Page Under  Construction

"Nothing in the world can take the place of persistence. Talent will not; nothing is more common than unsuccessful men with talent. Genius will not; unregarded genius is almost a proverb. Education will not; the world is full of uneducated derelicts. Persistence and determination alone are omnipotent."

Calvin Coolidge

Memo
To:      Colleague

From:   Ray Martin

Ref:     EPA/NSF Proposal Rejection

Date:    July 21, 1999

Subj:    Risk Assumptions in Mainstream Discounted Cash Flow Analysis


Introduction.  The Internal Rate of Return is Okay proposal was rejected by the National Science Foundation and the US Environmental Protection Agency. It was in response to their solicitation on Decision-Making and Valuation for Environmental Policy (DMVEP). 

NSF/EPA Proposal and Reviews.  The complete proposal and the NSF/EPA reviews are online in Adobe PDF.  Contact me if you prefer them via E-mail or on diskette.  The reviews were scanned for better text searches, but may have minor character recognition errors.  My review of the reviews is listed below, with the stem in default browser color and my response in navy blue.  The Discussion Questions at the end may assist in your review.

Informal Peer Review. I conducted an informal peer review July 1, 1997 through February 1, 1999.  It was informal in the sense that responses were voluntary and provided direct to me.   The original thesis and the results of the informal peer review are online:

Relevant Disciplines. The disciplines and subdisciplines with a stake in the issue include finance, economics, industrial engineering, accounting, business, management and project management, risk analysis and management, mathematics, insurance, and real estate. 

Peer Review Phases.  I tentatively propose a more formal peer review in phases.  The first phase is to address a risk assumption implicit in the traditional or mainstream discounted cash flow analysis:

  1. An implicit assumption in mainstream discounted cash flow (DCF) analysis that risk for different sized projects can reasonably and safely be assumed to be identical.   Ten-fold differences are typical, but 100-fold and greater difference are offered.   I contend that risk cannot reasonably or safely be assumed to be equal for any project, especially those of greatly different size or cost.
  2. An extension of the first assumption is that risk and return can be analyzed separately and a rational decision made using the discounted return (net present value (NPV)) only.  Or that risk is a secondary consideration.  I contend that although return and risk can be analyzed separately, they must be reconciled in order to make a rational decision and that risk is a key consideration.

Feedback.  All feedback is sincerely appreciated, especially critical feedback and especially from previous informal reviewers and those reviewing the NSF/EPA proposal.

Note: Principles of Corporate Finance is commonly cited by the the authors' surnames, "Brealey and Myers (B&M)"  Other than IRR/NPV treatment, I take no issue whatsoever with it.  However, considerable feedback from four continents from the informal peer review cited B&M most often as the source of disagreement with my  IRR is Okay thesis.  With one notable exception of which I am aware, other textbooks take the same or a similar position on IRR/NPV treatment.


Review of

NSF/EPA Reviews


1. RIRR is never really defined in the proposal.

RIRR, the Relevant Internal Rate of Return, is an algorithm. It is about 90 lines of computer code.  The algorithm analyzes cash flows in terms of the criticisms and implicit assumptions inherent in the mainstream way of viewing IRR.  It calculates RIRR the way it is explained in the proposal rather than the way it is commonly described.  RIRR is one discount rate at which net present value (NPV) is zero.  It is not any discount rate at which NPV is zero.


2. Is it just that one ignores the nonsensical results of IRR analysis?

No. And both IRR and NPV can be nonsensical. The problems with IRR have been repeated often, while the coexistent problems with NPV have been ignored.  Whenever RIRR is incalculable, NPV is suspect and in some cases nonsensical.  Although both IRR and NPV have minor problems, the two are consistent, sensible and complimentary.  


 3. For example, the proposal provides an example where two different discount rates may set net present value equal to zero (p. 8.), but the author argues that only one (the 25% rate) is reasonable, whereas the other answer (400%) is unreasonable. In this example it may well be easy to say that 400% is unreasonable, but what happens when the two rates are fairly close to one another?

