Botswana
Selected Issues and Statistical Appendix

This Selected Issues paper and Statistical Appendix on Botswana underlies that diamond reserves are not adequate to generate enough permanent revenue to support the current level of expenditure. Despite strong overall growth, in Botswana, a pattern of dependence on diamond revenue and high unemployment persists. Botswana, as a typical small open economy, is closely linked to a large neighboring economy. This linkage means Botswana’s monetary and exchange policies must consider the external economic environment, particularly the pula’s exchange rate against the rand.

Abstract

This Selected Issues paper and Statistical Appendix on Botswana underlies that diamond reserves are not adequate to generate enough permanent revenue to support the current level of expenditure. Despite strong overall growth, in Botswana, a pattern of dependence on diamond revenue and high unemployment persists. Botswana, as a typical small open economy, is closely linked to a large neighboring economy. This linkage means Botswana’s monetary and exchange policies must consider the external economic environment, particularly the pula’s exchange rate against the rand.

I. Are Diamonds Forever? Using a Lifecycle Approach to Analyze Botswana’s Reliance on Diamond Revenues1

A. Introduction

1. Diamond exports, Botswana’s main source of foreign exchange, accounted for an average of 75 percent of total annual exports over the past 10 years. Fiscal revenues also depend heavily on diamonds, 2 which account for 95 percent of minerals revenues. Over the same period, minerals revenues accounted for 63 percent of tax revenues. Botswana is expected to remain the world’s leading producer of diamonds well beyond the next decade.

2. This study assesses the fiscal sustainability of Botswana’s diamond sector using an analysis designed for oil-exporting countries (see Davoodi 2002, for a general presentation, and Barnett and Ossowski 2003 for an application to oil-producing countries). It applies the permanent income hypothesis (Friedman, 1957) to diamond-related fiscal revenues and the lifecycle approach (Modigliani and Brumberg, 1954) to different expenditure options.

3. This study does not address the optimality of the fiscal policy, since it is not derived from the maximization of an intertemporal utility function. Instead, Botswana’s current fiscal rule, which imposes a strict balance between expenditure and revenue, is considered as a given constraint. As a result, government expenditure, over the medium and long term, is constrained by fiscal revenue generated by the economy. Any level of expenditure permanently above fiscal revenue would be unsustainable.

4. Botswana’s economy faces a difficult challenge with the depletion of its diamond resources, which are expected to be exhausted around 2030 (Section B). Under the current fiscal rule, and without any additional saving effort, the depletion of diamond resources would imply a sharp fall in government expenditure. A reduction in expenditure of 1 point of GDP per year would generate savings high enough to smooth the adjustment imposed by the depletion of diamond resources. The main specific messages are as follows.

  • Strong fiscal adjustment may be necessary to ensure that diamond revenues benefit future generations. Revenue-raising measures would help preserve expenditure, either in level (Section D) or per capita (Section E).

  • The results are sensitive to the uncertainty regarding diamond prices and resources. One additional year of above-ground extraction with a high extraction rate would significantly increase Botswana’s capacity to smooth the adjustment (Section E).

  • If the economy fails to diversify and create additional sources of growth before the full depletion of diamond resources, then the contraction of fiscal revenue would be far greater, hence putting additional constraint on expenditure (Section E).

5. The rest of this paper is organized as follows. Section B presents an overview of Botswana’s diamond sector. Section C introduces a simple model of permanent income, which is applied in section D to evaluate different policy scenarios (e.g., saving to soften future adjustment), while alternative assumptions (e.g., diamond stock, prices, and GDP growth) are explored in Section E. Section F discusses some issues indirectly related to the analysis.

B. The Diamond Sector in Botswana

6. Botswana is one of sub-Saharan Africa’s few upper middle-income countries. The diamond industry has been a major contributor to Botswana’s growth. Over the past 10 years, the mining sector represented an average of 34.5 percent of GDP, with diamonds constituting nearly 94 percent of the sector’s total exports (see Table 1). The mining sector was also a main factor contributor to fluctuations in the GDP growth rate over the past decade (see Figure 1).

Table I.1.

Share of Diamond and Mining Industries in GDP 1995–2005

article image
Source: Botswana authorities and Fund staff computation.
Figure I. 1.
Figure I. 1.

