Chapter 3: Southern African Customs Union Revenue Volatility: Roots and Options for Mitigation

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
  • ShareShare
Show Summary Details
Alfredo Cuevas, Lars Engstrom, Vitaliy Kramarenko and Geneviève Verdier 

Botswana, Lesotho, Namibia, and Swaziland (BLNS) receive significant government revenues in transfers from the Southern African Customs Union (SACU). As a percentage of GDP and total revenues, these transfers were very large and rising during 2007–09. In Lesotho and Swaziland, SACU transfers exceeded a third and a quarter of GDP, respectively, in 2008/09 (Figure 3.1).1 The magnitude of these transfers makes public finances in BLNS highly dependent on their evolution. The very high volatility of SACU transfers significantly complicates BLNS’s public financial management.2 In contrast, SACU transfers to South Africa (about 1 percent of GDP) are not nearly as important as they are for BLNS.

Figure 3.1BLNS: Total Fiscal Revenue and Grants (Percentage of GDP)

Sources: IMF data and staff calculations.

A major factor behind the volatility of SACU transfers is the composition of the union’s common revenue pool, which consists largely of revenue from customs duties and, to a lesser degree, excise taxes (Box 3.1). Volatile customs duties are the main revenue source for the common pool. It is well known that trade tends to move procyclically and to display wider swings than output, making the ratio of customs duties to total output highly variable.3 In fact, the recent global recession was characterized by the severity of the contraction in trade that accompanied it. In SACU’s case, revenue pool volatility is amplified further by the high proportion of customs duties accounted for by duties on imported motor vehicles, which tend to be even more procyclical than other imports.

SACU transfers also suffer from an additional, technical source of volatility, arising from the existence of what, for lack of a better term, will be called the “T + 2 adjustment mechanism.” Transfers from the SACU common revenue pool at year T are made based on a forecast of revenue collections made at year T – 1.

If the transfer to a SACU member is larger (smaller) than it is entitled to receive by treaty in a given fiscal year (FY) T, the excess (shortfall) will be reversed two years later (T + 2) through an “adjustment” to the SACU transfer. In practice, this mechanism amplifies the variance of SACU transfers beyond the variance of the underlying common revenue over the medium term, even if it makes payments more predictable in the short term.

This chapter presents different options for reducing the volatility of SACU transfers from the common revenue pool. The motivation is to identify the significant benefits to mitigating and managing volatility at the origin. Ultimately, each country receiving distributions from SACU must be responsible for its own fiscal decisions, but reducing the volatility of payments at the origin would greatly facilitate fiscal management for each member country. Chapter 4 shows how fiscal reforms in BLNS can reduce dependence on SACU transfers, including through an appropriate fiscal adjustment and the possible use of fiscal rules.

Box 3.1Current Revenue Sharing within SACU

All customs and excise revenues collected in SACU are to be paid into a common revenue pool (administered through the South African Revenue Fund). The revenues are then distributed to each member according to a specific revenue-sharing formula. The current distribution formula was agreed to in 2002 and became operational in FY 2005/06. Payments to members for the next fiscal year are usually determined in late December, based on the following principles:

  • The forecast values of customs and excise revenues of the union for the next fiscal year (T).
  • Forecast customs revenues are distributed based on shares in intra-SACU trade at FY T − 2. Countries that import proportionally more 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.
  • Monies equating to 85 percent of forecast excise revenues are distributed based on the shares of each country in SACU GDP at FY T − 2. The remaining 15 percent of forecast excise revenues, the Development Component, is distributed according to a progressive formula favoring countries with lower per capita GDP at FY T − 2.
  • From the various forecasts, a distribution budget is drawn up and agreed upon by the SACU Council of Ministers. Revenues are distributed by South Africa, which acts as the pool administrator, in quarterly installments over the course of FY T, irrespective of what the actual common revenue pool turns out to be during that year.
  • Revenue pool forecast errors related to the distributions at FY T are corrected in FY T + 2 through repayments to the pool or member states proportionately to their original shares for FY T − 2.

