The noble Mankiw has a thoughtful reply to a question from me about the effect of fiscal and monetary policy on nominal GDP. (The question is based on one of my old posts):

How far off is the vertical LM case as a practical matter? One way to answer this question is to look at the fiscal-policy multiplier. In chapter 11 of my intermediate macro text, I give the government-purchases multiplier from one mainstream econometric model. If the nominal interest rate is held constant, the multiplier is 1.93. If the money supply is held constant, the multiplier is 0.60. If the LM curve were completely vertical, the second number would be zero. To return to Bryan’s WWII example, taking these estimates literally, if the Fed had held the money supply constant rather than keeping interest rates low, the WWII boom would about been about 1/3 as large as it was.

Where does Mankiw think I go wrong?

I disagree with Bryan’s suggestion that the LM curve is vertical, however. Introspection is not a particularly reliable way to measure elasticities. There is a substantial empirical literature on money demand that demonstrates that it is interest-elastic.

I agree that introspection is far from perfect. But in practice, economists form their beliefs based on a weighed average of introspection and econometrics. And we should, becase econometrics is far from perfect too. Even if econometricians agreed, for example, that prime-age male labor force participation was highly responsive to wages, I would still think “I’ve never heard of a prime-age male who dropped out of the labor market because wages were too low. How can that be?”

From another angle: What happens when econometricians reach counter-intuitive conclusions about elasticities? Example: There was a long period in macro when researchers kept finding that money supply shocks increased interest rates (the “interest rate puzzle”). The main result was that econometricians kept trying different approaches until they finally got the “right sign.”

Perhaps the econometrics that was consistent with introspection was of vastly higher quality than all the other approaches. But I doubt it. If we didn’t rely on introspection, we’d probably still be saying that “the balance of evidence shows that monetary shocks raise interest rates.”

None of this means that were should ignore econometric evidence. But when econometrics and introspection conflict, we should think harder about both – not give econometrics veto power.

P.S. One of Mankiw’s commenters writes: “If I didn’t know better, I’d have to conclude that Bryan Caplan was too young to have a bank account in the early 80s.”

Guilty, as charged. I’m willing to admit that money demand would respond to large changes in interest rates. But that’s still consistent with the LM curve being roughly vertical over the ranges that are relevant today.

Readers’ Query: Do you change your cash holdings when nominal interest rates change? Anyone you know?