Arnold Kling  

Scott Sumner vs. John Taylor: An Imaginary Dialog

Card and Krueger on Minimum Wa... Power Corrupts...

The purpose of this exercise is to test my understanding of Sumner's view of monetary theory and policy. I will do this by putting words in the mouths of Sumner and Taylor. It will be a long post, so it goes below the fold.

As background, Sumner and Taylor would each claim to be the intellectual heir to Milton Friedman. Yet Sumner comes down in favor of strongly expansionary monetary policy today, while Taylor does not.

Let us start by asking each to explain why he is the true heir of Milton Friedman.

Taylor: I am Milton Friedman's true heir because I believe in the importance of rules rather than discretion in monetary policy. Discretionary policy is policy that is not predictable. The important thing about money is that it provides the nominal anchor for the economy. Workers and firms make long-term arrangements based on assumptions about how the general level of prices will behave. When their assumptions are correct, macroeconomic outcomes will be good. When their assumptions are incorrect, outcomes will be bad. They are more likely to be correct if the Fed follows predictable rules than if it undertakes discretion.

Sumner: I am Milton Friedman's true heir because I believe that one of the biggest tragedies in economic policy is the failure to diagnose bad macroeconomic performance as a monetary phenomenon. Part of the reason that the Great Depression was as bad as at was is that policies were undertaken that exacerbated problems. The National Industrial Recovery Act, which cartelized big business and labor, was a prime example. It took several decades for the diagnosis of the Great Depression as a monetary contraction to be articulated and accepted in the profession. Because the monetary contraction was not understood at the time, not only was short term policy bad, but we lost precious economic freedom. I do not want the same thing to happen today, and yet I see it happening.

Next, let us ask each to explain why we have had such high unemployment in the past two years.

Taylor: We have high unemployment because from 2003 through 2006 the Fed deviated from the Taylor rule by being overly expansionary. They kept the nominal interest rate too low for too long. This created a housing bubble and was followed by a crash. We are paying the price today for the excesses of the earlier period.

Sumner. The nominal interest rate is not the correct indicator of monetary policy. The correct indicator is growth in nominal GDP. Growth in nominal GDP was reasonably steady from 2003 through 2006, relative to its previous trend. Therefore, it is wrong to characterize monetary policy in that period as being overly expansionary. If you look at nominal GDP growth, the most striking data point is the spectacular decline that started in the fall of 2008 and that has never been offset. This shortfall in nominal GDP growth is the Great Recession, pure and simple.

Next, we note that Taylor does not favor expansionary moves today, such as quantitative easing. Sumner does favor such moves. We ask each to criticize the stance of the other.

Taylor. Sumner is wrong because he fails to have monetary policy adhere to a rule.

Sumner. Taylor is wrong because he uses his rule to justify the monetary contraction that took place in 2008. If the rule that you are following leads to a horrendous decline in nominal GDP growth, then you are following the wrong rule.

Taylor. If you change the rule when you think that nominal GDP is too low, then you are not following a rule. You are using discretion.

Sumner. My rule isto target nominal GDP.

Next, we note that Milton Friedman did not argue for monetary policy that targets nominal GDP. He favored steady monetary growth. This would create a stable, nominal anchor. Friedman was known for fearing that, due to "long and variable lags" in monetary policy, the attempt to hit a target for GDP would result in overshooting. In James Tobin's phrase, Friedman feared a Fed that, like an amateur shower tuner, alternately scalded and froze the economy in reaction to past data.

Taylor. That's right. What my rule does is that it makes sure that the Fed does not over-react in one direction or another.

Sumner. The modern twist in macro is to focus on expectations. We think of the private sector as forming expectations for nominal variables. In addition, the Fed can use indicators of market expectations for prices and output growth. Because the private sector's expectations matter, the lags in monetary policy are not necessarily long. And because we can read market expectations for prices and output, we can avoid overshooting, because we are not reacting to past performance but instead we are reacting to expectations about the future.