It makes no difference whatsoever how close two rates are. All subsequent rates past the first have nonsensical implications for both NPV and IRR. Between any two rates, NPV must increase with the discount rate, implying increasing value with increasing cost. This is absurd.  I prefer relevant over reasonable, i.e., relevant for making a decision.


4. But the author uses the terms RIRR and IRR interchangeably throughout the proposal, so it is difficult to isolate exactly what makes RIRR different from IRR.

The intent was to use RIRR when calculated using the algorithm described at 1. above and IRR when calculated the mainstream way.   The difference is that RIIR is calculated the way it is explained in the proposal, whereas IRR is not. RIRR and IRR are the same value more often than not.


5. Further, the proposal never outlines exactly how RIRR and Net Present Value are complementary: what does one give you that the other does not?

RIRR is a rate whereas NPV is an amount. They are as complementary as knowing both the price of a home and the rate of interest on the mortgage loan when buying one. One thing they would give is broader acceptance by business people.


6. When do they (IRR and NPV) differ and why?

They do not differ, at least as far as conclusion or decisions are concerned. But one is a rate and the other an amount. 


7. The author also cites the number of hits on a web publication as evidence of interest in the paper, but it is not clear that his arguments have been subjected to formal peer review.

The peer review was extensive but informal.  The results are summarized in the paper Principles of Corporate Finance Questions


8. The proposal is based on the idea that mainstream project analysis based solely on net present value has got it wrong.

Not wrong. But mainstream discounted cash flow (DCF) analysis overemphasizes NPV at the expense of IRR. It uses strained assumptions. It separates risk from return, implying an investment decision can be made with return only. Discounted cash flow analysis is not as well accepted as it could be if the apparent NPV/IRR contradictions were simply understood and explained rather than limiting the analyses to NPV only because IRR can be troublesome.


9. The author of the proposal may or may not be on to something, but the proposal is written is such a way as to obscure his arguments.

I am looking for suggestions to make less obscure?  However, the criticisms of IRR must be addressed somehow, in my opinion.


10. If the RIRR arguments standup to external peer review, then the proposed research could be quite valuable.

It stood up pretty well to an external but informal peer review, but it took iteration to convert those that did.  Some expressed disbelief that the prevailing view, that has been around so long, could be flawed in any way.  However, most abstained after the first iteration, so their views are unknown. 


11. While the proposal tries to make th ecase that the IRR method should replace net present value considerations in environmental decision making, it also (correctly) observes that the two methods are equivalent.

No.  There was no attempt to replace NPV with IRR, or even an intentional suggestion that it should. They are best used together.


12. Thus it is not clear why the IRR method should be used instead of net present value for policy purposes.

RIRR should be used along with NPV, not instead of it. An important reason is that business people who actually make the decisions prefer a rate to an amount or insist on having both. NPV is commonly ignored in favor of undiscounted payback and accounting rate of return, neither of which considers the time value of money.   These are serious omissions. 


13. This project intends to determine the extent to which the use of a "relevant" or "revisited" internal rate of return (RIRR) might enhance environmental decision making (EPA programs and analyses in particular).

Correct.

The underlying hypothesis is that the use of the RIRR method is preferable to the use of net present value (NPV).

Incorrect. RIRR is preferable to the use of IRR as it is usually defined. RIRR and NPV are best used together.


14. NPV can lead to inappropriate policies because it "includes potentially contentious assumptions and calculations."

The actual statement was: "However, relegating time value of money calculations to something more akin to a mortgage loan than a maze of potentially contentious assumptions, should allow greater focus on benefits analyses (emphasis in original)." One contentious assumption that has been around so long that it is generally ignored, is that greatly different sized projects always have identical risk. A contentious calculation is the discount rate for public sector projects.   It has been disputed continuously for decades.

Using RIRR alone likewise has potentially contentious assumptions. RIRR and NPV are best used together in a dynamic framework that incorporates risk.


15. The IRR method will be used in the EPA Cleaner Technologies Substitutes Assessment Guide to evaluate selected historical programs.

