Thursday, 3 December 2009

Spurious correlation watch

Andrew Sullivan takes a break from spelunking in Sarah Palin’s reproductive tract to provide us with highly superficial social scientific analysis:

Ezra Klein asks:

Is there any evidence that financing wars brings them to a quicker close? Any papers examining this question?

From Bruce Bartlett’s column last week:

History shows that wars financed heavily by higher taxes, such as the Korean War and the first Gulf War, end quickly, while those financed largely by deficits, such as the Vietnam War and current Middle East conflicts, tend to drag on indefinitely.

How about a more plausible explanation: Korea and Gulf War I were conflicts against state actors that fought using traditional military tactics, while Vietnam and the Middle Eastern conflicts (particularly in Afghanistan) were/are conflicts mostly involving indigenous, non-state resistance movements or terrorist cells with some degree of local popular support (the Viet Cong, Iraqi Shiite and Sunni extremists and al-Qaeda, and the Taliban and al-Qaeda, respectively) that are engaged in unconventional warfare. The mode of funding would seem to have little to do with conflict length. Particularly since World Wars I and II were also funded by massive deficit spending, yet U.S. involvement in both conflicts was comparatively brief (although not on the order of Gulf War I).

Besides, the Johnson-Nixon era’s massive expansion of the deficit-financed American welfare state would be a serious conflating factor in attributing Vietnam’s success or failure to its funding approach, much as the effects of the Bush tax cuts likely dwarfed Iraq and Afghanistan spending as a source of the increased budget deficit over the past eight years and change; the liberal CBPP think-tank attributes the effects of one year (2004) of the Bush tax cuts as being $276 billion in reduced tax revenues (and thus increased debt), far more than the annualized cost to the Treasury of both conflicts combined even based on the most pessimistic estimates.