A. Uncertainty and Volatility in Natural Resource Revenues—Analytical Considerations¹

Two major questions for policymakers in natural resource-rich developing countries (RRDCs) are (i) how much of resource revenues to consume and how to save (invest) the remainder, and (ii) how to cope with the uncertainty and volatility of resource revenues. This chapter offers some analytical background on the second question. Many resource-dependent developing countries must deal with very volatile exports, revenues, and non-resource GDP growth.² In the short to medium term, designing policies to cope with that uncertainty and volatility may be a more pressing challenge than dealing with the issue of exhaustibility, especially when there is a relatively long extraction horizon. Uncertainty may relate to how large the resource reserves are, how much will be extracted in a given period, what prices will be on average, and how volatile prices are likely to be in the short term.³ These factors complicate macroeconomic management and stimulate demand for precautionary mechanisms.

1. Large swings in commodity prices deeply complicate resource and macroeconomic management. Commodity price swings can be large, long-lasting, and asymmetric. Shocks to prices for several commodities may take several years to dissipate, and for commodities like gold, natural gas, and oil they may be permanent (Box 1). It is therefore hard to forecast prices with a reasonable degree of confidence, even over the medium term. These factors also complicate the task of policy makers who wish to assess whether a shock is permanent (warranting adjustment) or temporary (warranting smoothing).

2. There can also be substantial uncertainty and volatility related to the volume of commodity production and exports. Production might be disrupted by technical difficulties, accidents, strikes, social and political unrest, and cross-border disputes. Production forecasts may prove too optimistic because of delays in investment or for economic reasons (e.g., drop in international demand, substitution of other commodities). These factors may call for a relatively conservative approach to projecting production volumes. However, in several countries there is also substantial upside risk to production, especially over the longer term. While it is hard to obtain estimates for numerous commodities, with new discoveries and better extraction technologies proven reserves for a number of commodities have been revised upward over time.⁴ For example, proven oil reserves in sub-Saharan Africa more than tripled between 1990 and 2010, to nearly 70 billion barrels (Figure 1). Moreover, high prices have made extraction in difficult-to-reach areas economically viable. Since estimates of reserves and extraction may be projected with varying degrees of confidence, there may be merit in using a probabilistic approach to get a sense of possible extraction paths and thus the robustness of baseline projections.

View Full Size

Changes in Proven Oil and Gas Reserves¹

Proven reserves of oil and gas have steadily been revised upward with new discoveries and better extraction technologies.

Citation: Policy Papers 2012, 071; 10.5089/9781498340007.007.A001

Source. BP Statistical Review of World Energy, June 2011.¹Excludes Canadian oil sands.

• Download Figure
• Download figure as PowerPoint slide

3. Revenue volatility creates a motive to save some of the resource revenues for precautionary reasons. Prudent policymakers who wish to cope with sharp unwanted and unforeseen swings in resource envelopes may wish to build up a liquidity fund to smooth consumption spending (see Figure 2).⁵ Saving for precautionary (prudential) reasons is conceptually different from other motives, such as intergenerational risk aversion (which aims to spread consumption spending over time because of lumpiness in the timing of revenue) and temporary parking of revenue to minimize absorptive capacity disruptions.⁶ Conceptually, the optimal size of the liquidity buffer will be larger if revenue volatility is higher and more persistent, and if society dislikes consumption swings more (i.e., is more prudent). Also, the optimal liquidity buffer will be larger when consumption using resource revenue is higher. Consider two countries with similar resource inflows, one where the inflows are short-lived and the other where they are long-lasting. Absent any uncertainty, and other things being equal, the country with short-lived resource inflows optimally will consume relatively little, saving more, and the other will consume relatively much, saving less, of the resource inflows, all for intertemporal smoothing reasons. When there is uncertainty, however, consumption is likely to be more variable in the country with the long-lasting resource windfall, and optimally it will choose to build up a relatively large liquidity buffer.⁷

View Full Size

The Precautionary Savings Motive Illustrated

Uncertainty about future resource flows may call for cautious spending plans and create a precautionary savings motive.

Incremental Consumption and Saving from an Uncertain Resource Windfall (Stylized example).

Citation: Policy Papers 2012, 071; 10.5089/9781498340007.007.A001

• Download Figure
• Download figure as PowerPoint slide

4. Self-insurance for precautionary reasons may be costly, especially in developing countries, which suggests that any impact from volatility needs to be managed rather than totally eliminated. A number of authors have derived orders of magnitude of precautionary balances for oil producers where persistence in oil prices implies need for a large precautionary buffer. Van der Ploeg (2010) calculates that 3.75-7.5 percent of an oil windfall would need to be saved to cope with unexpected oil price volatility within a single quarter. When the horizon is extended to four years, the precautionary motive suggests that 30-60 percent of an oil windfall be saved. Cherif and Hasanov (2012a) also find sizable precautionary saving by oil exporters, on the order of 30 percent of income.⁸ Finally, Bems and Carvalho (2011), using a sample of oil exporters to assess the importance of precautionary motives for the current account, found that the precautionary motive added about 1 percent of GDP (US$36 billion for the sample) to the current account balance in 2007, with substantial country variation depending on the degree of resource dependency.

5. In practice, policy makers may wish to build liquidity buffers based on what they consider a tolerable degree of confidence that the buffer will be adequate if there is a tail-risk shock. A value-at-risk (VaR) or DSGE model-based approach could be used to derive the buffer. These simulation methods make it possible to calculate the required liquidity buffer to absorb a tail-end adverse price shock across a spectrum of price paths over a period of time. Bartsch (2006) assessed a number of fiscal rules for Nigeria, with spending envelopes based on 3-5-year average oil prices, and applied a VaR approach to derive the size required for the liquidity fund. He found that a fund of US$16-18 billion (75 percent of 2004 revenue) would be tolerable, in the sense of having more than 80 percent confidence that it would not be exhausted within five years. A similar exercise was conducted for the Republic of Congo (see Chapter III.A). With an 8-year moving average oil price rule (5 historical years and 3 years of forward projection), it was found that with a high degree of confidence a stabilization buffer of about 48 percent of non-oil GDP would be needed to avoid depletion within three years. The DSGE model-based approach has a richer economic structure. An application to Angola found that under an aggressive public investment scaling up scenario, a stabilization buffer can still be built, but with a 30 percent probability that the fund is inadequate to prevent future spending cuts (see Chapter III.B and Box 3 in the main paper).