The section that follows discusses possible solutions for mitigating the impact of the “T + 2 adjustment mechanism.” One of the technically easiest options is to extend the period for correcting over- and underpayments to the member states. A collective approach to managing SACU revenue volatility is explored in the third section. There is significant merit in decomposing the stream of revenue collections into structural and cyclical components and then distributing only the structural component of the common revenue pool each year. The fourth section looks briefly at some current ideas, including moving toward closer economic integration and using the destination principle for distributing SACU revenues. Closer economic integration would open new options for managing SACU-related revenues and fiscal policy more generally—for example, by making it possible to rely on less-volatile revenue sources, eliminating fiscal frontiers, and collectively managing the spending of (some of) the common revenues. This would be, in the best of scenarios, a long-term proposition. Supporters of the destination principle argue that the common revenue pool could be abolished, and customs revenues could be distributed to the countries of final use of the imported goods originating outside SACU. However, this option would lead to stricter fiscal frontiers, undermining intraregional trade and destabilizing member countries.

Addressing Volatility Stemming from the T + 2 Adjustment Mechanism

When a BLNS country receives from SACU more (less) than it is entitled to receive in a given fiscal year, the excess (shortfall) is reversed two years later through a special adjustment. This practice increases the variance of SACU transfers. Possible options to eliminate or mitigate such excess volatility will be discussed.

In the words of the SACU Agreement itself (Annex 1, paragraph (g)):

Where revenue forecasts for year (t) are used to calculate the size of the customs component to be distributed over the course of year (t), adjustments will be made in years (t+1) and (t+2) to account for differences between the forecast and actual revenue collected.

Similar clauses apply to the excise and development components of SACU revenue as well. In practice, “differences between the forecast and actual revenue collected” cannot be corrected in year T + 1 because of lags in the availability of relevant information (the outcome of year T is known after year T + 1 has begun). Before T + 1 starts, a careful observer can form an idea of the approximate size of the adjustment that will be needed; but any legally valid correction has to wait for final information on T, that is, it cannot happen before T + 2.

To understand the effects of T + 2 adjustments, it is useful to consider payments over the period 2007/08 to 2010/11 (Table 3.1):

Table 3.1SACU Transfers to BLNS, 2007/08–2010/11
(Millions of rand)
Forecast share19,47924,42121,20819,479
Actual adjustment5,2374,5006,707−4,488
SACU payments24,71628,92127,91514,991
Sources: South Africa Budget Review (2011); and IMF staff estimates.Note: BLNS = Botswana, Lesotho, Namibia, and Swaziland; SACU = the Southern African Customs Union.
Sources: South Africa Budget Review (2011); and IMF staff estimates.Note: BLNS = Botswana, Lesotho, Namibia, and Swaziland; SACU = the Southern African Customs Union.
  • In FY2009/10, the total distribution to BLNS was nearly 28 billion rand (R), which was the sum of R21.2 billion, corresponding to the expected BLNS shares in the FY2009/10 revenue pool (as forecast in December 2008), and an adjustment of R6.7 billion, corresponding to the amount needed to correct an underpayment dating back to FY2007/08, itself caused by a positive surprise to the revenue pool that year.
  • In FY2010/11, the total distribution to BLNS declined by 46 percent to R15 billion, which was the sum of R19.5 billion, corresponding to the expected BLNS shares in the FY2010/11 revenue pool (as forecast in December 2009), and a negative adjustment of R4.5 billion, representing the amount needed to correct an overpayment that took place in FY2008/09.

The goal of the current SACU revenue-sharing formula was to provide certainty to BLNS about the payment for the upcoming fiscal year, but it amplified the revenue drop in FY2010/11. Serial correlation in revenues means that shocks in revenue data are normally followed by similar “aftershocks.” For example, lower-than-forecast revenue collections in FY2008/09 (negative shock) meant that lower revenue was forecast for 2010/11 (negative aftershock). However, because of the T + 2 adjustment mechanism, the negative shock was delayed and materialized at the same time as its own aftershock in FY2010/11. This coincidence in time of a shock and its own aftershock causes the variance of SACU transfers to rise above the original variance of the underlying revenues. The variance of SACU transfers under the T + 2 adjustment mechanism is estimated to be 38 percent higher than the variance of the actual revenues that BLNS are entitled to receive.4

In theory, the T + 2 adjustment mechanism gives the BLNS countries time to prepare for the shock. In practice, some of the BLNS countries, particularly the smaller countries, have had difficulties managing the volatility of SACU payments. Their ability to manage budgets is made significantly more difficult by very large swings in revenue. Even if the government can see the swings coming, it may be difficult to resist pressures to increase permanent commitments on the upswing, and even harder to cut back spending when SACU revenue drops.