Mining and Nonmining Sector Contribution to GDP Growth,1994/95–2004/05

Citation: IMF Staff Country Reports 2007, 228; 10.5089/9781451806465.002.A001

Sources: Botswana authorities and Fund staff computations.

7. Over the medium term, Botswana is expected to enjoy sustained growth of around 4 percent a year. Meanwhile, diamond production will continue at high levels until 2021, when a sharp decline will set in until diamond resources are depleted around 2030. The end of diamond production, which now accounts for one-third of GDP and half of government revenues, will severely affect real GDP growth (projected at 3.5 percent in the long run). However, Botswana’s efforts to build diamond processing capacity, for domestic as well as foreign mines, could yield additional revenues and help offset diamond production losses.

8. Because of the rapid depletion of diamond resources, fiscal revenues are expected to shrink by about two-thirds from 2021 to 2029. Still, diamond extraction will remain be the main contributor to mineral revenue. Despite expected new mine production (i.e., copper, nickel, and gold), other-mineral profits will likely be relatively modest.

9. When its new plant opens, Debswana, Botswana’s joint venture with De Beers, will increase its diamond extraction by 35 percent, though income is expected to grow by a more modest 12–15 percent. Using more modern techniques, the new plant will extract more, though smaller (and thus less profitable), diamonds from Debswana’s mine tailings. The plant will also be able to extract more diamonds from the site’s large stocks of waste.

10. Nevertheless, production is expected to decrease after 2017, as diamond reserves are drawn down, and then to decrease further as mining goes underground. Sometime between 2015 and 2020, when surface mines are expected to grow fallow, Debswana will need to shift to underground mining, resulting in lower profits (owing to higher extraction costs and decreased output). The profitability and economic feasibility of underground mining are open to question. Though there is some uncertainty about the stock of diamonds, most analysts expect reserves to be depleted by 2029.

C. The Structure of the Model

11. This paper draws on the permanent income hypothesis (PIH) of consumption, an approach widely applied to countries with significant natural resources (oil, gas, and minerals). The steps are as follows: (i) the total diamond resource is estimated; (ii) the net present value (NPV) of this resource is taken to be a financial asset generating a permanent income; and (iii) the income figure is then used to determine how much the government can spend without eroding its long-term fiscal position.

12. The dynamic structure of the approach is based on the intertemporal budget constraint, which can be written in period t as follows:

TtD+TtO+itAt1Gt=AtAt1,(1)

where TD is diamond-related revenue; TO is other (tax and nontax) revenue; it is the interest on net government financial assets, At; and G is government consumption.

13. Let WD be the NPV of diamond revenues. At date t, it is defined as

itWt1DTtD=WtDWt1D,or:WtD=Σj=t+1TTtDπτ=t+1j(1+iτ),(2)

where T is the period beyond which diamond resources are totally exhausted. Equation 2 accurately defines W,D which is inherently forward looking and based on both expected interest rates and expected revenues (see Figure 2). The results therefore depend not only on the structure of the model but also on the uncertainty surrounding the measurement of future variables.

Figure 1. 2:
Figure 1. 2:

Time Structure of the Model

Citation: IMF Staff Country Reports 2007, 228; 10.5089/9781451806465.002.A001

14. The total financial wealth of the government is used as an annuity, which is translated into a permanent and sustained flow of public expenditure. Let WO be the net financial asset generated by nondiamond revenues:

WtO=Σj=t+1+TjOΠτ=t+1j(1+iτ).(3)

The budget constraint of the government imposes a balanced budget over the long run. The intertemporal budget constraint can be derived from equation 1:

At=Σj=t+1+(TjO+TjDGj)Πτ=t+1j(1+iτ).(4)

That is, the net financial asset at period t should be equal to future net savings (NPV of future net revenue minus consumption). Assuming, for simplicity, that At=0, it follows that

Σj=t+1+GjΠτ=t+1j(1+iτ)=WtD+WtO.(5)

That is, the NPV of future consumption is equal to the total financial wealth at the government’s disposal.