Box 3.

Domestic Revenue Effort in Resource Exporters

We explore the possible effect of resource abundance on non-resource fiscal revenue in resource-rich SSA countries following Bornhorst, Gupta, and Thornton (2009), which focused on oil exporters; we expand the analysis to all types of natural resources. A first look at the data for these countries yields a negative and significant correlation between resource and non-resource fiscal revenue.

View Full Size

Resource-Rich Countries in SSA: Fiscal Revenue Excluding Grants, 2000-11

(Average,in percentof GDP)

Citation: Policy Papers 2012, 071; 10.5089/9781498340007.007.A001

Source: IMF staff estimates.

• Download Figure
• Download figure as PowerPoint slide

To control for other factors, we estimate a panel regression of the form

$(\frac{{\displaystyle NRR}}{{\displaystyle Y}}{)}_{it}={\alpha }_i+\beta (\frac{{\displaystyle RR}}{{\displaystyle Y}}{)}_{it}+control{s}_{it}+u_{it}$

where $(\frac{{\displaystyle NRR}}{{\displaystyle Y}}{)}_{it}$ and $(\frac{{\displaystyle RR}}{{\displaystyle Y}}{)}_{it}$ denote the ratios of non-resource and resource revenue to GDP, respectively, for country i at time t. As controls we use external financing, measured by the grants-to-GDP ratio; income (measured by the log of real per capita GDP); openness to international trade (exports plus imports to GDP); the share of agriculture to GDP; the share of urban to total population; and a corruption index (ICRG). In our regression we include all SSA countries from 2000 to 2011 and allow for country-specific (fixed) effects (a detailed description of the data used in this exercise can be found in IMF (2012)).

Resource abundance crowds out other sources of revenue after controlling for different factors. As shown in the Table 1 (columns 1 and 2), the coefficient associated with resource revenue is negative and significant. In particular, the fixed-effects contemporaneous estimation (column 1) implies that every additional point of GDP of resource revenue is associated with lower non-resource revenue amounting to 0.12 percentage points of GDP. The coefficient increases to —0.07 if we include the lag of the dependent variable on the right-hand side (column 2), which itself is positive and significant, consistent with the finding elsewhere that revenue ratios are persistent over time. This implies that in two years, the cumulative impact would be —0.11. Interestingly, when including the lags of both the resource and non-resource revenue to GDP ratios as regressors (column 4), only the latter is positive and significant. This suggests that the effect of additional resource revenue on collection of non-resource revenue is relatively short-lived.

Table 1.

Regression Results^1,2

Source: IMF staff estimates.

¹S. E. in parentheses

²***, **, and * indicate 99, 95, and 90 percent levels of confidence respectively.

Table 1.

Regression Results^1,2

Dependent Variable: Non-Resource Revenue to GDP
	(1)	(2)	(3)	(4)
Constant	-0.149***	-0.075***	-0.173***	-0.028
	(0.022)	(0.017)	(0.025)	(0.019)
Resource Revenue to GDP	-0.119***	-0.070***
	(0.033)	(0.006)
Resource Revenue to GDP (lag)			-0.078**	0.038
			(0.034)	(0.028)
Non-Resource Revenue to GDP (lag)		0.535***		0.581***
		(0.040)		(0.050)
Grants	-0.019**	-0.024**	-0.021**	-0.024**
	(0.010)	(0.009)	(0.011)	(0.009)
Real Per Capita GDP	0.020***	0.011***	0.017***	0. 001
	(0.004)	(0.004)	(0.005)	(0.006)
Trade Openness	0.048**	0.094***	0.138***	0.083**
	(0.020)	(0.035)	(0.026)	(0.037)
Share of Agriculture	-0.078**	-0.039	-0.049	-0.024
	(0.034)	(0.030)	(0.031)	(0.031)
Share of Urban Population	0.006***	0.003***	0.007***	0.003***
	(0.000)	(0.001)	(0.001)	(0.001)
Corruption	0.018***	0.012***	0.017***	0.013***
	(0.003)	(0.003)	(0.003)	(0.003)
	Weighted
R-squared	0.96	0.98	0.97	0.98
Adjusted R-squared	0.96	0.98	0.96	0.97
Sum Squared Resid	0.17	0.09	0.14	0.09
	Unweighted
R-squared	0.94	0.97	0.95	0.97
Sum Squared Resid	0.19	0.09	0.15	0.09

Source: IMF staff estimates.

¹S. E. in parentheses

²***, **, and * indicate 99, 95, and 90 percent levels of confidence respectively.

View Table

Table 1.

Regression Results^1,2

Dependent Variable: Non-Resource Revenue to GDP
	(1)	(2)	(3)	(4)
Constant	-0.149***	-0.075***	-0.173***	-0.028
	(0.022)	(0.017)	(0.025)	(0.019)
Resource Revenue to GDP	-0.119***	-0.070***
	(0.033)	(0.006)
Resource Revenue to GDP (lag)			-0.078**	0.038
			(0.034)	(0.028)
Non-Resource Revenue to GDP (lag)		0.535***		0.581***
		(0.040)		(0.050)
Grants	-0.019**	-0.024**	-0.021**	-0.024**
	(0.010)	(0.009)	(0.011)	(0.009)
Real Per Capita GDP	0.020***	0.011***	0.017***	0. 001
	(0.004)	(0.004)	(0.005)	(0.006)
Trade Openness	0.048**	0.094***	0.138***	0.083**
	(0.020)	(0.035)	(0.026)	(0.037)
Share of Agriculture	-0.078**	-0.039	-0.049	-0.024
	(0.034)	(0.030)	(0.031)	(0.031)
Share of Urban Population	0.006***	0.003***	0.007***	0.003***
	(0.000)	(0.001)	(0.001)	(0.001)
Corruption	0.018***	0.012***	0.017***	0.013***
	(0.003)	(0.003)	(0.003)	(0.003)
	Weighted
R-squared	0.96	0.98	0.97	0.98
Adjusted R-squared	0.96	0.98	0.96	0.97
Sum Squared Resid	0.17	0.09	0.14	0.09
	Unweighted
R-squared	0.94	0.97	0.95	0.97
Sum Squared Resid	0.19	0.09	0.15	0.09

Source: IMF staff estimates.

