Commodity Price Volatility and Inclusive Growth in Low-Income Countries

Chapter 6. Savings and Investment Decisions in Resource-Rich Low-Income Countries

Rabah Arezki, Catherine Pattillo, Marc Quintyn, and Min Zhu
Published Date:
October 2012
  • ShareShare
Show Summary Details
Paul Collier


Many of the poorest countries on Earth are in the throes of a double bonanza of high prices for their natural resource exports coupled with new discoveries. Over the next decade, the potential financial flows from resource exports will dwarf aid, remittances, and foreign direct investment (FDI), providing an unprecedented opportunity for development.

Directly, high prices disproportionately increase the rents available for governments. Although prices may have peaked, Asian growth seems likely to sustain them well above past levels, which is an aspect of the “new normal.” Indirectly, the high prices have triggered prospecting. Although Africa is considered resource rich, as of the millennium the value of discovered subsoil assets per square mile was only one-fifth that of OECD countries. It is unlikely that this is because there is less below the ground: Rather, there had been less prospecting. New discoveries are therefore concentrated in Africa and the other neglected, impoverished, and misgoverned parts of the world such as Central Asia. A reasonable assumption is that new searches will gradually bring discovered subsoil assets up to around the OECD level, which would be a fivefold increase.

This unprecedented opportunity requires distinctive and sometimes non-obvious economic policy responses, which will be discussed later in this chapter. The actual policy record suggests that taking the right decisions is politically difficult, so approaches that might bring policy choices closer to the theoretical ideal will be proposed.

The Distinctive Economics of Resource Depletion in Low-Income Countries

Resource-rich low-income countries face common challenges and need saving and investment policies that are broadly common. Further, these common policies are distinctive relative to other countries. All resource-rich countries need policies that differ from resource-scarce countries. However, those resource-rich countries that are low income need policies that differ from those that are high income. An important implication is that there are no OECD role models. Not only are OECD countries high income, but nearly all are resource scarce. As a result, the prevailing OECD policy discourse inevitably neglects the saving and investment policies that are of fundamental importance to resource-rich LICs. Even the few OECD countries that are resource rich, such as Australia, Canada, and Norway, are inappropriate role models for these decisions. The common challenges faced by resource-rich LICs are unique to them.

Because the saving and investment challenges are qualitatively common to the entire category of countries, the design of policies that are appropriate for them is an international public good, though not a global public good. As an international public good, it is appropriately supplied by IFIs rather than by each government individually: Most resource-rich low-income country governments are ill equipped for original economic policy design. However, in providing the public good of economic policy advice, it is vital that IFIs recognize that policy in this category of countries should be distinctive. Policies that are considered sound for other categories of countries would be fundamentally unsound for resource-rich LICs. Neither staff trained in OECD macroeconomics nor those trained in development economics are adequately equipped for this task, so specific training is required.

Rents and the Role for Government

The economic returns on the activity of resource extraction come partly in the form of factor incomes to capital and labor and partly in the form of rents. Some of these returns accrue to nationals of the country and some to foreigners. In African LICs, relatively little of the factor income is likely to accrue to nationals: The massive capital investments required for resource extraction can be financed only by international companies, and the sophisticated skills required are possessed only by foreigners. Hence, the returns available to accrue to nationals are disproportionately rents. However, rents initially accrue to the resource extraction companies—they accrue to nationals only to the extent that the government is able to transfer them to itself through taxation in various forms. Thus, the predominant means by which the country can benefit from resource extraction will typically depend upon the efficacy of government rent capture.

An implication is that for LICs, government is central in making resource extraction nationally beneficial. The rents must accrue as public revenue; once they are accrued, the government is then inevitably responsible for spending them. Resource-rich LICs should therefore have a large and active state. This contrasts both with those LICs that are resource scarce and with those resource-rich countries that are high income. The former will develop through the expansion of industry, services, and agriculture, in all of which rents are modest and much of the factor payments accrue privately to nationals. In the latter, exemplified by the United States, rent capture by the government is less important because factor incomes accrue largely to nationals, as do rents left with the extraction companies.

Savings: Depletion Matters

The extraction of mineral resources is unsustainable because endowments are not renewable. The appropriate savings rate from such natural resource revenues depends upon the horizon to depletion. For a given extraction rate, the shorter the horizon until expected depletion, the higher the savings rate should be. This follows directly from the permanent-income framework. For a given extraction rate, and hence a given annual revenue, the shorter the extraction period, the lower the present value of the resource endowment and hence permanent income, and the sustainable increase in consumption. With less consumption warranted from a given resource revenue, the higher is the savings rate.

A straightforward but important corollary is that for a given constant rate of extraction, the savings rate should rise as resources are depleted. Each year, the horizon to full depletion is shorter and so the appropriate savings rate is higher. By the final year of depletion, the savings rate out of the revenues from extraction should be 100 percent.

