU GDP (The Development Component)

    • GDPi = GDP of country i

    • GDPPi = GDP per capita of country i

    • SACUGDP = GDP of all SACU

    • SACUGDPP = average GDP per capita of SACU

    • M = total intra-SACU imports

    • Mi = total intra-SACU imports of country i

    APPENDIX 2 Botswana: Diminishing Returns to Public Expenditure19

    Botswana’s rapid economic growth during the 1970s and 1980s was largely the result of high public investment in human and physical capital. The associated gains—in terms of building the basic infrastructure and the provision of education, health, and other services that were virtually nonexistent at independence in 1966—are remarkable. Over time, however, public investment has become less efficient in promoting long-term growth. Returns have declined as the public sector has grown and engaged in a widening range of public investment that extends well beyond the provision of core public goods. Simultaneously, the attention paid to project appraisal and evaluation has weakened as project implementation capacity has become stretched and buoyant diamond revenues and repeated budget surpluses relaxed the government budget constraint. The postponement of maintenance expenditures and a loose definition of what constitutes investment have further debilitated the quality and efficiency of public investment spending.

    Education is one area where returns to public spending appear to be lower than in some other middle-income countries. Botswana allocates nearly a quarter of its annual public spending to education, and spending per student is significantly higher than in other middle-income countries. High rates of literacy and enrollment in secondary schools have been among the system’s accomplishments. However, less than 60 percent of those enrolling in secondary school complete their schooling. Other measures of educational attainment are also no better than average compared with other countries in the region, despite higher spending. Costs are also increasing, as measured by the high and rising implicit price deflator for education services.

    Social safety nets are another area where value for money appears weak. Social safety net programs account for about 10 percent of government spending, or 3—4 percent of GDP, and are designed to address risks related to malnutrition, HIV/AIDS, unemployment, disability, and old age.

    Middle Income Countries: Public Expenditure on Education, 1999–2009

    Source: Botswana Ministry of Finance, Development, and Planning and IMF staff estimates.

    CPI Inflaction and Education Services Volume and Prices

    (Annual percentage change)

    Source: Bank of Botswana and IMF staff estimates.

    However, they cover only a small portion of the poor (19 percent) and many beneficiaries are nonpoor households (57 percent). Better and more regular information (for example, poverty assessments and household income and expenditure surveys) is needed to improve targeting of beneficiaries.

    The large size of the public sector is undermining the competitiveness of the economy as a whole, by increasing costs, particularly wages. Government spending, averaging 35 percent of GDP during the past decade, is among the largest in Africa. The public sector outbids the private sector for available labor, exerting upward pressure on economy-wide labor costs, and contributes to high unit labor costs and unemployment. While government workers constitute 40 percent of the total formal workforce, they claim more than 54 percent of total wages paid. The reservation wage that workers demand before accepting employment has risen because workers expect salaries significantly higher than the market clearing levels in the nongovernment sectors of the economy.

    Botswana: Central Government Total Spending and Wage Bill

    Source: Botswana Ministry of Finance, Development, and Planning, and IMF staff estimates.

    Botswana: Central Government Employment and Wages

    Source: Botswana Ministry of Finance, Development, and Planning, and IMF staff estimates.

    APPENDIX 3 Public Debt Arithmetic

    The change in public debt can be expressed in the following form:

    Where I is interest payments; P is the primary surplus; and A is other items besides the budget deficit that affect indebtedness, for example, privatization receipts, devaluation losses, and issuance of bonds for recapitalizing banks.

    Equation (1) is useful in helping to identify the key determinants of the change in nominal debt. However, to facilitate an analysis of debt dynamics and the sustainability of debt, it is useful to rewrite (1) in terms of ratios to GDP. Dividing both sides of (1) by Y, the nominal GDP, and defining:

    where i is the nominal interest rate and g is the growth rate of nominal GDP, we obtain:

    Equation (3) can be rewritten as:

    where d = D/Y, p = P/Y, d-1 = D-1/Y-1, and a = A/Y

    On the basis of the debt dynamics in (4), the main contributors to large fiscal adjustments can be derived. Table 6 shows a decomposition of the main contributors to successful fiscal adjustments for advanced economies and emerging markets, subdivided into episodes with more or less than 10 percent inflation. What the evidence shows is that changes in the primary surpluses contributed the most in advanced economies to bring about the necessary fiscal adjustment, while the growth—interest rate differential was less relevant. This is less clear in episodes with high and low inflation in emerging markets, where increases in primary surpluses played a lesser role in the growth—interest rate differential.

