چارچوب آزمون چندگانه بهینه سازی شده برای انتخاب پروژه در بخش دولتی، با یک مثال دفع زباله های هسته ای موردی

Since 1967 the real (net of inflation) discount rate used to assess public sector projects in the UK has changed four times: from 8% to a peak of 10%, then to 5%, then to 6%, and then to the current position of 3.5% for years 1–30 declining in increments to 1% by year 300. This paper argues that the trend is in the right direction, but the associated decision process remains inappropriate, because the rationale for the changes has been flawed. Many other countries use discount rates of 6% or more in decision processes which are probably even more inappropriate. The fundamental problem stems from attempts by economists to embed too many conflicting considerations in a discount rate which is used in a single hurdle rate test. A multiple test approach is needed to address all the issues that the associated decision processes need to consider. This paper offers a way forward based on alternative economics perspectives, operational research perspectives, and established practice in a project risk management context. A case-based example concerned with the disposal of intermediate level nuclear waste in the UK illustrates how it works, and the implications of current UK practice, suggesting a very different view of a decision a decade ago which is a topical again. The proposed framework should be of interest to anyone interested in public sector projects, in the UK and elsewhere, and its generalisation has implications for private sector projects and private public partnerships (PPPs). Generic project risk management processes concerned with the whole of the project life cycle could embed a generalised form.

Investment appraisal to select projects in the public sector presents considerable difficulties. Potential investment projects are numerous, and they often involve large capital expenditures, benefits that are uncertain or difficult to quantify, and consequences which may be very long-lived. Decisions in any one year will be constrained by ongoing commitments, anticipated future levels of resources, and anticipated future demands on resources. Professor Sir Alan Budd, a distinguished former Chief Economic Advisor to HM Treasury, has argued that the public sector investment appraisal process is one of constrained optimisation [1]. In terms of constrained optimisation, the decision problem a government must address can be described as follows: 1. the primary objective of the government decision making process is maximising the expected net present value (NPV) of the portfolio of all current and planned public sector investments; 2. one secondary objective is seeking an optimal trade-off between this primary objective and associated downside risk; 3. further secondary objectives address optimal trade-offs between expected outcomes and associated risk in terms of all reasonable economic, political and social concerns not captured by a cost-benefit approach to NPV, and optimal trade-offs between these concerns; 4. secondary objectives include seeking an optimal size for the portfolio of public sector investments. From an operational research (OR) perspective, the only way to make this operational is a goal programming [2] approach. The distinguishing feature of a goal programming approach is dealing with all secondary objectives via appropriate constraints, and managing trade-offs between objectives via shadow prices. More specifically: 1. the objective function in the mathematical programming framework is maximising NPV in expected value terms (the primary objective); 2. constraints limit values for all secondary objectives; 3. constraints express any required limitations on the available choices beyond those imposed by secondary objectives (for example, only one of two alternatives may be feasible); 4. a basis for trade-off analysis is provided by shadow prices, defining the change in the value of the primary objective function if any constraint is relaxed by one unit; 5. trade-off analysis involves adjusting the constants assumed for each secondary objective constraint on successive iterations until the shadow prices for all constraints reflect appropriate trade-offs. The proposed role of this goal programming perspective is insight into simple decision rules. In principle it might be used directly. However, this paper argues that it should be used indirectly, as a ‘maximum insight process’, comparable to ‘background theory’, to develop a much simpler operational framework. This simpler operational framework should be based on a ‘minimal process’, for considering projects one at a time on a multiple test ‘traffic light’ basis, and a related ‘intermediate process’, to periodically review marginal projects on a portfolio basis and adjust the minimal process. This approach uses the concept of ‘constructive simplicity’, with origins in mainstream operational research [3], extensively developed in the context of project risk and uncertainty management [4]. Constructive simplicity implies starting operational modelling with minimal complexity, with the aim of facilitating transparency and understanding. Operational complexity is added only when it is useful. The concern is a flexible and general approach to optimising the decision process as well as the choices made using the decision process. This departs from most mainstream economic theory and practice associated with discounting for public sector investment selection. However, it is consistent with some very basic established economics principles, and operational research traditions which date back to the origins of OR. The latest advice from HM Treasury [5] can be interpreted as a first step in this direction. Viewed within this framework, public sector investment appraisal to date has not progressed much beyond the first requirement of those on either list above. The focus of investment appraisal in the UK prior to the 2003 ‘Green Book’ advice from HM Treasury [5] was the maximisation of NPV for individual projects using a single hurdle rate test. The 2003 Green Book considers multiple objectives via multiple tests explicitly for the first time, in line with our proposed framework. However, it does so one project at a time, without a basis for optimising the decision process. The limited international literature review undertaken by the authors suggests that the UK Treasury may be in the forefront internationally, in terms of both a realistically low discount rate and a multiple test decision process. The basis of our proposed approach, and all other approaches, is NPV. For present purposes define the NPV of a project’s costs and revenues as follows, using notation convenient for our example: equation(a) View the MathML sourceNPV=-C0+∑t=1…nStRtt+CnRtn Turn MathJax on where n, life of project in years; t, (1 … n); C0, initial capital expenditure; St, net cash flow in year, t, a saving in operating costs; Rt, discount factor for year t, 1/(1 + Dt/100); Dt, annual discount rate for year t (%); Cn, terminal value of the project. In practice C0, St, Cn and Dt may be uncertain, and Dt may be assumed to be constant D for all time periods. A project comprising a particular set of estimated cash flows is considered worth undertaking if NPV > 0 for the ‘correct’ discount rate(s). For this calculation to have an economic meaning, use of the ‘correct’ discount rate(s) is essential. Moreover, the use of ‘incorrect’ discount rate(s) can induce important bias in investment decisions. Economists have recognised the importance of this bias for a long time. For example, Hawkins and Pearce [6] argue IRR (internal rate of return) rates (obtained by setting NPV equal to zero and solving for the discount rate) should never be used to choose between projects, because they have no economic meaning beyond their use for sensitivity analysis of a given single NPV calculation, a fundamental principle we endorse. Unfortunately, differences of opinion on the ‘correct’ discount rate have a long and confusing history which has lost sight of Budd’s constrained optimisation view and the Hawkins and Pearce concern for a correct rate which does not induce bias. While the authors are critical of HM Treasury recommendations on investment appraisal, they believe considerable sympathy for the predicament of HM Treasury and comparable bodies in other countries is warranted. The issues involved are exceedingly complex, and the relevant economics literature is very confusing. Disputes about ‘correct’ discount rates often involve incompatible basic framing assumptions, different world views, and conflicting political agendas, and these differences are usually implicit. Private sector investment decision processes raise some similar issues, as do private public partnerships (PPPs). The differences are too significant to address here, but the common ground will be clear. A full synthesis of the relevant economics literature from the perspective of the proposed approach would be very useful, but this too is beyond the scope of this paper. The paper by Sir Alan Budd cited above [1] is titled ‘Economic Policy – Too Important to be Left to Economists?’, a question he answers in the affirmative. In a similar spirit, the authors believe that a discussion which engages project champions and their advisors, operational researchers, economists, actuaries, accountants, other finance specialists, economic policy makers, politicians and all other interested parties within the proposed framework ought to resolve the current confusion in the long run. This paper aims to initiate such a discussion. It also makes some outline recommendations to HM Treasury which could be adapted for other countries. In the short run project champions and their advisers can use the structure provided by this paper to engage in discussion with HM Treasury or comparable bodies in other countries about possible bias against their projects. In the medium term generalisation to deal with private sector and PPPs can be addressed. Guides concerned with project selection as part of a whole life cycle perspective on projects, like the Risk Analysis and Management of Projects (RAMP) guide [7], can embed generalised versions. The body of this paper starts with an overview of issues associated with identifying a ‘correct’ discount rate in the economics literature, linked to some of the different discount rates used by HM Treasury. Comparable bodies in other countries will probably be using one of the approaches adopted by HM Treasury over the past forty years, so this background should provide a starting point for understanding the relevant economics for most readers, whether or not they are UK based. The proposed decision process is then outlined. A topical case-based example is provided next, concerned with the disposal of intermediate level nuclear waste in the UK, to help explain the proposed ‘traffic light’ approach to individual projects and the implications of alternative approaches. The paper concludes with a brief discussion of the periodic review part of the proposed process.