Applying this to Africa, because really large deposits of natural resources are easier to find than smaller deposits, those deposits that were still awaiting discovery when the present commodity boom started were disproportionately smaller than those already discovered. Evidently, smaller deposits are depleted more rapidly than larger deposits, so the time frame for depletion is liable to be shorter for recent and forthcoming discoveries than the historical norm. For example, the oil fields typically discovered in Africa usually have an economic life of only around two or three decades. Hence, the initial savings rate out of the revenues from extraction should be higher than in the Middle East, where deposits typically have a longer life.

The physical rate of extraction is seldom constant. More typically, it follows a humped pattern, rising to a peak and then tapering off. Evidently, when extraction is at its peak and set to decline, some of the volume can be sustained only briefly and so the savings rate should be higher. Conceptually, each unit of volume regularly extracted has a specific time to exhaustion and hence a distinct path of the savings rate from the revenues generated by it. In the phase of rising extraction rates, as time passes, not only is the horizon for the existing volume of extraction getting shorter, but each addition to volume has a shorter horizon than the existing volume. Both influences imply that in the phase of rising extraction, the rate of savings should rise particularly rapidly.

A final proposition on the appropriate savings rate from depleting resources is that the savings rate varies inversely with the expected long-term rate of change in the world price of natural resources. According to the Hotelling rule, the price of natural resources can be expected to increase at the world rate of interest. In practice, however, prices of natural resources do not follow this predicted path, and it would be highly risky for the governments of resource-rich countries to assume that they would. However, to see the difference that the assumed path of prices makes, consider a scenario in which the world interest rate is 2 percent and the initial resource discovery will be depleted at a constant physical extraction rate over 50 years. In this benchmark case, in the first year of extraction, the appropriate savings rate out of the revenues from extraction would be zero—permanent income would be maintained purely by the rise in the value of resources remaining in the ground. With a constant extraction rate, the appropriate savings rate would rise linearly from 0 percent in the first year to 100 percent in the fiftieth year.

The other extreme from assuming that prices follow the Hotelling rule is to assume that they will collapse to zero next year and remain there. On this assumed price path, the appropriate savings rate is always 100 percent (the “‘bird-in-the-hand rule”). More generally, the more valuable that natural resources left in the ground are assumed to become, the higher the permanent income becomes that can be supported by their gradual depletion. Hence, for any given physical extraction rate, the higher is the consumption that is warranted, so that less of the revenue needs to be saved.

Whereas the Hotelling rule provides a reason to expect that natural resources will become continuously more valuable, the historical experience has been of eventual technological obsolescence. When the price of a commodity rises, there is investment in research to develop substitutes. Ultimately, the Hotelling rule gets overridden—a time is reached at which people expect the resource to become less valuable, but the remaining endowment cannot quickly be extracted and sold off because the costs of extraction would become prohibitive. Those who have gambled on continued rising prices get caught and suffer losses. The lags in technological obsolescence are long and variable, but because global technological research has intensified, they may well be shortening.

Perhaps a reasonable compromise between the cheery implication of the Hotelling rule and concerns about technological obsolescence is to assume that over the horizon of a generation, there will be no trend change in the real level of prices, but that thereafter, either physical exhaustion or obsolescence will become a serious prospect.

Savings: The Prospect of Convergence

A low-income society with significant natural resource endowments can reasonably expect that if it manages its opportunities effectively, it will gradually converge with richer societies. In effect, by appropriately saving and investing the revenues from its depletion of natural assets, the economy can rectify its initial shortage of capital, in the process enjoying a phase of growth in excess of global rates.

An important implication is that per capita consumption today is markedly lower than it will be in the future. On the usual utilitarian framework, this justifies some redistribution of consumption from the low-marginal-utility future to the high-marginal-utility present. In other words, it justifies having a lower savings rate in the early period of resource extraction than would be warranted by an application of the permanent-income framework (Collier and others, 2010; van der Ploeg and Venables, 2011).1

Of course, this is conditional upon convergent growth and therefore depends upon the society adopting and maintaining appropriate savings and investment policies from resource depletion. However, a virtue of the assumption of convergence is that it provides a coherent policy vision: If the policies are implemented, the outcome will validate the choice of policies. The apparently more cautious policy would be to assume that either savings or investment policies will not be implemented, so that the economy will not converge, in which case the society should rationally have a higher savings rate initially. Yet planning an optimal savings policy for current revenues conditional upon other policy mistakes rapidly leads into policy confusion; for example, if it is assumed that current savings will subsequently be squandered, it is better not to save at all. The approach taken here is first to establish what an optimal policy set would look like and then propose a political economy solution to the practical challenge of these decisions being made.