    Table 6.Decomposition of Large Reductions in Debt-to-GDP Ratios in Advanced and Emerging Economies
    EpisodesStarting Debt RatioDebt ReductionEnding Debt RatioPrimary SurplusGrowth-Interest Rate DifferentialResidual
    Ireland (1987–2002)
    Denmark (1993–2008)
    Belgium (1993–2007)136.953.084.070.2-25.28.0
    New Zealand (1986–2001)71.641.829.852.1-8.9-1.4
    Canada (1996–2008)101.739.062.739.3-19.218.9
    Sweden (1996–2008)
    Iceland (1995–2005)58.933.625.417.44.711.4
    Netherlands (1993–2007)78.532.945.627.5-8.313.7
    Spain (1996–2007)67.431.436.121.611.5-1.7
    Norway (1979–1984)56.521.435.124.211.7-14.5
    Emerging Market Economies
    Inflation > = 10 percent a year
    Serbia (2001–2008)114.582.831.6-3.874.811.9
    Bulgaria (1996–2007)96.477.718.737.521.518.7
    Poland (1993–1998)84.347.736.73.350.6-6.3
    Turkey (2001–2007)77.638.139.429.712.0-3.6
    Hungary (1993–2001)88.736.552.222.637.0-23.0
    Chile (1989–1998)46.833.912.935.729.5-31.2
    Ecuador (1991–1997)88.726.961.810.724.0-7.8
    Sri Lanka (1989–1997)
    Romania (1999–2006)30.311.918.41.822.1-12.0
    India (1993–1998)79.59.470.1-8.219.5-2.0
    Inflation < 10 percent a year
    Egypt (1991–1997)93.560.333.218.540.31.5
    Paraguay (1989–1997)72.951.521.3-1.839.413.9
    Thailand (1986–1996)95.544.551.031.613.4-0.6
    Tunisia (1987–1992)90.944.546.413.930.30.3
    Indonesia (2000–2008)53.643.310.333.419.1-9.1
    Uruguay (2002–2008)100.941.759.231.620.0-9.9
    Georgia (1999–2007)102.139.462.6-3.544.7-1.7
    South Africa (1998–2008)57.234.722.513.934.0-13.1
    Jordan (2002–2008)73.629.044.612.110.46.5
    Panama (1990–1998)48.521.227.330.23.6-12.6
    Sources: IMF, World Economic Outlook database and IMF staff estimates.Notes: Figures are in percent of GDP. The episodes listed represent the largest year-to-year reductions in the debt-to-GDP ratio over the past three decades that were separated by at least 15 years. The interest rate used in the computation of the growth–interest rate differential is the “effective” interest rate, calculated as the ratio of government interest payments to the previous period’s ending debt stock. For emerging markets, known episodes of debt default, exchange, or rescheduling were dropped. The inflation rate cut-off of 10 percent refers to the average inflation rate prevailing during the episode.
    APPENDIX 4 Model

    The BGPZ model features an infinite-horizon small open economy consisting of households, two production sectors with firms producing traded and nontraded goods, a government, and a central bank.

    Households: There are two types of households.

    • a. A fraction of households can smooth consumption by varying their asset holdings of government bonds, money balances, and foreign assets.

    • b. The remaining share of households is liquidity constrained and can essentially consume only its current labor income.


    • c. Nontraded-goods sector: Firms operating in this sector face monopolistic competition and nominal price rigidities. Nontraded goods are produced using labor and private and public capital.

    • d. Traded-goods sector: firms operating in this sector face perfect competition and flexible prices. Traded goods are produced using labor and private and public capital. There is learning-by-doing in the traded-goods sector. Increases in activity will temporarily increase sectoral total factor productivity. This captures potential Dutch disease effects from capital inflows and the resulting real appreciation.

