New York Times columnist David Brooks recently identified several policy ideas that, in his view, amount to “a completely obvious agenda to create more middle-class, satisfying jobs.” One of the ideas is to have the federal government “borrow money at current interest rates to build infrastructure, including better bus networks so workers can get to distant jobs.”

Now, if there is a pressing need for large-scale infrastructure projects, they can obviously make sense. In fact, the latest World Economic Outlook report from the International Monetary Fund argues that, if certain economic, fiscal, and monetary conditions prevail, debt-financed infrastructure spending can deliver output increases large enough to produce a long-term reduction in the public-debt-to-GDP ratio.

Sounds great! Except there are some rather important caveats, including this one: “If the efficiency of the public investment process is relatively low — so that project selection and execution are poor and only a fraction of the amount invested is converted into productive public capital stock — increased public investment leads to more limited long-term output gains.”

As Washington Post editorialist Charles Lane writes, “That’s a huge caveat.” Indeed, it helps explain why so many highly touted infrastructure projects have ultimately proven so disappointing. In the United States, such projects are notoriously expensive, owing to a variety of labor, environmental, and other policies. Moreover, they tend to encounter unforeseen contingencies that disrupt the plans — and thwart the ambitions — of politicians and developers.

Here’s Lane:

“Long-term costs and benefits of major infrastructure projects are devilishly difficult to measure precisely and always have been. The C&O Canal was supposed to revolutionize U.S. commerce; it ultimately proved a bust because its proponents (including George Washington) failed to anticipate the true hazards of building it or the advent of the railroad.

Today we have ‘bridges to nowhere,’ as well as major projects plagued by cost overruns and delays all over the world — and not necessarily in places you think of as corrupt. Germany’s still unfinished Berlin Brandenburg airport is five years behind schedule and billions of dollars over budget, to name one example.

“Bent Flyvbjerg of Oxford’s Said Business School studied 258 major projects in 20 nations over 70 years and found average cost overruns of 44.7 percent for rail, 33.8 percent for bridges and tunnels and 20.4 percent for roads.

“Even popular projects, like New York City’s subway, often require operating subsidies after completion. To be sure, that system yields abundant ‘positive externalities’; the Big Apple wouldn’t be livable without it. How that pencils out in macroeconomic terms, though, is anyone’s guess.

“In short, an essential condition for the IMF concept’s success — optimally efficient investment — is both difficult to define and, to the extent it can be defined, highly unrealistic.

“As Flyvbjerg explains, cost overruns and delays are normal, not exceptional, because of perverse incentives — specifically, project promoters have an interest in overstating benefits and understating risks.

“The better they can make the project look on paper, the more likely their plans are to get approved; yet, once approved, economic and logistical realities kick in, and costs start to mount. Flyvbjerg calls this tendency ‘survival of the unfittest.’

“Obviously, the U.S. and world economies need first-class roads, harbors, airports and electrical grids.

“Equally obviously, they must be upgraded to accommodate and facilitate growth, and it can be worth risking public funds, including, at times, borrowed funds, to make that happen.

“But this is an inherently imprecise business. Governments that invest in infrastructure on the assumption it will pay for itself may find out that they’ve gone a bridge too far.”