Scott Sumner  

My views on monetary economics

PRINT
What Is Resentment?... Will antitrust become a geopol...

In this post, I'll use "(e)" to denote a (market) expected value.

1. NGDP(e) is the single most important variable in macro; it should be the centerpiece of modern macro. Unfortunately, it often doesn't even appear in models.

2. NGDP(e) should be used as an indicator of the stance of monetary policy. Instead the highly misleading fed funds rate is often used. Some economists cite the difference between the actual and natural rate of interest, but the natural rate is unobservable, and hence not useful.

3. Interest rates and the price level should be dropped from macro models for "never reason from a price change" reasons. Changes in interest rates and changes in inflation don't tell us anything useful about the economy, at least nothing that cannot be better demonstrated with alternative variables. Regarding the price level, Keynes agreed with me, ustilizing a nice analogy in the General Theory:

But the proper place for such things as net real output and the general level of prices lies within the field of historical and statistical description, and their purpose should be to satisfy historical or social curiosity, a purpose for which perfect precision -- such as our causal analysis requires, whether or not our knowledge of the actual values of the relevant quantities is complete or exact -- is neither usual nor necessary. To say that net output to-day is greater, but the price-level lower, than ten years ago or one year ago, is a proposition of a similar character to the statement that Queen Victoria was a better queen but not a happier woman than Queen Elizabeth -- a proposition not without meaning and not without interest, but unsuitable as material for the differential calculus. Our precision will be a mock precision if we try to use such partly vague and non-quantitative concepts as the basis of a quantitative analysis.
4. There are welfare effects of changes in trend NGDP growth and also the volatility of NGDP. There are welfare effects from instability in hours worked. The "divine coincidence" that some economists believe applies to inflation and output, actually better applies to NGDP growth and hours worked. Low and stable NGDP growth minimizes the welfare costs of "inflation", and also leads to approximately optimal hours worked. Not exactly, but more closely than for inflation and output.

5. NGDP fluctuations are monetary shocks, they are not financial or real shocks. But they do have financial and real effects.

6. The lesson of the Great Recession is that macroeconomists need to spend less attention on the financial system. Bernanke attempted to integrate the financial system into macro in his famous 1983 AER article, which argued that banking problems directly depressed output during the early 1930s. That led Bernanke and others to wrongly conclude that fixing the banking problem was the way to stabilize the economy in late 2008. In fact, the Fed needed to use monetary policy to try to prevent a steep fall in NGDP. It did not even attempt to do so. Indeed the (contractionary) monetary policy of interest on reserves (adopted in October 2008) was aimed at depressing the economy, while the Fed worked to rescue the banking system.

7. The New Keynesian/NeoFisherian debate is focused on the wrong set of issues. It's not a question of who's right---they are both wrong to focus on interest rates---it's about how to determine which shocks have NK effects and which shocks have NF effects.

a. Sometimes monetary shocks have NK effects. This occurs when an easy money policy leads to lower interest rates, a weaker currency, and an increase in the expected rate of appreciation in the domestic currency (due to lower interest rates and interest parity). An example was the QE announcement of March 18, 2009.

b. Sometimes monetary shocks have NF effects. This occurs when a tight money policy leads to lower interest rates, a stronger currency, and an increase in the expected rate of appreciation in the domestic currency (due to lower interest rates and interest parity). An example was the Swiss currency appreciation of January 2015.

8. Most business cycles in large complex economies occur when NGDP moves unexpectedly in the presence of sticky nominal hourly wages. These monetary shocks cause a fluctuation in the ratio of NGDP to hourly wages, and this leads to a similar fluctuation in total hours worked---the essence of a business cycle:

Screen Shot 2017-06-26 at 8.14.47 PM.png
9. There may be a zero lower bound of interest rates (or perhaps a slightly negative lower bound), but this is not a problem. There may be a problem of a zero lower bound of eligible assets not yet purchased by the central bank. But if so, it is a self-imposed problem. Make more assets eligible, or raise the trend rate of NGDP growth. Socialism or inflation.

PS. Ramesh Ponnuru and David Beckworth have an excellent piece in National Review, which discusses both the logic of a higher inflation target, and why a NGDP level target is the superior option. Highly recommended.


Comments and Sharing






COMMENTS (19 to date)
John Brennan writes:

Scott--how is what Bernanke did in response to the "financial crisis" different from what Greenspan did in response to the "1987 Crash/SNL Crisis." And does the variance in responses support your claims? Or were the situations not comparable?