Essentially correct.  The EPA Cleaner Technologies Substitutes Assessment Guide is linked from the proposal.  The relevant portions of the Guide are suitable for evaluating historical programs using RIRR with NPV compared to NPV only.


16. The use of IRR and NPV are both predicated on the same underlying assumptions: they are simply alternative ways of determining whether a given investment (or program) is worthwhile.

They have been historically treated as alternatives or competing. They are instead complimentary. We see a mix of views--alternatives, competing, and complementary in these reviews. 


17. Thus, it is not clear why the use of IRR would be more appropriate than NPV (or any other method) in comparing cash flows over time.

Neither IRR nor RIRR alone would be more appropriate. Using RIRR together with NPV is.


18. It is not clear why the use of IRR would enhance receptiveness of time value of money considerations compared to NPV.

An important reason is that business people who actually make the decisions prefer a rate to an amount or want both. The alternative to IRR is the Accounting Rate of Return, which does not consider the time value of money. Excluding the time value of money is a serious and uncorrectable flaw, whereas the difficulties with NPV and IRR are minor and correctable.


19. The PI has numerous publications on the IRR method and has been promoting an understanding of its potential for public decision making for more than two decades.

Correct. The thesis was presented as information only for two decades. The first attempt at peer review and promoting the thesis began about two years ago.


20. The proposal states that this project "could change the way costs are analyzed at all levels of government." Not quite. While it has the potential of encouraging environmental decision-makers to consider an alternative method for determining a project's effectiveness, it need not change how project costs are measured or analyzed.

Discounted cash flow analysis in the public sector is cost analysis that focuses on the time value of money.   Costs over time within government are analyzed using NPV only.  With scant few exceptions, IRR is excluded.  For example, I found no relevant IRR hits on the Office of Management Budget WWW site. If RIRR were used in conjunction with NPV, that would be a major change.  DCF is not a good method for determining project effectiveness, only cost. 


21. The first phase of the project will familiarize EPA analysts and managers with the IRR technique, after which the RTRR method will be applied to selected EPA programs. Plans for disseminating results of the project to a wider audience are not explained, but it is likely that CD-ROM and on-line formats will be employed.

Other than standard NSF reports, disseminating results outside EPA was left to the discretion of EPA. Use of RIRR in conjunction with NPV has major policy implications.


22. There are no collaborators on the proposed project.

Correct. Collaboration with EPA or others invited by EPA was left open.


 23. I don't find this proposal to be intellectually interesting.

Okay.


 24. "Properly calculated and used, RIRR gives identical decisions to NPV." But then why is it needed?

An important reason is that business people who actually make the decisions prefer a rate to an amount. It is not that difficult to reconcile NPV and RIRR rather than ignore the difficulties with one and reject the other because of them.


25. I believe that the argument about mutually exclusive options is not correctly interpreted. Suppose that you use a single piece of property either for project 1, which has an internal rate of return of 25% or for project 2, which has an internal rate of return of 24%, but not both. IRR can't generally be used for selecting among the projects.

This is difficult to conduct discounted cash flow analysis without  the cash flows. But if the same amount is invested for the same amount of time and in the unlikely event that risk is identical on both projects, take the one with the highest return.

There is no reason to use RIRR alone. The decision should be made using both RIRR and NPV.  We should understand the implications of any apparently conflicting signals rather than exclude either RIRR or NPV because they can be  troublesome. Risk should not be ignored or assumed identical because it too is troublesome.


26. If project 2 is larger than project 1, it may be more desirable to invest in project two. For example, if project 1 requires an investment of exactly $10 (can can't be re-scaled), and project 2 requires an investment of $1,000 (and can't be rescaled), project 2 will be more desirable since it results in a larger return if 24% is greater than the opportunity cost of capital. It wouldn't make sense to forgo a return of $240 dollars per year (24% of$1,000) to obtain a return of $2.50 (25% of $10).

Yes, it may.  But not always.  Projects that are 100 times larger should be expected to have a larger NPV, ceteris paribus. This is simply another way of saying that more money is better than less, even when discounted. An implicit, strained assumption here is that risk is equivalent in both projects.