The assumption of convergence justifies an initial savings rate out of resource depletion that is somewhat lower than what would be implied by a simple application of permanent income. However, the appropriate initial savings rate is still likely to be substantial. Because the justification for a lower savings rate is the prospect of convergence, the investment rate must be consistent with convergence. To date, Africa’s investment rate of around 20 percent has been considerably below that of emerging Asia, which has mostly been above 30 percent. For Africa to assume convergence with emerging Asia and yet persist with an investment rate far below that of Asia would be to place too much reliance upon achieving a rate of return on investment well above global rates. The prudent course is therefore to raise the investment rate to above 30 percent as soon as practically possible. In turn, this implies that the initial savings rate out of resource depletion should probably be no lower than 30 percent. If the onset of resource revenues yields a quantum increase in overall government revenues, the implied increase in consumption might pose transitional problems that temporarily warrant a higher savings rate.

Investment: Capital Scarcity and Investment Capacity

Convergence depends upon investment that gradually rectifies capital scarcity. This has an important implication for the choice of assets acquired with the savings from resource depletion. Whereas a capital-abundant economy such as Norway should rationally acquire claims on capital, in other economies that are less capital rich, a low-income economy should acquire capital domestically.

There is an important caveat to such a policy. Whereas savings are generic, physical investment is always specific and so requires a decision process that designs, selects, and implements projects. One reason why low-income societies are capital scarce is that they do not have the capacity to manage domestic investment well. Further, if the investment rate is to be increased from around 20 percent to around 30 percent, the demand for the capacity to manage the investment process will increase by 50 percent. The likely consequence of such an increase in the demands placed upon an already weak capacity would be deterioration in the efficiency of the investment process. In turn, this would manifest itself as a decline in the rate of return on investment. This is not just hypothetical. Globally, episodes of large surges in the investment rate are not usually followed by accelerations in growth—high investment is instead dissipated in reduced returns.

Hence, a critical stage prior to an increase in domestic investment is to build the capacity to manage it. I term this process “investing-in-investing.” It has three components. The first component is the capacity to manage the process of public investment, that is, project design, appraisal and selection, implementation, and ex post evaluation. These are the capabilities assessed by the new Public Investment Management Index proposed by Dabla-Norris and others (2011). This index provides a useful benchmark to judge improvements and potentially can also can be used in a decision rule as to when the return of public money held abroad is warranted.

The second component is to improve the environment for private investment. Public and private investments are complements (for example, roads and trucks), so the return on either depends upon investment in the other. The environment for private investment is already reasonably well measured by the annual Doing Business ratings of the World Bank. The government of Rwanda has demonstrated how it is possible to improve performance on this rating very rapidly, as it has now overtaken several European countries.

The third component is for policy to reduce the unit cost of capital goods for both public and private investment. In Africa, capital goods are typically expensive.2 Structures are costly because construction costs are avoidably high—the market for urban land is restricted, there are impediments to imports of key inputs such as cement, and there has been little training in construction skills. Conceptually, government policies need to shift downward and flatten the supply curve of construction so that the large increase in demand (which is inevitable if depleting natural assets are to be converted into domestic capital) should not be dissipated in higher unit costs. It would be useful to build an international index to benchmark and monitor construction costs analogous to the Public Investment Management Index and Doing Business.

Equipment, though imported, is costly because national markets are too small to be competitive and trade barriers have inhibited the emergence of regional markets. Trade policies and behind-the-border measures could regionalize the market in capital goods, leading to lower markups. Again, a benchmarking index would be useful.

Investing-in-investing takes time, but it is concerned with a manageable set of tasks, processes, and skills. Emerging Asia was able to build its investment capacity, thereby ramping up its investment rate without the return on investment collapsing. Once policymakers recognize that investing-in-investing must precede an increase in domestic investment in the sequence of policies for development, then it should be feasible over the course of a decade. Until investment capacity is built, the increase in savings generated from resource depletion should be parked abroad in financial assets. The timescale for such saving abroad is, however, much shorter than that adopted by sovereign wealth funds (SWFs): The objective is to preserve asset value until investment capacity has expanded sufficiently for the money to be brought back and invested domestically. Hence, the asset composition should be more conservative than in an SWF.

Absorption and Temporary Saving Abroad

These concerns about the capacity to invest have been microeconomic. Potentially, as investment spending is ramped up, it may also generate macroeconomic effects. However, in LICs, capital spending has a high import content and so the repercussions for the domestic economy are largely limited to the construction sector. Here, sectoral policies targeted at breaking supply constraints are more appropriate than generalized macroeconomic restraint of demand such as would be achieved by increasing the proportion of savings allocated to foreign financial assets. From the perspective of asset accumulation, the rationale for an initial phase of accumulating foreign assets is fundamentally microeconomic rather than macroeconomic. It is not demand that has to be managed, but rather supply.

Macroeconomic considerations may, however, matter for the warranted increase in consumption. This will depend upon the scale of the increase and the composition of demand. Except with major discoveries, the initial increase in warranted consumption may be quite modest—even the initial savings rate out of revenues from natural resource depletion should be markedly higher than that from sustainable revenues so that the implied percentage increase in consumption will be lower than that for investment. Nevertheless, even modest increases in warranted consumption may have substantial effects on the real exchange rate: In many low-income economies, the share of consumption met by imports is low, whereas the increase in the supply of consumer goods warranted by resource revenues consists only of imports. This is especially the case for public consumption, which is the component of consumption most likely initially to be increased by a step increase in government revenue.