Jake writes:

Prof Sumner, two questions if I may. You said:

That led Bernanke and others to wrongly conclude that fixing the banking problem was the way to stabilize the economy in late 2008. In fact, the Fed needed to use monetary policy to try to prevent a steep fall in NGDP.

When you say "try to prevent" is there any reason to believe that they could not have succeeded? At least over a reasonable period of time. I recall you stating before that no country has ever tried to inflate, and failed... I'm guessing this case would be no different if the Fed really committed to it.

It did not even attempt to do so. Indeed the (contractionary) monetary policy of interest on reserves (adopted in October 2008) was aimed at depressing the economy, while the Fed worked to rescue the banking system.

Are you saying that they intentionally caused a deeper recession? If so I can't see what benefit they saw in that, or what they hoped to achieve with the new IOR policy.

Jake writes:

Also, the Ponnuru/Beckworth article is excellent but there might be one more benefit of switching to an NGDP(e) target.

As they mention, the public perceives inflation as bad. If the Fed says "we want to raise inflation" then a lot of folks will get upset.

But NGDP(e) is something most people have no concept of. I bet an announcement of "we want to raise NGDP" would generate much less criticism, even though it is essentially the same thing.

So in addition to being a technically superior policy instrument, NGDP(e) targeting might actually help depoliticise the Fed and make it easier for them to do their jobs.

Michael Sandifer writes:

John Brennan,

It's interesting you bring that up, because my best guess is that the markets were predicting a 1.4% recession during the crash in late '87. This is evident both in falling Treasury rates and in the 23% drop in the S&P 500 index. In the case of the stock market drop, the P/E was around 20, so one should expect a drop of that magnitude with a 1.4% decline in NGDP.

I think Greenspan sent some signals that markets either misinterpreted(he was a new Fed Chairman), or that he quickly changed his mind about the stance of monetary policy and saved the day.

If this isn't what an averted recession looks like, what is? From a market monetarist perspective, which is one that respects at least a semi-strong EMH interpretation, isn't this the most parsimonius explanation?

bill writes:

Another difference between 1987 and 2008 is this. Greenspan added liquidity for a (brief) period of time and then withdrew it once it was no longer needed. Bernanke sterilized his moves. So he was withdrawing credit from everyone to give it to entities that were failing.

Scott Sumner writes:

John, Greenspan kept NGDP growing at a fairly steady rate, which is very different from what happened in 2008.

Jake, Central banks have virtually unlimited ability to inflate. If they try hard enough they will always succeed.

You asked:

Are you saying that they intentionally caused a deeper recession?"

Yes, in a sense. The Fed was worried about high inflation in late 2008 (even though deflation was the real problem.) This fear led them to adopt contractionary policies like IOR, which made the recession deeper.

Michael, I don't really have an explanation for the stock crash of late 1987.

Bill, I agree.

Michael Sandifer writes:

Scott,

The crash of '87 was not just a stock crash, of course. It was a world-wide phenomenon that affected all markets. I'm curious as to why you don't think the market was incorrectly predicting recession.

Thaomas writes:

Several questions/comments:

On 1. I’m inclined to agree, but I think we need some actually need some models that demonstrate under what assumptions that is the case.

On 2. Why do we need to discuss “stance?” Why not just rules for adjustment of policy instruments? Or is “stance” just a judgment on the configuration of policy instruments at a particular time?

On 3. As with 1 we need to see some models in which this is demonstrated. Since, at least right now, people do seem to care about inflation/price level, having models in with these variable do not appear would seen unnecessarily revolutionary. Moreover, I suspect that uncertainty about inflation/future price levels is an important variable in determining investment and hence real GDP and probably should NOT be dropped from macro models. And without inflation/price levels in the model, how could one determine the optimal target for inflation/price level or NGDP?

On 4. Quite possibly. What about a model to show the superiority?

On 5. Is this just a definition? Do you mean that there cannot be shocks that arise from the financial system – a change in the demand for safe assets, for example -- or sudden changes in “animal spirits?” If the latter, again, can we see that in a model?

On 6. I do not understand the contradiction you see in “fixing the financial/banking system and running appropriately simulative monetary policy in 2008. I agree that there was not enough attention to monetary stimulation but I do not see the bailouts/stress test/ etc. as being constraints on the former. Rather, things probably go the other way. The reluctance (still unexplained in political economy terms) would have made some of the “fixing” unnecessary. Do you just mean the vague fears of “financial instability” that got in the way of “whatever it takes” monetary policy?