Projects do not necessarily have to be re-scaled. Simply redo the $10 project 100 times (NPV = $250) and evaluate all projects dynamically in that context. Alternatively, take the first $10 generated by the larger project and execute the small one as well.  Static analysis is another shortcoming of traditional DCF analysis.


27. If you know the net present value, you don't also need to know the IRR in order to make a decision. For most cases the two results are identical, so we don't need both methods.

An important reason for using both methods is that business people who actually make the decisions prefer a rate to an amount when deciding. The alternative to RIRR is the accounting rate of return, which ignores the time value of money. This is more undesirable than excluding RIRR because it can be troublesome.  It is also not a good idea to ignore the inherent difficulties with NPV because they are less obvious.  


28. If the two measures are not consistent, NPV is the better approach to use.

Properly viewed, the two methods always give consistent answers.  They approach the problem from a different perspective.


29. The stated objective of the proposed project is to determine whether the evaluation and prioritization of EPA programs and "investments" can be improved through the use of a revised internal rate of return (RIRR) concept.

Correct.


30. This would serve as an alternative to the more traditional net present value (NPV) concept more commonly used in cost-benefit analysis.

No. RIRR is a supplemental or complementary calculation. It is not an alternative to NPV.


31. The PI indicates that he has already developed and disseminated (through the worldwide web) the conceptual basis for relying on this alternative evaluation paradigm; therefore, the purpose of the project would be to communicate these ideas specifically to an EPA audience and apply them to specific EPA programs.

Communication within EPA was to have been determined by EPA. Using RIRR in addition to NPV has major policy implications that should have consensus and would need approval prior to change. The proposal incorporated a method for immediate cancellation at any time without cause.  Simply determining that the needed consensus or approval were unattainable would have been sufficient.  That is already done by rejecting the proposal.


32. Although the fundamental assertion - that RIRR should receive more attention as an evaluation tool - has some merit, the project, as it is proposed, has several important deficiencies which should make it a low priority for funding.

Deficiencies listed below: 33. through 43.


33. Although the proposal discusses in general terms why RIRR can be viewed as a useful alternative to NPV for evaluating private investments, it does not show a good understanding or provide a good explanation of why this would be specifically useful for EPA programs.

No. RIRR is not an alternative to NPV. It is a supplement to NPV and has essentially equivalent utility. An objective of the research was to determine usefulness in EPA.   The suggested starting point was Energy Star.


34. The proposal mentions a few programs, such as Energy Star, and indicates that such programs would, in effect, serve as useful project case studies for demonstrating the application of RIRR; however, it does not provide any specifics about why these programs would be appropriate candidates for RIRR.

Any or all Energy Star programs, to be determined by EPA, would have been good candidates. RIRR has significant policy implications, meaning that EPA should be convinced first.

Energy Star might have been the largest beneficiary, but would have been a policy decision. Current Energy Star instructions use IRR because business people prefer a rate.   However those EPA instructions call for considerable caution when using IRR, citing the criticisms addressed in the proposal.

For example, EPA instructions exclude IRR for phased project expenditures because they can give negative flows after the first year. One algorithm used by EPA actually changes the cash flows so they will be less troublesome.


35. The proposal indicates that because RIRR is "not a monetary value" it circumvents problems of placing a monetary value on benefits; however, this does not appear to be a valid claim - calculation of RIRR requires monetary estimates of future streams of costs and benefits.

The public sectors in free market economies do not generally make investments for profit; therefore, DCF analyses used in the private sector for that purpose are generally inappropriate.  I do not recommend using DCF for analyzing public sector benefits or the positive cash flows imputed from them.  However, if benefits are equivalent, a wash, we can compare cost streams directly.  This is done using differential cash flows for mutually exclusive projects as explained most textbooks and summarized in the proposal.  See "Incremental Cash Flows" in Principles of Corporate Finance Questions  or tab "3. Incremental" for mutually exclusive projects in the Excel 5.0 version of the Questions.


36. Furthermore, RIRR may well be appropriate for evaluating future cash flows, but this is not the problem usually faced by the Agency because such a large portion of the benefits of its programs are non-pecuniary in nature.