Different options, with their own specific advantages and disadvantages, are available for dealing with the excess volatility problem:

  • Let the forecast be final. Any under- or overpayments arising from differences between expected and realized revenues would be deemed bygones. This option eliminates short-term uncertainty and overall volatility would be limited to the volatility of the revenue forecasts, which is lower than the volatility of revenues themselves. However, this option would create incentives for countries to manipulate the inputs that go into revenue forecasts, and thereby create an atmosphere in which even technically sound forecasts may be suspected of manipulation.
  • Pay as you collect. A simple option that can be implemented without much complication. The obvious cost is the loss of certainty of the amount that will be received the upcoming year. Adopting this option would imply accepting short-term uncertainty for the sake of reducing overall variance over the medium term.
  • Spread adjustment over more years. SACU members could modify the current system by requiring that adjustments for the under- or overpayments experienced in year T be spread over a number of years. The higher the number of annual installments, the lower is the variance in distributions that can be achieved. However, this option has an implicit cost. Unremunerated credit would be constantly, implicitly extended and extinguished. If the forecast undershoots the true revenue, the loan is from BLNS to South Africa; if the true revenue falls short of the forecast, South Africa lends to BLNS. In other words, an adjustment spread over many years also represents a commitment to “lend” or “borrow” the excesses and the shortfalls in receipts for many years, as opposed to the current adjustment, in which the implicit loan is paid back in two years. Therefore, a very long adjustment period may not be acceptable in practice—at least not without the introduction of proper terms for these implicit loans, such as explicit interest charges (which should be relatively easy to handle in practice).

Mitigating SACU Revenue Pool Volatility: A Cooperative Approach

The previous section dealt with the volatility-amplifying effects of a technical aspect of the revenue-sharing mechanism, but the more fundamental problem of dealing with the volatility of the revenue pool itself would remain. As noted earlier, significant benefits could accrue from managing the volatility of the revenue pool collectively rather than at the level of individual countries.

To deal with the pronounced procyclicality of the common revenue pool, SACU member countries could decide to distribute each year only the structural component of the common revenue pool. When revenues are above potential, the member countries would save the cyclical component of revenues in a stabilization fund. When revenues are below potential, the countries could cover cyclical shortfalls by making withdrawals from the stabilization fund. In practical terms, at year T – 1, the member countries would need to reach consensus on the size of the structural component of the revenue pool to be distributed the following year (T). This system would have no need for further adjustments in future years. The agreed-on distributions should also take into account the minimum level of assets in the stabilization fund needed for precautionary reasons; thus, during an initial accumulation phase distributions would need to be lower than the estimated structural component of revenue.

This approach raises two fundamental questions. First, what are the advantages of the collective approach over management of SACU revenue volatility at a country level? And second, what methodology should be used for estimating the structural component of the common revenue pool?

Pros and Cons of a Collective Approach

Key advantages of the collective approach include the following:

  • Commitment to countercyclical fiscal policy would be more credible (although not necessarily guaranteed). This advantage would be particularly important for countries that need to further develop their fiscal institutions.
  • Fiscal policy could be better coordinated, which could be beneficial, especially for the four SACU members belonging to the Common Monetary Area.5
  • Member countries, especially those needing to strengthen their institutions, would be less likely to find themselves in extreme situations in which they would need external financial support.
  • A larger precautionary fund may earn a higher gross rate of return with lower management costs, benefiting all member countries. Also, pooling the cyclical component of common revenues would allow for joint use of scarce management skills and would avoid duplication of country-level management efforts.

The key disadvantages of the collective approach to managing volatility of SACU revenues include:

  • a certain loss of sovereignty over the use of the cyclical component of the pool and
  • difficulties in reaching consensus on the structural component of the pool and the optimal size of the precautionary deposit.

Whether the advantages of the collective approach to managing volatility outweigh the disadvantages is ultimately for policymakers to determine.

