Capital Budgeting

A CRITIQUE OF A CAPITAL PROJECT APPRAISAL TECHNIQUES Simon M Keane CONTENTS 3.1 Introduction 3.2 Underinvestment bias in conventional capital budgeting techniques 3.3 Wealth creation versus wealth accretion 3.4 Nonrationing versus Rationing 3.5 Does the NPV method need support? 3.5.1 Payback 3.5.2 Internal Rate of Return 3.5.3 The significance of Scale 3.6 Investment fallacies 3.6.1 The positive-NPV fallacy 3.6.2 The reinvestment fallacy 3.6.3 Investment efficiency fallacy. 3.7 Does IRR have any role? 3.7.1 Can IRR ever serve as an alternative to NPV in ranking projects? 3.7.2 Can IRR serve as a supplement to NPV? 3.7.3 Is IRR equal to NPV as an accept-or- reject criterion? 3.7.4 Is IRR useful in rationing conditions? 3.7.5 Is IRR a useful aid to nonfinancial managers? 3.7.6 Can IRR serve even as a preliminary screening method? 3.7.7 Is IRR not a better link to the rate of return in subsequent financial reporting? 3.8 Why is the yield approach acceptable in the securities market but not in the product market? 3.9 Conclusion 3.10 Summary Tutorial questions Problems Suggested further reading

3.1 INTRODUCTION Most Finance text books agree that the Net Present Value (NPV) method is the optimal investment selection criterion, but they also tend to suggest that the Internal Rate of Return (IRR) has a significant role to play, and to a lesser extent the Profitability Index (PI) and even Payback (PB). The argument is often made that investment decisions are too important to be left to a single method of appraisal. Decision-makers need to see the problem from more than one, and possibly several, perspectives. This multicriteria approach to capital budgeting is widely reflected in practice, in that most large companies use several methods to make investment decisions, except that IRR tends often to be preferred over NPV as the appropriate Discounted Cash Flow method. The purpose of this chapter is to review this multi-criteria approach and to show that: A)

The IRR method is fundamentally flawed as an investment appraisal method and, like Payback and PI, has no defensible role in capital budgeting decisions,

B)

The use of the above measures, together with the popular interpretation of the Positive NPV rule, have a propensity to encourage underinvestment,

C)

The widespread practice of using two or more methods for investment appraisal purposes is more likely to confound rather than enrich the decision process. The NPV method should be used as the sole criterion under all conditions, being the only method consistent with the primary financial objective of the firm.

3.2 UNDERINVESTMENT BIAS IN CONVENTIONAL CAPITAL BUDGETING TECHNIQUES

The chapter adopts the corporate objectives developed in Chapter 2 where it was shown that the appropriate goal is “to maximise the value of the firm subject to maximising the share price” (MAX VtMAX Pt). This translates to a capital budgeting context as “to maximise the present value of the firm’s investments subject to maximising their net present value” (MAX PVtMAX NPVt ). To develop the argument we first need to show why there is an underinvestment bias in standard investment selection criteria, in particular a propensity for conventional capital budgeting decision rules to be influenced by the logic of capital rationing even when nonrationing conditions are explicitly assumed to hold. Symptomatic of this bias are three widely held misconceptions that will be later discussed in some detail. These are the assumption i) that, for a project to be acceptable, it must in practice have a positive, as distinct from a nonnegative, net present value (the "positive-NPV fallacy") ii) that the reinvestment opportunities for a project's intermediate and terminal cash flows are relevant to the project's degree of acceptability, (the "reinvestment fallacy") and iii) that a project is more desirable the more efficient its use of capital, in the sense that, for a given NPV, a quick payback is preferred to a slow payback, and a high IRR or Profitability Index preferred to a low one (the "investment efficiency fallacy"). To illustrate these fallacies consider the two mutually exclusive projects, A1 and A2, in Table 1 available to a company operating in normal, nonrationing conditions. It is assumed that neither project will be replaced on completion. The projects have identical positive NPVs and both are obviously acceptable. Conventional theory would suggest that they are equally attractive or, possibly, that the project with the lower outlay and shorter life should be favoured. To test the intuitive appeal of this interpretation, the above problem was presented to a group of students who had successfully completed a standard introductory course in Business Finance. 92% of the students indicated either that A1 is less attractive than A2 or that, at best, the two are equally desirable. The main reasons given for favouring A2 were that the investment capital "saved" might be better employed in other projects, and that A2 is more desirable because it employs less capital to generate the same amount of NPV.

This reasoning illustrates all three of the above misconceptions. First, the preference for A2 over A1 implicitly negates the desirability of zero-NPV investment, given that A1 is equivalent to A2 plus a zero-NPV investment of 8M. Even the more orthodox view that the choice is a matter of indifference indicates a failure to acknowledge the positive worth of the incremental investment of £8M. Secondly, it implies that the extra capital invested in A1 might be better employed in financing additional projects, thus failing to recognise that such additional projects could equally be financed from an issue of new capital and undertaken in conjunction with A1. Thirdly, it implies that, for a given NPV, a project is more desirable the more "efficient" it is, by having, for example, a higher profitability index or a shorter payback. It will be shown in this chapter that these considerations have no bearing on the respective attractiveness of the two competing projects under normal, nonrationing conditions. It will also be argued that, unless project A1 is instinctively perceived to be more desirable in a nonrationing environment, an underinvestment bias can be assumed to exist. It will subsequently be shown that the ultimate source of this bias is a failure of traditional investment theory to recognise the effects of scale. 3.3 WEALTH CREATION VERSUS WEALTH ACCRETION Before proceeding it is necessary to define some relevant terms. "Market criterion" denotes the fundamental requirement that a project should earn no less than the minimum acceptable rate of return for the level of risk, or, equivalently, that it should be expected to have a nonnegative impact on the share price. The term "wealth creation" is used to denote the process of transforming investible funds into productive investments that satisfy the market criterion (the zero-NPV component of an investment). The term "wealth accretion" is used to denote a project's positive-NPV component, if any. Hence a zero-NPV project consists only of the primary wealth creation process, whilst a positive-NPV project contains both the wealth creation and wealth-accretion components. This distinction is important because a key assumption here is that "wealth creation" is not only desirable, but may in practice be the principal activity of many seasoned companies in mature industries. Wealth accretion, it will be argued, is a desirable adjunct but should not be seen as a pre-condition of investment.