Discussion of investment appraisal approaches which underlie project selection are traditionally dominated by arguments about a ‘correct’ discount rate assuming only one rate will be used in a simplistic decision process which uses a single hurdle discount rate test. Each school of thought attempts to determine an appropriate discount rate by considering a subset of the issues, but, as Baumol put it, ‘no one discount rate will deal with all the issues’[15]. HM Treasury have moved to a discount rate which is so low it has become self-evident further tests are needed. However, they have chosen a basis for their discount rate which involves STPR issues, which should not be in the discount rate, ignoring ACC considerations in the process. Further, they have not considered the SOC or risk issues appropriately. Further still, they have not directly considered optimising the portfolio of projects selected or optimising the effort involved in the selection process. A constructively simple approach is concerned with optimisation at both levels, in a framework which keeps the practice as simple as possible. The authors’ position is that the discount rate should be defined by the actual cost of capital, with an ACC approach used as the basis of the bond test, the basic financing test. All other issues need separate tests which do not operate via the discount rate. A simple sequence of separate tests will be effective for most projects, provided the issues are not too complex and the choices are reasonably obvious. Marginal choices and complex choices will require simultaneous consideration in a more complex review process. The authors have no doubt that many of their suggestions can be improved upon, and some of the assumptions used in this paper may be successfully challenged. However, the authors believe the basic framework proposed is a sound way forward. It should yield the benefits of low discount rates that the sustainability lobby have quite properly been seeking (via STPR approaches), but also meet the concerns of the SOC lobby, who quite properly wish to see a favourable return on rationed public investment. It should allow risk to be addressed properly. It should facilitate simple tests for simple projects, and complex analysis in a portfolio framework for complex issues like UK energy policy. It should make political priorities operational in a transparent manner. It should do all this in a framework that non-economists can understand, allowing them to contribute to the key process decisions and the key trade-off decisions. Further, it should generalise to private sector and PPP projects, and this should facilitate comparison of alternative approaches to funding some projects and organisations.