Here is Marty Feldstein, through some vias:
[F]inancing additional government spending by an across the board rise
in all marginal tax rates would make the cost per dollar of government
spending equal to $1.76.
These two facts – that the actual revenue is only 57 percent of
the static gain and that the deadweight loss is 76 cents per dollar of
revenue – should be central to any consideration of tax policy. And yet
they are not.
What is talked about even less is that most government programs are, for better or worse, irreversible investments. They don’t fade away very easily, even if they have been proven ineffective or harmful. Do you really think that the Departments of Energy and Education have to pass a market test every year? In fact government programs, on average, grow almost automatically.
Do you know the literature on irreversible investment and option value? Under plausible parameter specifications, you need a benefit-cost ratio of 3 to 1 or more before it makes sense to proceed with the irreversible investment. Otherwise it is better simply to wait. That’s not 1.3 to 1, that’s 3 to 1.
The main argument for not waiting is simply that the political process cannot do much useful with any new information that is generated by waiting. But viewed more broadly, that is hardly an argument for proceeding in the first place.
The bottom line: Cram all those numbers together into your noggin’ and keep them there.