Next, we ask whether the sharp drop in nominal GDP growth in 2008-2009 was due to monetary contraction or a velocity shock.

Taylor. It was a velocity shock. You don't try to offset velocity shocks, because that is what leads to over-reaction.

Sumner I don't think that distinction matters. It was a nominal shock, and that is all we care about. If the Fed is targeting expected nominal GDP growth, then it will naturally act to offset velocity shocks.

Next, we ask how they would address the argument that the "zero bound" made monetary expansion ineffective in the Great Recession.

Taylor. The zero bound did not matter, because we did not need more monetary expansion. The Taylor rule, as I formulate it, did not call for negative nominal interest rates.

Sumner. The zero bound is a red herring. There is nothing about interest rates that impedes the Fed from hitting a target for nominal GDP. To argue otherwise is to suggest that we have reached the point in a fiat money economy where the government can no longer debase its currency. That is clearly false. It is absurd.

This concludes our imaginary dialog. At this point, Sumner would appear to be the winner, in my imagination. Next, we need to have him take on those of us who would argue that monetary debasement would not, under present circumstances, do much to relieve unemployment.

Comments and Sharing

COMMENTS (17 to date)
effem writes:

I don't understand Sumner's link between NGDP and employment. In theory higher NGDP = higher corporate profits = higher employment. However, we now have near-record corporate profits with no major change in employment. There must be something else at work.

It cannot be possible that real wages are too high when corporate profits are at a record. Therefore why would we expect lowering real wages to increase employment?

Further, I can understand lowering real wages when both profits and employment are low - some "burden sharing" is normal. However, why should labor shoulder all of the burden. Corporates are sharing none of the burden - they are no worse off now then they were in 2007. Asking labor to take yet another downgrade via inflation seems wrong.

Until economists can demonstrate why firms hire or don't hire, all the theories seem rather meaningless.

Ed Hanson writes:

No, Sumner is not the winner.

Simply put, it was never necessary to expand monetary stimulus after TARP had more or less resolved the banking crisis. The Fed should and did mop up a lot of the dead weight of overhanging dpet that could not be priced rapidly, this gave the economy time to price many.

Beyond that no monetary expansion has helped but instead has hindered. The type of nominal GDP that Sumner is advocating is a phony indicator. When it comes down to it, only real GDP matters and the economy has not been able to adjust because of the false signals being generated. It is past time to end central planing. It is past time to give incentives to the millions of ideas to the private sector to undo the harm.

ed writes:

Effem, what if Garret Jones is right that most workers produce organizational capital? Getting rid of these workers, effectively reducing investment in organizational capital, both lowers employment and raises (in the short term) corporate accounting profits.

mktlogic writes:

Sumner's view seems to be that the damage of the recession is optional. It isn't.

For the better part of a decade labor, capital and loanable funds were consumed in producing houses that turned out to be worth less than the economic costs of producing them. No monetary rule or NGDP target will ever change the fact that the resources consumed in the housing sector have already been consumed. A NGDP target might well have prolonged the illusion that all of the homebuilding was actually an efficient use of resources and made things worse.

Employment won't return until people identify new projects to invest in, but that's not something that can be helped by boosting NGDP either. It just makes it harder for people to identify profitable projects when the monetary policy further distorts relative prices. A Taylor rule approach would be less bad, since it would call for less debasement but an even better approach would be no debasement.

B writes:

Sumner: The zero bound is a red herring. There is nothing about interest rates that impedes the Fed from hitting a target for nominal GDP. To argue otherwise is to suggest that we have reached the point in a fiat money economy where the government can no longer debase its currency. That is clearly false. It is absurd.

Hasn't the fed shown their willingness to screw with interest rates other than the short term rate when they started dealing in treasuries?

Doesn't that move the goal posts of what a zero bound is? Now there's not just one zero bound. They have a whole basket of nominal rates they can affect.

