Chapter

3 Principles of resource taxation for low-income countries

Author(s):
International Monetary Fund
Published Date:
April 2010
Share
  • ShareShare
Show Summary Details
Author(s)

1 Introduction

The taxation of extractable natural resources poses complex design problems-and indeed the chapters of this book address many of these in detail. These complexities arise because natural resources are not akin to most other economic activity: their distinctive features make government central. In low-income countries the problems that are generic to the taxation of natural resources in all contexts are compounded by important additional features which make the solutions appropriate for a high-income country inapplicable.

The chapters in this volume largely focus on this distinctive low-income context. To date, most of the work on tax design has been for high-income countries, and I will try to set out why the distinctive features of low-income countries change the policies that are appropriate. The new website www.naturalresourcecharter.org complements both this chapter and this book in setting out for resource-rich low-income societies the entire decision chain involved in harnessing natural assets for transformative development. However, as a preliminary it may be helpful to set out the four generic features of natural resource extraction that make it distinctive from normal economic activity. These are that the ownership of natural assets is rightly vested in citizens; that extraction is a process of asset depletion rather than merely production; that investment in extraction requires high sunk costs and long periods of payback; and that the prices of depleting assets are volatile. Since the rents from extraction belong, in their entirety, to citizens, the government as their agent needs a tax regime which captures these rents, over and above the standard taxation of profits. If the tax system does not discriminate between rents and returns to other factors of production then it is sure to be misdesigned. In practice, this implies that the taxation of resource extraction is likely to look quite different from that of most other economic activities. Because resource extraction is depleting an asset it is not sustainable, and so the savings rate out of these revenues should be higher than that out of ordinary taxation. Finally, because prices are volatile, rents and profits will also be volatile.

In Section 2A lays out the distinctive features of low-income resource-rich countries. Section 3A suggests how these features make the policies that are conventional in high-income resource-rich countries inappropriate. Section 4A sketches what more appropriate policies might look like, although it should be evident that to do this thoroughly is an undertaking well beyond the scale of this brief chapter.

2 Four distinctive features of low-income countries

A Discovery is key

The first distinctive feature of low-income countries is that the discovery process is likely to be far more important than in the high-income resource-rich countries. A snapshot of discovered natural assets for the year 2000 assembled by the World Bank brings this out. In the OECD the average square kilometre possesses known sub-soil assets to the value of $125,000, whereas the figure for Africa is only $25,000. Since both land masses are enormous such a large difference is unlikely to reflect differences in luck: the original endowments of sub-soil assets were probably not very different. Further, since the OECD has been depleting its natural assets for far longer than Africa, a reasonable expectation is that Africa has more sub-soil assets remaining than the OECD. Of course, even in the OECD by no means all natural assets have yet been discovered: discovery is costly so there is little incentive to prove reserves that will not be exploited for decades, and as the technology of discovery improves more becomes economic. The implication is that a large majority of Africa’s natural assets remain undiscovered. The predominant reason for this is presumably that the incentive regime is less conducive to discovery. This is supported by the substantially lower density of drilling in the major sedentary basins of Africa compared to those in the OECD. Since Africa has radically less invested capital, physical and human, than most other regions, its successful management of its extensive undiscovered natural assets is both absolutely and relatively far more important: the design of an appropriate tax regime for resource extraction is a first-order issue.

B Commitment problems

The second distinctive feature of low-income countries is that their institutions are less robust. They lack the sanctity of time, and any particular institution is likely to be less well-defended because other institutions are weak or missing, and because there are fewer supports from the neighbourhood. If institutions are not robust then the credibility of government commitments is impaired: even if everything is currently satisfactory it is less likely to stay that way. There is only a limited amount that a government can do to reduce doubts about the future and so it is necessary to recognize the consequences of the limited credibility of commitments. There are two respects in which this is particularly pertinent in respect of natural resources.

The first is that the extraction process typically requires massive initial investment which need not then be renewed. In this respect the time profile of investment in the extractive industries is highly distinctive. For example, in manufacturing it is likely that investment gradually builds up over the decades, so that a wise government knows that should it attempt to expropriate accumulated investment through heavy taxation it will kill the valuable process of future investment. In contrast, investment in resource extraction faces a time-consistency problem: the initial investment is so large relative to all future investment that once made it is rational for the government to confiscate it. Fearing such an eventuality the extraction company decides not to make the investment in the first place and the government, despite being worse off than if it could credibly commit not to impose such taxation, is unable to do so.1

The second respect in which the lack of a commitment technology matters is that the government may find it difficult or even impossible to commit not to spend all the revenues from asset depletion on consumption. Yet the inability to make such a commitment may imply that it is wiser to leave the assets unexploited until the commitment problem has been overcome.