If we now turn to the PIH, assuming that consumption grows at a constant rate g:τt+1Gτ=(1+g)τt1Gt+1. With a constant interest i, the result is:

WtD+WtO=Σj=t+1+(11+i)jtGj=Gt+1Σj=t+1+(1+g)jt1(1+i)jt=Gt+1Σj=t+1+(1+g1+i)jt,(6)=11+gGt+1(1+g1g)Gt+1=(ig)(WtD+WtO)

and

τt+1Gτ=(ig)(1+g)τt1(WtO+WtD).(7)

15. The total wealth generated by diamond revenues can also be used as quasipermanent income to generate income for a certain time horizon, T’. Let YD be this revenue flow:

Σj=t+1T(11+i)jtYjD=WtD.(8)

Assuming that this flow is used to generate quasipermanent income that grows at a rate of g, it follows that:

Yt+1D=(ig)(1(1+g1+i)Tt)1WtD,(8)

and

Tτt+1YτD=(1+g)τt1(ig)(1(1+g1+i)Tt)1WtD.(10)

Equation 7 would then be rewritten as follows:

{Tτt+1Gτ=(ig)(1+g)τt1TtO+YtD.τ>TCτ=(ig)(1+g)τt1TtO(11)

Thus, the quasipermanent income approach consists in concentrating the annuities generated by diamond revenues in a finite period, which induces a higher revenue for this period but necessitates an adjustment beyond it. Section D shows how this approach might be relevant to Botswana, where diamond resources may not provide permanent income high enough to both ensure fiscal sustainability and meet the authorities’ development objectives.

D. Application of the Model to Botswana

16. Diamond revenue’s projections results from projections of diamond production and prices (Figure 3 and 4). The profile for diamond production and prices summarized in Figure I. 3 is a stylized representation of information gathered from Debswana and the Ministry of Minerals and Water Resources. 3 The price trend is a result of three assumptions: (i) a constant price in real terms of about US$100 a carat, (ii) a nominal 2 percent inflation rate, and (iii) a discounted average price for 2012–18 to reflect the reprocessing of waste using newer extraction technologies. Production increases gradually, from 32 million carats in 2005 to 44 million carats in 2017, mostly reflecting diamond production from reprocessed mine waste. Them from 2017 production will fall as resources accesible from above-ground resources are depleted, until 2021 where another decline will occur as underground resources are much less rich and more difficult to extract, making mining less productive.

Figure I. 3.
Figure I. 3.

Projected Diamond Production and Nominal Prices 2005–29

Citation: IMF Staff Country Reports 2007, 228; 10.5089/9781451806465.002.A001

Source: Fund staff computations based on government information.
Figure I. 4.
Figure I. 4.

Diamond Revenues (million dollars) 2005–29.

Citation: IMF Staff Country Reports 2007, 228; 10.5089/9781451806465.002.A001

Source: Fund staff computations based on government information.

17. Diamond resources are not adequate to permanently finance a significant share of government expenditure. Applying the PIH with an infinite time horizon, projected diamond revenues represent only 1 percent of GDP, compared to 19.5 percent for nondiamond revenues. Using the assumptions summarized in Table 2, and assuming, for simplicity, that interest rates are constant, the NPV of diamond revenues in 2005 would be WD = 158 billion pula. Once converted into a permanent income growing at the same rate of GDP, this sum leads to a long-run annuity of 1 percent of GDP, compared with the 2005 values of 19.8 percent of GDP for nondiamond fiscal revenues and 19.5 percent of GDP for diamond revenues.

Table I.2.

Changes in Diamond Prices and Production

article image

Percentage change from baseline.

Percentage of GDP change from baseline. The long-term value refers to the constant annuity applying once diamond resources are exhausted.

Assumes that production in 2022 is the same as in 2021 and that production from 2022 to 2030 decreases at the same rate as that between 2021 and 2029 in the baseline scenario.

18. When applying the model, the main assumptions made for the long term are as follows.

  • In the long run, all real variables, notably the GDP grow at the same rate of 4.5 percent a year.

  • Purchasing power parity holds in relative terms. World inflation is 2 percent, while domestic inflation is 4.5 percent.

  • As of 2005, the population is 1.7 million and grows at a constant rate of 2 percent a year.

