The risk premium for evaluating public projects.
The other solution [to highway finance] is to finance the project wholly in the public sector, either with government or multilateral funds. It is, after all, more expensive to raise debt on a project finance basis. When considered alongside the guarantees and commitments which have to be provided to attract commercial finance, the best approach would be to borrow on a sovereign basis. (Euromoney, 1995)
The view that `private-sector capital costs more' is naive, because the cost of debt both to governments and to private firms is influenced predominantly by the perceived risk of default rather than an assessment of the quality of returns from the specific investment. We would lend to government even if we thought it would bum the money or fire it off into space, and we do lend to it for both these purposes. (John Kay, 1993)
These two quotations provide a flavour of the opposite positions that are being advocated for the finance of infrastructure projects. In an ironic twist, a trade magazine for the private sector advocates government solutions, whereas a prominent advisor to the current leader of the Labour Party in the United Kingdom, takes the opposite view.
One may look at the issue as follows. For example, when a toll-road is financed on the basis of sovereign borrowing, the tolls will be lower than if the road had been funded by private investors. But the taxpayers will have assumed a contingent liability for which they are not remunerated, i.e. there is no risk premium in the sovereign borrowing rate. In fact, the taxpayers play the role of investors bearing the risk of failure. The question is, who is better placed to bear (as opposed to `manage') the risk -- taxpayers or investors? If taxpayers are just as good or bad at bearing risk as investors, then they should ideally receive a remuneration equivalent to the risk premium demanded by investors. If they are better at bearing risk, then project finance by way of sovereign borrowing is `truly' cheaper than private funding, because the cost of risk-bearing is lower, i.e. taxpayers would be less interested in buying insurance -- via charging risk premiums -- than investors.
Obviously, the view one takes on the `true' cost of sovereign funds affects one's attitude toward the role of government in financing projects. If government funding is really cheaper than private finance, then the government should have a greater role. At the same time, it is clear that the cost of funds is only one argument among others that determines what governments should or should not do.
This paper takes the view that the apparent cheapness of sovereign funds reflects the fact that the taxpayers,(2) who effectively provide credit insurance to the sovereign, are not remunerated for the contingent liability they assume. If they were to be remunerated properly, then the advantage of sovereign finance would -- almost by definition -- disappear.
If one takes that view, then the government should, in principle, not participate in the financing of projects, either with debt or with (quasi) equity. There are other useful roles for the government (see Annex 1), but there is no argument for government funding of projects based on the cheapness of government funds.
Also, there is then no argument for the typical `lobbyist' position on private finance, which argues for privatization of some sort (because the private sector is more efficient), but then requests financial support from the government or an agency underwritten by government (because government funds are cheaper).
The implication of the argumentation that government funding is not cheaper than private funding runs counter to the worldwide practice of using risk-free rates to discount cashflows in public-sector project appraisal (cost-benefit analysis), whereas risk-adjusted discount rates are used in private appraisals. Current practices of cost-benefit analysis introduce a bias against private solutions in decisions about projects.(3)
The paper starts with a review of both basic positions and some of their implications, followed by an analysis of the rationales for each view. The policy implications of this view are drawn out. Annex 1 sketches some other arguments affecting the role of government in project finance, using, in particular, examples from highway finance.
The arguments are crucial for many current policy debates. The governments of various Australian states, Canada, New Zealand, and the United Kingdom are grappling with them in the context of promoting all sorts of private investment, for example, under the British private finance initiative. In France and Brazil, the arguments are currently debated because they would imply that politically favoured nuclear and hydro plants would become uneconomic. In Britain, the non-governmental organization, Friends of the Earth, has drawn the conclusion that advocacy of privatization is the way to go for groups opposed to nuclear plants -- thus implicitly supporting the arguments presented here. In Colombia, the arguments pop up as the government is introducing a system to value contingent fiscal liabilities. To value them, one needs to discount cashflows and doing so at the risk-free rate would favour public over private investment. Some implications for the role of development finance institutions ate sketched in the conclusion.
II. THE BASIC POSITIONS STAKED OUT
(i) `Private-sector Capital Costs More'
The view that private-sector capital costs more starts from the observation that, in general, sovereign borrowing is cheaper than borrowing by private firms.(4) This reflects the lower default risk of the sovereign, which tends to be able to raise money by way of taxation including the inflation tax. As a result, the interest rate on sovereign debt is commonly called `risk-free.'
To put this in the context of the example of an independent power generation project, a privately funded project would have to charge higher tariffs than a publicly funded project -- everything else being equal. Why then is anyone advocating private power projects?
`The public sector cannot fund all the project needs'
One commonly heard answer is that the government cannot be expected to fund the enormous projected requirements for new projects. But somehow the private sector is expected to be able to do so. This answer is not convincing per se. Total saving (domestic and foreign) available to fund investment is unlikely to be affected by the shift from public to private projects. Why should the private sector be better able to tap these savings than the public sector -- for the very same projects?
For there to be substance to the argument the government must be credit-rationed, while the private sector is not. In other words, government debt would need to be more risky than private debt. This is conceivable when government default is a possibility, i.e. when the government is not able to print the means of payment itself.(5) In a sense government is, then, like a firm that is not able to borrow for the great new project it has identified but has to let others invest in it, because it is already overextended. Overextension means the government (the firm) has financial obligations owing to other projects, which may render service of new financing difficult. So the new project needs to be incorporated separately and independently -- the definition of project finance. Alternatively, a creditworthy firm could fund the project on-balance-sheet. In any case, the government is a riskier, i.e. more expensive borrower -- contrary to the basic assumption that `private-sector capital costs more.'