Because a substantial temporary increase in the real exchange rate is liable to be disruptive to the real economy, there is therefore a good case for delaying much of the increase in consumption to provide the time for domestic production of nontradable consumer goods to increase. Evidently, such an increase in supply is predominantly a consequence of investment (public and private). This returns us to the investing-in-investing agenda. Once the capacity to invest is in place, domestic investment can scale up without reducing the efficiency of investment. In turn, this will increase the supply of consumer goods and services, enabling the demand for consumption to be increased without a substantial appreciation in the real exchange rate.

Hence, analogous to the temporary accumulation of foreign assets out of the revenues eventually to be used for domestic investment, there may also need to be temporary accumulation of foreign assets out of the revenues to be used for consumption. As domestic investment ramps up, the continuing flow of such temporary savings can be tapered out and reversed, with the accumulated savings-for-consumption repatriated.

This macroeconomic approach of countering Dutch disease by the sequencing of investment can be complemented by a sectoral policy concerned with the composition of investment. Commonly, a key policy concern is that real appreciation will damage nonresource exporters. By skewing the composition of public investment toward projects that lower the costs of nonresource exporters, the sector can even be advantaged by resource revenues rather than squeezed. An example is the active promotion of Penang as an export zone for electronics by the government of Malaysia. During the 1970s, much of government revenue still came from natural resources; this helped to finance the physical and social public infrastructure that transformed Penang (see Collier and Venables, 2011).

A further compositional policy is that between government, firms, and households. Depending on the scale of resource revenues, there is a case for transferring decisions over some spending to firms and households. In the rare cases in which a poor economy receives very large resource revenues, the case for some transfer to private households is overwhelming; otherwise, households will have an imbalance of abundant public goods and inadequate private goods. Similarly, because public capital and private capital are complements, there is a case for some transfer to firms. However, there is a good case for delaying such transfer until both public infrastructure and the policy environment for private investment have been improved (as tracked by Doing Business). Ideology-driven assumptions that the private sector is inherently better at handling the investment process are probably misplaced. For example, in Kazakhstan, a prudent government saved resource revenues abroad but permitted the banking sector to borrow heavily on the implicit security of these public foreign assets. The banking sector thereupon channeled the resource revenues into a disastrously excessive property boom.

Instead of using the banking system, the transfer of decisions over investment from the government to firms can potentially be done through reducing taxation. However, there are two powerful political economy arguments against this: The taxation of business gives the government a stake in private sector growth, and it also provokes scrutiny. In practice, if the domestic banking system is unreliable, it may be better to induce private investment indirectly through improving infrastructure and the policy environment.

Managing Volatility

Commodity prices are volatile, and their path is deeply unknowable. For example, in January 2008, the bounds of the 95 percent confidence interval for the 12-month market-based forecast for the world oil price were around US$210 and US$65. Two features of this forecast are equally striking—its range is so wide as to make it useless for practical budgeting purposes, and the actual price, at US$37, was far outside it. What is important, therefore, is not to forecast prices as accurately as possible, but to smooth spending in the face of price volatility.

Revenues are even more volatile than prices; not only does supply naturally respond to price, but markets are not fully cleared by price. The failure of prices to clear commodity markets became evident during the collapse in demand of 2008 and 2009 when the market for commodities started to resemble that for manufactures with mines temporarily closed.

There is a very strong case to avoid revenue volatility generating corresponding volatility in expenditure. Often the costs of volatile expenditure are thought of with reference to Dutch disease. However, the fundamental reasons for expenditure smoothing in the face of volatility are microeconomic rather than macroeconomic. Large fluctuations in consumption are inefficient even in the most rudimentary single-good framework (that is, with no real exchange rate). Due to both habit formation and political costs, substantial reductions in consumption alternating with substantial increases yield less utility than a constant level around the same mean. Although volatility in investment does not face these difficulties, beyond a point, large swings in investment are liable to impair its efficiency. If investment spending faces steep unanticipated reductions, then projects will be abandoned uncompleted, and if it is increased very fast, then project selection and implementation are likely to deteriorate. Hence, the true diagnostic for managing volatility is not the real exchange rate, but the level of aggregate spending and its decomposition into consumption and investment, both public and private.

Smoothing spending in the face of unknowable volatility in revenues is necessary but costly. Were the path of fluctuations in revenues fully known, the least costly way of smoothing expenditure would be to accumulate liquid savings during periods of above-average revenue and run them down when the revenue was below average. However, because the path of revenues is not known, smoothing spending is a matter of managing risk. The appropriate way of coping with risk is not savings, but rather insurance. In the face of price risk, governments can, in principle, either fully insure themselves by locking into long-term contracts at an agreed price or set bounds on the degree of risk through hedging.