3.4 NONRATIONING VERSUS RATIONING A further distinction that needs to be made is between rationing and nonrationing conditions. Students often assume that capital rationing is the normal state for companies. Most firms, it is assumed, have a limited amount of capital available to them to undertake new projects. But when we consider what capital rationing means we have to conclude that it is in fact an irrational state that should rarely, if ever, exist for a listed company. It is true that intuitively it is appealing to assume that most companies probably face a shortage of capital and could undertake more investments if they could only find the finance. The significance of being listed on the stock market, however, is that a company acquires access to the capital needed to finance any project that satisfies the market criterion. The implication of a project with a nonnegative NPV is that it is expected to satisfy the cost of capital criterion. It is, therefore, a contradiction in terms for a company to face capital rationing. The company cannot rationally argue that it is unable to finance a specific project that meets the market criterion since the NPV is measured by discounting the cash flows at the cost of raising the necessary capital. If a company discounts its projects at a particular discount rate and then claims that it cannot find the funds to finance the project then it is clear that the discount rate is understated. It should raise the discount rate to the level at which it can raise capital. Capital is rationed not be quantity but by price. If the NPV calculation reflects this price then it follows that the normal condition for listed companies is one of nonrationing. Companies, in effect, face a limited number of acceptable projects rather than a shortage of capital to finance them.

3.5 DOES THE NPV METHOD NEED SUPPORT? The central issue addressed in this chapter is whether NPV should be the sole criterion or is more effective if supported by other selection criteria. The following analysis will focus on the IRR because it has the strongest claim to be treated as an alternative to NPV. Payback, however, is still the most widely used technique and deserves some comment.

3.5.1 PAYBACK The textbooks base their criticism of Payback on the fact that it ignores a) the cash flows after the payback period and b) the time value of money. The underlying assumption, therefore, is that, were it not for these defects, the principle of seeking an early payback is fundamentally sound. But if we accept the premise that all investment is desirable provided it satisfies the present value acceptance criterion then it becomes clear that an early payback is not in itself desirable. It is often defended as indicating the risk and the liquidity implications of the project. Apart from the fact that it is wholly inadequate measure of risk or liquidity it is essential to recognise that the object of investment is not to minimise either risk or illiquidity. The essence of investment is to assume risk and to part with liquid funds. The object is to earn a return that compensates for the risk and commitment of funds, and Payback does not even pretend to measure return. It is, in effect, an anti-investment measure, the logic of which is that the optimal investment is to keep all cash in the bank. It is important to recognise that, for a given NPV, the longer the payback period the better. If it is agreed that acceptable investment is good then it is better that a project lasts two years than one year, and so forth. For example, the most common assumption is that investment A3 in Table 1 is preferable to investment A4 because it generates the same NPV in two years instead of six. But investment A4 is equivalent to investment A3 plus an additional 4-year investment of £5M with a zero NPV. Now since by definition a zero NPV investment earns the minimum acceptable return, then it is desirable and should be undertaken. The argument that A4 releases capital for reinvestment is fallacious because, if a further investment presents itself, the company can raise the necessary capital and undertake it in conjunction with investment A3. To argue otherwise is to suggest that the company is operating under capital rationing. The weakness with PB, therefore, is not that it ignores the time value of money but that it is based on the misconception that long-term investment is intrinsically bad. If we accept that investment is good provided it satisfies the market criterion, other things being equal, then the longer an investment lasts the better. Payback is irrelevant and misleading.

3.5.2 INTERNAL RATE OF RETURN

IRR answers the question “ what is the maximum cost of capital the project can bear?” The issue is, can this statistic ever replace or reinforce the net present value? The traditional arguments in favour of IRR are: 1. That businessmen think in percentage terms, and 2. It does not require a precise calculation of the cost of capital The first argument is irrelevant if it can be shown that IRR is fundamentally invalid, and the second argument is a fallacy because, although IRR can be calculated without knowing the cost of capital, it cannot be used without it since the cost of capital is the benchmark for deciding whether the project is acceptable. Traditional arguments against IRR: 1. IRR assumes reinvestment at a rate equal to the IRR. 2. Some projects have more than one IRR In this chapter we will show why the case against IRR cannot validly be made on the basis of either of these arguments and is flawed for a more fundamental reason than either. The generally held view is that IRR is inferior to NPV but useful for comparing projects unless the projects a) have unequal lives b) have unequal outlays or c) have significantly different cash flow patterns. In effect, this view states that, to be confident that IRR is valid in comparing two or more projects, the projects must be virtually identical. If they are not identical there is no way of being sure that IRR gives the correct ranking without reference to the projects’ NPVs. 3.5.3 THE SIGNIFICANCE OF SCALE The key factor in explaining the fundamental weakness in IRR is the issue of investment scale. Commonsense suggests that a relative measure of worth such as IRR is inappropriate when comparing projects of different scale. A project costing £1,000 may have an IRR of 40% but this is immaterial when compared to a competing project costing £20M with an

IRR of 30% (assuming both have the same cost of capital, say 15%). The only thing that matters is the respective NPVs of the projects. Although most textbooks recognise the effect of a size difference in judging the relevance of IRR, they fail to recognise that size is not simply a function of initial investment but of both the length of period over which the investment lasts and the pattern of the cash inflows over the project’s life. Consider the projects in Table 2. The conventional explanation for the different rankings given by IRR and NPV for projects B1 and B2 is the difference in cash flow patterns, and for projects B1 and B3, the difference in lives, and for projects B1and B4, the difference in size. But these differences are all symptoms of the same phenomenon, the scale effect. Equating the size of a project with its initial outlay, without taking account of the duration or pattern of the cash flows, is analogous to equating the size of a building with the area of land it occupies without taking account of the height of the structure or the configuration of its architecture. The outlay reflects the initial allocation of capital to the project but it does not capture the scale of the capital invested throughout the project's life. To derive a common measure which fully reflects the three dimensions of scale - outlay, cash flow pattern and life - the equivalent investment of capital over a single year can be calculated, as in Table 3. Continuing the previous analogy, this measure is comparable to the aggregate square footage of a building, which clearly better reflects the scale of the building than the area of land it occupies. Applying the same measurement technique to the other projects, this reveals that project B1 in Table 2 is significantly greater in scale than any others. Even B4 is greater than B2 or B3, despite the lower outlay. It has a lower return than B3 but a higher NPV as a result of being significantly greater in scale. The Equivalent Capital Investment over a single year may have little practical significance,1 but it demonstrates the central principle that every project has a unique scale. Whatever the outlays, cash flow patterns, or lives of the projects, conflicts between IRR and NPV can be attributed wholly to differences in scale. Not only are reinvestment opportunities for the intermediate cash flows irrelevant in interpreting the two methods, but there is no need to seek an alternative explanation for any conflict between them.