C Capital and consumption scarcity

The third respect in which low-income countries are distinctive is that both consumption and capital are scarce. As the economy gradually converges with richer ones the marginal value of consumption will fall, but the society is unable to borrow for consumption now as much as would be appropriate because it is rationed in capital markets. Similarly, the rate of return on capital is likely to be high because capital is so scarce.

D Asymmetric information

The final distinctive respect of low-income countries is that their governments are likely to be a severe informational disadvantage vis-à-vis resource extraction companies. Governments are not able to recruit civil servants with the requisite specialist knowledge, due both to a shortage of nationals and the inability of government pay-scales to match private rewards. Specialist information can be purchased on the global market and is typically well worthwhile, but because it is expensive and hires non-nationals, many governments do not buy enough of it.

3 Principles appropriate only for high-income countries

I now set out three conventional principles and explain why they are only appropriate in the context of high income countries.

A Integrated budgets

The principle of an integrated budget is Fiscal Economics 101. The advantage of pooling all revenues without any prior earmarking is evident: it enables the marginal benefit of public spending to be equated across all components of spending, and it enables flexible responses to unanticipated circumstances which change the relative values of the components. These are powerful arguments, not to be dismissed lightly.

However, they presuppose a context in which the government is able to function extremely well. In particular, the preservation of flexibility, which is the great achievement of an integrated budget, comes at no cost. Yet in other contexts the case for commitment technologies is now fully accepted in both academic and policy circles. In particular, the independence of central banks has, over the past three decades, become a standard commitment technology against inflation. In resource-rich low-income countries the key need for a commitment technology is not monetary but fiscal, and the key fiscal issue to be addressed is the replacement of depleting natural assets with other assets, real and financial. Where it is possible, the equivalent of a constitutionally independent central bank might be a fiscal constitution. Essentially, what such a constitutional provision would need to do would be to ring-fence a substantial part of the revenues from natural resources from expenditure on consumption. However, as discussed below, it will normally be appropriate to spend savings on domestic investment, and so the Future Generations Fund model in which revenues do not even reach the budget is not appropriate. Rather, the revenues need to be earmarked for investment. As discussed below, this still leaves an important role for periodic accumulation of foreign financial assets, but the role is essentially to buy time, putting a brake upon the rate of increase in domestic investment until the capacity to invest is enhanced.

Why might such a fiscal commitment technology be necessary? The clear answer is that there are strong day-to-day political pressures for subverting resource revenues from investment into public consumption. The interest of the future is at best only fitfully represented in the political market place. A far-sighted Finance Minister, acting in the long-term national interest, would indeed want to create commitment technologies for defending the future against the potent special interests of the consumption lobbies. In the OECD societies political institutions and the sophistication of electorates may have evolved to the stage at which such commitment technologies are unnecessary. In the resource-rich, low-income societies this is clearly not the case.

Such institutions for earmarking some revenues to savings and ultimately to investment are only in their infancy and have suffered from substantial design flaws. The College in Chad attempted to ring-fence resource revenues but earmarked them not for investment but for particular social uses. These social priorities were rapidly weakened by the government. The Nigerian Fiscal Responsibility Bill attempted to earmark a proportion of oil revenues for savings, but initially ran foul of constitutional requirements to share revenues with the state governments. In general, earmarking a substantial proportion of resource revenues for asset accumulation curtails a degree of flexibility, which is undesirable. The need for a commitment technology, however, overrides concerns about the loss of flexibility. Nevertheless, as earmarking becomes more specific as to which assets should be accumulated, it increasingly contravenes the valid principles of an integrated budget.

