From David Wessel in the Wall Street Journal: "The president measures Social Security liabilities not over the actuaries' usual 75-year horizon but over eternity -- which generates a frightening $10 trillion in unfunded promises. But when he measures the cost to the Treasury of his tax cuts, he prefers the five-year horizon."
I agree with Bush's measurement stance on Social Security, which is that the infinite horizon offers the right perspective, subject only to the point that claims about our current intentions regarding future policy may be highly dubious. (Congress could play games with infinite horizon forecasting by pretending that totally unfeasible tax increases and spending cuts would take effect in the far future.) But people who reject the infinite horizon on the ground that we don't know what, e.g., the economy will look like in the future overlook the fact that our uncertainty may be symmetric. That is, if reputable independent forecasters reasonably deem X the median likely outcome, then presumably this means worse-than-X and better-than-X have equal weight as probabilities. It would thus be a silly error to assume that our uncertainty means we must act as if we are forecasting zero problem for the future.
It is certainly worth pointing out how phony and hypocritical Bush is being about the Social Security fiscal gap when it is small compared to the holes he has shot in the budget through tax cuts, current spending increases, and an unfunded $16.6 trillion Medicare prescription drug entitlement. But the problem isn't infinite horizon forecasting; it's inconsistent use thereof. Let's project everything on an infinite-horizon basis (though also with 5 year, 10 year, and 75 year estimates) and require that the forecasts ignore phony policy changes that are claimed for the future (such as officially sunsetting the Bush tax cuts) but that the enacters are not willing to forthrightly embrace as their intended policy.