The effect of conventionally treating these three dimensions of scale as distinct phenomena is that, when conflict between IRR and NPV happens not to arise, IRR tends to be interpreted as valid. But the absence of conflict does not validate the use of a relative measure in circumstances where it is inherently wrong. The issue of scale is present even when its effect does not manifest itself in a conflict the two measures. This is evidenced by the fact that it is never possible to be certain that conflict is absent except by first estimating the respective NPVs of the projects. If the NPV is not calculated, the IRR can be used with confidence to rank projects only when the projects are substantially identical. It may appear an academic nicety to argue that, even when a method gives a correct ranking, it remains incorrect in principle. The point is stressed, however, for two reasons. First, if a company uses IRR when it accords with NPV, and then, at a later date, avoids the method when it clashes with NPV, it will cause unnecessary confusion for the method to have been declared "valid" in the one set of circumstances and invalid in the other. Second and more fundamentally, even when IRR happens to accord with NPV, it gives a misleading signal about the project's degree of desirability. The significance of this effect will be examined shortly. It is sufficient at present to note that one project is never superior to another by virtue of simply of having a higher IRR. If it happens to have both a higher NPV and IRR this is coincidental, and indicates simply that the differences in scale happen to be insufficient for the fundamental flaw of IRR to manifest itself. In summary, the fact that all capital projects are different in scale implies that an absolute measure of added value is the only relevant criterion of project worth. IRR as a ratio is fundamentally unsuited for comparing competing projects. In effect, provided a project's return meets the minimum required rate for the risk class, the project’s degree of "profitability" is irrelevant, only its contribution to value. 3.6 INVESTMENT FALLACIES It is now possible to put in perspective the three fallacies identified at the beginning of the chapter. 3.6.1 THE POSITIVE-NPV FALLACY

The primary fallacy consists of assuming that, because some positive NPV projects are available, a project must have a positive NPV to be desirable. This would imply that a project must be more than acceptable to be desirable. Some texts are quite explicit about this: "Financial managers should only accept those actions which are expected to increase the share price" (Gitman, 1991, p18) A greater NPV is, of course, always preferable to a lesser one, but the only precondition of investment is that the project should have a nonnegative price reaction, this being the visible manifestation of the nonnegative-NPV rule. The primary object therefore is to create wealth not to achieve wealth accretions, in effect not to maximise NPV but to maximise long-term PV subject to maximising NPV. A positive NPV is a bonus not a precondition of investment. Indeed, if it is contended that it is a pre-condition, it is a short step to assuming that the magnitude of the positive NPV should be significant, with the result that it is the Profitability Index that becomes the criterion of acceptance rather than NPV, leading to the possibility that the firm may reject some positive-NPV projects. It follows that, even in a world where the opportunity to generate wealth accretion exists, the conventional goal of share price maximisation fails to represent the primary role of investment. The alternative goal of maximising the value of the firm subject to maximising the share price (MAX VtMAX Pt) implies that, although every opportunity to increase the share price should be taken, the expectation of a rise is not a prerequisite of investment. The significance of the share price constraint is that projects should not be undertaken which are expected to reduce the price. Only in a grossly uncompetitive world where there is a universal excess of positive-NPV projects is it possible to make wealth accretion the exclusive goal of business. 3.6.2 THE REINVESTMENT FALLACY The overemphasis on achieving excess returns has led in turn to the assumption that the possibility of using a project's cash flows to finance additional excess-return opportunities is relevant to its perceived worth. Underlying the reinvestment fallacy, therefore, is the assumption that the discounting process penalises later cash flows more severely because early cash flows are available for reinvestment. Early flows, however, are

rewarded in the discounting process not because they make possible new investment but because they give the option of paying off the capital used to finance the original investment. This is why we discount at the cost of capital. It is irrelevant whether the firm actually uses the cash to repay the capital or chooses to use it to finance new investments, however profitable these might be. In a nonrationing environment, it is a matter of indifference whether the firm uses cash inflows from an old investment or new capital to finance its new investments, since the necessary capital can readily be raised at the appropriate market rate.2 For example, even if the £11M cash inflow in period 1 of project C2 in Table 4 could be reinvested in a new project C3 with a positive NPV of £5M, C1 would remain the more desirable of the two because C3 could be financed independently at 10% and undertaken in conjunction with C1. A project, therefore, is not more desirable for releasing capital earlier rather than later, except under capital rationing. The classic manifestation of the fallacy is, of course, in the IRR versus NPV debate. Thus, the conventional criticism of the Internal Rate of Return is that it assumes reinvestment at a rate equal to IRR. Although this issue has been identified in the literature (Dudley 1972, and Keane 1979), the misconception about the relevance of reinvestment opportunities has persisted, and seems to have become entrenched in the investment literature. The criticism of IRR is specious because it is meaningless to impute to IRR such an assumption, since, as already noted, the destiny of the cash flows is irrelevant except under rationing conditions. Hence, the fundamental flaw in IRR has nothing to do with assumptions about reinvestment opportunities for the intermediate cash flows but can be explained wholly by the effects of scale. Applying the scale-measurement technique illustrated in Table 3, it can be shown that all investments are different in scale, whatever their initial outlay. Therefore, IRR, as a ratio, is fundamentally irrelevant for comparing investments, even when the outlay is the same. If IRR happens to accord with NPV in ranking particular projects this is pure chance. The technique is wrong in principle for comparing capital projects. 3.6.3 INVESTMENT EFFICIENCY FALLACY. A natural extension of the assumption that early recovery of cash is intrinsically desirable is the belief that a project is more desirable if it is efficient in the use of capital. This belief is evidenced by the near universal practice by companies of using efficiency criteria such as IRR