B Permanent income and future generation funds

If the government and firms of a country can borrow on world capital markets at an interest rate very close to that at which they can lend, then the country will already be developed. In particular, it will have borrowed sufficient to drive down the rate of return on domestic investment to the world interest rate. As Ploeg and Venables (2008) argue, this is the condition necessary for the permanent income hypothesis to be the appropriate guide for policy. With this condition fulfilled, on the discovery of a natural resource consumption would leap to a permanently sustainable level and as natural assets were depleted they would be offset by the accumulation of foreign financial assets. Note that even in this scenario the discovery would be followed by an initial phase of borrowing: consumption should leap on the discovery while revenues will take time to come through.

Manifestly, this is not the context for a low-income country. Such countries are not able to access world capital markets sufficiently to finance the massive investment needed to drive down the return on domestic capital to world levels: they are capital-scarce. This has two important corollaries. First, because current generations are much poorer than future generations, some of the revenues should be consumed: the permanent income approach of consuming only the sustainable income from the natural assets no longer has a sound analytic foundation. In low-income countries the appropriate use of natural assets is to accelerate the evolution towards the eventual level of sustainable income, whereas under the Permanent Income Hypothesis (appropriate for a developed economy) it is to raise that eventual level. Second, because domestic rates of return are above world rates, such savings as are appropriate should gradually be directed into domestic investment rather than foreign financial assets. This needs at once to be qualified. As the pace of investment is increased the returns on investment fall below the returns on installed capital because of congestion and inefficiencies in the investment process. Hence, the pace of investment needs to be set by the capacity to absorb it efficiently. However, the accumulation of foreign financial assets is not the solution to this problem; it merely buys the time in which to address it. In these economies development is fundamentally about raising the capacity to invest productively. The process can be thought of as ‘investing in investing.’ It is an agenda for the real economy: improving bureaucratic procedures to design and implement public investment; enhancing the efficiency of the capital goods producing and distributing sectors; and increasing incentives for private investment. A policy of financial asset accumulation should not detract from this by weakening the sense of urgency. Nevertheless, it is often necessary to buy time. A classic instance of the consequences of attempting to ramp up investment ahead of the capacity to implement it efficiently was the Nigerian ‘cement armada’ of 1975. In this instance the uncoordinated and excessive purchase of cement encountered the bottleneck of limited port capacity and dissipated expenditures on investment in avoidably high costs.2

C Excess profits taxes

Natural resource extraction generates both normal profits and rents: the latter need to be captured by the government. Since both normal profits and rents are aggregated into reported profits, the first-best is to decompose reported profits into its two components, applying the normal corporate profits tax to normal profits and imposing a very high ‘excess profits’ tax on the rents. The alternative of a royalty payment on resource revenues, however, structured, is second-best because it cannot target the rents as precisely as the excess profits tax. For example, as full depletion approaches and extraction costs mount the company will choose not to extract those resources which incur a royalty in excess of the diminishing rents and so some rents will be left unexploited; other resources may be left unexploited.

The problem with any form of taxation is that information is costly and held asymmetrically: the company knows the true division between rents and profits but has no incentive to reveal it. On the contrary, where the government has little information the company has considerable scope for concealing profits altogether by reclassifying them into costs. While these problems are generic to all forms of taxation, they are far more acute with the taxation of resource rents.3 Whereas tax rates on profits that result from capital and risk are typically around 25 percent, in principle the tax rate on excess profits should approach 100 percent. The incentives to cheat are thus radically greater, and the scope for cheating is increased by the co-existence of two conceptually distinct forms of profit. As a result, whereas within the OECD the first-best is unambiguously the right policy, in the context of small, low-income countries it is at least debatable. The choice in tax design therefore reduces to one between an excess profits tax that will be gamed by companies unless resources are spent to counter it, and a royalty which, though inefficient, may be harder to game because revenues are more observable than profits. In this situation it may no longer be possible to navigate by the simple principles which rank the excess profits tax as analytically superior to a royalty, and a good system may combine elements of both.4

4 Rethought principles

If the principles that are appropriate for a resource-rich country in the OECD are not appropriate for the typical resource-rich low-income country then policies should look different. Norway and Timor-Leste both have oil, but their policy responses should be different. How different should they be?

A The discovery process

Recall that the discovery process is far more important in low-income countries than in the OECD: there is much more to be discovered. However, at the discovery stage the lack of a credible commitment technology imposes compounded risks onto investment in prospecting. The company is uncertain both as to whether anything will be found, and what the eventual tax regime will be. A pre-commitment to a tax regime which is based on inadequate geological information will lack credibility. As a result, if the incentive for discovery is that the company will acquire extraction rights to whatever it discovers, the expected value of these rights will be heavily discounted by these uncertainties. Further, the rate of discount used by the typical resource extraction company is very high.