¹S. E. in parentheses

²***, **, and * indicate 99, 95, and 90 percent levels of confidence respectively.

The signs of other coefficients are generally consistent with previous studies. A larger inflow of foreign financing (grants) in proportion to GDP has a negative effect on non-resource-related revenue mobilization (this result is robust to the specification). The coefficient of —0.02 implies that an additional 10 percent of GDP in resource revenue reduces grant funding by 0.2 percent.

Interestingly, the development of the economy as a whole (proxied by real per capita GDP) is positively and significantly associated with higher revenue (though it becomes insignificant when lags of the revenue ratios are included, column 4). Trade openness and the share of urban population each have a positive and significant effect on non-resource revenue, whereas the share of agriculture is significant only for the baseline regression (column 1). Overall, this suggests that less developed countries tend to exhibit lower revenue mobilization. By construction, the measure for corruption (ICRG) increases as corruption diminishes, hence the positive sign, and is robust to all specifications (we used an indicator of rule of law as an alternative for the ICRG, which rendered insignificant in specifications 2 and 4).

6. Developing countries might also consider using market-based instruments to manage commodity price volatility. Countries can enter into over-the-counter forward contracts to lock in prices or hedge price risk with options.⁹ While several countries have used market-based hedging instruments—Mexico, for example, has bought put options on oil to insure against a fall in prices— they are probably still under-used by developing countries, including commodity exporters. A number of factors could explain why such instruments are not used more often; for instance, these contracts can be technically complex, costly, hard to communicate to stakeholders, and politically risky.¹⁰

7. RRDCs also need to take into account uncertainty about the return on investment. While capital scarcity calls for ramping up spending for domestic investment, taking into account absorptive capacity constraints, the additional savings required for precautionary purposes will slow the scaling up of investment. In reality, the return on investment is risky, and could even be negative sometimes. Where there is risk aversion, uncertain returns on investments will slow the speed at which investment spending is ramped up; instead, countries will want to invest relatively more in safe assets or to repay debt faster.¹¹

8. When there is both uncertainty and volatility, the approach to natural resource management should be holistic. Policy makers will need to carefully balance consumption spending and savings when resource revenue flows are uncertain and volatile. Spending can be safeguarded by setting aside a liquidity buffer as a precaution. Additional savings can be used to pay down debt, ramp up domestic investment spending, or invest in external financial assets to benefit future generations (for example, when absorptive capacity constraints make it impossible to invest faster).

B. Fiscal Analysis for Resource-Rich Developing Countries¹²

Fiscal Sustainability Analysis

This section provides a fiscal sustainability framework for countries with non-renewable natural resources, adapting some simple identities to the realities of resource rich countries.

9. The overall fiscal balance (OB_t) in year t can be decomposed into resource revenues (RT_t) and non-resource revenues (NRT_t), primary expenditure (E_t), income from the initial stock of financial assets (A_t-1), and interest payments on the initial stock of liabilities (D_t-1):

where $i_t^a$ and $i_t^d$ are the net interest rates associated with the stock of assets and liabilities, respectively. The overall fiscal balance is also equal to the change in the net financial assets:

The non-resource primary balance can then be defined as:

10. Resource-rich countries often run overall fiscal surpluses, which can facilitate the accumulation of substantial financial assets over time, but the NRPB is often in deficit. The law of motion of net financial assets (A_t — D_t) is given by:

11. As in other countries, the government’s inter-temporal budget constraint requires that the initial stock of net debt equals the present value of the cumulative future primary balances. For countries with exhaustible natural resources, this condition can be disaggregated into: the (i) NRPB and (ii) the resource revenue (accruing only for a fixed period of time, N, which is a function of the reserve horizon). Assuming for simplicity that all assets and liabilities are discounted at the same constant rate i, after imposing the no-Ponzi game condition,¹³ this can be expressed as:

12. Equation 5 describes the inter-temporal budget constraint consistent with fiscal sustainability. It states that the stabilization of net wealth is a function of the present discounted value of future NRPB and the present discounted value of future resource revenues. Net wealth is stabilized when the present discounted value of future resource revenues is equal to the present discounted value of future non-primary balance deficits (or future non-primary balances, with a negative sign).

13. The assets the government holds in the form of natural resources can be viewed financially as the present value of the future path of resource revenue (the “resource wealth”). Thus, the net wealth (W_t-1) of the government at the end of period t — 1 is given by the net financial assets accumulated by the end of period t—1 (A_t-1 - D_t-1), plus the present value of the natural resources asset in the ground (V_t-1). Hence, the following identity:

14. From this framework, the change in the government’s wealth in nominal terms is determined by the NRPB and the net return on the previous period’s wealth. This can be decomposed into the returns on financial assets and on the natural resources in the ground, minus the interest payment on debt:

15. The change in government wealth can also be presented as a share of nominal non-resource GDP, Y_t, which can be defined as Y_t = Y_t-1(1 + γ_t). Nominal GDP growth can be decomposed into baseline growth ${\gamma }_t^{base}$ plus a potential growth premium in response to frontloaded investment ${\gamma }_t^{premium}$, such that:

16. Equation [9] is useful to consider the impact of higher investment on growth and the effect on fiscal sustainability (see below).

The associated ratio of non-resource primary balance to non-resource GDP is defined as:

17. where nrpb is the NRPB simply expressed as a percent of non-resource GDP. All remaining variables can also be redefined as ratios to non-resource GDP. Hence, the change in government wealth is determined by the NRPB (relative to non-resource GDP) and the growth-adjusted return on net assets. Interest and growth rates can be stated in nominal (i, γ) or real (r, g) terms. Assuming a constant non-resource growth rate gives¹⁴:

18. The intertemporal budget constraint can therefore be restated in percent of non-resource GDP:

The PIH-Based Rule

19. There are many alternative paths for the NRPB that are fiscally sustainable, i.e., satisfy the no-Ponzi game condition.¹⁵ A simple, but restrictive, path is the benchmark provided by the permanent income hypothesis (PIH), which assumes constant net wealth starting immediately (in real terms, or as a percent of non-resource GDP, or in real per capita terms).

20. To be sustained for an infinitely long period, the annual level of the primary balance should be equal to the return on net wealth, adjusting for inflation; the notional real return on wealth is the real interest rate $(\tilde{r}=\frac{{\displaystyle i-\pi }}{{\displaystyle 1+\pi }})$, where π is the constant long-term inflation rate.