`Government funding is cheaper, but private management is better'
When governments are not credit-rationed, they tend to be able to raise funds at lower rates than private firms. Still, to advocate private projects in this situation one would need to assume that the government's funding advantage is more than offset by the private sector's management efficiency.
An example of this argument can be found in the World Bank's World Development Report 1994:(6) `In infrastructure projects, the cheaper credit available to governments needs to be weighed against possible inefficiencies in channelling funds through government.' The report cites an example where a 3 per cent interest-rate advantage of government funding would need to be offset by cost savings of more than 20 per cent for private ownership to be advantageous.
There are a number of cases where efficiency gains of 20 or more per cent have, indeed, been achieved by transferring management to the private sector.(7) But efficiency gains vary widely. They may amount to only a few per cent and, in some important cases, it is not clear whether private is really more effective than public management, particularly when private enterprise is subject to monopoly regulation. For example, the most thorough study of the relative efficiency of private- versus public-sector power utilities in the United States suggests that -- controlling for differences in cost of capital, tax treatment, and other factors -- public utilities are more efficient than investor-owned ones (Kwoka, 1996). Should the United States, therefore, nationalize all power utilities? If we believe in the basic arguments about the cost of government borrowing, the case would seem strong, because public-sector utilities deliver slightly cheaper service, even when one does not add in the borrowing cost advantage.
If one believes that a trade-off between low-cost government finance and greater private-sector efficiency exists, then one needs to be careful with privatization unless it is reasonably clear that private management will yield sufficient benefits to offset the lower cost of sovereign funds. By the same token, private management is then justified in clamouring for government financial support up to the point where its own incentives to perform efficiently are undermined.
This line of argument also lies behind calls for government or multilateral financial participation in private projects. Such calls are, for example, made by the Institute for International Finance (IIF),(8) which suggests World Bank lending to the private sector. As is well known, a number of infrastructure projects benefit from some type of government financial participation (beyond guarantees for policy failures), for example, Hungary's M5 motorway project.
Extending this argument, then, means that privatization of infrastructure should often be reserved for countries with highly inefficient government enterprises. But countries with good public enterprises could gain relatively little from privatization. Given the cost advantage of public borrowing, they should then not privatize. In particular, the cost of government funds would appear cheaper than private equity.
(ii) `There is No Real Cost Advantage to Public Borrowing'
How can one deny that sovereign borrowing is not cheaper than private finance? A task force of the Australian government on private infrastructure has formulated the position thus:
[The task force] rejected the argument that the cost of government debt is necessarily cheaper than the private-sector cost of capital (which would have implied that government should finance most infrastructure investment). The task force argued that governments' lower cost of funds largely reflects the fact that taxpayers are providing an implicit guarantee for project risks under public ownership. Thus, it concluded that much of the difference in the private and public cost of capital is apparent rather than real. (Economic Planning Advisory Commission, 1995, p. 37)
In other words, if the taxpayers were to receive appropriate remuneration for the insurance they provide, the apparent cost advantage of government finance would disappear. If this were so, there would be no trade-off between the cost of sovereign funds and efficiency gains from private management. It would still be true that in countries with good public enterprises the benefits from privatization may be lower than in countries with bad ones -- assuming equally good private firms. Yet, privatization could still be advocated, even if it brought only very small improvements. By the same token, there would be no rationale for private firms to ask for government financial support in the form of debt or equity or guarantees of credit obligations. To take the example of World Bank instruments, there would be no rationale to use straightforward loans or credit guarantees -- only policy guarantees and like instruments would be justified.(9)
Government efforts should be directed to improving macroeconomic stability, improving the policy environment for private projects, and deregulating and liberalizing the financial sector to allow private entrepreneurs to put together the best possible deals. Multilaterals should support governments in these efforts but stay away from taking on commercial risk. This position has, for example, been forcefully expressed by Charles Frank of GE-Capital (Frank, 1995) and underlies the philosophy of the World Bank's new policy guarantee programme.(10)
III. THE UNDERPINNINGS OF THE BASIC POSITIONS EXPLORED
Is one of the basic positions the right one or is there some middle ground? In the following discussion the analytical underpinnings for a possible answer will be explored. To do so, a central position will be taken, namely that -- from a normative point of view -- the government has no interests apart from those of the individuals it represents. Government policy should thus express and represent the interests of individual citizens. It may not do so in practice, but such deviations from the will of the citizens would be illegitimate and could not be adduced to justify policy.
(i) The Cost of Risk-bearing and the Choice of Discount Rates
The core issue can then be recast as follows. Differences in the cost of finance or capital reflect differing abilities to bear risk or cope with it. In a seminal article written by Nobel prize winners, K. Arrow and R. C. Lind (1970), the essence of the controversy is introduced as follows:
It is widely accepted that individuals are not indifferent to uncertainty and will not, in general, value assets with uncertain returns at their respected value. For example, risk aversion implies that an individual would be interested in trading a gamble with a 50-50 chance of winning or losing US$5,000 for the certainty of neither gaining nor losing anything and would be willing to pay some positive sum -- not choose investments to maximize the present value of expected returns, but to maximize the present value of returns properly adjusted for risk. The issue is whether it is appropriate to discount public investments in the same way as private investments.
This issue is equivalent to the basic question here of whether the cost of sovereign borrowing is cheaper than that of private finance, because the cost of capital determines the discount rate for projects.(11) Projects have to earn at least the cost of capital to be acceptable to investors. Discounting project cashflows by the cost of capital, therefore, is an appropriate method to identify viable projects. If the cost of risk-bearing in the public sector, i.e. of the taxpayers, were the same as in the private sector, then the discount rates would be the same. The lower cost of sovereign borrowing would simply be due to the fact that government forces taxpayers to provide funds without remunerating them for the risks they take. If the cost of risk-bearing of the taxpayers were, however, lower than that of private parties, then the lower cost of sovereign borrowing would be due to a `real' advantage, i.e. a lower cost of risk. Note that each individual, whether in the role of taxpayer or investor, has identical attitudes to risk, although such attitudes will vary among individuals.
