An appraisal is only useful if it actually improves a decision. This final module covers the step from analysis to decision: how to present an appraisal honestly, why estimates are so systematically wrong, how to correct them, and how the political incentives around project selection corrupt the whole exercise. It closes the course and connects directly to the public-investment-management problem of the Public Budgeting course.
The appraisal report
A good appraisal report does not just announce an NPV; it exposes the reasoning so a decision-maker can judge it. That means: a clear statement of the counterfactual and the options compared (including the do-nothing and do-minimum); the key assumptions laid out explicitly; the sensitivity and switching-value analysis showing what the answer depends on; the distributional analysis showing who gains and loses; the risks and their treatment; and an honest account of what could not be quantified. The UK Green Book's 'five-case model' embeds this — a project must make a strategic, economic, commercial, financial, and management case, not just clear an NPV hurdle. The discipline is transparency: a reader should be able to see and challenge every judgement, which is the difference between appraisal that informs and a number that launders a decision already taken.
Optimism bias
Projects systematically cost more and deliver less
Bent Flyvbjerg's research across hundreds of megaprojects worldwide documents a brutal regularity: costs are systematically underestimated and benefits systematically overestimated at appraisal, by large margins, again and again. This is optimism bias, and it has two distinct mechanisms: • The planning fallacy (Kahneman-Tversky) — a genuine cognitive tendency to underestimate the time, cost, and risk of our own plans, taking an 'inside view' (this project's specifics) and neglecting the 'outside view' (how projects like this actually turn out). • Strategic misrepresentation — deliberate low-balling of costs and inflation of benefits by promoters who want the project approved, knowing that once it is begun and money is sunk, it is politically too committed to cancel. This is not cognitive error; it is incentive (the political economy below). The two are hard to tell apart in any single case but both push the same way, which is why appraisal optimism is so persistent and why uncorrected appraisals are unreliable.
Reference-class forecasting
Take the outside view
The corrective, pioneered by Kahneman and operationalised by Flyvbjerg and now mandated in the UK Green Book and elsewhere, is reference-class forecasting. Instead of building a cost/benefit estimate bottom-up from this project's assumed specifics (the inside view), look at the actual outturn distribution of a reference class of similar completed projects (the outside view): how much did comparable railways/dams/IT systems actually cost and deliver versus their original appraisals? Then base your estimate on that distribution, applying an 'optimism-bias uplift' to costs and a discount to benefits drawn from the real track record. Reference-class forecasting corrects both the planning fallacy and (by anchoring on independent data rather than the promoter's claims) strategic misrepresentation. It typically raises cost estimates substantially — and turns many apparently-attractive projects marginal or negative, which is exactly the point: it replaces hope with history.
The political economy of appraisal
Appraisal does not happen in a political vacuum, and this is where the course connects back to the Political Economy & Governance area. Project sponsors have incentives to game the appraisal — to commission favourable numbers, choose generous assumptions, pick a low discount rate, and inflate benefits — because the appraisal is the gate to a project they want for reasons (visibility, rents, patronage, the white-elephant logic of the Public Budgeting course) that have nothing to do with the NPV. The appraisal can thus become advocacy dressed as analysis. The defences are institutional, not technical: independent appraisal (the analyst should not report to the promoter), a credible appraisal gate that political fiat cannot override, standardised assumptions (mandated discount rates, reference-class forecasting) that remove discretion to game, and transparency that exposes the assumptions to challenge. Good appraisal is as much about who does it and under what rules as about the technique.
Closing the loop: ex-post evaluation
Finally, appraisal should not end at the decision. Ex-post evaluation — going back after a project is built to measure what it actually cost and delivered — is what builds the reference classes that correct future optimism, holds promoters accountable for their forecasts, and lets the system learn. Yet it is rare: once a project is approved and built, the political and bureaucratic incentive to revisit whether it was worth it is weak (no one wants to document their own white elephant), so the same mistakes recur. Closing the loop — institutionalising honest ex-post evaluation and feeding it back into appraisal (the rigorous version of which is the Impact Evaluation course) — is the unglamorous discipline that turns a one-off appraisal into a system that gets better over time. It is the final lesson of the course: appraisal is not a number you produce once, but a loop you close.
Exercise
A ministry presents a flagship metro project with an appraisal (commissioned by the project's own promoter) showing a benefit-cost ratio of 2.2, based on optimistic ridership forecasts and a low discount rate, with no reference-class analysis. Senior politicians strongly back it. (1) Identify the reasons to distrust the headline BCR. (2) Explain how reference-class forecasting would change the appraisal and roughly which direction the BCR would move. (3) Diagnose the political-economy problem and why technical fixes alone won't solve it. (4) Recommend the institutional safeguards that should govern the decision, and the role of ex-post evaluation.