Many governments of LICs lack the credibility to be able to commit to long-term price contracts (except at deeply discounted prices), and so the practical alternatives are saving and hedging. Markets are not yet sufficiently deep to permit hedging many years out, and so the most efficient structure is likely to be to hedge prices over a short horizon (at a minimum of the price assumed for the annual budget process), relying upon savings for the longer term. The cost of hedging is evident. The cost of liquid savings is the opportunity cost of domestic investment forgone.

Because both methods of smoothing expenditure are costly, the objective should not be literally to stabilize public spending. Rather, governments should aim to keep rates of change of spending (both increases and decreases) within manageable bounds. Politically, the key issue is to set maximum rates at which public consumption and public investment will be permitted to rise so that revenues in excess of these ceilings on expenditure will be used to smooth future spending, whether by purchasing hedges or by saving. What is manageable is itself endogenous to policy—the more public investment can be varied without damaging its efficiency, the less liquidity is needed. Public investment can potentially be designed so as to be able to cope with a degree of volatility. A high average level of investment should make fluctuations easier; for example, if investment is on average 30 percent of GDP, a 10-point swing (25 to 35 percent) is proportionately less drastic than if it averages 20 percent (15 to 25 percent). Further, in periods of low investment, project preparation can continue so that there is a shelf of projects ready for implementation as financing becomes available.

The task of smoothing public spending is essentially driven off the upper and lower bounds placed on changes in public spending. Initially, all the work is done by the upper bounds on public consumption and public investment. These bounds determine how much money is channeled into expenditure smoothing, whether by the purchase of hedges or the accumulation of savings. Until there has been a substantial phase during which the upper bounds on expenditure growth are binding, there can be no counterpart phase in which expenditures can be permitted to decline less rapidly than resource revenues are declining. Hence, with the benefit of hindsight, 2000 to 2010 was the ideal time to put such a policy in place.

Although this opportunity has gone, an equivalent opportunity still pertinent for many countries is at the onset of a resource discovery. As previously discussed in this chapter, in such circumstances, there is a case for restraining the increase in consumption until investment has increased supply to meet new demand. Rather than this being merely a transitional pot of liquidity, a first call on it can be to fund the long-term task of smoothing public spending. In other words, the transitional need to accumulate liquidity until productive capacity has been enhanced so as to avoid Dutch disease can be matched to the other transitional need to accumulate liquidity to cope with volatility.

In the steady state, the liquidity pot, combined with hedges, is judged sufficient to meet all likely calls on it. Because the path of commodity prices is not known, once the pot is being drawn down, there is an inherent risk that low prices will persist for long enough to bankrupt it. To guard against this, the bound on the maximum rate of decrease in public consumption should be endogenous to the available financing, with hedges used to limit the risk of further price declines. As the available funds are depleted, the cushioning limit on the rate of reduction in public consumption would be reduced.

The Desired Policy Package

A useful way of bringing these disparate aspects of savings and investment policy together is to think of three policy clocks. The government of a resource-rich LIC needs to be conscious of all of these clocks, each of which ticks at a different speed.

The slowest-ticking clock is that of resource depletion: over the course of a generation, natural assets should be converted into productive assets, implying the need for substantial savings out of resource revenues. As noted, the proportion saved will depend upon several assumptions, but a common feature is that the savings rate should rise over the course of depletion and rise especially rapidly in the phase during which the rate of extraction is increasing. This policy clock earmarks, year by year, the proportion of revenues appropriate for long-term asset accumulation, with remaining revenue being earmarked for warranted consumption.

The second clock is that of investing-in-investing—building the capacity to invest a high proportion of income domestically while maintaining a good rate of return. If this is given its proper priority, then this clock should run much faster than that of depletion: Perhaps it might take a decade to put in place the systems and human skills to manage the investment process well. During this phase, both the revenues earmarked for long-term asset accumulation and much of the revenues earmarked for warranted consumption should temporarily be invested in foreign financial assets. As capacity is built and the productive capital stock is increased, both of these temporary pots of foreign assets can be repatriated, financing domestic investment and consumption, respectively.

The third clock is that which rides the tiger of commodity prices. This clock has an initial phase during which revenues are being set aside to accumulate liquid assets and hedges, followed by a steady state in which spending is cushioned relative to revenues. The transition phase requires that the upper bound on the rate of increase of public consumption be set sufficiently conservatively that the revenue warranted for consumption exceeds the ceiling on expenditure for a sustained period. For the steady state to be sustainable, the maximum permitted rate of reduction in public consumption should adjust as funds are depleted.

Because these three clocks are all running in real time, the required policy dance is complicated. Sometimes one clock will be indicating a time for high savings, while another will be indicating a time to run savings down. Faced with such complexity, ad hoc decisions taken day by day are liable to go wrong. There is a need for some policy rules, but policy rules are political—they work only if they are politically realistic.