and PB to support the NPV method, (Indeed the evidence indicates that about 70% of smaller companies exclusively use one or more of the standard efficiency measures in preference to the absolute NPV method (Runyon 1983). It is clear, however, that, provided an investment satisfies the market criterion, its degree of efficiency is irrelevant. Indeed, in the sense that investment efficiency is conventionally understood, the less efficient a project is in generating a given (nonnegative) NPV the more desirable it is. Hence, if a project is acceptable, a slow payback has the merit that it prolongs the wealth-creation process. The notion that early cash recovery is not intrinsically desirable may seem at variance with ordinary business liquidity considerations, but, in nonrationing conditions, the discounting process should fully reflect the liquidity implications of an investment. To discount each project's cash flows at the acceptable rate of return, and then mentally to penalise some projects for being less efficient than others in their use of capital, is double counting for cost-of-capital effects. It is not simply that efficiency indicators fail to support the NPV method. They positively militate against the primary wealth-creation process. Given the premise that the object of investment is to create wealth and, wherever possible, to generate wealth accretions, the only relevant measures are those that signal absolute wealth effects. For example, consider the projects in Table 5. The basic NPV rule favours D2 over D1 and D3, but there is a risk that the tendency to attach significance to investment efficiency might lead some decision-makers to prefer D1. D2, of course, is preferable, both because of its higher NPV and its greater scale. But D3 is preferable to both because it is the most consistent with the MAX VtMAX Pt and MAX PVtMAX NPVt rules established in the objectives chapter. Most decision-makers versed in conventional capital budgeting criteria would probably favour D2 over D3, but D3 puts to use £500M of capital and still satisfies the market criterion. In effect we are conditioned to perceiving an “efficient” return per pound invested as important, but the efficiency of a project is in fact irrelevant provided it meets the MAX PVtMAX NPVt criterion.3

3.7 DOES IRR HAVE ANY ROLE? It is now possible to consider whether IRR has any valid role in capital budgeting. The focus is on IRR rather than the alternative competitors for NPV because it is the most popular DCF method in practice and the measure with the best claim to provide to have a valid status. The arguments that follow, of course, apply even more forcefully to Payback. For the purposes of the discussion that follows it is useful to distinguish between “investment selection methods” and “investment decision tools”. The case against using IRR, PI and Payback as investment selection criteria does not imply that the numbers represented by IRR, PI and PB should never be used by the financial manager. An investment selection method provides the benchmark for acceptance or rejection of a project, whether primary or secondary. For example, a company might employ each of NPV, IRR, and Payback as investment methods, without even stating which is the primary method. Or it might use Payback as a preliminary screening method. An investment decision tool, on the other hand, is a statistic or device that might be used in applying a given investment method in a particular set of circumstances. For example, the Profitability Index may be useful under capital rationing conditions to help identify the combination of projects that maximises NPV, without PI having the status of a separate selection method. In assessing the desirability of a multiple-criteria approach, therefore, we are concerned with the practice of employing multiple investment methods rather than multiple investment decision tools. The latter statistics may be important elements in deriving the former but they are not themselves the criteria by which investment worth is judged In order to decide whether IRR has any role in investment appraisal it is necessary to consider the full range of potential uses, from a stand-alone measure of investment worth to a simple aid to nonfinancial managers.

3.7.1 CAN IRR EVER SERVE AS AN ALTERNATIVE TO NPV IN RANKING PROJECTS?

We have argued that IRR can never be viewed as a valid alternative to NPV in ranking projects because of the scale effect. Projects always come in parcels and the rate of return per pound invested can never be an alternative to the absolute measure of added value. INCREMENTAL APPROACH Textbooks often suggest that the scale problem of IRR can be partially overcome by using an incremental approach where the cash flows of smaller project are subtracted from those of a larger project so that the incremental cash flows can be evaluated as a simple accept-or-reject decision. The use of this approach, however, is severely limited: a) the projects must have identical risk otherwise it is like deducting apples from oranges, since the quality of the respective cash flows is different. b) they have must identical outlays, otherwise the incremental IRR relates to year 1 or later when the cash flows first differ. It is difficult to conceptualise the significance of an IRR relating to cash flows commencing in the year ahead. c) there is a significant chance that subtracting the cash flows of one project from those of another will give rise to more than one sign change and therefore to the multiple yield problem. d) the approach is clumsy if there are several projects since each has to compared incrementally with all the others. e) more significantly, even if the above restrictions are met, the resulting measure gives no information that the NPV would not give, and indeed less since it fails to indicate the incremental value of the respective projects. The aim of the incremental approach is to convert a ranking of projects into an accept-or-reject decision by comparing the incremental cash flows of two or more projects. It will be shown later that this objective is of dubious value since, even in an accept-or-reject context, the IRR is misleading. Not only is the incremental approach an unsuccessful technique for the purposes of salvaging IRR but it will be argued that the attempt to salvage IRR is itself a fruitless endeavour.

3.7.2 CAN IRR SERVE AS A SUPPLEMENT TO NPV?

The most common defence for IRR is that, as a secondary measure, it provides additional insights that act as a useful supplement to NPV. The investment decision is too important, the argument goes, to be left to a single criterion, and IRR has the most convincing claim to provide support. The NPV of a project may be a sufficient benchmark in principle, but in practice the method is only as good as the input data, and it might seem less risky to use supplementary measures even if they are theoretically flawed. However, although the problems of measuring NPV are real ones, they cannot be alleviated by importing measures that are conceptually unsound. IRR draws on the same data and the same cut-off rate as NPV, but the conversion of the data into a ratio rather than an absolute number cannot improve the quality of the data. The ultimate test of the value of a supplementary method is that it is validly capable of qualifying the signal of the primary method, either by a) reinforcing it, or b) by overriding it when the latter fails to reflect some relevant dimension of investment worth. If it cannot perform either of these two functions, and has a tendency at times to conflict with the primary measure, then it serves no substantive function, and may in practice have a negative impact on the decision process. Since IRR gives a misleading signal about the acceptability of a project, its use as a supplement must be clearly open to doubt. The issue, therefore, is whether an estimate of the "maximum cost of capital that a project can bear" can validly be used to qualify the signal contained in NPV. The state IRR > k is a precondition of a positive NPV, and therefore to argue that it reinforces the latter is tautological unless it can be shown that the degree to which IRR exceeds k has relevance to the interpretation of NPV. But this has already been shown not to be the case. Both B2 and B3 in Table 2 have a greater IRR-k spread than project B1 but this does not make b1 any less desirable. It is true that the spread between IRR and the equilibrium return indicates the degree of abnormality in the investment’s rate of return, but this in itself is irrelevant to the incremental significance of the investment. If IRR cannot reinforce the signal contained in NPV still less is it validly capable of countervailing it. However, one of the more plausible arguments in favour of the method is that the IRR-k spread provides an