19. The authorities might thus consider (i) smoothing the path of diamond revenues over the medium term, (ii) continuing to diversify the economy, and (iii) creating new sources of income to help finance Botswana’s development needs. As implied by the PIH exercise, unlike some oil-producing countries, Botswana cannot rely on its diamond resources to finance long-term expenditure. The exercise also suggests that, given the expected short-term horizon of remaining diamond resources, a strategy of revenue saving must be implemented before (or in addition to) other revenue-generating measures.

20. Under a quasipermanent income approach, an alternative to the infinite horizon PIH, diamond resources could generate income of 2.7 percent of GDP in 2021–49,4 while preserving much higher revenue through 2021 (see Figure 4). The revised assumptions are as follows:

  • The time horizon is shortened to 2049, leaving 20 years of annuities once diamond production ends. In addition, along with a gradual partial adjustment in savings, long-term reforms would ensure that public expenditures are fully sustainable.

  • Instead of assuming that the current NPV of diamond stocks is directly transformed into a financial annuity growing at a constant rate, the authorities are assumed to gradually increase their saving of diamond revenues. The first years of revenue mostly finance expenditures, though some savings could be used to finance an annuity until a certain date (2049 is proposed here), after which further adjustment would be needed.

  • More gradual adjustment in savings would limit the impact on development and poverty-related expenditures. The pure PIH approach, by contrast, implies a sharp drop in revenues in the first few years.

  • The quasipermanent income approach implies a three-phase adjustment: (i) during the diamond extraction period, savings are increased gradually, though by much less than in the PIH; (ii) during a protracted period (between 2021 and 2049 here), these savings generate a higher annuity (2.7 percent of GDP) than in the PIH (1 percent of GDP), owing to the finite time horizon; and (iii) beyond 2050, additional adjustment would be needed, including structural and other measures to support growth and create more public spending resources.

21. The adjustment needed for a soft landing requires a sharp reduction in spending as a share of GDP. Of the adjustment options open to the authorities, the ideal would: (i) avoid cutting expenditure too steeply in the short term, which could be disruptive; and (ii) cut enough to ensure a soft landing. The simple approach adopted here evaluates two policies—reducing expenditure by 1.0 percent of GDP a year and by 0.5 percent a year. As shown in Figure I. 6, an adjustment of 0.5 percent per year, a considerable amount, would allow the economy to sustain high government expenditure over a longer period, though deficits would accumulate even before diamond resources were exhausted, creating the need for greater adjustment later on. Thus, in the absence of offsetting revenues from economic diversification, only an adjustment of 1 percent a year or more would likely ease the impact of diamond depletion.

Figure I.5.
Figure I.5.

Permanent and Quasipermanent Income Approaches, 2005–2100

Citation: IMF Staff Country Reports 2007, 228; 10.5089/9781451806465.002.A001

Source: Botswana authorities and Fund staff computations.
Figure I. 6.
Figure I. 6.

Two Alternative Quasipermanent Income Approaches, 2005–49

Citation: IMF Staff Country Reports 2007, 228; 10.5089/9781451806465.002.A001

Source: Botswana authorities and Fund staff computations.

E. Sensitivity Analysis

22. Because Botswana is a price maker in the diamond market, lower production is likely to lead to higher prices, hence offering a partial compensation for a lower production. Because a sharp drop in Botswana’s production would imply a major structural change in the diamond market, and because of a lack of data, it is almost impossible to quantify Botswana’s market power. However, being in a situation of a quasi-monopoly it would definitely be possible to exploit this situation and to extend the production period, hence increasing the profit per carat, in order to maximize the net present value of diamond-related fiscal revenues. Bostswana’s market power might also vanish as production decline, because the production of other countries could increase and impose greater competition.

23. Because we cannot precisely quantify Botswana’s market power, we present a partial analysis showing, everything being equal, where a 1 percent price increase, during the period of diamond depletion, induces a increase of 0.1 percent of the NPV of diamond revenues (see Table 3). The impact on the long-term value of the annuity is rather modest, with an increase of 0.01 percentage point of the GDP. While the absolute impact, particularly on the annuity, is small, the precise relationship between world diamond prices and Botswana’s production needs further investigation.