21. The following rule is therefore consistent with keeping the real NRPB constant: $(NRP{B}_t=-\tilde{r}{W}_{t-1})$, where NRPB_t and W_t-1 are now expressed in real terms. Alternative (more restrictive) benchmarks could be to keep real spending constant per capita or constant as a share of non-resource GDP.

The Modified PIH (MPIH) with Scaling up of Capital Spending

The PIH framework can be modified to accommodate scaling up of public investment.

22. The modified PIH assumes that government front-loads investment spending above the baseline forecasts by $I_t^{\prime }$ until the last year of investment front-loading, year F (e.g., F = 2018). The additional front-loaded capital spending could be financed by “saving” less natural resource revenue during the scaling up period. In this case, the accumulation of financial assets (A_t) would be lower during the scaling up period than in the baseline. The higher capital spending would also directly reduce the NRPB during the front-loading years, in line with the scaling up period by the amount of $I_t^{\prime }$.

23. Two additional assumptions characterize this simple fiscal framework. First, in order to assess the government’s intertemporal spending choice, the potential growth premium in response to the front-loaded investment is assumed to be equal to zero, i.e., ${\gamma }_t^{premium}=0$. Second, over the long run (e.g., year T), the level of financial wealth from this front-loaded investment scenario $W_T^{\prime }$has to be equal to the level from the usual PIH fiscal framework:

24. These two assumptions together imply that the front-loaded investment has to be fully compensated by a fiscal adjustment in the medium term (spread over T-F years). This adjustment would directly increase the NRPB and generate an extra accumulation of financial wealth $A_T^{\prime }$ during the period F to T Hence, the level of financial wealth after year T would be the same for the two alternative fiscal paths.

25. In other words, the future value of the total front-loaded investment spending (and related decrease in the NRPB) in year T has to be equal to the future value of the total fiscal adjustment (increase in the NRPB in the same year). In terms of financial wealth (W), this can be represented by:

26. The MPIH approach provides an ex ante measure of possible future fiscal adjustment needs if the scaling up of investment does not have an impact on growth. It does not imply that policy makers explicitly believe that scaling up will not have an impact on growth. Rather, it assumes that given the uncertainty of higher investment on growth, policy makers should not incorporate this impact ex ante in their medium-term plans. It therefore provides a future fiscal adjustment path in a worst case scenario where higher public investment has no impact on growth and hence provides a measure of the potential implications for future fiscal adjustment (see simulation below).

Fiscal Sustainability Framework (FSF) for Resource-Rich Countries

27. This approach allows for a stabilization of net wealth at lower levels than with the PIH or MPIH framework. Higher investment is assumed to have a positive impact on growth, ${\gamma }_t^{premium}>0$, which generates higher non-resource revenues $\left({\tau }_t^{\prime }\right)$, but also leads to an increase in operation and maintenance expenditures (σ). Both variables affect the NRPB_t, but the key feature is that this framework allows for a fiscally sustainable level of financial wealth that is lower than in the PIH or MPIH. The specific stabilizing target for net wealth (and the time horizon by which it is to be achieved) is country-specific since it involves estimating the interactions between government spending needs and non-resource growth.¹⁶ As a percent of non-oil GDP, wealth evolves before being stabilized according to the law of motion described by equation 14. Once wealth is finally stabilized at year F, the associated constant non-resource primary balance is given by

28. Hence, in the FSF, the following effects are to be expected:

• An increase in public investment will have a positive impact on non-resource GDP growth through γ^premium > 0. The higher non-resource GDP growth rate will lower the growth-adjusted asset return. Using equation 15, and assuming no debt and positive a and v, if (i - γ) > 0, then nrpb* <0. On the other hand, if (i — γ) <0, then nrpb* >0. In all cases where investment increases growth, the wealth-stabilizing NRPB will increase for a given nominal or real interest rate.

• Non-resource tax collections will also increase if growth is higher. The fiscal impact will depend on the non-resource tax ratio/effort, NRT_t = τY_t-1(1 + γ^base + γ^premium). For a given size of the net wealth-stabilizing NRPB, higher growth will allow for a higher primary expenditure path.

• Higher investment is likely to be associated with higher operation and maintenance expenditures (σ), worsening the NRPB_t.

29. Two important differences between the MPIH and the FSF are therefore worth noting:

• The FSF allows for a positive impact of investment on growth γ^premium > 0, but also on tax collection, and on operation and maintenance spending (σ).

• In contrast with the MPIH, the FSF wealth is stabilized at a lower level after the front-loaded investment, $\left(W_F^{\prime }<W_F\right)$. This new level, $W_F^{\prime }$, is computed using the usual PIH framework (present value of remaining natural resources assets) after F periods of front-loaded investment, $(V_F={\mathrm{\Sigma }}_{s=F+1}^N\frac{{\displaystyle R{T}_s}}{{\displaystyle (1+i{)}^{s-F}}})$.

• Since financial wealth is allowed to decrease in this framework, the implicit adjustment path of fiscal variables can be smoother than with either the PIH or the MPIH framework. The intuition behind this framework is that instead of accumulating higher financial savings, the country has accumulated higher physical assets that also provide a fiscal return.

Practical Considerations in Selecting Fiscal Anchors

30. This section analyzes some of the key issues discussed in this section from a practitioner’s point of view. It uses a stylized example of a resource-rich country and simulates the effects of alternative fiscal rules. These simulations are generated with an Excel-based template developed by the Fiscal Affairs Department (FAD) of the IMF in consultation with other departments. Using a series of data inputs, this template generates simulations of fiscal policy dynamics. It is designed to help IMF teams and country authorities analyze the tradeoffs associated with alternative fiscal rules. Visualizing these trade-offs in practice and assessing the implications of alternative assumptions and scenarios can help inform policymakers about the best fiscal anchors given country-specific circumstances. These templates are still in an experimental phase and will evolve further during the second phase of this project.

31. The stylized example uses an RRDC with a non-oil sector growing at a constant annual rate of 3 percent in real terms and contributing about 20 percent of non-oil GDP to government revenues. The oil revenue share accruing to the government is assumed to be a constant 50 percent. Inflation is constant at 5 percent a year, while the average real rate of return on financial assets is assumed to also be 5 percent. For the sake of simplicity, the country has no outstanding liabilities when the framework/rule is initiated.