    The central bank: The central bank manages money growth in a way consistent with price stability and intervenes in the foreign exchange market to control the level of reserves in three ways: (i) it converts the value of SACU transfers in domestic currency; (ii) it buys and sells reserves in a way that is consistent with the exchange rate regime (for example, in the case of Swaziland, the central bank must buy and sell reserves to maintain the fixed exchange rate regime); and (iii) it buys and sells reserves in a way that is consistent with its long-run desired (‘optimal’) level of reserves.

    Government: The government spends on consumption and investment (of both traded and nontraded goods—ptggt

    ) and interest payments on government debt held by consumers ((it11)bt1cηπt)
    . It can finance this spending by taxing labor income (τtwtlt), using the domestic currency value of SACU transfers (stAt*)
    , drawing down its assets (drawing on deposits held at the central bank (dtgdt1gηπt)
    or issuing debt (btbt1ηπt)

    where ptg

    is the price of government consumption, gt is the basket of nontraded and traded goods consumed by the government, τt is the labor income tax rate, wtlt is labor income, st is the exchange rate, At*
    is the foreign currency value of SACU transfers, η is the gross rate of technological progress, πt is the gross inflation rate, bt is domestic debt, it is the interest rate on domestic debt and btc
    is domestic debt held by consumers. SACU transfers follow an AR(1) process:

    where 0 < ρA < 1 and tA

    is an exogenous decrease in SACU transfers at time t. Government spending can be used for consumption or investment. There are two types of public investment. xtgs=μsg
    is a fixed fraction of steady state government spending and xtgA=μA(gtg¯)
    is the fraction of public investment associated with SACU transfers. The scenarios considered in this paper are based on modifications of the government budget constraint.

    In Scenario (i), the labor income tax rate τ is fixed and a negative shock to SACU transfers (tA<0)

    must be offset by a reduction in government spending. The reduction in public capital can be shut down by setting μA = 0 when At falls so that the government must maintain the steady state level of investment in public capital after the shock and concentrate spending cuts on recurrent expenditures. In Scenario (ii), gt = ḡ is fixed at its steady state level, and the labor income tax τ must adjust to satisfy the government budget constraint. Finally, in Scenario (iii), the labor income tax only partially responds to the fall in SACU revenues:

    so that government spending must also offset the remaining shortfall necessary to satisfy the government budget constraint.

    APPENDIX 5 Calibration

    The model for Swaziland is calibrated based on the steady-state values described in Table 7 and the parameters in Table 8.

    Table 7.Steady-State Values
    National Income accounts (as a share of GDP)
    Traded sector37.0
    Nontraded sector56.5
    Private investment10.2
    Traded sector4.0
    Nontraded sector6.2
    Government Spending30.7
    Government consumption23.2
    Government investment7.5
    Government spending on traded goods9.2
    Government spending on nontraded goods21.5
    Trade Balance-3.0
    Value added in the nontraded sector50.2
    Value added in the domestic traded sector84.2
    Government accounts (as a share of GDP)
    SACU transfers20.3
    Interest payments1.0
    Government debt19.0
    Held by the central bank6.8
    Government deposits at the central bank0.6
    Central Bank Accounts
    Government debt held by the Central Bank3.2
    Government deposits at the Central Bank0.6
    Net Foreign Assets (Reserves)19.9
    Assets (as a share of GDP)
    Real money balances (Base money/Broad money)22.5
    Foreign assets held by the private sector0.1
    Government bonds held by the private sector0.0
    Annualized Inflation, nominal depreciation7.5
    Annualized Nominal interest rates9.5
    Annualized Real interest rates5.2
    Trend growth7.0
    Annualized real interest rate9.1
    Annualized short-term intreste raate17.3
    Sources: Swaziland authorities and IMF staff estimates.
    Table 8.Parameter Calibration
    Discount rate for the rule of thumb consumers0
    Labor share in nontraded sector0.7
    Labor share in traded sector0.7
    Public capital share in nontraded sector3
    Public capital share in traded sector3
    Depreciation rate (nontradable sector)0.015
    Depreciation rate (tradable)0.015
    Depreciation rate of public capital0.02
    Efficiency of public sector0.4
    Efficiency of SACU-financed investment0.4
    Investment adjustment costs, nontraded sector25
    Investment adjustment costs, traded sector25
    Learning-by-doing (LBD) externality parameter0.1
    Productivity in the traded sector at the steady state1
    Persistence of LBD externality0.1
    Inverse of the Frisch labor supply elasticity1.5
    Interest semielasticity of money demand8.5
    Measure of optimizing consumers0.666666
    Elasticity of substitution between traded and nontraded goods0.89
    Elasticity of substitution between varieties12
    Elasticity of substitution between the two types of labor1
    Degree of capital mobility500
    Persistence of SACU transfers0.9
    Average size of SACU adjustment (in percent of steady state GDP)
    over quarter of adjustment-5
    Number of quarters of adjustment40
    Spending response to the aid surge. Preferred calibration = 11
    Persistence of real deposit accumulation0.9
    Persistence of real debt accumulation0
    Persistence of tax increases (ρT)0.2
    Responsiveness of tax increases to changes in SACU transfers (yT)-0.1
    Sale of dollars from the aid surge.1
    Weight of exchange rate target on reserves100000
    Persistence of reserve accumulation0.9
    Inflation targeting coefficient from implicit interest rate rule1.5
    Degree of sterilization0 or 0.5
    Sources: Swaziland authorities and IMF staff estimates.

    This paper benefited from very useful comments and technical assistance on the simulations from Susan Yang and Felipe Zanna from the IMF Research Department. Useful comments were also provided by Thomas Baunsgaard, Andy Berg, Fabian Bornhorst, Hamid Davoodi, Jack Grigg, Mark Horton, and Vitaliy Kramarenko. We are grateful to Tara Iyer and Xin Li for excellent research assistance and to Karen Coyne and Breda Robertson for editorial assistance. The authors are also grateful to Ahmed Ndyeshobola, the Namibian authorities, and the SACU secretariat for very useful comments on an earlier draft of the paper. All errors are those of the authors.

    Appendix 1 provides a description of the history of SACU and the corresponding revenue-sharing formula.

    Excises were not affected as much. Given that South Africa gets almost 98 percent of the excise component of the revenue-sharing formula, the impact on South African revenue transfers has been significantly smaller.

    The fiscal year in each of the SACU countries starts on April 1.

    See, for example, Tsibouris et al. (2006) and IMF (2010a).

    Tsibouris et al. (2006) define as large those fiscal adjustments that meet either of two criteria: (i) a continuous improvement of the primary balance of the consolidated central government in excess of 6.3 percent of GDP, or (ii) a reduction in the initial size of government expenditure larger than 21.8 percent. These large cases are therefore highly relevant for the BLNS countries.

    See, for example, Cottarelli et al. (2009).

    See, for example, Mongardini and Samake, 2009.

    Scenarios (i) and (iii) require a modification of the original BGPZ model.

    Note that the steady state is invariant to the policy experiments considered. The adjustment path will differ depending on the policy scenario but all paths will lead to the same steady state.

    Swaziland is a member of the Common Monetary Area with Lesotho, Namibia and South Africa.

    For example, the tax rate could rise so much that government spending could increase following the fall in SACU transfers.

    The model, however, analyzes only the effects of changes in labor income tax rate. Analyzing the effects of consumption or profit tax rates is left for future research.

    For instance, to guide fiscal policy during the medium-term expenditure framework—MTEF (2010/11—2012/13), the Namibian authorities adopted the following fiscal rules: (i) a public debt-to-GDP ratio of 25—30 percent annually; (ii) an average budget deficit of 5 percent of GDP; (iii) public expenditure levels below 30 percent of GDP; (iv) interest payments capped at 10 percent of government revenue; and (v) contingent liabilities below 10 percent of GDP annually.

    See, for example, Gaomab and Hartman (2006), Grynberg (n.d.), and Senatla, Chankuluba, and Chepete (2010) on the history of SACU.

    In practice, the amounts distributed are net of administration costs from running the SACU Secretariat, the Tariff Board, and an ad hoc tribunal set to settle disputes arising from the application of the agreement.

    As noted in Appendix 1, the members’ shares in the development component are adjusted for differences in GDP per capita, that is, adjusted by a factor of 10 to control the bias in favor of the poorer member countries and ensure revenue stability for all parties.

    Source: Botswana: Staff Report for the 2010 Article IV Consultation.

      You are not logged in and do not have access to this content. Please login or, to subscribe to IMF eLibrary, please click here

      Other Resources Citing This Publication