Arrow and Lind argued that government finance was indeed cheaper than private finance. They claimed that the government discount rate should be a risk-free rate reflecting risk-neutrality on the government's part. Private discount rates, however, should reflect risk-aversion and therefore contain a risk premium rendering private finance more expensive than public finance.(12)
Before looking at the arguments in detail, it may be worth mentioning that the Arrow-Lind view of government discounting has in fact dominated the debate for years. Recently, a new World Bank handbook (World Bank, 1996) states that `the accepted view is that, save for very special cases, governments should not be concerned with the probability of failure or with the variance of outcomes [of projects]. . . . Government decision-makers should be "risk-neutral".'
In other words, governments should care only about expected values, not variances of outcomes. That is, they should be indifferent between a project with a certain value of US$100, and a project, with an expected value of US$100, which is based on a 50-50 chance of losing US$100 or winning US$300.
(ii) Risk-pooling and Risk-spreading
As stated by the World Bank (1996), the position of Arrow and Lind
is based on the concepts of `risk pooling' and `risk spreading.' If a country's portfolio has many projects whose outcomes are mutually independent, the country need not be concerned with the variability of the net present value (NPV) of a project around its expected values, as measured, for example, by the `variance' of the probability distribution of the NPV. The reason for this is that while many projects will result in lower-than-expected NPVs, others will result in higher-than-expected NPVs; if the projects are small and do not systematically reinforce each other's outcomes, then the negative and positive effects will tend to cancel out to a large extent.
In the case of zero correlation between all projects the variance of the portfolio is reduced towards zero by diversification.(13) If project risks are correlated, risk-pooling will not eliminate all risk. The size and sign of the covariance of returns on prospective projects is an empirical question. It seems, though, that a large number of projects undertaken by the state have outcomes correlated with national income. Electric power, highways, waterways, airports, and postal services, for example, all facilitate ordinary commerce and will thus be positively correlated with national income, allowing for a lower degree of risk reduction.
Arrow and Lind had mentioned the risk-pooling concept in their article, but did not present it as crucial. The private sector, too, can pool risks. Any advantage the state may have in pooling diverse risks can be transferred to the private sector by privatizing the projects in question.
The crucial argument by Arrow and Lind is about risk-spreading. They suggest that any financing system that is able to spread risks in tiny amounts over very large numbers of investors can tap funds at the risk-free rate. Increasing the number of investors will diminish the overall costs of risk-bearing at a faster rate than the summation of the individual costs will increase the overall costs. With an infinite number of investors the costs of risk-bearing will therefore converge to zero. Figure 1 illustrates the main point of Arrow and Lind. Investors with declining marginal utility of income will value losses more highly than gains. The utility of a project with uncertain outcome will thus be lower than the utility of a certain outcome equal to the expected value of the project. By subdividing the same project in equal shares among two parties the cost of risk-bearing will be reduced by more than half, because the slope of the marginal utility curve is steeper for smaller amounts. Consequently, when the number of investors bearing the risk becomes very large the total cost of risk-bearing approaches zero. Thus, the total cost of risk-bearing of a project which is independent of all other projects in the economy(14) is made negligible by the spreading of the risk over a large number of individuals.
Figure 1 illustrates the basic Arrow-Lind argument. The curve implies that the utility of income grows at a slower rate, the higher the income. Losses thus count more than gains of equal risk, a definition of risk aversion. The diagram depicts a gamble which yields income of [Y.sub.1] = 6 or [Y.sub.2] = 10 units with equal probability for an expected income of Y= 8. Subdividing the gamble among two persons with identical utility functions would yield two gambles yielding incomes of [Y.sub.'1], = 3 or [Y.sub.'2] = 5 with equal probability for an expected income of Y' = 4. The expected utility of the gambles EU(Y) is smaller than the utility from a certain income equal to the expected value of the gamble. The expected utility of the gamble may be set equal to that of a certain outcome, Y(*), the certainty equivalent, i.e. a sure return of Y(*). The difference between the expected income from the gamble and the certainty equivalent is the cost of risk-bearing. The cost of risk-bearing is the price the individuals in question would be willing to pay to exchange the gamble for the certainty equivalent. The cost of risk-bearing declines faster than the number of individuals sharing in the risk and approaches zero as the number of individuals increases.
Arrow and Lind argue that the government is actually able to spread risks in tiny amounts over millions of taxpayers, thus exploiting their `near-zero' cost of risk-bearing. Thus, the government could help the economy exploit the willingness of taxpayers to insure projects with non-negative NPVs without asking for remuneration, given that the cost of risk-bearing is zero. In this way the government provides a good substitute for missing insurance markets.(15) By relying on taxpayer guaranteed finance, the economy could reach the level of investments that is technically possible.
The government would thus be able to tap the low cost of risk-bearing of individuals carrying small amounts of risk efficiently. But then why could large corporations not achieve the same result? Arrow and Lind argue
that in order to control the firm, some shareholder may hold a large block of stock, which is a significant component of his wealth. If this were true, then, from his point of view, the costs of risk-bearing would not be negligible and the firm should behave as a risk averter . . . even though from the stockholder's point of view, risk should be ignored, it may not be in die interest of the corporate managers to neglect risk. Their careers and income are intimately related to the firm's performance. From their point of view, variations in the outcome of some corporate action impose very real costs. In this case, given a degree of autonomy, the corporate managers, in considering prospective investments, may discount for risk when it is not in the interest of the stockholders to do so.'
