The exercise needs JavaScript to deal the six proposals, to let you fund three of them and to show the outcome of each proposal afterwards.
How this works
The six proposals are invented and all numbers are generated in your browser. Each proposal has an outcome, meaning the benefit-cost ratio that it delivers if it is built. The six outcomes are drawn from one lognormal distribution, a distribution that gives positive values only and has a longer tail on the high side. Its median is 1.25 and its sigma, the standard deviation of the logarithm of the value, is 0.11. Because all six outcomes come from the same distribution, the proposals are of similar quality and the project names serve only as labels. The exercise also shows what the proposals left out would have delivered, which a committee does not observe in practice.
Each estimate is the outcome multiplied by a random error drawn from a lognormal distribution with a mean of 1 and a sigma of 0.25. Because the error has a mean of 1, an estimate equals the outcome on average, which is the sense in which the estimates are unbiased. Each ratio is rounded to two decimals before the averages of three ratios are calculated. In a benefit-cost ratio, benefits and costs that fall in later years are converted to their value today, called the present value, before the ratio is calculated.
The figures for 10,000 rounds come from the same model, with a committee that funds the three proposals with the highest estimates in each round. They are calculated in your browser from a fixed starting number for the random draws, so that they are the same on each visit. Each new deal takes its starting number from the clock.
Source: Eliasson, J., & Fosgerau, M. (2013). Cost overruns and demand shortfalls – Deception or selection? Transportation Research Part B, 57, 105–113.