Various writes:

Hmmm, that is interesting. I don't see either of these arguments articulating what I would imagine to be Friedman's views. I would guess that Friedman would focus on the quantity of money foremost, and consider velocity to be less important. But he was a pragmatist, so I imagine he would adjust for velocity if he believed that our approaching zero nominal interest rates is causing a semi-perminent shift in velocity.

I think, and I am only guessing here, I think that Friedman's beef would be that the current mechanisms we have to measure "money" have become obsolete in the face of shadow banks, off balance sheet financing of financial institutions, etc. He may also say that Treasury securities are a form of money also, although I'm not sure about that. In summary, I think Friedmant would say that "money matters" still, and that one must focus on the supply of money above all else. He would describe the problem as a technical one, i.e., how does one accurately measure the supply of money in an increasingly complex financial world

What I take away from Friedman is that he considers the supply of money to be the "forest" and all other considerations to be the "trees". His overarching critique of macro was that many economists were failing to see the forest through the trees. Both of your players are describing the trees and not the forest.

Richard A. writes:

One can view nominal GDP as the velocity corrected money supply.

effem writes:

ed, i'm willing to buy that story on employment but i still don't see how higher NGDP = more employment. Corporates have a singular focus on profits and they have now learned that shrinking, rather than growing, the workforce (at least in the US) may be the best path. Employment seems to have become independent of profit levels.

So I'm told the answer is to produce inflation to lower real wages and boost corporate profits. So in response to corporations deciding fewer workers is the optimal business strategy we are going to "reward" them with ever lower real wages and higher profits? Seems to me that's how you impoverish the middle/lower classes and keep the wealthy on their upward trajectory. Not healthy.

Charles R. Williams writes:

Friedman admitted that the quantity theory of money was discredited. He was in favor of rules rather than discretion in the exercise of monetary policy. Trouble is no one has a plausible rule.

Taylor's mistake is to postulate that an empirical relationship during a stable prosperous period involving endogenous variables implies that one of these variables can be manipulated according to the empirical rule to generate a stable economy. It is not clear that the fed can even manipulate the federal funds rate outside of a narrow band.

Sumner needs to demonstrate that the fed can manipulate NGDP onto a stable growth path and that this will put the real economy on a stable growth path. Nothing short of trying his proposals will demonstrate these things. He seems to think that if the economy collapses it must be because the fed is too tight. But the explanation of our current recession and its persistence lies in the collapse of consumer net worth since 2006. People needed to make a drastic reduction in consumer spending to secure their retirements, increasing savings at the same time investment opportunities were evaporating. Sumner would have us believe that if something causes real output to shrink 3% per year this can be cured by targeting an 8% inflation rate. Sumner thinks the gradual pickup in the pace of economic growth since last summer is driven by QE2 and its impact on expectations. Well, I don't think so and Sumner is unpersuasive. Why did growth slow last summer? The simple answer is fear of a financial crisis originating in Europe. Why did growth speed up a little bit in the fall? Simple answer: the European crisis didn't materialize and the Bush tax cuts were extended.

Patrick R. Sullivan writes:

It's hard to see how Taylor can claim to be Friedman's natural heir when Friedman was famously skeptical of using interest rates as a guide to monetary policy.

John Hall writes:

To be precise, Sumner's monetary policy indicator is expected nominal GDP growth, not nominal GDP growth.

Some versions of policy rate rules (not sure if this is true for the one Taylor uses) were calling for negative interest rates.

I think one problem with Sumner's line of thought is his belief in nominal GDP level targeting rather than growth targeting. If there is a shock to productivity that changes long-run trend GDP growth, then the central bank might get fooled. I don't think this is the case for something like price level targeting.

I also disagree with Sumner's assessment of the housing market and monetary policy in the early 2000s. He doesn't view policy as easy because he is wedded to the level targeting approach, which would say that monetary policy was appropriate in the early 2000s (based on say 1-year SPF consensus nominal GDP forecasts). However, using those same forecasts, you would get a different story if you used growth rates (ie. that monetary policy WAS too easy in the period). The fact that his approach fails to understand one of the most significant bubbles in the past 100 years is definitely a negative, even if he gets the collapse of the bubble correct.