If one accepts the Arrow and Lind arguments, then decisions on projects will be biased against private participation. This is not based on considerations of fairness and equity in the economy or arguments about so-called public goods, such as the rule of law or defence, where some form of government may be required. The argument is valid even for pure private goods, i.e. those where benefits and costs can be fully appropriated by individuals.
(iii) Risk Allocation and Incentives of Intermediaries
What are the possible counter-arguments? First, the Arrow-Lind argument as originally presented is overstating its case. There are cases where large projects are highly correlated with national income. Dixit and Williamson (1989) highlight some of these, for example, oil production projects in Nigeria. More fundamentally, later advances in theory have shown that the theorem at best holds under rather restrictive conditions. Gardner (1979) shows that the theorem only holds for government projects, which are insignificant in relation to overall income in the economy. The capital-asset pricing model as presented, for example, in Brealey and Myers (1991) implies that there remains a remnant of undiversifiable or systemic risk, which requires discount rates to be risk-adjusted.
Spackman (no date) of the UK Treasury tried estimating the undiversifiable risk after optimal pooling of projects, which is correlated with incomes or wealth of taxpayers and thus imposes some positive cost of risk-bearing on taxpayers.(16) When estimating the cost of risk-bearing for taxpayers, he finds that it affects the discount rate -- estimated at some 6 per cent annually in real terms -- only marginally, adding 0.1 per cent.(17) The discount rate of small countries or countries with a small tax base would, of course, be affected more dramatically. Dixit and Williamson (1989) find that, in small economies, risk-adjustment to discount rates may often require more substantial adjustments for risk, but normally not more than 1 percentage point.
The Arrow-Lind argument may thus hold only in a small number of cases. Yet, more often than not, the risk-adjustment to discount rates that is required may not be large when based on calculations of the covariance of project income with general income. Applying private-sector discount rates as derived, for example, in a capital-asset pricing framework, would appear to result in larger risk premiums. Thus the bias in favour of public financing still appears to obtain.
Any remaining bias in favour of public financing tends to become questionable when one considers explicitly not only risk-diversification issues, but also associated agency costs. To begin with, one may note that one large shareholder may not be necessary for reasons of corporate control. If there are private benefits of control, the threat of exit to the manager (with a concurrent drop in share prices) may serve for control purposes. Moreover, the major shareholder could be an equity fund consisting itself of a large number of shareholders, thus spreading the costs of risk-bearing over an even greater number of shareholders, given that the internal incentives mechanism of the fund manager are appropriately designed. Furthermore, bureaucratic managers might be just as risk averse and reluctant to undertake risky projects as corporate managers are. The rationale for the better risk-spreading capabilities of the government is thus rather doubtful.
More fundamentally, one may ask what it is about the tax system that allows it to tap lots of `investors' with low costs of risk-bearing. Is it the unique `technology' of the tax system, or the fact of coercion, or maybe both? If individuals are willing to part with small amounts of money in return for no more than the risk-free rate of return, why are private players not able to exploit this? Many financial systems have large banks, which raise small deposits from a large number of people. However, when investing in projects, the first-in-line risk-takers are the equity investors, for many of whom the project risks are not negligible. They therefore require risk-adjusted rates of return.
If large banks could collect `deposits' from their customers and invest them without exposing the bank's equity holders, then they should be able to mimic the tax system's ability to tap low-cost funds. Indeed, `trust' accounts would be the equivalent of such a transaction being mere pass-throughs. But we observe no project finance based on such pass-through mechanisms. Apparently, depositors will want to be reasonably sure that their money will not be abused. If the trustee is not exposed and small depositors cannot directly assess and monitor project performance, it is not likely that people will provide money. Almost by definition, small depositors will not be able to monitor projects. An exposed agent is thus necessary to gain confidence. As a result, depositors in banks are willing to provide money at low interest rates, but the bank will require risk-adjusted rates when investing in projects or portfolios of assets.(18)
One could easily test the hypothesis that providers of small amounts of money would require an exposed agent. On income tax returns a field could be created where taxpayers could voluntarily indicate the amount of money they would be willing to invest in projects via the government. The money could be collected with the tax bill and returns from such voluntary investment could be automatically credited to future tax bills.
Somehow we might expect that not too much money would be raised that way. If so, this would support the view that the financial advantages of government finance are based on coercive powers and not on a superior system of exploiting the low-cost of risk-bearing of tiny investors. In fact, it would become clear that the low cost funds cannot be tapped unless the intermediaries have good incentives to use the money well, which by necessity gives rise to significant costs of risk-bearing on the part of the exposed intermediary. In turn, this requires risk-adjusted rates of return on financial instruments. Crucially, we cannot consider the cost of funds independently of the incentive system, under which intermediaries collect them. This is similar to modem theories of intermediation, which emphasize the monitoring problem of small investors in intermediaries and not only optimal. risk-diversification issues (Hellwig, 1990).(19)
(iv) Incentives of the Government as Intermediary
For the Arrow and Lind argument to hold, the government as intermediary would need to have better incentives to invest well than private parties. Arrow and Lind state that
many of the uncertainties which arise in private capital markets are related to what may be called moral hazards. Individuals involved in a given transaction may hedge against the possibility of fraudulent behaviour on the part of their associates. Many such risks are not present in the case of public investments and, therefore, it can be argued that it is not appropriate for the government to take these risks into account when choosing among public investments.
