The planning fallacy, much emphasized by behavioral scientists, suggests that people systematically underestimate the time it will take to complete projects. However, the planning fallacy is not limited to time; it applies to the costs and benefits of projects as well.
Benefit-cost ratios are typically overestimated by 50-200 percent
This points to an important problem for cost-benefit analysis. Due to the planning fallacy, ex-ante estimates of costs and benefits are likely to be biased, with costs underestimated and benefits overestimated. If the biases were small, this would not matter too much. But the biases are substantial, as documented by the table below, which measures bias in cost and benefit estimates by cost overrun and benefit overrun, respectively, for eight different types of large capital projects. The data show that benefit-cost ratios produced by conventional methods are typically overestimated by between 50 and 200 percent, depending on project type.


This means that cost-benefit analysis will typically be a poor basis for decision making. In fact, ex-ante benefit-cost ratios for capital projects are so misleading as to be worse than worthless, because decision makers are led to believe they are being informed when in fact they are being misinformed. As a consequence, decision makers may give the green light to projects that should never have been started.
Cost-benefit analysis must be de-biased
This does not prove the uselessness of cost-benefit analysis as such, needless to say. It demonstrates, however, that cost-benefit analysis must be effectively and independently de-biased before serving as a basis for decision making, and incentives must be in place to make this happen. Fortunately, good de-biasing tools and incentive structures exist and have proven their worth in practice, in both government and business.

See full study by Bent Flyvbjerg and Cass Sunstein here: http://bit.ly/1hCsaOg
On how to de-bias cost-benefit analysis, see here: http://bit.ly/1ly0JDu and http://bit.ly/1oB2bb4