Scott Sumner has paid me the enormous compliment of linking to my recent post on monetary policy, and commenting thereon with his usual clarity and intelligence. Allow me to attempt to return the favor, and keep the conversation going.

One point I made in my first piece: if we were to establish a new NGDP target, how would we know what trend we’re trying to revert to? And how would this have played out at other periods in history (such as after the “Great Inflation” of the 1970s)?

To which Sumner responds:

Millman raises a very good question; when we introduce a new NGDPLT policy, where do we start the trend-line?  It seems to me that this decision unavoidably involves some degree of discretion.  My basic operating principle is to use a trend line that minimizes disequilibrium in the labor market.  Thus in 2009 I assumed that wages had been negotiated on the basis of the preceding 5% trend line, which had lasted for several decades.  I favored returning to that trend line, as that would bring the labor market back close to equilibrium.  As more time has gone by, I’ve now come to believe that it is no longer wise to return to the original trend line.  Too many wage contracts have been signed on the expectation that we will never return to the old trend line.  Similar discretion would have been appropriate in the early 1980s.

This may seem to conflict with my preference for a fully automatic system with no discretion.  I still favor a rigid rule-based policy, but only after the policy has been implemented.  It is impossible not to use discretion when implementing a new policy.  Choices always must be made.

I think on this point we are mostly, but not completely in agreement. The policy in the 1980s and 1990s wasn’t to return to a trend established some time in the 1970s. It was to change the trend. It took us about a decade to get to what looks like a fairly stable 5% NGDP trend. If Sumner agrees that this was appropriate in the 1980s, then why doesn’t he think the same is true now – that is to say, that we can’t simply bounce back to trend when we fall off, but need to slowly make our way back?

The answer lies in his response to another part of my post, where I suggest that his 5% target is really the combination of a stable 2% inflation rate and a 3% estimate of real growth potential, and that therefore implementing NGDP targeting still requires some knowledge of what the real growth potential of the economy is – otherwise you could wind up implicitly targeting higher inflation or, if the real economy grows faster than was thought possible, negative inflation.

Sumner replies:

There is no good economic argument for price stability.  Zero inflation is just another number.  On the other hand there is something very special about zero wage growth (due to money illusion) and zero interest rates (because of cash, nominal interest rates can’t fall below zero.)  It turns out that what most people regard as the “welfare cost (and benefit) of inflation” is actually better described as the welfare costs and benefits of faster NGDP growth.  The problem with high NGDP growth is that it raises nominal returns on capital, and this increases the tax rate on capital (which should be zero.)  The benefits from higher NGDP growth are that it moves you away from the zero wage boundary, and thus there is less labor market distortion caused by workers refusing to take nominal wage cuts, when market conditions would call for them (and when they would accept real wages cuts via inflation.)  And of course you are less likely to get stuck at the zero interest rate bound.  On the other hand zero inflation is almost a meaningless number.  It doesn’t mean product prices are not rising, it means the prices of computers, cars, TVs, etc, rise at a rate equal to the BLS estimate of quality improvement.  But why would that number matter in a welfare sense?  It doesn’t.

So NGDP growth matters, not inflation.  Thus I didn’t pick 5% because it was 3% plus 2%, rather I picked it for two reasons.

1.  It was the historical trend before 2008.  So it seemed like that was Fed policy.  I argued they should have stuck to their old policy, not changed radically in 2008.  Policy stability leads to macroeconomic stability.

2.  It seemed high enough to keep us away from those two black holes (zero wage boundary and zero nominal interest rates) but not so high as to put a heavy tax burden on capital.  It was a compromise.  I’d be fine with 4% or 6%, as long as it was stable, and level targeting.

Taxes are one problem with high inflation numbers. The other problem is that large numbers exhibit more variance, which translates into uncertainty, and economic actors react to uncertainty by trying to reduce risk, which over time reduces real growth. If we did not tax capital – or, alternatively, if we indexed capital gains taxes to NGDP – I don’t think Sumner would say an NGDP target of 30% was sensible. Or perhaps I’m wrong and he would – but I wouldn’t, because I would expect the variance around 30% to be higher than the variance around 2%.

But at this point, it seems to me, we’re debating whether, over the long run, an NGDP target would have to make some reference to the real growth rate of an economy. The NGDP target that makes sense for China or India is different from the NGDP target that makes sense for Japan – not just because of different population growth rates, but because the Chinese and Indian labor forces have the opportunity to become substantially more productive in a short time by playing industrial catch-up, which is not an option for the Japanese work force.

If we’re agreed on that, then our disagreement just relates to the degree to which the Fed needs to think about the long-term growth potential of the economy, and what is happening elsewhere on the policy front to affect that. Greenspan, in the mid-1990s, kept monetary policy looser than some thought wise on the grounds that the information technology revolution had made the economy fundamentally more productive, and therefore we could plausibly have a higher nominal growth rate – and a lower rate of unemployment – without higher inflation. I think Sumner thinks that the Fed shouldn’t be asking itself questions like that, and I think this is a point of disagreement between us. Precisely because I think the Fed does need to think about these questions, I think NGDP targeting wouldn’t be as “automatic” as Sumner does.

Two final points.

First, Fed policy arguably went off the rails twice in the 2007-2008 period. First, the Fed reacted to higher commodity prices by worrying about inflation, even as there were signs of economic weakness. Then, when we entered full-scale crisis mode, the Fed arguably didn’t open the spigots wide enough to prevent the precipitous fall in GDP that happened in late 2008 and early 2009. Sumner has a very solid answer to the first mistake. But I question the notion that with the right tools, the Fed would have known that GDP was about to contract at a 9% annual rate, and that it would have immediately opened the floodgates of nontraditional monetary policy to prevent such an occurrence. The Fed governors are human, and human beings don’t react to a nearly unprecedented crisis by sticking to whatever the computer program says. They would have been freaked out by the massive expansion in the Fed’s balance sheet in this alternative world, just as they were in our world, and that would have prevented the Fed from being as aggressive as Sumner would have liked. That’s what I think anyway.

Second, Sumner properly calls me out for talking about monetary policy as a “distraction.” He’s absolutely right: you have to do monetary policy; there’s no such thing as a “no policy policy.” This is a point I’ve made myself many times in the past in arguing with goldbugs. And Sumner himself is abundantly clear that getting monetary policy “right” isn’t a solution to all our economic problems. So on this point, I’m really only arguing with people who think that monetary policy is the main way we’ve gone wrong, and where I agree with Stiglitz is in his views of what’s happened to the real economy in the 2000s, and how that erosion was masked by the speculative housing bubble. Sumner and I may disagree on the relative importance of the nominal and the real in our current situation, but I think that’s the extent of our disagreement on this point, and I accept his chiding with good humor.