24. With one more year of high-return above-ground extraction (production of about 32 million carats), the NPV of diamond revenues would increase by 3.8 percent and the long-term annuity by 1.1 percent of GDP (see Table 3). As noted earlier, the exact magnitude of diamond resources has not been fully assessed yet. Although both Debswana and the Ministry of Mineral and Water Resources expect production to end around 2029, above-ground extraction could last another year or more. An additional year of above-ground extraction would put production at a much higher level, say, the same as in 2021 (32 million carats), compared with 12.8 million carats in the baseline scenario. In addition, the indirect impact of the authorities’ saving effort (owing to the cap on government expenditure) would boost revenue. Indeed, in 2022 the saving rate of diamond revenue would be 24.9 percent, compared with 2.5 percent in the baseline scenario.

25. Reducing government expenditure by 1 percent of GDP a year would also imply reducing real government expenditure per capita, which might have a major impact on the economy. Balancing expenditures and revenues is a necessary condition for fiscal sustainability and Botswana’s fiscal law imposes a strictly balanced budget with no debt accumulation. If revenues as a percentage of GDP, are falling, the increased savings will carry a social cost—a net reduction in expenditure per capita—unless GDP growth increases sharply. Figure I. 7 shows the patterns implied by (i) the baseline scenario, which features no adjustment before diamond production declines, and (ii) the scenario assuming a gradual decline in public expenditure of 1 percent of GDP a year. Not surprisingly, the decreased public expenditure leads to a sharper decrease in expenditure per capita. However, greater savings leads government expenditure per capita to recover more quickly during the post-diamond period. The minimum expenditure per capita is also higher in this scenario.

Figure I. 7.
Figure I. 7.

Real Government Expenditures per Capita Index (100=2005), 2005–49

Citation: IMF Staff Country Reports 2007, 228; 10.5089/9781451806465.002.A001

26. Unless the economy is diversified, projected GDP growth is projected to fall sharply after 2021 (by –24.2 percent in 2022, followed by several years of stagnation). All the scenarios proposed so far assume for total GDP a constant long-term growth of 3.5 percent a year, even with the sharp decline in diamond production. Thus, the baseline projection assumes that greater diversification of the economy would lead to a strong nondiamond-realted GDP growth, which fully compensates for the diamond-related GDP loss, while the alternative scenario assumes that the GDP loss would not be offset in any way. Potential growth, however, is likely to be somewhere in between, since diversification would likely only partly compensate for the GDP loss. For instance, diversification could come partly from diamond processing, since Botswana has invested in this activity, but it would suffer as well from the depletion of diamond resources hence partly compensating for the loss. Figure I. 8, which summarizes both scenarios, shows that the GDP contraction would almost double the cost of the end of diamond production, assuming expenditure per capita in 2021 holds throughout the period. Thus, in this scenario, annual government expenditure per capita would be only 48.7 percent of its 2005 level during the 2022–32 period, before gradually recovering.

Figure I. 8.
Figure I. 8.

Impact of a Recession Induced by the End of Diamond Production, 2005–49

Citation: IMF Staff Country Reports 2007, 228; 10.5089/9781451806465.002.A001

27. With a gradual adjustment in expenditure, the authorities could save additional resources, which would considerably lower the long-term cost of the GDP loss. Additional savings, even without economic diversification, would generate enough quasipermanent income to compensate for the GDP loss. For example, expenditure per capita in 2022–32 would be 55.2 percent of the 2005 level, compared with 48.7 percent (see Figure 8).

28. Another way to smooth the per capita expenditure would be to impose increase savings during the intial years of the transition (Figure 10). With an stronger adjustment in expenditure, namely a decrease by 2 percent of GDP until 2013, Botswana would be able to accumulate greater savings, which would.then be used after 2013 to avoid a too sharp decrease in per capita expenditure. However, since the short term effort would be a lot higher, it would require particularly strong political and social support in order to be implemented.

Figure I. 9.
Figure I. 9.

Preparing for Recession by Saving More, 2005–49

Citation: IMF Staff Country Reports 2007, 228; 10.5089/9781451806465.002.A001

Figure I. 10.
Figure I. 10.

Alternative adjustment, 2005–49

Citation: IMF Staff Country Reports 2007, 228; 10.5089/9781451806465.002.A001

29. For simplicity, the precise link between diamond and nondiamond sectors and/or revenues are not explicitly modeled in this study. The purpose of this analysis is just to provide some benchmarks on the impact of no compensation for diamond production loss. However, if the economy fails to diversify, the fall of diamond production could spread to the rest of the economy, hence leading to a sharper recession. This would put the economy further below than the scenario with no compensation for GDP loss provided here, which reinforces the need for early adjustment and further diversifying the economy.

F. Lessons from the Simulation Exercises and Sensitivity Analysis

30. Diamond reserves are not adequate to generate enough permanent revenue to support the current level of expenditure. Under the current fiscal rule, i.e. no accumulation of debt, higher savings will be necessary in order to avoid a too sharp adjustment over the medium term. However several points should be underlined.

  • To be viable, higher savings in the short and medium term require a strong political consensus, given that the current economic situation does not require such an adjustment. Thus the viability of any adjustment will be judged by the ability of the authorities to gather a large political and social support for it, which could possibly imply further constraints to keep pro-poor expenditures per capita very stable in real terms.

  • Economic diversification implies further attracting foreign investors, so as to improve the growth potential further. Moreover, new partnerships with the private sector could be considered for infrastructure financing.

  • Budget contingency planning might be necessary to accommodate uncertainty about the actual stock of diamonds.

  • The nondiamond fiscal deficit demands continued scrutiny. For example, if the nondiamond deficit widens, a tighter fiscal policy would be appropriate even if diamond-related revenue is high, precisely because of the economy cannot rely on diamond resources over the long term. Moreover, the nondiamond deficit might be associated with lower long-term growth (see Sachs and Warner, 2001, for a literature review of countries dependent on hydrocarbon resources).

  • Finally, the nondiamond fiscal stance might be of better use for assessing the soundness of fiscal policy relative to the business cycle On the expenditure side, a greater focus on productivity would alleviate additional pressures on fiscal policy.

31. In conclusion, it should be also noted that the HIV/AIDS epidemic is likely to affect both the growth potential of the economy (by lowering it) and government expenditure (which would need to be increased). Although a specific analysis of these linkages is beyond the scope of this study, it reinforces the need for initiating the adjustment rapidly, since the epidemic could only complicate further the adjustment.

References

  • Barnett S., and R. Ossowski, 2003, “Operational Aspects of Fiscal Policy in Oil-Producing Countries,” in Fiscal Policy Formulation and Implementation in Oil-Producing Countries, ed. by J. Davis, R. Ossowski, and A. Fedelino (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Davoodi, H. R., 2002, “Assessing Fiscal Vulnerability, Fiscal Sustainability, and Fiscal Stance in a Natural Resource-Rich Economy,” IMF Country Report No. 02/64: 7–31 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Friedman, M., 1957, A Theory of Consumption Function (Princeton, N.J.: Princeton University Press).

  • Modigliani, F., and R. Brumberg, 1954, “Utility Analysis and the Consumption Function: An Interpretation of Cross-Section Data,” Post-Keynesian Economics ed. by Kenneth K. Kurihara (New Brunswick, N.J.: Rutgers University Press).

    • Search Google Scholar
    • Export Citation
  • Sachs, J. D., and A. Warner, 2001, “The Curse of Natural Resources,” European Economic Review, Vol. 45, pp. 827– 38.

1

Prepared by Olivier Basdevant.

2

Botswana’s diamond production is composed for half of industrial diamonds and half of jewelry diamonds. The latter makes about 80 percent of Diamond revenues.

3

To protect the confidentiality of information, the actual data were smoothed and stylized, although more precise calculations were discussed with the authorities.

4

The time horizon in the quasipermanent income approach is fixed arbitrarily to 2049, i.e. a time-horizon rather long. The main point here is to propose an adjustment in two steps, where 2049 represents this second step, i.e. the end of the diamond-related annuity, which is far enough in the future to prepare an adjustment over a time horizon longer than the one of diamond depletion.

Botswana: Selected Issues and Statistical Appendix
Author: International Monetary Fund