It seems Arrow and Lind assume that the government will, in fact, do what it ought to do, namely act benevolently and efficiently. Clearly, that is not the case in practice for many governments throughout history. So Arrow and Lind may be right that there are low-cost funds out there, but they may err in jumping to the conclusion that the tax system is a good way of tapping them.(20)
But maybe one could argue that government should collect `cheap' money and pass it on to projects which have private sponsors, who have first-risk-of-loss and will want to invest well. This line of argument may be seductive at first sight. However, any such rule `invest only when there is serious private equity' or `invest only pari passu in senior debt and not more than 5 per cent of total project costs' can be adopted by private pass-through trusts, the trustees of which may have quite reasonable incentives to maintain a decent reputation, arguably better than government officials, whose reputation is much less dependent on any particular project than a trustee's reputation. But we know that in practice investors are wary of trustees who are allowed considerable discretion in interpreting rules of a trust.(21)
Full government funding combined with private service provision weakens incentives for project selection. When the government selects projects, privatization may amount to managing the wrong thing efficiently. Yet, to paraphrase Peter Drucker, it is much more important to do the right thing in the first place -- the fundamental argument behind using financial markets rather than central planning to guide investment decisions.
The preceding arguments apply to all sectors. It has been argued that the case of infrastructure is different, to the extent that there are elements of natural monopoly requiring price regulation. Regulation, by definition, implies the exercise of discretion by some form of governmental authority. The regulatory agency may then create `artificial' risks leading to high private-sector risk premia. The argument continues that public funding or ownership would reduce artificial risk and thus be preferable. However, the argument neglects to recognize that regulatory discretion of some sort will also be exercised in the case of public enterprises. In fact, complaints about political interference in state-owned utilities are rampant. So we are once again left with the question whether risk -- in this case regulatory risk -- can better be diversified by financial markets or the tax system. The arguments presented above thus apply. In addition, the best way for governments to commit themselves to certain policies may often be to limit the tax powers of government, because this reduces options on the government side for alternative solutions.
So far, the argument is that there is no a priori reason to believe that taxpayers can better diversify risk than capital markets. In a world where taxes are imposed by small states in a large world economy with liberalized cross-border capital markets, it would seem that all sorts of risks can actually be better diversified by capital markets than taxpayers. This would also apply to regulatory risk, which varies among countries and jurisdictions. More broadly, policy risks, such as foreign-exchange-related ones, should not be imposed on taxpayers of -- often small -- countries, but left for investors. This goes against widespread practice in project finance, where governments and investors often agree to impose, for example, the risk of currency convertibility on taxpayers.
Finally, the argument that government has a lower cost of risk-bearing than private investors would appear to entail absurd consequences, for example that government should invest in projects and funds with high expected values, such as venture capital, or that companies such as General Motors should benefit from free government credit guarantees on all their borrowings.(22)
The preceding points suggest that the apparent benefits of sovereign finance are all due to coercive powers with no social value. To construct an argument in favour of socially useful coercion, one would need to find some market failure that can be remedied by coercion. One such argument derives from adverse-selection problems in private financial markets. Very large projects may require sponsors to tap a large set of investors, many of whom may not be very informed investors. Such badly informed investors may be afraid of being systematically at a disadvantage compared to others, and thus shy away from investing. In a sense, they may think that any project seeking funds from them must be of inferior quality. Otherwise the well informed investors would have snapped it up. If the government were to recognize correctly those projects which would not be funded owing to adverse selection, then coercive financing via the tax system could get investors to put up money without being afraid of being saddled with a project nobody else wanted.(23) Of course, the problem remains that the government may neither be capable nor interested in correctly judging project quality and identifying when a serious adverse selection issue exists.
The discussion suggests that government through the tax system cannot be expected to do better than private financial markets -- unless government policy places constraints on financial markets such as restrictions on capital movements or interest-rate controls. As a result, all the financial advantages of sovereign finance are purely due to coercive powers. Because very often there is no free-rider issue here, which can be remedied, such coercion rarely has social value.(24) As argued by the Australian task force cited earlier, under government finance the taxpayers would bear a contingent liability, which if properly remunerated would wipe out any cost advantage of sovereign borrowing. Governments should then refrain from investing in projects or firms, whether with equity or with debt. They should not cover commercial risks.
In particular, one cannot argue that there is a trade-off between the lower cost of government finance and higher private efficiency. Private markets will do the best they can to tap low-cost funds while maintaining project discipline. They solve whatever trade-off there is. The government cannot do better by raising funds. A corollary is that discount rates for private- and public-sector projects would not be expected to differ -- contrary to standard practice.
Arguing that the government cannot be expected to improve on the outcome of free financial markets is not to argue that all is for the best in the best of all possible worlds and that there is no role for government. Private markets may not always find the best solution. Market participants constantly search for better ways of trading risks.(25) But on average we could not expect governments to do better.(26)
More importantly, governments can significantly reduce the cost of risk-bearing by conducting prudent macroeconomic policies, supporting secure property rights, and deregulating and liberalizing the financial markets, so that private players can do their best to take advantage of low-cost funding opportunities. But it is not efficient to offset the risks created through bad policy by taxpayer-supported funding. That would amount to stealing from investors and compensating them by taking away from taxpayers.
In principle, multilateral finance institutions should apply their instruments to support the development of better government policies, for example by granting insurance against policy failures, where new policy regimes are not yet credible, but not by investing in projects or by guaranteeing the full credit risk of loans.
In this paper examples have been drawn from infrastructure projects, but none of the arguments about risk allocation and management is specific to infrastructure. In competitive markets it is by now fairly widely accepted that there is no clear role for government finance. The debate is, however, still alive for services, which are provided by monopolies
ANNEX 1: SPECIAL POLICY ISSUES FOR NATURAL MONOPOLIES
In this section, some special issues relating to infrastructure projects are sketched, which are sometimes used to argue for government funding of infrastructure projects. In sectors such as gas and water pipelines, power transmission and distribution, or road networks, competitive markets do not exist. First, consumers may be `taxed' through regulation or by monopolists. Second, governments might need to pay subsidies(27) to ensure efficient pricing of infrastructure services.