Simulation of the Structural Component

Numerous approaches can be used to estimate the structural component of the common revenue pool. One option, used in this chapter, is to determine the structural component of the pool using a Hodrick-Prescott filter in combination with an autoregressive moving average forecast based on the information available at the time of the forecast.6

This approach is admittedly simplistic, but the volatility of SACU transfers would have been significantly lower if the countries had distributed the structural component of the SACU common revenue pool derived through the recursive Hodrick-Prescott filter (Figure 3.2). The standard deviation of the quarterly structural distribution to BLNS as a share of GDP would have been 25 percent of the standard deviation of the quarterly actual cash payments to BLNS during 2007–10. Other more sophisticated or elaborate modeling strategies could be adopted and the member countries could potentially invite independent experts to help them produce estimates of structural revenues. A more rigorous approach to modeling should improve on the performance of the type of rule discussed here.

Figure 3.2Simulations of SACU Revenue Payments, 2007–10 (Percentage of GDP)

Sources: National Treasury of South Africa; and IMF staff estimates.

Note: HP = Hodrick-Prescott; SACU = Southern African Customs Union.

But would the proposed scheme have been feasible? This depends on the starting balance of the stabilization fund. If the stabilization fund had started operations in 2007/08 with a balance of R10 billion, and enjoyed a 6 percent nominal return on assets, the fund would not have run out of money during the period of the analysis, which includes the 2009 recession and ends in the third quarter of 2011. This hypothetical performance highlights the need for the members to consider an initial accumulation period during which they would receive less than the full structural component of the pool, to save and permit the stabilization fund to attain the size necessary to withstand large cyclical shocks. It also underscores the need for conservative management of the portfolio of assets in the stabilization fund to ensure value preservation and liquidity. A detailed discussion of the trade-offs that would be faced in setting up and growing the stabilization fund goes beyond the scope of this chapter.

Finally, in addition to the structural distribution rule, the member states could agree to review import shares every three to five years (rather than annually), which would also contribute to more stable revenue flows, in particular to the smaller member states.

Options Beyond the Current Revenue-Sharing Framework

The discussion so far has centered on ways to reduce volatility within the current revenue-sharing framework. However, the framework is currently under review by the SACU member countries, and several concrete ideas for its modification have been floated. This section reviews two such ideas.

Toward an Economic Union

The collective approach to managing SACU revenue discussed in the previous section could be a stepping stone toward closer integration. In fact, the SACU member countries have proclaimed an ambitious long-term objective of moving toward an economic union (see SACU, 2010) whose institutions could further promote countercyclical policies and quicker convergence among member states. Closer integration could be achieved by exploring options for less volatile pooled revenue sources, eliminating tax-related border formalities (fiscal frontiers), and taking different approaches to spending the common revenue pool. The latter could include non-earmarked transfers to member states, earmarked grants similar to European Union (EU) structural funds (Box 3.2), or investment in regional infrastructure conveying collective benefits.

The various distributional aspects will need to be resolved by member states. Mechanisms that will ensure that transfers to member countries do not fluctuate significantly despite the inevitable volatility of the revenue pool would also be important. The union could, for example, seek to have a structurally balanced budget.

Allocation of Customs Revenues to Countries of Final Use of Imported Goods

Some observers and economists in the region have proposed moving to a destination principle in distributing customs revenues. They argue that the common revenue pool could potentially be abolished, and customs revenues could be distributed to the countries of final use of the imported goods originating outside SACU. Customs revenues could be collected in either of two possible ways. First, the country of first entry of the imported good collects revenues to be remitted to the country of final use after deduction of a collection fee. Or second, fiscal frontiers are reinforced, and customs revenues are collected directly by the country of final use.

Box 3.2European Union Structural Funds

The EU comprises 27 member countries, with 271 regions, forming an economic community and single market of about 500 million citizens. The structural funds were set up to reduce regional disparities in income, wealth, and opportunity. Europe’s poorer regions receive most of the support, but all European regions are eligible for funding under the policy’s various funds and programs. The structural funds comprise the European Regional Development Fund (ERDF), the European Social Fund (ESF), and the Cohesion Fund.

The structural funds are targeted to specific regions whose per capita GDP is less than 75 percent of the EU average and to member states with a per capita gross national income of less than 90 percent of the EU average. However, other regions can also qualify for projects that aim to improve competitiveness and employment.

The Community Strategic Guidelines—a framework approved by the European Council (heads of state or government of the EU member states)—establish all actions that must be taken to gain access to funds. Within this framework, each member state develops its own National Strategic Reference Framework (NSFR) and Operational Programs for each region within the member state.

