Confirmation Bias and Stimulus

Will Wilkinson and Ezra Klein notice that people tend to cherry pick the papers they like on big macroeconomic questions.  This is perhaps the greatest reason why I am a stimulus agnostic.  Oh, I believe that if you increase one component of a compound variable, you can produce a notional increase in that variable; this is virtually a truism.  But I am less convinced that it produces a large net increase in public welfare.




Wading through the online debates, I note that opinions on stimulus are nearly 100% correlated with the composition of that stimulus, and the opinionator's prior view of that activity.  So when Democrats are in power and stimulus is mostly spending, liberals think that the stimulus is an issue of fierce moral urgency stymied by venal greed and rank idiocy, while conservatives develop deep qualms about budget deficits.  When Republicans are in power, and stimulus consists mostly of tax cuts, Democrats get all vaporish about deficits and the income deficit, while Republicans suddenly realize that the normal rules don't apply in an emergency.  When out of power, both sides will grudgingly concede that some small amount of highly temporary stimulus might be all right, but note (correctly) that the other side seems to be trying to make permanent as much of this "stimulus" as possible.

For me, then, this mostly ends up as a proxy war over the level of government spending, a war I'd rather fight honestly on value grounds rather than attempting to disguise my preferences with a shoddy veneer of "scientific" logic. 

Yet people only seem to be able to recognize the instrumental arguments on the other side.  For example, we're now in year seven or eight of a popular liberal parlor game:  find the macroeconomics textbook written by a conservative that discusses how stimulus might work.  Matt Yglesias' commenters are quick to note item three from this Mankiw excerpt and jump to charges of bad faith:



Yet many of them miss the obvious rejoinder:  that outside of economics models where you can hold every variable but one constant, ceteris is rarely paribus.  Unfortunately, when the government increases spending in the real world, it cannot simply pluck the (non-inflationary) cash off the money tree out back of the White House.  Either taxes, inflation, or default risk must eventually go up as a result of this new spending.   All of these things have negative impacts on output which mitigate the benefits of the spending--and depending on how forward-looking you believe consumers and businesses are, that mitigation may start even before you manage to get the cash out the door.  Mankiw himself notes all of this today.

Now, the stimulus critics may in fact be wrong--but they are not insupportably hypocritical.  Probably they're engaging in what cognitive scientists call "motivated cognition"--seeking out facts that affirm what they already believe. 

The temptation is simply to seek the messy middle--perhaps to my credit, I was as skeptical of stimulus when it was composed of tax cuts, as I now am when it is composed of spending. (Though agnostic might be a better word for my attitude; I genuinely don't know how well it works.)  But I've seen no great evidence that the universe is a centrist, so I'm not sure we're any more likely to be right.

Nor am I sure it much matters.  There is no political possibility of a stimulus, at this point, so all of these arguments have a strong flavor of Laputan arcanology.