(i) Risk-sharing between Investors and Consumers
In infrastructure sectors, private incentives to operate efficiently may be weakened, because all forms of regulation or contracting tend to contain elements of cost-plus pricing. Wherever cost-plus pricing methodologies are pronounced, risks are shifted from investors to consumers. The cost of capital for investors and their projects is accordingly reduced. A number of US investor-owned utilities are effectively able to raise finance at little more than the risk-free rate of return (Alexander, 1995). or exclusive providers, such as many infrastructure services.
As sketched in Annex 1, there may be cases where subsidies are advisable, for example, to enable prices to be set at efficient levels, i.e. marginal cost. For example, on uncongested toll-roads the government may choose to pay subsidies in the form of shadow tolls, so as not to push drivers off the toll-road by high average cost-covering tolls. However, the policy-maker should not offer to fund construction of the road with either debt or equity.
Shifting risks to consumers may yield similar implications for the cost of capital as shifting it to taxpayers. But at least consumers have the option of not buying the service. Their effective willingness-to-pay places a limit on the scope for exploitation, which does not exist in the same way for taxpayers. Even if one interpreted the high cost to consumers as a tax, it would be a tax with some optimality features as it is imposed on price-inelastic consumption.
(ii) Subsidies to Support Efficient Pricing
Those elements of infrastructure that contain strong elements of natural monopoly have by definition low variable and high fixed costs, such that average cost will generally be above marginal cost. For investors to amortize their investment they need to receive prices that are at least equal to average cost. However, optimal pricing should be marginal-cost-based. For example, in the case of a toll-road, if the road is uncongested, the cost an additional passenger car imposes on the road is minimal. Ideally, all cars should be allowed to drive on an uncongested road as long as they pay for marginal cost. Tolls would thus collect prices corresponding to marginal cost and uncongested roads would not recover sufficient funds to service debt and equity.(28) The difference between marginal and average cost should be paid as a `subsidy' or shadow toll to the operator of the toll-road.
The problem is, of course, that demands for subsidies may well get out of hand. In the case of toll-roads it might well be that all sorts of uneconomic roads would get built if taxpayers were to foot the bill, the Mexican toll-road programme being a recent example. While no perfect solution exists, a number of methods are available to limit undue discretion of government authorities while trying to retain as many optimality features as possible.
The first may be called the `balanced budjet rule.'(29) Adam Smith (1776) noted that marginal cost pricing for roads and bridges might tempt the monarch into excessive construction. This is a reason why we may see infrastructure funded fully by consumers, for example, water in France or power in the United States. Utilities do not receive government transfers and are thus subject to a tighter budget constraint. As mentioned before, `taxation' via high consumer prices may also have certain optimality properties. A further rationale for a balanced budget pricing rule hinges on different incentives to monitor the government. When the firm overstates its true fixed costs under marginal cost pricing, taxpayers will bear the costs in the form of subsidies. Under average-cost pricing, an over-report of fixed costs is passed to consumers in the forms of higher prices. If consumers are better organized than taxpayers, average-cost pricing might induce better monitoring of the firm's activities and prove superior to otherwise first-best marginal-cost pricing.(30)
Greater transparency is another method to limit governmental discretion by exposing decisions to greater scrutiny by interest groups. Cross-subsidies are often criticized for not being transparent and for being open to abuse. In many countries, cross-subsidies benefit the middle classes, while the poor do not get any service at all.
Elements of competition
Finally, multiple players with conflicting objectives might be introduced to limit abuse. Auctions among competing bidders may, for example, be used to obtain a quotation for the lowest possible subsidy.
(1) The author, now Chief Economist at Shell International, was until recently Manager, Private Participation in Infrastructure at the World Bank. The views presented in the paper do not necessarily reflect the views of Shell International or the World Bank.
Basing this on a previous working paper, I received very insightful comments from Jean-Jacques Laffont and Tyler Cowen, as well as from Chris Lewis, Colin Mayer, Ignacio Mas, and Sweder van Wijnbergen. My colleagues Phil Gray, Tim Irwin, and Neil Roger at the World Bank helped clarify many points in tireless discussions, as did other colleagues, including Pedro Belli, Mansoor Dailami, Dale Gray, Ulrike Hoffman-Burchardi, Ashok Mody, Shanta Devarajan, and Sethaput Suthiwart-Narueput.
(2) Taking into account that one form of taxation is inflation.
(3) Standard arguments in cost-benefit analysis on the relative discount rates to be used for public versus private investments are surveyed in Spackman (no date). He concludes for the case of the United Kingdom that public and private discount rates are similar essentially because the risk premium of private rates happens to correspond to the adjustment to public rates required to correct for taxation of private investment. He believes the tax system can almost completely diversify risk for public investment.
(4) In fairly stable economies with well-functioning capital markets (e.g. the United Kingdom, the United States, and other major OECD economies), `risk-free' interest rates on sovereign debt are currently about 3-4 percent in real terms. Private corporate debt may be priced one percentage point higher, and structured project debt yet another one or two percentage points above this. On average, private equity tends to trade at some 2-4 percent above the risk-free rate on sovereign debt, the so-called equity premium. Rates tend to be higher in other economies owing to higher country risk, which affects all rates, or higher policy risk in a particular sector, which affects firms and projects in that sector (Blanchard, 1993; Spackman, no date).