32. The exercise has two parts. The first analyzes long-term sustainability, assuming that the resource horizon is 35 years and that production of natural resources rises steadily from 2012 to 2029. The objective of this part is to simulate three alternative ways of computing fiscal sustainability benchmarks:

• Traditional PIH, where the NRPB remains constant over time and is financed with the rate of return on accumulated financial assets plus the net present value of projected resource revenues;

• The modified PIH (MPIH), which allows for a scaling up of public investment during 5 years, but requires substantial fiscal adjustment to offset the initial scaling up and satisfy the intertemporal budget constraint; and

• The Fiscal Sustainability Framework (FSF), which incorporates the impact of higher public investment on growth, and non-resource revenues and, hence, generates a fiscally sustainable path that is consistent with a lower level of financial wealth. As noted above, the FSF requires a higher NRPB than the PIH because the steady state level of financial wealth over the long-term is lower, but since part of the resource wealth has been transformed into physical assets, growth and non-resource revenues are higher; so this is also consistent with a higher level of expenditures.

33. The second part of the exercise analyzes trade-offs in the operation of different short-to-medium-term fiscal rules. It looks at the effects of different price-based rules targeting a zero percent structural primary balance as a percent of non-resource GDP. A higher/lower primary balance can be targeted in countries that want to increase or decrease the level of financial savings over time. The exercise simulates how these rules would have performed in the past, using the evolution of oil prices for the past 40 years and showing how they would have performed in terms of smoothing out volatility and generating different levels of financial assets. It also simulates how the interaction of a price-based rule with a long-term sustainability benchmark can be implemented in practice and shows a comparison between the primary balance target and the accumulation of financial savings over time.

34. The country presented in this exercise is precisely at the boundary of the resource revenue horizon and hence could either have started with an NRPB rule associated with a sustainability benchmark, or with a price-based rule calibrated to take into account sustainability issues. The precise implication of these issues requires working with several fiscal rules at the same time. For example, where countries confront exhaustibility issues, this could take the form of an NRPB rule to maintain sustainability. In other countries a price-based rule could be used to smooth out volatility but calibrated in a way that ensures sustainability (i.e., higher structural primary balance target or using “prudence” in setting long-term prices, or both). This would also affect the path for the NRPB.

Assessing Long-Term Sustainability in Practice

35. Long-term sustainability should be accessed through the use of a PIH, MPIH, or FSF (see technical description above). The importance of this step is a function of the length of the resource horizon. Sustainability benchmarks are more relevant the shorter the estimated reserve horizon. A simple practical rule of thumb would involve the following steps.

• Compute the PIH sustainability benchmark. This benchmark is given by the NRPB that stabilizes net financial wealth over the long run $(NRP{B}_t=-\tilde{r}{W}_{t-1})$. This is an indication of the non-resource primary deficit that could be maintained indefinitely once resource wealth is exhausted.¹⁸ Figure 3 compares the three frameworks for a stylized country in which oil reserves last until 2046. In this example, the present value of oil wealth $(V_{t-1}\equiv {\mathrm{\Sigma }}_{s=t}^N\frac{{\displaystyle R{T}_s}}{{\displaystyle (1+i{)}^{s-t+1}}})$ at the end of 2010 corresponds to 679 percent of non-resource GDP.¹⁹ Assuming again a constant real interest rate of 5 percent and a constant real growth rate of non-resource GDP of 3 percent, the nrpb* consistent with the PIH sustainability benchmark is equal to -14.3 percent of non-resource GDP, as shown by the red dashed line in the first chart of Figure 3. Figure 4 shows the paths for primary expenditure and non-resource revenue for all three fiscal frameworks. With the PIH, the constant nrpb* combined with an assumption of constant non-resource revenue at 19 percent of non-resource revenues, stabilizes primary expenditure permanently at 33.3 percent of non-resource GDP.

  

  

  View Full Size

  

  Fiscal Sustainability Frameworks for Resource-Rich Countries: Non-Resource Primary Balance and Financial Wealth

  Citation: Policy Papers 2012, 071; 10.5089/9781498340007.007.A001

  Source: IMF staff estimates.

  • Download Figure
  • Download figure as PowerPoint slide

  

  

  View Full Size

  

  Fiscal Sustainability Frameworks for Resource-Rich Countries: PIH, Modified PIH, and FSF

  Citation: Policy Papers 2012, 071; 10.5089/9781498340007.007.A001

  Sources: IMF staff calculations.

  • Download Figure
  • Download figure as PowerPoint slide

• Assess the fiscal implications of a modified PIH (MPIH) scenario. This step estimates the intertemporal trade-off between a short-term increase in spending and future fiscal adjustment needs. Fiscal adjustment needs will be such that the future value of the financial wealth in year T = 2028 after the start of the front-loaded investment period is equal to the financial wealth level with the usual PIH $(W_T^{\prime }=W_T)$, where $W_T^{\prime }$ is the level of financial wealth following a fiscal path with an initial front-loading of investment followed by fiscal adjustment (see section I, C2 above). In the example below, an MPIH scenario allows for an increase in annual capital spending equivalent to 7.9 percent of non-resource GDP on average for the 5 years. The period of front-loaded investment then needs to be compensated by an annual improvement in the NRPB of 5.4 percent of non-resource GDP on average, smoothed over 10 years. The path for the NRPB under the MPIH is given by the solid green line in the first chart of Figure 3. For the financial wealth to be the same in both the MPIH and PIH after T periods, the future value of the cumulative increase in investment during the front-loading period has to be equal to the future value of adjustment after T. This strategy maintains wealth constant after year T = 2028. Figure 4 shows a similar path for the NRPB with an initial increase owing to the investment front-loading and then an adjustment returning to the PIH level of 33.3 percent of non-resource revenues after T = 2028. Table 1 also reports the average values of revenues and spending under the three frameworks.

  Table 1.

  Average Fiscal Variables Under the Fiscal Sustainability Frameworks

  

  Source: IMF staff estimates.

  Table 1.