easily understood indicator of the degree to which a positive expected NPV is vulnerable to errors in estimating k. This will now be considered. IRR as an indicator of cost-of-capital uncertainty It has already been noted that the argument that IRR has an advantage in apparently not requiring a precise estimate of the cost of capital is specious because, although it is calculable without reference to the cost of capital, it cannot be used effectively without it. More persuasive, perhaps, is the argument that because IRR indicates the maximum cost of capital a project can bear, the IRR-k spread indicates the degree of tolerance of the NPV to errors in estimating the cost of capital. This argument raises three questions: a) is cost-of-capital risk relevant to project worth, b) if so, is IRR-k spread the best indicator of it, and c) should the indicator be factored into the calculation of NPV, or should it be treated as an addendum to NPV? A project may be viewed as having two types of risk, cash flow risk and cost-of-capital risk, but standard NPV calculations may typically reflect only the former. The possibility that the latter may also be relevant, however, depends presumably on whether the risk can be shown to be systematic. But, even if it is systematic, it is not clear that the IRR-k spread is the best indicator of the effects of estimation errors. Every project's NPV has unique sensitivity to changes in k, depending on the configuration of the cash flows, and a more useful indicator is likely to be achieved by sensitivity analysis of the respective NPVs under different assumptions about k. More significantly, however, even if IRR-k spread is a useful indicator of cost-of-capital risk, the question arises whether this information should be factored into the NPV calculation or whether IRR should be presented as a supplementary method for the decision-maker to judge its impact on NPV. If the former, then IRR is simply input into NPV, one of several investment tools, and not a distinct measure of investment worth. If the IRR-k spread is first incorporated into NPV, and IRR is also presented as a separate investment statistic, this is clearly double counting for cost-of-capital risk. If the assumption is that IRR-k should not be factored into NPV but presented as a separate statistic this raises the question why cost-of-capital

uncertainty should have such a unique role in the investment decision process, given that the integrity of NPV is undermined when any relevant aspect of risk is omitted from the calculation.4 It would be difficult to claim that it is too problematic for the financial manager to incorporate cost-of-capital risk into the calculation of NPV since the task would be even greater for the decision-maker, who now is left with the task of modifying the unadjusted NPV to capture the implications of the IRR-k spread. The decision to omit any factor relevant to a project's incremental value and then to present this factor as a rider to NPV not only imposes a significant burden on the decision-maker but accords to that factor a definitive role in the decision process. Even if IRR is intended by the financial manager to be treated only as an addendum to NPV and not as a full investment method, it would be unrealistic to assume that the significance of this distinction would be recognised in practice without the risk of provoking the dysfunctional effects associated with a multi-criteria approach. It is perfectly legitimate to consider the sensitivity of NPV to variations in the cost of capital, but NPV remains the only valid measure of the project's incremental value. Presenting a range of NPVs may create its own problems for the decision-maker but at least it remains consistent with the single criterion ideal.

3.7.3 IS IRR EQUAL TO NPV AS AN ACCEPT-ORREJECT CRITERION? When the issue is not to rank projects but simply to decide whether a particular project is acceptable it is sometimes assumed that IRR is the better method because, being expressed in percentage terms, it is likely to be more effective than NPV in communicating a project's degree of desirability. The conventional assumption is that the conceptual differences between the absolute and relative form of the present value model are unimportant in an accept-or-reject situation on the grounds that IRR nearly always gives the same accept-or-reject decision as NPV (except when the multiple yield problem arises). But the signals given by the two measures are quite different. IRR is correct only in a yes-no sense. An essential test of the theoretically correct measure of investment worth is that a higher rating under the measurement system should denote a higher degree of desirability. NPV meets this test because greater (risk-adjusted) wealth is unequivocally more desirable than less wealth. IRR fails the test because a project's desirability is not a function of the excess of its IRR

and the project's minimum acceptable rate (k). The fact that IRR exceeds k for a given project is no more than a statement that the project has a positive NPV. IRR may give a correct signal, therefore, about the acceptability of a project, but it gives a misleading signal about the degree of acceptability. It may seem unimportant whether decision-makers misinterpret a project's degree of desirability if the project is acceptable in any event, but equating IRR with the degree of desirability may have serious dysfunctional consequences. Thus, if the notion is conveyed that, in an accept-or-reject situation, IRR signals the degree of project desirability, it would be difficult to convince nonfinancial managers that, when mutually exclusive projects are being compared, the project with higher IRR is not necessarily more desirable. Secondly, and more fundamentally, the ranking problem, and therefore the scale issue, cannot be avoided even in a single-project context. Capital budgeting is a dynamic process, and throughout a project's life the firm will continually seek to enhance the project's investment value, since it may always be possible for management to influence the subsequent configuration of the cash flows. When an enterprise has some degree of control over the risk or timing of a project's future cash flows, this is essentially no different from the problem of discriminating between separate, competing projects. An individual project in effect competes with itself. If the IRR-k spread were to be interpreted as an index of desirability it should logically be assumed that any strategy which would increase this spread would be desirable, and vice versa. This could inhibit management, when faced with a high IRR, from seeking out alternative strategies that would increase the project's NPV, or when faced with a low IRR, induce them to sacrifice NPV in pursuit of a short-term yield advantage. The logic of wealth-maximisation, however, is that any opportunity to reschedule a project's cash flows in such a way as to increase its NPV is beneficial even if the effect is to sacrifice IRR. For example, if, having accepted a project with the cash flows equal to B2 in Table 2, the firm found subsequently that it could alter the timing of these flows to replicate those of project B1, then clearly this would be advantageous, despite a lowering of the IRR. The firm would be disinclined to make this change, however, if it perceived IRR as a measure of desirability.