(5) Government is unable to print the means of payment when a project contract is, for example, denominated in foreign currency. In some extreme cases governments may have been so abusive that their own citizens are unwilling to pay taxes and to hold domestic currency. That, first of all, reduces the scope for viable contracts drastically and, second, means that new contracts may often be concluded in `foreign currency', which government cannot debase, e.g. payment in commodities such as gold or cigarettes.
(6) World Bank, 1994, Box 5. 1, p. 91.
(7) See, for example, Carnaghan and Bracewell-Milnes (1993) and Domberger et al. (1994).
(8) Institute of International Finance (1995, p. 6). The IIF suggested World Bank lending to the private sector without a host government counter-guarantee. Private companies would thus benefit from the low-cost funds the World Bank can raise owing to its backing from the best credit risks of the world, i.e. the major governments of this world.
(9) In the case of World Bank credit, taxpayers in industrial countries underwrite risks. When loans or guarantees are counter-guaranteed by the host government, then taxpayers of that country also underwrite, leaving taxpayers in developing countries to assume risks as a last resort.
(10) The policy risk guarantees are also called partial risk guarantees by the World Bank -- as opposed to partial credit guarantees, which guarantee full loan obligations.
(11) For a standard exposition see Brealey and Myers (1991).
(12) A higher private discount rate would also imply that the private sector would tend to choose projects and technologies with quicker pay-back or shorter gestation periods than the government.
(13) Paul Samuelson (1966) clarified the debate about the benefits from pooling by pointing out that pooling per se may not help to make a gamble more attractive. Pooling of uncorrelated risks does create a new gamble with a larger mean and reduced relative variance. But,to paraphrase the argument, flying with a dangerous airline many times does not help reduce risk. It may just increase the chance that very big losses will occur. But pooling combined with subdivision of risk would definitively help. Thus, for an investor with a given budget, risk is reduced by buying a share in a pool of uncorrelated projects, rather than buying a single one of the projects with the same expected value.
(14) Risk-spreading crucially depends on the returns from the project being independent of other components of national income. As discussed in the case of risk-pooling, this is not necessarily the case for infrastructure projects since (a) they are often correlated with other components through the business cycle, or (b) they constitute a large fraction of national income.
(15) Imagine a hypothetical bench-mark case in which insurance markets exist for every conceivable risk and circumstance. If such a complete set of insurance products existed, everybody could lock in riskless values of investment (Arrow and Hahn, 1971). Financing such investment would then be possible at the risk-free rate, which would also be the appropriate discount rate. A complete set of insurance markets does not, however, exist, a major reason being that the existence of insurance would make people behave more recklessly (moral hazard) and thus change the underlying risk profile of the projects. Owing to moral hazard, total investments in the country will be lower than they ideally could be if people were to `restrain themselves better'.
(16) Analysing the correlation of project cash-flow with national income is similar to the analysis underlying the standard capital-asset pricing model in finance theory. However, society is treated as if it were an individual, which would require strong assumptions about comparability of utility or compensation schemes among individuals.
(17) The social cost of the risk that taxpayers bear would be higher if one accounted for the distortions that are typically associated with taxation. The full cost of raising a dollar of tax revenues may well be much higher. Studies suggest ranges from 20 to over 100 percent higher. However, accepting Spackman's calculations, that would still leave the discount rate essentially unchanged, perhaps rising to 6.2 per cent.
(18) Because depositors cannot monitor intermediaries easily, they like contracts with specified interest rates, so that they can easily recognize default (Diamond, 1984).
(19) It might be argued that the attitude of individuals to risk may, in practice, not be rational in the sense of standard economic theory. Prospect theory is an attempt to come to grips with observed departures from `rational' behaviour (Thaler, 1992). To some extent, the rules of thumb and habits that people use to deal with risk may still be rational in the sense that they make life more easily manageable in the face of overwhelming complexity. However, whatever the reason, small departures from rational behaviour can, in principle, have large aggregate effects (Akerlof and Yellen, 1985). If individuals are dealing with risk `irrationally' there might be a role for government, if only the wise and benevolent were in charge. But if politicians are `just human' there is no a priori reason to believe they would deal with risk better than capital markets. Even if policy-makers are benevolent and wise to the ways of risk, it is unclear whether they can accurately predict the distortions arising from `irrational' behaviour and whether they can, in practice, design interventions to offset them.
(20) We know that, in some cases, people are willing to invest in opportunities with negative expected values, for example lotteries. If one could create a lottery for project finance this should lower the cost of funds. After investment in a project, investors' lots could be drawn, assigning very high proportions of the project return to some investors and nothing to others. The fact that we do not see such mechanisms may have to do with the fact that abuse under normal lotteries is easier to monitor. But perhaps there are unexploited ways of setting up mutual funds here. The equity contributors to projects such as Eurotunnel may, in fact, have been driven by a lottery spirit. In earlier days lottery bonds were issued, for example, to finance the Panama Canal.
(21) See, for example, Millman (1995).
(22) Note that modem tax theory assumes that governments are not risk-neutral, with important implications for optimal asset-liability management (Kaplow, 1994).
(23) I am indebted to Colin Mayer for this argument.
(24) The real issue is not whether governments should be treated as risk-neutral or not, but whether there is a problem of collective action and whether it requires tax powers to solve it. In this sense, the real debate is simply about whether privalization makes sense or not. The cost of capital is determined simultaneously by the decision about the appropriate form of governance for an activity, i.e. the best possible allocation of risks and incentives among various parties. There is no separate trade-off between the cost of capital and efficiency of an arrangement.
(25) In fact, if financial markets were fully efficient, i.e. all arbitrage opportunities (profitable trades) had been exhausted, there would be no incentive left to look for better deals. That, in turn, would imply that arbitrage stops and we could not assume that markets would really be efficient.