To the extent possible the government should not sell extraction rights until geological uncertainties have been reduced. The objective is not for the government itself to take on all the risk of prospecting, but to narrow likely outcomes to a sufficiently narrow range that contingent tax arrangements are regarded as credible. The government can collate and commission seismic data. Since the rate of return on private prospecting is typically high, these costs would be an appropriate use for aid: the donor is able to bear the risk, and the aid will on average have a high return.5 This implies that the government should, to the extent possible, separate the prospecting process from the extraction process.

B Auctions for price discovery

Once the government has good geological information it can then auction the rights to extraction. The auction would essentially reveal the appropriate rate of taxation or royalty. The design of auctions is complex,6 but they are the best way of tackling the acute asymmetry of information, and also, if properly supervised, of tackling the scope for corruption inherent in negotiated deals. Auctions are particularly appropriate where citizens are suspicious of government because, if verified by independent international scrutiny, they can enable a government to signal to its citizens that their suspicions are unwarranted.

There is likely to be a need for pre-screening of bidders. Typically the ideal number of bidders is around four: many more than this and no company invests enough in information to judge true value so that bids are liable to be opportunistic; much less than four and there is a risk of collusion. Since the exclusion of bidders is replete with opportunities for corruption this stage should also be subject to international verification.

C Geared royalties

If information is sufficiently asymmetric then a royalty may be the best option. In this case can we say anything about its design? It would need to be conditioned upon those observables which cannot readily be gamed, such as the price of the commodity and some basic features of geology. Since what can be observed depends upon the expenditure of the government upon monitoring, as monitoring is enhanced profits themselves become observable. Where, however, profits are not realistically observable, the royalty will generate less grounds for dispute the more it is anchored to those observables with clear consequences for profits. For example, in respect of the world price of the commodity, one feature that is at once apparent is that rents will be increasing more than proportionately in the price: there is some unobservable but positive price at which rents are zero. Hence, a (second-best) efficient royalty should be highly geared to the price of the commodity. The conventional practice of setting the royalty at a flat rate of 3 percent fails to satisfy this design rule.

D Pace exploration by absorption of investment

Above, I have discussed the need to pace investment by the rate at which it can be absorbed. What should be done with resource revenues that are substantially in excess of this level? The answer may well be that they are best not generated: resources can simply be left undiscovered. The advantage of leaving some resources undiscovered is that the economic pace of extraction of those resources that have been discovered, which is gradual, provides an automatic commitment mechanism. In contrast, resources accumulated in foreign financial assets can be no more robust than the constitutional provisions which protect them from rapid liquidation, and in low-income countries constitutional provisions have often proved to be fragile. However, building up financial assets has offsetting advantages: in particular it diversifies the asset portfolio away from dependence upon the commodity that is being extracted. Hence, the appropriate strategy is determined by a balance of risks. The risks that commodity prices will appreciate by less than the world interest rate can at least be estimated from the past history of prices; the risk that a future opportunistic regime will liquidate accumulated financial assets cannot be readily estimated but may reasonably be judged so substantial that it dwarfs the additional risk implied by the lack of portfolio diversification.

In this case, the rate of resource exploration should be matched to the ability of the economy to absorb domestic investment. Evidently, the latter is amenable to policy, and so augmenting the capacity to invest is a high priority.

E Borrowing, but only for appropriate uses and with appropriate signals

The conventional concession to the special conditions of low-income countries is to advise their governments not to borrow in anticipation of resource revenue. Indeed, the most conservative variant of this advice is to use all the resource revenues to accumulate foreign financial assets, and to increase consumption only by the rising income stream from these accumulating assets, this being the ‘bird-in-hand’ rule.

In practice, governments try to avoid the need for borrowing by advancing revenues through signature bonuses. For reasons discussed above, the true interest rate on signature bonuses is likely to be high (though lower than non-securitized borrowing which may well be prohibitive) and so they are a poor form of borrowing compared to loans from public agencies. Some borrowing can be appropriate and it would be useful if the international financial institutions developed financial instruments to support this need: for example, an International Bank for Reconstruction and Development window. However, the problems for the government are partly of prudence and partly of signalling to its own citizens.

