Timothy Taylor writes about “The Iron Law of Megaprojects vs. the Hiding Hand Principle“. He begins by quoting a piece by Bent Flyvbjerg in Cato Policy Report (January 2017):

Megaprojects are large-scale, complex ventures that typically cost a billion dollars or more, take many years to develop and build, involve multiple public and private stakeholders, are transformational, and impact millions of people. Examples of megaprojects are high-speed rail lines, airports, seaports, motorways, hospitals, national health or pension information and communications technology (ICT) systems, national broadband, the Olympics, largescale signature architecture, dams, wind farms, offshore oil and gas extraction, aluminum smelters, the development of new aircrafts, the largest container and cruise ships, high-energy particle accelerators, and the logistics systems used to run large supply-chain-based companies like Amazon and Maersk.

For the largest of this type of project, costs of $50-100 billion are now common, as for the California and UK high-speed rail projects, and costs above $100 billion are not uncommon, as for the International Space Station and the Joint Strike Fighter. If they were nations, projects of this size would rank among the world’s top 100 countries measured by gross domestic product. When projects of this size go wrong, whole companies and national economies suffer. …

If, as the evidence indicates, approximately one out of ten megaprojects is on budget, one out of ten is on schedule, and one out of ten delivers the promised benefits, then approximately one in a thousand projects is a success, defined as on target for all three. Even if the numbers were wrong by a factor of two, the success rate would still be dismal.

So far, so good. But then Taylor says this:

A common comeback to the Iron Law of Megaprojects is that if we pay attention to it, we will be so dissuaded by costs and risks of megaprojects that nothing will ever get done. Alfred O. Hirschman offered a sophisticated expression of this concern in his 1967 essay, “The Hiding Hand.” Hirschman argued there there is rough balance in megaprojects: we tend underestimate the costs and problems of megaprojects, but we also tend to underestimate the creative with which people address the costs and and problems that arise.

I will come to the irrelevance of Hirschman’s argument, but first a few more tidbits from Taylor:

[Flyvbjerg] argues that a number of prominent megaprojects have been completed on time and on budget. When choosing which megaprojects to pursue, it is useful to avoid underestimating costs and overestimating benefits. [Wow, what an astute observation.] …

Further, Flyvbjerg offers a reminder that even when a megaproject is eventually completed, and seems to be working well, project may still have been uneconomic–and society may have been better off without it.

The second comment brings Taylor close to the heart of the matter. But he never gets there. Like most economists, he overlooks the major flaw in the application of cost-benefit analysis to government projects: Costs and benefits usually have different distributions across the population. At the extreme, benefits that accrue only to the indigent are borne almost entirely by the non-indigent. (The indigent may pay some sales taxes.)

Cost-benefit analysis (applied to government projects) effectively rests on the assumption of a social welfare function. If there were such a thing, then it would be all right for people to go around punching each other (and worse), as long as the aggressors derived more gains in “utility” than the losses suffered by the victims.