RIRR would be of little or no value in analyzing benefit streams converted to imputed monetary values in any public sector undertaking in my opinion. RIRR is valuable for analyzing only cost streams in the public sector.  This is done using incremental cash flows.  NPV on incremental cash flows is well accepted and widely used.  IRR works the same.


37. The PI does not demonstrate exceptionally strong qualifications for conducting the project either in the form of previous grant experience or in terms of peer-reviewed publications.

The PI directed all and managed some of over 200 major defense cost, budget.schedule and program management project totaling over $25 million.  They were competitive contracts, not grants.  The analyses supported multi-billion dollar acquisition decision.

The thesis in the proposal has been available for peer review at the Social Science Research Network and online for two years.  The detailed results of the peer review were offered in the proposal, but greatly exceed the total page count.  Results of the informal peer review, Principles of Corporate Finance Questions, are summarized in and linked from the proposal.


 38. This is somewhat mitigated by the fact a primary objective of the proposal is to communicate and educate EPA regarding RIRR (rather than to conduct innovative research).

Other than seeing how completed research might enhance EPA programs and analyses, research is largely complete. The innovation now lie in conveying the results and achieving consensus or identifying reasons why it is unachievable.


39. Even so, despite the fact that the proposal is generally well-organized, the written discussion generally lacks clarity, which casts some doubt on the PIs effectiveness as a communicator.

Okay.


40. There is an important inconsistency in the stated objective of the research. On the one hand the PI wishes to "make no value judgments or advocate particular choices" regarding RIRR v. NPV. On the other, the PI states clearly "The current goal is to promote RIRR rather than simply inform." This does not inspire confidence in the objectivity of the proposed research.

The "Objectives" of the project conclude: "This translates to first validating RIRR for environmental decision-making, then incorporating it where appropriate."

The "Relationship to PI Longer-term Goals" concludes: "The current goal is to promote RIRR rather than simply inform."  This is my goal, not a project objective. They are complementary, not identical. 


41. The proposed budget seems excessive -- about $90K in each year, all for the PI Given that it proposes to expand completed research, the tasks and corresponding budget should be better defined and justified to merit this amount of funding.

Budget justification is separate from the proposal per NSF Fastlane instructions. It includes institution overhead, publications, multimedia distribution costs, travel, computer services, support from a data technician. The PI portion is less than half of the total.  The PI portion is well below the NSF budget guidance.  It is well below my last salary on government contracts.  It is one-third of my last billing rate on government contracts.


42. In accordance with the objectives stated in the NSF/EPA proposal solicitation, the project does propose to significantly advance understanding through training and educating EPA (through various media) about the potential applications of RIRR. Unfortunately, the proposal does not provide a very strong justification for why this is an important issue for EPA.

From what is available online and in a draft EPA Economic Analysis guide, there is no consideration of using IRR (or RIRR) in conjunction with EPA programs and analyses. All of the supposed theoretical problems with IRR are repeated throughout various Energy Star decision-making instructions.


 43. The proposal does not include a very strong description of how the results of the analysis would be disseminated beyond the EPA audience.

Given the policy implications, that should be determined by EPA. Options were proposed, primarily for electronic multimedia and online.

 


Discussion Questions

A. Are there different views regarding IRR in the stems above?   How many of the stems frame the issue in terms of RIRR and NPV being alternative rather than complementary methods?  Vice versa?

B. In the reviewer stems above, does IRR (or RIRR) give the same decision as NPV?  

C. Is it reasonable to assume that 100-fold differences in project size ($10 versus $1,000 in an example above) always have identical risk?  What about 10-fold?  Two-fold?  Same amount invested?

D. People making major financial decisions in business more often that not ask for a rate.  If this can be accomplished without changing any decision rules or measures, other than not excluding one measure, would that be a positive contribution?

E. Other than EPA-peculiar, are the arguments in the stems above above different from those at Principles of Corporate Finance Questions?  If yes, which ones?


 Click Please contact me if you find bad links, need additional information--for any reason.

Access Since July 21, 1999: