tim's experiment

A llttle snippet stuck in my brain from the Legco debate (as I call it for want of a better word) on the third runway. The government spokesperson gave a figure for the total cost of the new runway, and followed it with a figure for the runway’s contribution to the local economy over the ensuing 30 years.

And the question is, of course, what was this figure worth. Look at it this way. The runway will not be finished for six years. So the official figure purports to tell us the share of the Hong Kong economy contributed by the new white elephant between six years from now and 36 years from now. This prediction comes to you from a government which has consistently failed to predict its own income and expenditure over the ensing 12 months.

There are good reasons why predictions of this nature should be regarded as a joke. Over 36 years your errors proliferate. If you make an error of one percent a year in your estimate, this does not mean that at the end of 36 years your estimate is 36 per cent out. This works like compound interest: the accumulated error will be over 50 percent. Inevitably, there will also be unforseen events of a one-off kind: banking crises, epidemics, volcanic eruptions or whatever. These will blow large holes in your forcast. A serious prediction would at least come with a range. One could pretend to take seriously a figure which came with a caveat that the outcome might vary by 30 per cent either side. But I fear on topics of this kind the official figures are not intended to be honest: they are intended to be propaganda for what the government has already decided it wants to do.

Well I came across an interesting piece of research the other day. An academic had studied all the railway projects in the world over the last three decades of the 20th century. He came up with some interesting figures. The average cost over-run for these projects – not the maximum, the average, was 120 per cent. The average shortfall in the passenger numbers compared with the pre-project predictions (once again note this is an average) was 50 per cent. In other words, as a rough rule of thumb you can expect any railway project to cost twice as much as it was supposed to, and then to attract half the predicted number of passengers. Shares in the High Speed Rail Link, anyone? I have seen no comparable figure for new airports or runways but it does show you that there is a large margin of error here, and most of the errors tend to be in the same direction.