The Political Economy of Saving and Investment Decisions

There is now a large and contentious literature on the natural resource curse. As this topic has been ably and recently covered, it will not be reviewed in this chapter (see Ross, 2011). The discussion here will be on some new and disturbing direct evidence of a policy-based “resource curse” through public investment.

Whatever else the government of a resource-rich LIC does with the revenues from natural resources, one prescription is unambiguous: Over the medium term, it should accumulate public capital. Of course, all governments should accumulate public capital, but the case for resource-rich LICs is overwhelming as they start chronically short of public capital and their revenues are coming disproportionately from depleting the society’s natural capital. Yet, in general, such governments do the opposite. Bhattacharyya and Collier (2011) bring together a new international time series data set on the stock of public capital with data on resource rents. They find that by controlling for per capita income and fixed effects, resource rents significantly and substantially reduce, the public capital stock.

Faced with this gulf between the savings and investment policy appropriate for resource-rich low-income societies and that which their governments have actually implemented, it is evident that managing resource depletion has proved to be politically difficult. Instead of accelerating the accumulation of public capital, resource rents have enabled the few to plunder what should have benefited the many and the current generation to plunder what should have benefited the future.

However, such outcomes are by no means inevitable—some resource-rich societies have indeed harnessed natural resources for an accelerated transition out of poverty; examples of this are Botswana and Malaysia. Furthermore, societies can and do learn from economic policy mistakes—they are not condemned to repeat them. In many low-income societies, citizens are well aware of past plunder. This is a potential political constituency for policy change.

However, the saving and investment policies required by resource depletion in a LIC are long term: Good decisions need to persist for a generation. An episode of good savings decisions is not enough. The Nigerian experience is salutary. Following a prolonged period of plunder, in 2003, Ngozi Nkonjo-Iweala was appointed finance minister. Inheriting a fiscal deficit, she rapidly turned the budget around and accumulated large foreign savings. But after three years, she was removed from her post, and within five years of her departure, these savings had been dissipated. The lesson from this frustrating experience is that the priority for an episode of reform is to build commitment technologies that can survive their creator.

Building commitment technologies is difficult. First, they have to be designed. The rules appropriate for resource-rich low-income societies are distinctive and so cannot simply be copied from some OECD template. IFIs have an important role here. Even well-designed new rules can easily be ignored or not survive their initiator. Three features are likely to make them more robust—legislation, an institutional champion, and broad popular support.

Legislation has the major advantage that laws are costly to reverse. Constitutional laws are particularly costly to reverse, and so ideally, the policy rules concerning resource depletion should be embedded in the constitution. In turn, getting an appropriate law adopted is a political process. However, the difficulty of reversing the legislation lends itself to a temporary big push such as is common in NGO campaigns. The cost of coordinating political pressure has fallen dramatically, and so legislation has potentially become more feasible.

A new law is inherently fragile: It can easily be overridden, ignored, or reinterpreted by the pressures of personal interest. To counter these pressures, it needs a credible existing organization within government that is tasked with implementing it. For the custodial role of natural assets, the most appropriate public institution is likely to be the central bank.

An institutional rule is likely to survive only if it has broad popular support. For this, it needs to be understood by ordinary citizens. To take a currently pertinent example from another context, Germany’s constitutional opposition to funding the fiscal deficits of other euro area countries is clearly underpinned by the strong memory that Germans have of the hyperinflation of 1923. Resource-rich Africa has already been through equivalently searing experiences of resource plunder. But social learning from mistakes is not automatic; economic events need to be appropriately interpreted. Such interpretation is a public good and has to be supplied. Small low-income societies lack the market size to support high-quality information media, and so citizens are not naturally well informed about economic issues. This is also an appropriate role for central banks.

The law-cum-institution that is most closely analogous to what is needed is the SWF. Many governments have sought to commit both themselves and their successors to prudent decisions through the creation of such funds and they are now becoming fashionable in low-income resource-rich societies. Typically, an SWF has three sets of rules. One determines the size of the flows going into the fund. A second protects the accumulated stock of its assets from being dissipated. The third concerns the composition of the stock, namely, foreign financial assets.

Such a rule structure is not appropriate for a poor country. Whereas in a capital-rich country, it is better to accumulate foreign than domestic assets, in countries that are chronically capital scarce, the investments should ultimately be domestic. Instead of an SWF, there are two rule structures that between them incorporate the three policy clocks. One is a sovereign resilience fund (SRF), and the other is a sovereign development fund (SDF).

Sovereign Resilience Funds

The purpose of an SRF is to ride the tiger of commodity price volatility. It therefore needs one set of rules for the flow of money going into the fund, one for the composition of assets, and one for the flow out of the fund. In steady state, these two flows will balance out over time.