The European Commission negotiates and approves the NSFR and the Operational Programs proposed by the member states and uses these as a basis for allocating budget resources. The member states and their regions manage the programs by selecting individual projects, and controlling and assessing them. The European Commission is responsible for overall program monitoring, pays out approved expenditures, and verifies the national control systems.

The EU’s annual budget is projected to be 1.1 percent of member countries’ gross national income (GNI) over 2007–13. Annual spending on structural funds amounts to 0.5 percent of GNI. The funding of the EU budget has gradually shifted from customs levies and collection of value added tax to direct payments by the member states based on their GNI.

Adopting revenue-sharing based on the country of final use would be destabilizing for BLNS. First, BLNS would need to build the capacity to collect customs duties at their intra-union frontiers or at least a system of tracking goods in transit would need to be established. These changes would require sufficiently long transition arrangements. Second, assuming that institutional and transition challenges are addressed, revenue losses for BLNS would be large because import duties collected on extra-SACU imports based on the destination principle are likely to be significantly lower than the average level of distributions under the current system;7 further losses would follow from efforts to undertake massive fiscal adjustments and from stricter fiscal frontiers. Such an adjustment may destabilize poorer countries, creating significant humanitarian challenges for the region. Third, the erection of stricter fiscal frontiers between members would increase the cost of doing business and discourage intra-union trade, inflicting significant costs on private sector participants in both larger and smaller member countries. The empirical study presented in Chapter 2 finds that all SACU members have benefited from the trade creation induced by SACU; reform of the revenue-sharing formula should, therefore, support greater integration and avoid introducing new barriers to trade.


SACU members are rightly interested in finding ways to reduce the volatility of distributions under the union’s revenue-sharing arrangement. This chapter has explored various options:

  • In the near term, a straightforward option would be to reduce the additional volatility by modifying the existing “T + 2 adjustment mechanism.” Under- and overpayments could be settled over a longer period than the current two years. This option would be easy to implement within existing administrative and analytical structures; in fact, it could be implemented immediately, if member countries so wished. An important feature of this option is that it would maintain certainty about the amount of the upcoming payments to BLNS and still reduce the volatility of payments over time. Alternatively, SACU members could revert to a pay-as-you-collect system (simple, but reintroduces short-term uncertainty) or use forecasts to determine final payments (difficult to implement because it would create incentives to manipulate the forecasts and thus undermine the credibility of any forecasts, however sound).
  • Over the near to medium term, a more fundamental option would be to separate the structural and cyclical components of the SACU common revenue pool, and distribute only the structural component every year. To make this operational, it would be necessary to set up a stabilization fund, into which excess (positive) cyclical revenues would be deposited in good times and saved to cover future shortfalls created by low cyclical revenues in bad times. The challenges are not insurmountable, but implementation would require consensus building and preparatory work to design and establish the necessary institutions and mechanisms.
  • Over the long term, a move toward deeper integration might open up new ways to manage both the volatility of SACU transfers and the use of these resources in a broader sense, reinforcing countercyclical principles of fiscal policy. However, the SACU member countries first need to consider and reach consensus on difficult political questions related to the ownership and distribution of union revenues and the sovereignty of member countries. In this context it is also important to note that trade liberalization is likely to erode revenue collections over the long term. Therefore, SACU members would be wise to prepare themselves by reducing their reliance on SACU transfers and developing domestic sources of tax revenue.
  • Moving to the destination principle in allocating customs duties, in particular if implemented quickly, would undermine intraregional trade, destabilize public finances of BLNS, and may increase the risk of major humanitarian problems in the region.

The fiscal year begins April 1.


For a comprehensive discussion of revenue sharing within SACU, see Mongardini and others (2011).


In addition, some volatility is attributable to changes in effective duty rates. For example, duties were waived on imports related to World Cup infrastructure investment, leading to a significant shortfall in SACU revenue.


The analysis of the volatility mechanisms is presented in Cuevas and others (forthcoming).


Lesotho, Namibia, South Africa, and Swaziland.


Technical details of estimates of structural and cyclical components are provided in Cuevas and others (forthcoming).


Given existing shortcomings of trade statistics, in particular on re-exports, it is difficult to estimate potential direct revenue losses from a hypothetical move to the destination principle. Based on official statistics on trade with non-SACU countries and assuming an average tariff of about 20 percent, estimates of possible direct revenue losses for BLNS vary from about 6 percent of GDP to more than 20 percent of GDP, depending on the country.

    Other Resources Citing This Publication