I think Sumner needs to do a whole paper evaluating level vs. growth nominal GDP targeting, including simulations from the same models that economists use to justify price level targeting rather than inflation targeting.

effem writes:

Richard A, I'm sure there remains some link between NGDP and employment as your chart suggest (although %change exaggerates the employment gains). The way I look at it is when you stimulate you get two things: employment and inflation. In this cycle it seems we are getting more inflation per unit of employment than in other cycles - something has changed.

Perhaps chart the misery index against NGDP. I dont believe the misery index has improved much from the bottom despite the large NGDP gains. And my sense from watching unemployment claims and inflation expectations is that the misery index is set to get worse, not better.

James A Donald writes:

Here is an alternate account of recent economic events:

We have had an explosive growth in regulation, as measured by pages of legislation, number of regulators, by any reasonable way of measuring it. It is growing hyperexponentially - the doubling time is getting shorter and shorter, and any reasonable extrapolation leads to the unreasonable prediction of infinite amounts of regulation generated in zero time some time in the current decade or early in the next decade. It is a regulator singularity.

This has produced real decline in GDP, which is concealed by creative statistics that understate inflation, artificially producing a housing bubble thus overstating production, and a multitude of other means.

Flat out simply stimulating the economy will eventually lead to hyperinflationary shocks. From time to time, there simply will not be any stuff in the shops, and new stuff, when it appears, will appear at much higher prices, leading people to conclude that stuff will get you through times of no money better than money will get you through times of no stuff.

Any realistic, non hyperinflationary adjustment has to be a third worldization of the economy, in which the majority of people are hungry, get electricity only irregularly and infrequently, do not have access to safe water, and die when there are unusual weather extremes - which weather cataclysms will be blamed on global warming, though there will be nothing unusual about the weather extremes, merely the deaths will be unusual - Katrina foreshadowed this. We have had worse weather than Katrina many times before, but previously we had the capability to keep dikes in repair, evacuate threatened areas, and so forth.

Anon writes:

Sumner ignores the impact of relative price distortions underlying constant NGDP growth in the early 2000's. As long as NGDP hits his target, he is indifferent as to how you get there. But how you get there matters: the Fed produced a rate of NGDP growth that required "payback" when house prices maxed out relative to incomes. In other words, the Fed produced an unsustainable NGDP trend, rather than an organic, sustainable one.

One way to describe it: the Fed caused middle income consumers to bring forward consumption; the result is that higher-income consumers made more income (due to financialization), so they also increased consumption. Once the middle income borrowers found they could not service their debt, the process worked (is working) in reverse.

Jon writes:

Anon hit just the right note of concern: "As long as NGDP hits his target, he is indifferent as to how you get there."

Scott tells a great story that I mostly believe--I was banging the drum on policy being tight in 2008--be he goes flaccid and distant on certain questions.

1) How exactly does NGDP below trend cause unemployment per se?

2) What determines the mix of inflation and real growth? Does that have implications for the stability of inflation expectations?

3) If inflation uncertainty increases, there are risk premiums in the nominal rates which drag on investment. Is this necessarily better?

4) Is he talking about rate targeting or level targeting? He mentions both. Trouble is both have issues, level targeting forces the Fed to contract hard (cause a crisis) follow past excursions above trend. Rate targeting doesn't motivate the Fed to restore growth to trend.

Philo writes:

Why do you make Taylor criticize Sumner for not advocating monetary policy according to a rule (as opposed to "discretion")? You immediately allow Sumner the line: "My rule is to target nominal GDP." More accurately, Sumner's rule calls for steady 5% per annum growth in expected NGDP; this is reminiscent of Friedman's rule of steady 3% per annum growth in the money supply. Obviously, it's a *rule*; it doesn't even look like "discretion."

Comments for this entry have been closed
Return to top