  Average Fiscal Variables Under the Fiscal Sustainability Frameworks

  		PIH			MPIH			FSF
  	Scaling up	Adjustment	Steady state	Scaling up	Adjustment	Steady state	Scaling up	Adjustment	Steady state
  Non-resource revenue	19.0	19.0	19.0	19.0	19.0	19.0	19.0	20.6	22.7
  Total Primary Expenditure	33.3	33.3	33.3	41.1	27.9	33.3	41.1	31.5	33.6
  Current Spending	29.9	29.9	29.9	29.9	27.9	29.9	29.9	29.9	29.9
  Capital Spending	3.3	3.3	3.3	11.2	0.0	3.3	11.2	1.6	3.7
  Non-resource primary balance	-14.3	-14.3	-14.3	-22.1	-8.9	-14.3	-22.1	-10.9	-10.9

  Source: IMF staff estimates.

  View Table

  Table 1.

  Average Fiscal Variables Under the Fiscal Sustainability Frameworks

  		PIH			MPIH			FSF
  	Scaling up	Adjustment	Steady state	Scaling up	Adjustment	Steady state	Scaling up	Adjustment	Steady state
  Non-resource revenue	19.0	19.0	19.0	19.0	19.0	19.0	19.0	20.6	22.7
  Total Primary Expenditure	33.3	33.3	33.3	41.1	27.9	33.3	41.1	31.5	33.6
  Current Spending	29.9	29.9	29.9	29.9	27.9	29.9	29.9	29.9	29.9
  Capital Spending	3.3	3.3	3.3	11.2	0.0	3.3	11.2	1.6	3.7
  Non-resource primary balance	-14.3	-14.3	-14.3	-22.1	-8.9	-14.3	-22.1	-10.9	-10.9

  Source: IMF staff estimates.

• Consider the fiscal implications of an FSF scenario. This involves the calculation of the new (lower) financial wealth value in the FSF scenario assuming a positive impact of front-loaded investment on growth. The precise impact of higher public investment on growth is difficult to measure. However, one can make an assumption of the growth impact and compute the new stable financial wealth after the front-loaded investment, such that $(W_T^{\prime }<W_T)$. This exercise is shown by the blue dashed lines in Figure 3. After the period of front-loaded investment, a new PIH exercise is performed for the remaining oil wealth and financial assets accumulated at that point, but taking into account the growth impact of the additional public investment. Assuming a permanent additional effect of 1 percent on real growth and a reduction in the positive value of r — g,²⁰ after the front-loading investment period F = 2018, the new lower wealth level becomes 633 percent of non-oil GDP and the nrpb*is stabilized at the lower deficit of - 10.9 percent of non-resource GDP. As Figure 4 shows, the primary expenditure path depends on the multiplier effects of the front-loaded investment on the economy. For this exercise, the assumption is that the fiscal multiplier becomes larger than 1 just after the front-loaded investment and returns to its steady state level equal to 1 in the long run. This way, even if the nrpb* is lower in the FSF than in the PIH, the level of primary expenditure can be the same or even higher depending on the additional growth impact of public investment and the multiplier of the economy, given that this will increase non-resource revenues. A final computation could be finding the interest-growth rate differential that allows to be the same as in the PIH model.

Dealing with Price Volatility

36. We assume in this simulation that a price-based rule has been chosen (i.e., structural primary balance rule with price smoothing). As the main fiscal rule, this approach would be more appropriate for countries with relatively long reserve horizons. This rule is also more helpful in practice when resource dependence is high and thus the use of the NRPB in a fiscal rule becomes less politically attractive due to the difficulties of communicating large negative numbers to the public. Figure 5 shows the performance of alternative price rules in terms of smoothing out volatility, both when used in isolation and when combined with an expenditure growth rule. Figure 6 shows how a price-based rule could be combined with a long-term sustainability benchmark, and how the choice of structural balance target generates higher, or lower, financial buffers. Possible steps could include:

View Full Size

Expenditure and Savings Outcomes Under Various Price Rules¹

Citation: Policy Papers 2012, 071; 10.5089/9781498340007.007.A001

Source: IMF staff estimates.¹ The numbers associated with the price rules correspond to the number of years in the past, present,and future used in the smoothing calculation. For example,the “12/1/3” rule uses 12 years of past prices,the current year price, and a 3-year projection for the calculation.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Volatility and Sustainability Under Various Fiscal Rules¹

Citation: Policy Papers 2012, 071; 10.5089/9781498340007.007.A001

Source: IMF staff estimates.¹ The numbers associated with the price rules correspond to the number of years in the past, present, and future used in the smoothing calculation. For example, the “12/1/3” rule uses 12 years of past prices, the current year price, and a 3-year projection for the calculation.

• Download Figure
• Download figure as PowerPoint slide

• Calibrate the rule based on past historical experience and the desired policy objective (higher or lower financial savings). In this example (Figure 5), we illustrate the trade-offs available to a country that has decided to use a price-based rule using different computation formulas. The country would start by observing how different calibrations of the price-based rule smooth out volatility and generate fiscal buffers. In this case, a rule based on a long moving average of resource prices (the black line in panel 1) does the best in terms of expenditure smoothing but generates a relatively high level of financial savings (panel 2). This may be more or less appropriate depending on whether the country is using this rule just to smooth out short-term volatility or, like Chile, also to generate financial savings over time. A rule with a combination of past and future expected prices (red line) generates more volatility but would be consistent with less financial savings. The key trade-off is between volatility and the level of financial savings. To be sure, given the uncertainty about the future path of resource prices, past performance provides only limited guidance. The intent, however, is to have a mechanism to make simulations about the implications of alternative choices.

• Decide whether a complementary rule is needed to further smooth out volatility. Where avoiding volatility and fiscal procyclicality is the most important objective, an expenditure growth rule can be used in combination with the price-based rule. Figure 5 (Panel 3) shows how the use of the expenditure growth rule manages to smooth out expenditure to a much greater degree. However, in such cases where commodity prices rise for a long period, the use of this rule may be conducive to higher savings than desired (panel 4) consistent with sustainability concerns.

• Assess trade-offs in managing volatility and sustainability issues. Price-based rules can also be visualized in combination with long-term sustainability benchmarks. Figure 6 (Panels 1-3) simulate a price-based rule under different paths for the evolution of resource prices. Following price-based rules could result in expenditure paths far from those suggested by the sustainability-focused benchmarks, suggesting that investment scaling up plans may need adjustment or implying a drawdown of wealth. Panel 1 assumes high oil prices, which leads to deficits substantially higher than those implied by the sustainability benchmark. Panel 2 assumes very low oil prices over the next 20 years, which indicates that investment scaling up would happen just at the time when price rules suggest a reduction in deficits. Panel 3 presents a more favorable scenario, with oil prices evolving in such a way that the price-based rule balance follows the sustainability benchmark relatively closely. Panel 4 simulates different levels of financial savings for a given price-based rule, under alternative structural primary balance targets. The results suggest that small changes in the target can have large impacts on savings over time and indicate the need to reassess existing targets periodically.

Priorities for Public Financial Management in Resource-Rich Countries

37. While PFM requirements are broadly the same across countries, certain dimensions need to be emphasized in the resource-rich countries (see also http://blog-pfm.imf.org and Dabán, Sánchez, and Hélis, 2010).

38. Forecasting resource revenues. Forecasting revenue flows that are large, volatile, and uncertain require capacity and appropriate tools. The complexity of tax or other fiscal arrangements also complicates revenue forecasting—progressivity in taxation could also increase volatility. Forecasting frameworks would ideally have the following features:

• Bottom-up forecasts. Revenue forecasts that build on individual extraction projects better capture the evolving revenue path over the life of the project and any project-specific fiscal terms. A prudent approach is only to include proven reserves in the revenue forecasts

• Aggregation of individual project forecasts. Revenue forecasts from individual projects need to be integrated into an aggregate resource revenue forecast for the medium term fiscal frameworks.

• Capacity for sensitivity analysis. The revenue forecasting framework should allow for sensitivity analysis and alternative scenarios under different assumptions regarding price, costs, and volume.

39. Medium-term budget planning. A medium-term orientation to budget planning would help to ensure that natural resource revenues are utilized in a sustainable manner. Key components are

• Medium-term fiscal framework (MTFF). The budget should be underpinned by an MTFF with forecasts for broad fiscal aggregates for revenues and expenditures. The MTFF should be formulated in line with a fiscal rule.

• Medium-term budget and expenditure frameworks. Aggregate expenditure forecasts need to be translated into disaggregated expenditure ceilings including the investment program. The level of disaggregation could initially be done with less granularity, focusing on larger expenditure categories before moving toward component programs.

40. Investment appraisal, selection, and implementation. Countries need to ensure they have the capacity to efficiently appraise, select, and implement public investment, including public-private procurements, so they can overcome absorption constraints that would otherwise prevent productive scaling up of capital and investment spending. The following components are essential (Rajaram et al., 2010):

• Investment plan integrated with the budget, including medium-term expenditure plans and budget ceilings. Where a country faces absorption and capacity constraints, the investment plans and ceilings should provide a gradual path for scaling up investment rather than large changes.

• Project appraisal, selection, and implementation. A clear process is needed to appraise project proposals, which also requires developing sector expertise in key infrastructure areas. Capacity for executing and overseeing projects needs to be enhanced. This also requires having in place rules-based and transparent procurement procedures (including competitive bidding).

• Ex-post evaluation. A process needs to be in place for an objective evaluation of the selection and implementation of projects to ensure that lessons of both success and failure can be applied to strengthen future project planning and implementation.

• Multiyear funding for investment projects. For large investment projects, it is important to fully account for the full costs of project implementation, which may extend over several years. Likewise, recurrent costs also need to be factored in.

41. Transparency. The integrity of resource revenue utilization depends upon the transparency, quality, and timeliness of government accounting and reporting. Best practice guidance is available in the Fiscal Transparency Manual and the Guide on Resource Revenue Transparency (IMF, 2007).

Additional Fiscal Indicators

42. In light of the unique policy challenges faced by resource-rich countries, the fiscal policy framework needs to be based on a multiple indicator approach. Focusing exclusively on conventional indicators, such as the overall and primary balances, would result in incomplete and potentially misleading conclusions about the sustainability of public finances and the impact of fiscal policy on the economy. In addition to the NRPB discussed above, the following indicators can be considered:

43. The overall (or primary) balance, common in countries without natural resources, measures the change in net financial assets. This indicator provides a measure of related fiscal vulnerabilities as well as net financing needs in the event of declines in resource revenue. However, this indicator can often be procyclical in resource-dependent countries: with rising resource revenues, a fiscal expansion (increase in spending) can be masked by an improving overall balance. At the same time, a persistently negative overall fiscal balance in a resource-rich country would have to be assessed carefully, as it would imply, under certain conditions, that the country is continuously increasing its net debt.

44. Other complementary indicators might be the current balance (which excludes public investment); the domestic balance (which excludes resource revenues and the import content of government expenditures); and the break-even resource price.

• The current balance,²¹ which is equivalent to public savings, is a useful measure of the budgetary resources available for public investment (i.e., net acquisition of non-financial assets). It therefore indicates the maximum amount of public investment that could be executed without incurring an overall fiscal deficit. The non-resource current balance is similar but excludes resource revenues. A practical drawback of these indicators is that they do not provide a clear anchor for fiscal policy. They also ignore difficulties in classifying current and capital expenditure, which leads to incentives for creative accounting.

• The domestic balance can be helpful in assessing short-term demand pressures that can give rise to inflationary concerns. When the domestic balance deteriorates, the demand for domestic goods increases, so an abrupt deterioration of this indicator (for example, when a large expansion of public investment does not have a large import content) could indicate limited absorptive capacity for scaling up. However, this indicator is even more difficult to compute than the current balance because data on the import content of government spending is often not readily available.

• The break-even resource price, which is the price at which the overall balance will be zero, can also be used in resource-rich countries to determine the short-term vulnerability of public finances to lower resource prices. It does not take into account any financial savings the country may need to deal with shocks, but a constantly rising break-even price would suggest that the budget is becoming increasingly vulnerable to declines in resource prices.

C. Managing Natural Resource Revenues: A Dynamic Stochastic General Equilibrium Framework²²

45. Dynamic stochastic general equilibrium (DSGE) models are increasingly being used to analyze the effects of macroeconomic policy.²³ Besides being applied to short-term issues, they have also been used to evaluate policies with longer-term implications.²⁴ Based on Berg et al. (forthcoming), this section introduces a DSGE framework for assessing fiscal strategies for managing natural resource revenues. It focuses on the scaling up of public investment in developing countries using volatile resource income.

Specification of the Model

46. The model is a small open real economy that has a closed private capital account. The government cannot borrow in international financial markets but can save externally by holding financial assets.²⁵ To allow for the low quality of public institutions and governance, it builds in public investment inefficiency and absorptive capacity constraints, which cause substantial waste in producing public capital. The model also features learning-by-doing externalities in production of traded goods to capture potential Dutch disease from spending large amounts of foreign exchange.

Households

47. A representative household chooses composite consumption, c_t, and labor, l_t, to maximize expected utility,

subject to the budget constraint:

σ and ψ are the inverses of the elasticity of intertemporal substitution for consumption and labor supply; κ is the disutility weight on labor; w_t is a real wage index; ${\tau }_t^c$ and ${\tau }_t^l$ are consumption and labor tax rates; rm* is remittances in units of foreign consumption (denoted by *); z_t is government transfers; s_t is the real exchange rate, the price of the foreign consumption basket relative to the domestic; and ${\mathrm{\Omega }}_t^T$ and ${\mathrm{\Omega }}_t^N$ are profits from traded and from non traded goods. The only financial asset a household can hold is domestic government bonds, b_t; R_t is the domestic gross real interest rate.

Composite consumption, c_t, consists of non traded and traded goods, combined in a CES (constant elasticity of substitution) basket. The price of one unit of composite consumption is

where $p_t^N$ is the relative price of non traded goods to composite consumption, χ is the intratemporal elasticity of substitution, and φ is the degree of home bias. Assuming the law of one price holds for traded goods, s_t is also the relative price of traded goods.

Total labor input of a household is

where $l_t^T$ and $l_t^N$ are labor supplied to the traded and to the non traded sectors; δ is the initial labor share in the non traded sector; and ρ is the elasticity of substitution.²⁶

Non-resource Production Sectors

48. The economy consists of three production sectors: a non traded goods sector, a non-resource traded goods sector, and a natural resource sector (denoted by 0). As the vast majority of resource output in developing countries is exported, in the model resource output is solely for exports.

A representative goods firm, non traded or traded, produces using the technology,

where $z_t^j$ is sector-specific total factor productivity (TFP); $K_t^G$ is public capital; and α^G is output elasticity with respect to public capital.

Private capital evolves by the law of motion,

where κ^j ≥ 0 governs investment adjustment costs.

A representative firm maximizes its net-present-value profit weighted by the marginal utility of households (λ_t),

subject to (6). Note that $p_t^T=s_t$, and ι captures distortion that discourages firms in developing countries from investing and hiring further; ι acts like an implicit tax on firms but the revenues collected remain in the private sector; and $Y_t^j$ denotes the aggregate output of nontraded or traded goods.

Following van Wijnbergen (1984), TFP in the traded goods sector is subject to learning-by-doing externalities in the form of

Non traded TFP is assumed to be constant: $z_t^N=z^N,\forall t$, ∀ t.

Resource Production Sector

49. The model takes the world price of a resource commodity as given and assumes that $p_t^{o_{*}}$ follows a random walk,²⁷

where ${\epsilon }_t^{po}\sim i\mathrm{.}i\mathrm{.}d\mathrm{.}N(0,{\sigma }_{po}^2)$ is the resource price shock.

To simplify resource production, resource capital in the model is financed by foreign direct investment (FDI), which follows an AR(1) process,

where ${\epsilon }_t^{FDI}\sim i\mathrm{.}i\mathrm{.}d\mathrm{.}N(0,{\sigma }_{FDI}^2)$ is the FDI shock. Resource capital evolves according to

Resource output is produced by the technology

where $z_t^o$ is resource TFP. It also follows an AR(1) process,

where ${\epsilon }_t^{zo}\sim i\mathrm{.}i\mathrm{.}d\mathrm{.}N(0,{\sigma }_{zo}^2)$ is the resource TFP shock.

Resource production is subject to royalties at a rate of ${\tau }_t^o$ and profit taxes at a rate of τ^div. Because intermediate inputs in resource production are omitted here, the profit is calculated as

Resource taxes paid each period are

Fiscal Policy

50. One main policy issue for a resource-abundant country is how to allocate resource revenues among saving and spending options. The model considers the options of external saving in a resource fund $\left(F_t^{*}\right)$, public investment $\left(G_t^I\right)$, government consumption $\left(G_t^c\right)$, and transfers to households (Z_t).²⁸

Let ES_t be external savings from resource revenues and C_t and L_t be aggregate consumption and labor. The government’s flow budget constraint can then be written as

where r* is the constant net foreign interest rate, and $p_t^G$ is the relative price of government purchases.²⁹

The evolution of a resource fund is

where $E{S}_t^{*}\equiv \frac{{\displaystyle E{S}_t}}{{\displaystyle s_t}}$ is external savings in units of foreign consumption.

51. To center the analysis on different approaches to investing resource revenues, policy specifications here lay out a “spend-as-you-go” approach maintaining a resource fund at the initial low level and a “gradual scaling up” approach with external savings. Both approaches keep consumption and income tax rates constant.

• The “spend-as-you-go” approach: In each period the government maintains fixed ratios of government consumption and public investment to resource revenues. Transfers to households adjust to clear the government budget constraint. When resource revenues grow, government purchases and hence non-resource GDP also grow. The feedback effect of more government spending generates higher non-resource tax revenues. Since transfers adjust to clear the budget, the additional non-resource revenues also drive up transfers to households. As government consumption, public investment, and transfers all rise when resource revenues increase, this implies a procyclical policy stance on managing resource revenues.³¹

• The “gradual scaling up” approach: The government plans a gradual scaling up path for public investment and government consumption. Public investment is scaled up gradually despite the possibility of a revenue surge—the slow pace makes it possible to shore up a fund to build a stabilization buffer against future resource revenue shocks. Assuming that transfers to households are also kept at the initial level, when a resource fund has insufficient funds to allow continued financing of the predetermined periodic investment, investment expenditures for the period are reduced to maintain a nonnegative value in the resource fund. (This reflects the model’s assumption that the government cannot borrow externally to finance public investment.)

52. Public investment accumulates as public capital subject to absorptive capacity constraints and investment inefficiency. To model absorptive capacity constraints, public investment expenditure $G_t^I$ is distinguished from effective public investment ${\tilde{G}}_t^I$. Specifically,