Commodity prices are so volatile that the safe assignment of revenues to consumption is very low. For example, in the first quarter of 2008 when the current oil price was $115, based on its past volatility the 95 percent confidence interval for the forecast of the price in the first quarter of 2009 was in the very wide range $65-$200. Hence, the ‘safe’ revenue estimate would have been only around half the current price. Yet even this proved to be far from safe, the actual price being only around $43. The prudent approach to this extraordinary volatility is that borrowing for consumption should be kept to very low levels. However, borrowing to finance investment is far less risky. The government is not taking on a liability backed only by the highly uncertain future value of its natural assets: the borrowing is also backed by its new investment. The rationale for borrowing for investment is that the country can thereby get started on ‘investing in investing’ several years earlier than if it were to wait for the natural resource revenues to come on-stream.

Two types of governments would wish to borrow in anticipation of future resource revenues, the very good and the very bad. The very good government astutely recognizes that consumption now is much more valuable than consumption in the future because of current poverty. The very bad government simply wishes to plunder the future so as to enrich its members. Since citizens can be presumed to be well aware of the dangers of borrowing for plunder, the problem facing the very good government is to signal to its own citizens that it is indeed not of the plundering type. In the standard theory of signalling, the solution to this problem is for the good government to adopt a strategy that would not be imitated by the bad government: what might this be in the present instance? The most promising approach is for the spending from borrowing to be earmarked to uses which cannot directly benefit members of the government, but which clearly directly benefit ordinary citizens. An example of such expenditures is a bursary paid directly to school children. By linking the borrowing to such a use the good government reveals its type.

F An application: China in Africa

How might these rethought principles affect the assessment of what is surely the single most important new resource-related phenomenon: the deals being struck between China and various African governments for infrastructure in return for extraction rights?

On the conventional principles these deals are unambiguously undesirable. They are non-transparent, and instead of revenues flowing into the budget they are earmarked for a particular form of spending. On conventional principles the deals would be far better unbundled into an extraction contract, with revenues going into the budget, and then construction contracts financed by all or part of the public spending supported by the revenues.

How might the issue look differently given the issues raised above? First, the Chinese approach offers a new commitment technology: resources extracted are, with certainty, offset by the accumulation of a domestic asset. A wise Finance Minister may reasonably decide that this is much safer than letting the revenues flow transparently into the budget and then hoping to emerge triumphant from the subsequent political contest for spending. Second, the Chinese approach bypasses both the civil service and domestic construction companies and so relaxes the constraint upon domestic absorption of investment. Of course, this bypass may in some contexts be undesirable: it might be better to generate local employment in the construction sector even if this slows down the pace of investment.

These two advantages are real and substantial: in effect, the Chinese have innovated rather than merely undermined existing practices. The appropriate response is therefore to learn from the innovation and to improve upon it. It would, in fact, not be difficult to improve upon the current Chinese model. Its limitation is not that the extraction and construction contracts are bundled, but that China is currently a monopolist in this form of packaged contract. The appropriate response is therefore for other consortia of resource extraction companies, construction companies and donors to compete with China. Competition could then be fitted into the framework proposed above, namely auctions. Where a government determined that a packaged approach would be advantageous the auction would be conducted in terms of the amount of infrastructure provided for a predetermined set of extraction rights. Prior to the auction the government would set out a prioritized listing of desired infrastructure. The auction would reveal the best value: the bid that undertook to go furthest down the ranked list. Transparency would come about not through unbundling the contract, or insisting on its components being individually priced, but through the process by which the packaged contract was awarded. As with other auctions, bids would need to be screened for credibility. Additionally, there would need to be a specified and credible process for monitoring the quality and timeliness of infrastructure provided, including penalties for non-performance. Such matters are not trivial and may sometimes make the entire process so unsatisfactory that the unbundled approach is clearly superior. The ability to manage the process might be enhanced if an agency such as the World Bank provided loans available to winning consortia in return for standardized procedures and verification.

5 Conclusion

In this brief overview my purpose has been to highlight the implications of the profound differences between those resource-rich countries that are at OECD levels of income, and those that are impoverished. The economic principles for taxing resource extraction imply that the way in which natural assets are harnessed for society should differ considerably in Australia, Canada and Norway on the one hand, and in Angola, Chad and Timor-Leste on the other.