As discussed previously, in a steady state, the rules governing the inflows should be by means of ceilings on the rate of expansion of public consumption and public investment (these ceilings should not be the same since investment needs to grow on average more rapidly than consumption). The SRF should also be used to manage the transition, such as guarding against Dutch disease by avoiding demand for consumption rising ahead of investment in supply.

The composition of the assets should clearly be conservative and liquid as the whole purpose of the fund is to liquidate its assets when times are bad. However, subject to this, the SRF should be free to manage how it meets the calls upon it by combining liquid assets with commodity hedges. Delegating this to a public organization tasked specifically with delivering resilience has a major political advantage. Although hedging is generally the best way of protecting against the risk of price changes, it is very rarely used. The reason for this is well understood: The political costs of hedging are too high. If a finance minister uses precious budget revenue to purchase a hedge, in any particular year, the chances that this will pay off are below 50 percent. As a result, the minister is exposed to the charge that money has been wasted. Because all politicians have enemies, it is a certainty that this charge will be made. Further, there is a considerable risk that if the policy is maintained, it will again fail to pay out, exposing the minister to the added charge of “I told you so.” The best prospect of hedging being adopted is therefore for it to be removed from the day-to-day political arena and delegated to an organization whose sole purpose is to use money for resilience, partly through the acquisition of foreign assets and partly through hedges. There is less political mileage in attacking professional technocrats than politicians, and in any case, the organization can reasonably defend itself on the grounds that it must be judged on its mandate, which is to protect public spending during difficult times.

The rules governing the outflows are correspondingly the maximum permitted rates of reduction in public spending on investment and consumption. As discussed previously, there is a case for endogenizing these rules on the evolving capacity to finance periods of persistent withdrawal.

In good years, an SRF top-slices the revenues from natural resources, whereas in bad years, it supplements them, providing money to sustain both investment and consumption.

Sovereign Development Funds

An SDF is a variant of an SWF. The first two rules of SDFs are the same as in an SWF—they govern flows into the fund and restrict the scope for flows out of it. As to the flow into the SDF, the available flow is preadjusted to the commodity cycle through the SRF. In good times, the SDF has the second call on resource revenues, whereas in bad times, it receives some of its inflow from natural resource revenues and some from the SRF.

Despite being preadjusted for the commodity cycle, the proportion of available resource revenues appropriately devoted to savings is not constant. As previously discussed, through the course of depletion, the savings rate should rise. This may prove to be one complexity too many for the rules of an implementable SDF, in which case a constant-savings approximation will need to be adopted. However, a rule of a rising savings rate is potentially very attractive for a government. Based on the idea of “God make me good but not yet,” it imposes less pain on the government that introduces the legislation than on future governments that are stuck with abiding by it. This enables the average rate out of savings over the course of depletion to be set higher than if the full burden of prudence had to be borne by the initiating government. For once, such a rule is fully justified by the underlying economics.

All moneys flowing into the SDF are irreversibly for the accumulation of assets. This is one reason why it is important to have two distinct funds rather than a single fund—the SRF has to permit withdraws to finance public consumption, whereas the SDF should expressly forbid such withdrawals. The other reason is that the asset composition is very different.

A key decision for the SDF is the balance between the accumulation of foreign financial assets and domestic investment. Whereas the liquidation of foreign assets to fund public consumption is forbidden, liquidation to fund domestic investment is permitted, subject to conditions. The guiding principles in the allocation between foreign and domestic investment should reflect the considerations discussed in this chapter; that is, the pace of implementing the investing-in-investing agenda and any remaining concerns about Dutch disease in the construction sector.

For those funds assigned to domestic investment, a fundamental part of the SDF is to police professional standards of project assessment—design, selection, implementation, and evaluation. In other words, the SDF is the agency enforcing a rising performance in the Public Investment Management Index.

A potentially effective way to strengthen adherence to the rules of the SDF and particularly to the microeconomic procedures for investment projects is to open the fund to contributions from aid donors.3 Most resource-rich low-income countries will continue to receive significant aid inflow to fund development, and channeling aid-for-investment through an SDF would avoid duplication and, in effect, import good economic governance into domestically financed investment.

The Roles of Central Banks

Where, then, should the SRF and the SDF be lodged within the government? The dismal experience of anticorruption commissions illustrates that small, new, and isolated-purpose-designed agencies of restraint are liable to be unsuccessful. Hence, the SRF and the SDF should not be free-standing.

In the typical small low-income society, there are few institutions that are within government but have some independence to restrain economic policy. By far, the most important is the central bank. Both the SDF and the SRF could be agencies under the umbrella of the central bank, reporting through the governor to the government.

The SRF is a natural extension of the current functions of a central bank, that is, reserve management and risk management. The SDF is an extension of this role, but is philosophically still within the overall purpose of custodianship. Just as the central bank is the custodian of the currency and of debt sustainability, in a society in which the major assets are natural resources, it is reasonable for it to have a role in the prudent management of these assets.