(26) The argument is not simply tautological. It just says that we do not know whether governments have discount rates that are systematically different from those of private investors. One reason why they might be systematically different is that private entrepreneurs may be protected by limited liability provisions, which limit the feasible set of risk diversification and incentive schemes. Governments, on the other hand, can to some degree circumvent this constraint by imposing and adjusting taxes. However, we currently do not know how to weigh these considerations.
(27) Traditionally, it has been argued that there are cases where financial subsidies from the state are justified on other grounds than the taxpayer's low cost of risk-bearing. Social justice may argue for supportingthepoor. Environmental concerns may argue for subsidies for clean production processes. In all these cases, economic reasoning suggests that the best possible outcomes for society cannot be expected to result from the free interplay of private parties. Instead, properly designed `subsidies' should improve the outcome. In this respect the arguments are similar to the Arrow-Lind argument. However, the Arrow-Lind thesis assumed that taxpayers were willing to provide small amounts of insurance for 'nearly' free. As argued, intermediaries would have the option of attracting many mini-investors into an insurance scheme. Subsidies, on the other hand, are unlikely to be paid voluntarily, because individuals might not directly gain from them and/or are likely to hope that others will pay and try to free-ride on them.
(28) The development of transport infrastructure may have positive effects on adjacent land development, i.e. benefits owing to the infrastructure but not captured by its builders. Therefore, the infrastructure developers should be able to obtain land at its value prior to their investment so as to provide optimum incentives to them for the expansion of infrastructure. Alternatively, they might benefit from the rise in property taxes owing to appreciation of property consequent to infrastructure development.
(29) See Laffont and Tirole (1993, chs 15 and 16) for a discussion.
(30) Laffont and Tirole (1993), pp. 591-613).
Akerlof, J., and Yellen, J. L. (1985). `Can Small Deviations from Rationality Make Significant Differences to Economic Equilibria?', American Economic Review.
Alexander, I. (1995), `Regulatory Structure and Risk: An International Comparison', Washington, DC, World Bank, mimeo.
Anderson, J. R., and Quiggin, J. C. (1990), `Uncertainty in Project Appraisal', paper presented at the World Bank Annual Conference on Development Economics, Washington, DC.
Arrow, K. J., and Lind, R. C. (1970), `Uncertainty and the Evaluation of Public Investment Decisions', American Economic Review.
Bishop. M., Kay, J.,and Mayer, C. (1994), Privatization and Economic Performance, Oxford, Oxford University Press.
Blanchard, O. J. (1993), `The Vanishing Equity Permium', in Finance and the lntemational Economy 7, Oxford, Oxford University Press.
Brealey, R. A., and Myers, S. C. (1991), Principles of Corporate Finance, 4th edn, McGraw Hill.
Carnaghan, R., and Bracewell-Milnes, B. (1993), `Testing the Market: Competitive Tendering for Government Services in Britain and Abroad', London, Institute of Economic Affairs.
Diamond, D. (1984), `Financial Intermediation and Delegated Monitoring', Review of Economic Studies.
Dixit, A., and Williamson, A. (1989), `Risk-adjusted Rates of Return for Project Appraisal', Policy, Planning and Research Working Paper 290, Washington, DC, World Bank.
Domberger, S., Meadowcroft, S., and Thompson, D. (1994), chapter in M. Bishop et al. (eds) (1994).
Economic Planning Advisory Commission (1995), Private Infrastructure Task Force Report', Canberra.
Euromoney (1995), `Where, When -- and How Much?', March.
Frank, C. R., Jr (1995), `The Role of the World Bank in Infrastructure Investment', paper presented at Forbes Infrastructure Conference, New York, 18 January.
Gardner, R. (1979), `The Arrow-Lind Theorem in a Continuum Economy', American Economic Review, June.
Hellwig, M. (1990), `Banking, Financing Intermediation, and Corporate Finance', Discussion Paper 9015, University of Basel.
Institute of International Finance (1995), `Increasing the Supply of Bankable Infrastructure Projects m Developing Countries', Washington, DC.
Kaplow (1994), `Taxation and Risk-taking', National Tax Journal, 47, 789.
Kay, J. (1993), `Efficiency and Private Capital in the Provision of Instructure', in Infrastructurepoliciesforthe 1990s, Paris, OECD.
Kwoka, J. E., Jr (1996), `Privatization, Deregulation, and Competition: A Survey of Effects on Economic Performance', World Bank, Washington, DC, mimeo.
Laffont, J.-J., and Tirole, J. (1993), `A Theory of Incentives in Procurement and Regulations', Cambridge, MA.
Millman, G. (1995), `In Whom Should Projects Trust?', Infrastructure Finance, June/July.
Samuelson, P. A. (1966), `Risk and Uncertainty: A Fallacy of Large Numbers', in J. Stiglitz (ed.), The Collected Scientific Papers of Paul A. Samuelson, Cambridge, MA, MIT Press.
Smith, A. (1776), The Wealth of Nations.
Spackman, M. (no date), `Discount Rates and Rates of Return in the Public Sector', Government Economic Service Working Paper 113, London, HM Treasury.
Thaler, R. H. (1992), `The Winner's Curse -- paradoxes and Anomalies of Economic Life', Princeton, NJ.
World Bank (1994), World Development Report 1994, Washington, DC, World Bank. _____ (1996), Handbook on Economic Analysis of Investment Operations,
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||goverment funding for investment projects and risk premium for such projects|
|Publication:||Oxford Review of Economic Policy|
|Date:||Dec 22, 1997|
|Previous Article:||Investment appraisal in the public sector.|
|Next Article:||Government debt management.|