This point is important because to date virtually all the serious analysis has been conducted with reference to the OECD economies. Currently, those Finance Ministers from low-income countries who are most concerned to manage opportunities well look to the OECD models for guidance: for example, this is manifested in the application of what is often wrongly imagined to be the ‘Norwegian model’ to contexts which are wildly different from that of Norway. In recent years some 50 governments of resource-rich countries have approached the government of Norway for advice. Yet, as the government of Norway is careful to explain, there is no ‘Norwegian model.’ For example, the high-profile Sovereign Wealth Fund was not begun until some 30 years after natural resource revenues had started: until then they were deployed domestically.

It is one thing to criticize the inappropriate application of an OECD model, it is quite another to replace it with principles that are appropriate. In this paper I have merely sketched the outlines of what needs to be a substantial undertaking.

1Several chapters in this book focus on this time consistency issue: Boadway and Keen (2010) review what theory has to say about possible responses, Daniel and Sunley (2010) focus on experience with one of these – fiscal stability agreements – and, an interesting illustration of the importance of strong institutions in this context, Osmundsen (2010) discusses how Norway has managed to achieve substantial credibility in its petroleum tax regime.
2The appropriate use of resource revenues in low-income countries is discussed more fully in Collier et al., 2010.
3Experience with the design and implementation of rent and other resource taxes in low income countries are discussed elsewhere in this volume by Calder (2010)Land (2010).
4See Boadway and Keen (2010) for a formalization.
5This possibility was raised by a few commentators in response to earlier mineral price booms, see Garnaut and Clunies Ross (1983: 61).
6Cramton (2010) provides a detailed treatment of auction design for the resource sector.
References

    BoadwayRobin and MichaelKeen (2010) ‘Theoretical Perspectives on Resource Tax Design’ inPhilipDanielMichaelKeen and CharlesMcPherson (eds.) The Taxation of Petroleum and Minerals: Principles Problems and Practice.

    • Search Google Scholar
    • Export Citation

    CalderJack (2010) ‘Resource Tax Administration’ inPhilipDanielMichaelKeen and CharlesMcPherson (eds.) The Taxation of Petroleum and Minerals: Principles Problems and Practice.

    • Search Google Scholar
    • Export Citation

    CollierPaulRick vander PloegMichaelSpence and AntonyVenables (2009) Managing Resource Revenues in Developing EconomiesIMF Staff Paper advance online publicationJuly212009. Available online at http://dx.doi.org/10.1057/imfsp.2006.16, doi: 10.1057/imfsp.2009.16. Last seen: March22010.

    • Search Google Scholar
    • Export Citation

    CramtonPeter (2010) ‘How Best to Auction Natural Resources’ inPhilipDanielMichaelKeen and CharlesMcPherson (eds.) The Taxation of Petroleum and Minerals: Principles Problems and Practice.

    • Search Google Scholar
    • Export Citation

    DanielPhilip and EmilSunley (2010) ‘Contractual Assurances of Fiscal Stability’ inPhilipDanielMichaelKeen and CharlesMcPherson (eds.) The Taxation of Petroleum and Minerals: Principles Problems and Practice.

    • Search Google Scholar
    • Export Citation

    GarnautRoss and AnthonyClunies Ross (1983) ‘Taxation of Mineral Rents,’The Economic Journal Vol. 94 pp. 427428 (Oxford: Clarendon Press).

    • Search Google Scholar
    • Export Citation

    LandBryan (2010) ‘Resource Rent Taxation: Theory and Experience’ inPhilipDanielMichaelKeen and CharlesMcPherson (eds.) The Taxation of Petroleum and Minerals: Principles Problems and Practice.

    • Search Google Scholar
    • Export Citation

    OsmundsenPetter (2010) ‘Time Consistency in Petroleum Taxation: Lessons from Norway’ inPhilipDanielMichaelKeen and CharlesMcPherson (eds.) The Taxation of Petroleum and Minerals: Principles Problems and Practice.

    • Search Google Scholar
    • Export Citation

    PloegR. van der and Anthony J.Venables (2008) ‘Harnessing Windfall Revenue in Developing Economies’ Discussion Paper No. 6954CEPR (London, United Kingdom).

    • Search Google Scholar
    • Export Citation

    Other Resources Citing This Publication