Consulting by RPM || Free Advice Blog

Scott Sumner Assumes a Can Opener

It’s possible that I have blogged about this before, but in any event one of the things that bothers me about Scott Sumner’s proposal to have the Fed target nominal GDP (NGDP) is that he often argues almost from definitions, rather than the economics. This is analogous to a Keynesian justifying stimulus spending by pointing out Y = C + I + G + Nx.

Now before we get going, a note on definitions: Nominal GDP is the actual dollar amount of total spending. Real GDP, in contrast, is the amount of spending, after the “price level” has been normalized to some reference year.

So for example, suppose this year total spending is $10 trillion, while the price level is 100. Then next year, total spending is $11 trillion, while the price level rises to 107. NGDP went up 10%, but RGDP only went up about 2.8%. So the actual increase in real output–the increase in the number of cars, iPods, apples, etc. produced in the second year–was only 2.8%. The rest of the measured increase in total spending was simply due to the fact that price tags increased on average by 7%.

OK back to Sumner and his recommendation that the Fed set a target of (say) NGDP growing by 5% per year. I am arguing that Sumner argues from definitions sometimes, and ignores the economics of what he is saying. The most blatant example I’ve seen came in November when Scott (yes, I think we are on a first-name basis) was taking Robert Hall to task.

In one component of the critique, Scott first quoted Hall who had the audacity to write:

The worst financial crisis in the history of the United States and many other countries started in 1929. The Great Depression followed. The second-worst struck in the fall of 2008 and the Great Recession followed. Commentators have dwelt endlessly on the causes of these and other deep financial collapses. Less conspicuous has been the macroeconomists’ concern about why output and employment collapse after a financial crisis and remain at low levels for several or many years after the crisis.

Scott didn’t like this. Instead, he said Hall should have written:

The worst depression in the history of the United States and many other countries started in 1929. The Great Banking Panics followed. The worst recession since the 1930s struck in December 2007, and dramatically worsened in July 2008. The Great Financial crisis of the fall of 2008 followed. Commentators have dwelt endlessly on the causes of these and other deep financial collapses. Less conspicuous has been the macroeconomists’ concern about why output and employment collapse before a financial crisis and remain at low levels for several or many years after the crisis.

Now here’s the important part: Look at the reasoning behind Scott’s suggested alternative paragraph:

We know that severe declines in NGDP are likely to cause severe declines in RGDP. The reasons are murky, although I believe sticky wages are an important transmission mechanism. There is more controversy about what causes NGDP to plunge. But given the recession began in December 2007, and given the severe plunge in NGDP occurred between June and December 2008, it seems a bit hard to believe that the financial crisis of the fall of 2008 was the cause.

First, look at the part I put in bold. Isn’t that a bit like saying, “We know that severe declines in numerators are likely to cause severe declines in quotients. The reasons are murky, although I believe sticky denominators are an important transmission mechanism.” ?

Second, look at how Scott is re-writing history to vindicate his theory. In his suggested paragraph, Scott wanted Hall to write that the worst recession since the 1930s “dramatically worsened in July 2008.” If that were true, then it would bolster Scott’s theory, since NGDP started plunging in June 2008.

But actually, I don’t remember people flipping out about the recession in July 2008. In fact, a whole month later, Bryan Caplan–who Scott admits is smarter than him, or at least sounds smarter at a lunch table–was talking about how great central banks were.

No, it was not obvious that the recession had taken a turn for the worse, until September 2008, i.e. exactly when Hall said it happened. I doubt Caplan would have blogged the above post, in late September.

Skip to Scott’s most recent post. Scott is arguing with Arnold Kling, who had said that Scott’s idea of targeting NGDP might lead to “high and variable inflation.” Scott disagreed, saying:

I believe…that inflation would not become high and volatile. In order for inflation to be high (on average), RGDP growth would have to average much less than 3%. Let me repeat; much less, not slightly less. A trend rate of 1% or 2% RGDP growth will not get you high inflation on average, unless you think inflation was still high after Volcker brought it down to low levels.

So there we have it again, the trick of using definitions to defend his policy. Let me paraphrase what Scott is saying, “If my plan doesn’t screw up economic growth, then my plan will be good for the economy.”

Well, yes, I grant you that Scott. But I go you one better: The Murphy plan calls for the Fed to buy and sell RGDP futures contracts, to hit a target of 5% growth per year. Don’t get me wrong, I’m not saying we’ll always get real growth of 5% per year, but that’s what we’ll hit on average. The years when we’re below target, the Fed will do what it has to do with the monetary base, in order to grow extra fast the next year and catch up with it.

Finally, let me simply say that I disagree with Scott when he writes:

Much more importantly, I don’t think high or variable inflation is a problem as long as NGDP growth is on target. All of the problems that are widely believed by economists to flow from high and variable inflation; actually result from high and variable NGDP growth:

1. Excessive taxation of capital in a non-indexed tax system results from high nominal rates of return, associated with high NGDP growth.

2. Unfair borrower/lender redistributions actually result from volatile NGDP, not volatile inflation.

3. Distortions to the labor market (when nominal wages are sticky) are caused by NGDP shocks.

4. Even the “shoe leather” cost of inflation may be better described by NGDP growth, assuming real interest rates and real GDP growth rates are strongly correlated.

I’m not going to bother going into each one. I claim, without proof, that Scott is simply mistaken. For example, if I have a ten-year bond paying me $25,000 per year that I’m supposed to live on, I don’t really care what happens to NGDP. I get killed when P goes up faster than I expected. Yes, it’s true that in most cases, that’s going to happen when NGDP goes up more than I expected, but that’s because of the definitional trick. Notice that I had to DEFINE NGDP in the beginning of this post, since I couldn’t be sure most readers would even know what it was. So Scott really should stop acting is if everybody “really” cares about NGDP, not P.

Last point: This post should not be construed as me saying Scott’s idea is dumb. It is actually elegant, from a theoretical perspective, and for all I know it might be better than some other rule-of-thumb we could give the Fed, like a Taylor Rule.

All I am saying is that Scott often retreats to an argument from definitions, when he seems to think he is appealing to economic analysis.