In summary, the excess of a project's IRR over the cost of capital indicates that the project has a positive NPV, but does not reflect the project's degree of desirability. IRR is misleading even for accept-orreject decisions because it promotes the short-term perspective implicit in the IRR method, and may lead to a subsequent mismanagement of the project. IRR in discriminating between projects with identical NPVs This insight into IRR’s limitation as a measure of desirability allows us to address one other claim for IRR, that it is useful in discriminating between projects that happen to have identical NPVs. Of course, the same could be said for Payback, or even for the toss of a coin, except that IRR shares the same DCF framework as NPV. If it is accepted, however, that IRR is a misleading indicator of a project's degree of desirability, it is incorrect to assume that it can validly be used to discriminate between projects with identical NPVs. It is clearly convenient to have a tie-breaker when NPVs are the same, but the project with the higher IRR is only by chance more desirable, just as is the one with the lower payback. Moreover, if the firm were to treat IRR as the ultimate benchmark when NPVs are the same, it makes it difficult subsequently to present a credible case to managers to ignore IRR when the NPVs are different. The important point is that it is not a matter of indifference when two projects have the same NPV. When faced with dilemma, the choice should be made on the basis of the primary decision rule, MAX PVtMAX NPVt. The primary task of business enterprises is to create wealth. Hence the project with the greater scale should be chosen rather than one indicated by an otherwise irrelevant criterion. Hence the reason why D3 is to be preferred over D2 in Table 5. 3.7.4

IS IRR USEFUL UNDER RATIONING CONDITIONS?

It has been shown that, for listed companies, conditions of capital rationing should normally not exist. They could, of course, arise if selfimposed, or in unlisted companies without access to outside capital. We have shown that that the reinvestment opportunities for intermediate cash flows are irrelevant under nonrationing conditions, but clearly they become relevant if, for whatever reason, the firm does operate under rationing. Does the relevance of reinvestment under these circumstances justify the use of

IRR? Is there not then the need for an efficiency measure rather than for the absolute measure of NPV. Even in rationing conditions the fundamental objective of maximising PV subject to maximising NPV remains unchanged. The problem facing the firm is to select the combination of investments that achieves this end. Certainly this involves taking into account the reinvestment opportunities created by the intermediate cash flows from its possible projects, but the solution requires a sophisticated mathematical programming approach outside the scope of this chapter. Simpler decision tools, such as IRR or PI are sometimes suggested and these may be sufficient under simple rationing conditions. The object is to maximise the combined NPV given the budgetary constraints, and, at best, the PI and IRR are simply arithmetic tools to aid the selection process. It is important to stress, therefore, that even in these circumstances, the only relevant acceptance criterion is NPV. Once the optimal combination of projects is selected their individual IRRs, PBs and PIs are irrelevant.

3.7.5 IS IRR A USEFUL AID TO NONFINANCIAL MANAGERS? The common argument that a ratio, and therefore IRR, is more understandable to nonfinancial managers confuses familiarity with understandability, and ignores the fundamental issue of decision relevance. Although the basic concept of yield is well understood, the application of the concept to the product market exacts a heavy learning burden on those facing its conceptual idiosyncrasies for the first time - the reinvestment issue, the problem of multiple yields, the incremental yield approach, the problem of accommodating differential short and long-term interest rates, the problem of comparing yields of projects in different risk classes, and the issue of scale. The notion that the concept is more understandable is a myth. Coming to terms with these complexities might arguably be worthwhile if the method could be shown to be an acceptable substitute for, or to add relevant information to NPV. But, because IRR is liable to contradict the decision indicated by NPV, then annexing it to the latter allegedly to help the nonfinancial manager is difficult to defend as a prudent or harmless expedient. The notion that the method is a useful supplement on the grounds of being more understandable is therefore illusory because paradoxically it can be

properly understood only when its fundamental irrelevance to the product market is fully recognised. More importantly, the assumption that NPV lacks intuitive appeal is misguided. If you were to ask the average finance student to define NPV the answer would be something like “If the investment outlay is deducted from the discounted present value of the project’s future cash flows the difference represents the net present value which, if positive, indicates that the project is acceptable.” In effect, the natural tendency is to define the concept by describing the method of its computation. This has become the practice in textbooks and may make sense for examination purposes but it is of doubtful help to the nonfinancial manager unfamiliar with the concept of discounting cash flows. If we are going to make NPV more accessible to the general user we must learn to define the method in terms of its significance rather than of its method of computation. The fact that a project is expected to add £5M to the value of the firm has more significance to the decision-maker than how the measure has been derived. Indeed the very name net present value suffers from the criticism that it focuses on the measurement process rather than on the significance of the measure. If we denoted the measure “Added Value” or “Incremental Wealth” it would become clearer to the nonfinancial manager why the measure should take priority over all others, and why a higher added value is always better than a lower added value whatever the rate of return per pound invested.

3.7.6 CAN IRR SERVE EVEN AS A PRELIMINARY SCREENING METHOD? It is often argued that it is unnecessary to use the NPV method to screen potential projects. IRR, it is claimed, and, even more commonly, the Payback method, could be satisfactorily used to screen out projects before the final analysis by the NPV method. This, of course, makes little sense. To screen out some projects using a theoretically invalid method cannot be defended even if could save a little time. It would have been a serious matter if North Sea oil exploration companies had used Payback to screen their long-term investments in the first instance. Payback and IRR both involve estimating the same future cash flows as the NPV method. The incremental time and effort in discounting those cash flows to determine the respective NPVs is a small price to pay to avoid the risk of rejecting worthwhile long-term projects. It is ironic that, on the one hand, it should be

claimed that investment projects are too important to be left to one method and, at the same time, capable of being screened out by one inferior method.

3.7.7 IS IRR NOT A BETTER LINK TO THE RATE OF RETURN IN SUBSEQUENT FINANCIAL REPORTING? The conclusion that capital investment appraisal should be based exclusively on an absolute measure of wealth unaccompanied by a relative measure may appear overprescriptive given that the corporate financial reporting system is rooted in profitability and rate-of-return measures designed to help review the firm's past investment decisions. The inclusion of IRR in the ex ante decision process has arguably the advantage of creating a useful link with the subsequent review process since this tends to be based primarily on rate-of-return concepts. It would be undesirable, however, to allow ex ante investment choices to be driven by ex post performance evaluations systems. It is possible, moreover, that, although there should be a fundamental consistency between ex ante investment choices and ex post analysis, a somewhat different perspective is appropriate. It is outside the scope of this paper to consider optimal performance evaluation or financial reporting systems, but the use of achieved rates of return in retrospective analysis may be partly defended on the grounds that the scale of the investment program has been substantially determined for the period under review, that ex-post analysis is concerned more with measuring the effectiveness of investment choices that have been made than with evaluating choices that might have been made, and the primary purpose of periodic review is to evaluate managerial performance in respect of a given set of investments within a given time-frame. Ex ante investment selection is concerned with evaluating possible project alternatives over their total future lifespan. Therefore, the conventional bias towards rate-of-return concepts in financial reporting cannot be used to defend the use of a rate-of-return measure in capital budgeting, otherwise it would be a persuasive argument for using the Accounting Rate of Return.

3.8 WHY IS THE YIELD APPROACH ACCEPTABLE IN THE SECURITIES MARKET BUT NOT IN THE PRODUCT MARKET?

Having dismissed IRR as an inappropriate under all circumstances some comment is needed about the fact that the yield of an investment is widely used in the securities market. Indeed it was from the bond market that the concept of discounted cash flow was first imported in the last century. IRR is simply another name for the yield to redemption of a bond. Why is it that a percentage measure works satisfactorily in the bond market but not in the product market? The explanation, of course, lies in the scale effect. The scale problem does not arise in the securities market because investors can choose to invest any amount they wish in a given security. Investments in the real market, however, come in parcels, and it is not possible to buy a quarter or half of a project. Therefore, whilst it is quite legitimate to compare securities on the basis of the return per pound invested, only an absolute measure of added value is relevant in the real market to compare projects of different scale. The fact that IRR appears to work at times for real investments does not alter the principle that a yield approach is fundamentally unsuitable for capital projects.

3.9 CONCLUSION A casual survey of most standard finance textbooks makes it evident that investment appraisal is rendered unnecessarily complicated by attempts to explain the problems associated with IRR and PB and even by efforts to salvage them. This chapter has argued that these methods are not potentially acceptable measures of worth that happen to have some flaws, but totally irrelevant measures that confound rather than enrich the decision process. The investment appraisal process could be made significantly more straightforward if PV/NPV analysis were presented in the finance literature as the only valid approach to the measurement of investment worth.

APPENDIX

TABLE 1 PROJECT A1 A2 A3 A4

OUTLAY

NPV

PAYBACK

£10M £2M £5M £5M

£1M £1M 0 0

5 YEARS 3 YEARS 2 YEARS 6 YEARS

MUTUALLY EXCLUSIVE PROJECTS UNDER NONRATIONING CONDITIONS

TABLE2 B1

B2

B3

B4

10%

10%

15%

13%

0

-2000

-2000

-2000

-1000

1

+10

+2100

+2500

-

2

-

-

-

-

3

+3100 _____ +345 16%

+330 ____ +156 17%

Discount rate YEAR

NPV IRR

____ +174 25%

+1725 ____ +198 20%

MUTUALLY EXCLUSIVE PROJECTS WITH CONFLICTING SIGNALS FROM NPV AND IRR

TABLE 3 PROJECT B1

PROJECT B2

PV FACTOR

PV FACTOR

YEAR 1 INVESTMENT £2000 YEAR 2 INTEREST

CASH FLOW

PV

1.00 £2000

200

____

____

2200 10 _____

2200 2100 _____

219 _____ INVESTMENT 2409 _____

.909 1991

100

.826 1899 ____

10 ____ 110 .826 ____

YEAR 3 INTEREST

EQUIVALENT ONE YEAR INVESTMENT

£2000 1.00 £2000

200

INVESTMENT 2190

5980 ____

PV

.909

91

90 ____ 2181 ____

MEASURING THE SCALE OF AN INVESTMENT BY CONVERTING IT INTO THE EQUIVALENT OF A ONE-YEAR INVESTMENT

TABLE 4 PROJECT

OUTLAY CASH INFLOWS ________________________________ YEAR 1 YEAR2 YEAR3

NPV @10%

C1

-£10M

-

-

+£13.3M

0

C2

-£12M

+£11M

-

+£2.7M

0

INTERMEDIATE CASH FLOW AND THE ISSUE OF REINVESTMENT

TABLE 5 PROJECT

OUTLAY

NPV

PB

IRR

D1

£5M

£1.4M

3 YEARS

23%

D2

£100M

£1.5M

4 YEARS

14%

D3

£500M

£1.5M

10 YEARS

10.5%

THE “EFFICIENCY“ OF INVESTMENTS VERSUS THEIR NPV

3.10

SUMMARY

Conventional investment selection procedures tend to be influenced by the logic of capital rationing even when nonrationing conditions are explicitly assumed to obtain. This tendency has serious dysfunctional implications and can be attributed to a number of misconceptions and inconsistencies that have entered into capital budgeting theory. These can be summarised as follows: 1. The scale of an investment is a function not only of initial outlay but also of cash flow pattern and of project life. Consequently, every project has a unique scale. In normal, nonrationing conditions the greater the scale of a project for a given NPV the better. 2. Zero-NPV investments are positively worthwhile in nonrationing, and possibly even in rationing conditions. In a reasonably competitive environment, they may even form a significant component of a firm's investment opportunity set. Indifference to zero-NPV investments is, therefore, a form of self-imposed capital rationing and may lead to underinvestment. 3. Indifference to zero-NPV investments, and therefore a tendency to underinvest, can be traced ultimately to the traditional share price maximising objective, which directs managerial attention exclusively to the wealth accretion aspect of investment. The alternative goal of “maximising the long-term value of the firm subject to maximising the share price” is more appropriate for competitive markets because it encompasses both the wealth-creation and wealth-accretion dimensions of investment. 4. Reinvestment opportunities for intermediate cash flows are irrelevant in assessing project worth under nonrationing conditions. The common assumption that a project with a given NPV is more beneficial for having a smaller rather than greater outlay, or a quicker rather than a slower payback period on the grounds that capital is thereby released for further investment is fallacious. When choosing between competing projects with identical NPVs, the project that employs most capital is generally to be preferred because it creates more wealth for the same wealth accretion. 5. The fundamental flaw in IRR as a criterion for evaluating competing projects is therefore not related to any reinvestment assumptions inherent in the method, but can be attributed entirely to the scale factor in investment. Because all projects are different in scale, IRR, as a ratio, is intrinsically invalid, even when it happens to give the correct ranking. Even for accept-or-reject investment decisions IRR is misleading because it suggests that the investment's cash flow recovery should be managed, wherever possible, in such a way as to give the investment a high IRR rather than a high NPVs. 6. A multi-criterion approach to investment appraisal is indefensible. The common practice of using "efficiency" indicators, such as IRR and Payback, to supplement the NPV method is misplaced, given that such measures confound rather than enrich the decision process. NPV is not the best measure of investment worth, it is the only acceptable measure.