Further, because the typical small LIC is short of technically qualified senior public sector officials, it is important to economize on their use. If both the SRF and the SDF are lodged within the central bank, they can both draw readily upon central bank expertise and, indeed, their memberships can overlap. There is no inherent conflict of interest that requires walls between personnel. The importance of keeping the agencies distinct is that the money under their control needs to be subject to different rules and different objectives.

There is a final reason for lodging the SRF and the SDF with central banks. Recall that their authority rests ultimately upon building support among a critical mass of citizens who understand their purpose. Central banks are appropriate for building this constituency of support. As trusted public authorities, they are in a position gradually to build a critical mass of economically literate citizens (a critical mass being a group large enough for these key economic decisions to be well taken). In OECD countries, central banks have increasingly communicated directly with ordinary citizens. In Africa, where there are far fewer other sources of trusted economic information, this role is more important but less developed. The aspirations of Africa’s central banks need to extend beyond the technocratic.

International Support

Finally, the wider international community has a useful supplementary role in setting global standards. The EITI, launched in 2002, was swiftly adopted in Nigeria by Dr. Nkonjo-Iweala, becoming the NIETI. However, although the EITI’s focus on transparency in the reporting of resource revenues was the right place to start, it does not address the savings and investment decisions. The more recent Natural Resource Charter ( covers the entire decision chain, from discovery to investing-in-investing. Already adopted by the New Partnership for Africa’s Development (NEPAD) as a flagship program, endorsed by the African Development Bank, and supported by IFIs, it is designed both for public officials and for citizens. Although the EITI and the National Resource Charter are voluntary codes, there is also potential for enforcement of more ethical practices in resource extraction; a global extension of the Cardin-Lugar Amendment is surely a priority for the Group of Twenty (G-20).4

Africans are well aware of their history of resource plunder. Courageous politicians, responsible central banks, and new international standards must try to make a reality of “this time it’s different.”


    Bhattacharyya, S., and P.Collier,2011, “Public Capital in Resource-Rich Countries: Is There a Curse?” Working Paper No. 2011–14 (Oxford, U.K.: Center for the Study of African Economies, Department of Economics, Oxford University).

    • Search Google Scholar
    • Export Citation

    Collier, P., and A.Venables,eds.,2011, The Plundered Planet: Why We Must—and How We Can—Manage Nature for Global Prosperity (Basingstoke, U.K.: Palgrave Macmillan).

    • Search Google Scholar
    • Export Citation

    Collier, P., R.van der Ploeg, M. Spence, and A.Venables,2010, “Managing Resource Revenues in Developing Countries,” IMF Staff Papers, Vol. 57, No. 1, pp. 84–118.

    • Search Google Scholar
    • Export Citation

    Dabla-Norris, E., J.Brumby, A.Kyobe, Z. Mills, and C.Papageorgiou,2011, “Investing in Public Investment: An Index of Public Investment Efficiency,” IMF Working Paper 11/37 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    Ross, M.,2011, The Oil Curse: How Petroleum Wealth Shapes the Development of Nations (Princeton, New Jersey: Princeton University Press).

    • Search Google Scholar
    • Export Citation

    van der Ploeg, R., and A.Venables,2011, “Harnessing Windfall Revenues: Optimal Policies for Resource-Rich Developing Countries,” Economic Journal, Vol. 121, pp. 1–30.

    • Search Google Scholar
    • Export Citation

Paul Collier is affiliated with the Centre for the Study of African Economies, Department of Economics, Oxford University.


The utilitarian framework sees no ethical problem in a poor current generation raising its consumption at the expense of future generations by using up the national endowment of natural assets. However, although this ethic is common among economists, it is probably not shared by many ordinary citizens. In particular, many young people may feel that the generation in power has a responsibility of stewardship toward natural assets—if natural assets are exploited, an equivalent value should be passed on to future generations. As I argue in The Plundered Planet (Collier and Venables, 2011), this need not imply a reversion to the permanent-income implication of a 100 percent savings rate. In a capital-scarce economy, properly invested savings should be able to generate a return above any return earned by leaving the natural assets in the ground (the world rate of interest if the Hotelling rule were to be relied upon). For example, if the rate of return is double the world interest rate, the present generation can satisfy the obligation of passing on equivalent value to future generations with a savings rate of only 50 percent. Hence, both utilitarian ethics and the ethics of stewardship imply qualitatively the same strategy: Savings rates out of income from natural resource extraction should be above that on other income but less than 100 percent.


The Penn World Table provides some evidence of high unit costs but is not suited to benchmarking for tracking improvement.


I am indebted to Kerfalla Yansane, Minister of Finance, Government of Guinea, for this suggestion.


The G-20 is an informal group of 20 major economies (19 countries plus the European Union) and representatives from the International Monetary Fund and the World Bank. It has met yearly in a central bank governors and finance ministers format since 1999, and in a chiefs of state and head of governments format (G-20 summits) since 2008.

    Other Resources Citing This Publication