QUESTIONS 1. What characteristics should be possessed by the appropriate acceptance criterion in capital budgeting decision-making? 2. In what sense can it be claimed that Payback is an anti-investment criterion? 3. The internal rate of return is often criticised as a) sometimes generation multiple yields and b) assuming reinvestment of the intermediate cash flows at the internal rate. Explain why these criticisms are misplaced. 4. The Internal Rate of Return is widely used in the securities market in the form of the redemption yield on bonds etc. Explain why the concept cannot not validly be extended to the capital budgeting context. 5. Explain the importance of “scale” in assessing the significance of the IRR. 6. It is commonly assumed that IRR is a perfect substitute for NPV in a Simple accept-or-reject situation where no ranking is involved. Explain why IRR does not provide a reliable index of project desirability even in these circumstances. 7. The argument is sometimes made that more information is always better than less, and therefore that the use of several acceptance criteria in support of NPV always enriches the decision process. Discuss whether this logic justifies the use of PB and IRR as ancillary measures of investment worth. 8. How would you explain the significance of NPV to a manager who is unfamiliar with the discounted cash flow concept?

PROBLEMS 1. Company One operates under normal, nonrationing conditions and is presented with two competing projects. Which should it choose? Explain your decision, and indicate what additional information you might require? A £50m £4m 18%

Outlay NPV IRR

B £80m £4m 16%

2. Company Two is also presented with two competing projects under nonrationing conditions. Which should it choose and why? C £100M 0 5 years

Outlay NPV Payback

D £100M 0 3 years

3. Company Three, again under nonrationing, is presented with the following projects. YEAR 0

Outlay

E £100M

YEAR 1 YEAR 2

Cash Flows -do-

£144M

IRR

20%

F £70M 60M 40M 30%

The company is confident that the cash inflow of £60M from project F in year 1 could be reinvested at 30%. Consider which project the company should choose, explaining your decision.

READINGS TEXT BOOKS All standard finance texts. These rather than journal articles provide the best insight into the conventional interpretation of the share price maximising objective and its implications for financial decision-making. These include: Brealey, R. and S. Myers, Principles of Corporate Finance, McGraw Hill. Brigham, Eugene F., Fundamentals of Financial Management, Orlando, Florida, Dryden Press. Copeland, T.E., and J.F. Weston, Financial Theory and Corporate Policy, Addison Wesley Gitman, L.J. Principles of Managerial Finance, Harper Collins Publications Moyer, R., J McGuigan and W. Kretlow, Contemporary Financial Management, West Publishing Company. Ross, S., R. Westerfield and J. Jaffe. Corporate Finance, Irwin.

ARTICLES Dudley, E, Jr, (1972) "A Note on the Reinvestment Assumption in choosing between Net Present Value and Internal Rate of Return", Journal of Finance, September. Gitman, l. and V. Mercurio, (1982) "Cost of Capital Techniques used by Major U.S. Firms: Survey and Analysis of Fortune's 1,000," Financial Management, Winter, pp. 21-29.

Haka, S.F., L. Gordon, and G. Pinches, (1985) "Sophisticated Capital Budgeting Techniques and Firm Performance," The Accounting Review, October, pp651-669. Keane, S. M., (1979) "The Internal Rate of Return and the Rein-vestment Fallacy", Abacus, Vol 15, No 1, pp 48-55. Kim, S. H. (1982), "An Empirical Study of the Adoption of Sophisticated Capital Budgeting Practices and earnings Performance” Engineering Economist, Spring, pp. 185-196. McIntyre, A. and N. Coulthurst, (1987) "The Planning and Control of Capital Investments in Medium-sized Companies", Management Accounting, March, p.39. Moore, J.S. and Reichert, A.K. (1983) "An Analysis of the Financial Management Techniques currently employed by large US Corporations", Journal of Business Finance and Accounting, Vol 10, No 4 pp623-645. Mukherjee, T. (1987) "Capital-Budgeting Surveys: The Past and the Future", Review of Business and Economic Research, Spring pp 37-56. Runyon L. R. "Capital Expenditure Decision Making in Small Firms", Journal of Business Research (September, 1983) pp 389-397. Solomon, E., 1956 "The Arithmetic of Capital Budgeting Decisions", Journal of Business, April, pp 124-129.

FOOTNOTES 1. As a measure of project scale it is noteworthy perhaps that the ratio of NPV to ECI gives a better indication of the relative wealth-creating efficiency of the project than IRR, which fails to reflect the riskadjusted wealth increment, and than the Profitability Index, which incorrectly equates initial outlay investment scale. 2. See Dudley, E, Jr, "A Note on the Reinvestment Assumption in choosing between Net Present Value and Internal Rate of Return", Journal of Finance, September 1972, and Keane, S. M., "The Internal Rate of Return and the Reinvestment Fallacy", Abacus, Vol 15, No 1, pp 4855, 1979 3. Of course if the purpose were say to compare the efficiency of two divisional managers operating with different capital bases, then a ratio is appropriate. But when the object is to measure the contribution of each division's decisions to the value of the company, an absolute measure of incremental value is the only acceptable criterion.