On 7. Possibly true, but, again, you can’t beat a model with no model.

On 8. Why sticky hourly wages instead of a broader number of prices? The fact that machines, real estate, services become unemployed during recessions implies that many prices fall out of equilibrium.

On 9. I agree. What is the political economy behind the resistance to monetary policy manipulation of interest rates of longer term, foreign currency assets? For that matter what is the political economy behind treating the inflation target as a ceiling, failing to use the price level as the target?



Scott Sumner writes:

Michael, It might have been, but I'd like to see more evidence of that. What happened on October 19, 1987 that dramatically increased the probability of a deep recession? (A mild recession should not have a dramatic impact on stock prices.)

Thaomas, There are quite a few models supporting the claims I am making. For instance, my claims about NeoFisherianism vs. New Keynesianism are supported by the interest parity model and the Dornbusch overshooting model.

I agree that we don't need to discuss "stance", that rules for instruments are enough. But the discussion of stance by other economists tends to lead them astray.

There are also a number of recent models showing that NGDP targeting is superior to inflation targeting.

James writes:

"There are welfare effects of changes in trend NGDP growth..."

Do you think the relationship between welfare and NGDP would remain the same if central banks began targeting NGDP directly?

What about if private citizens who believed in NGDP targeting formed two nonprofits and each nonprofit paid the other equal amounts for "consulting" services? That is, if NGDP is $X below where they think it should be, each nonprofit could write the other a check for consulting services for $X/2 and bring NGDP up to any desired level. Would the relationship between NGDP and welfare still hold?

Michael Sandifer writes:

Scott,

What might have happened to make markets think a recession was coming? Here's a quote from a federal reserve paper about the '87 crash produced in '06:

“The macroeconomic outlook during the months leading up to the crash had become somewhat less certain. Interest rates were rising globally. A growing U.S. trade deficit and decline in the value of the dollar were leading to concerns about inflation and the need for higher interest rates in the U.S. as well.”

Also, here's a quote from finance professor Charles Geisst who wrote the book Wall St: A History:

“The market had experienced very high interest rates between 1980 and 84, and was spooked by talk of a return to those rates,...”

I found both these references in this Time magazine article:

http://business.time.com/2012/10/22/25-years-later-in-the-crash-of-1987-the-seeds-of-the-great-recession/

Here's a really nice quote from the article:

"Alan Greenspan assumed the role of Federal Reserve Chairman in August 1987, just a few months before the crash. The dollar had been declining for several years due to an international agreement in 1985 to devalue the dollar in order to help American exporters. Fearing that the dollar had fallen too far, Greenspan took measures to raise interest rates to defend the dollar. According to Geisst, this action spooked the markets, which had gone through a painful period of high rates in the early 80s. “Greenspan learned his lesson,” Geisst says. “He didn’t want to get blamed for something like that again.”

Now, as for what you said about a mild recession not having a dramatic effect on stock prices, I think history disagrees with you.

There was a very mild recession that triggered the tech crash in 2001. NGDP fell about 2.8% from trend, or twice as much as the 1.4% the 1987 market crash implied was expected. And the 2001 crash was about twice as severe for the S&P 500, as it fell about 46%, compared to about 23% for the '87 crash.

Market P/E ratios seem to help to predict well the magnitude of stock market reactions to falling NGDP. For example, see my blog post where I show this relationship holding up with considerable precision in 4 out of 5 recessions mentioned:

https://wordpress.com/post/thehonestbrokernet.wordpress.com/207

Higher P/E ratios obviously reflect higher future NGDP expectations and their effects on earnings, hence the higher the ratio before a recession, the steeper the stock market crash.

Here's a back of the envelope calculation for the magnitude of the '87 crash:

The S&P 500 P/E was roughly 21 or so.

The S&P 500 fell about 23%

Treasury Yields fell about 1.4%

1 - [1/(1+(21)(.014))] = ~22.7%

Todd Kreider writes:

Wasn't greatly expanded program trading a major cause for Black Monday? Not just to accelerate sells on the way down but also to accelerate them higher prior to October 19th as the DJIA had increased 45% in the ten months prior to the crash.

Scott Sumner writes:

James, That's an interesting question. Of course if this strategy were feasible, it could be done with or without NGDP targeting. Thus someone could do it in order to win the Hypermind prediction market.

In addition, someone could do this to distort inflation data. After all, this would not affect real GDP.

I think the big problem here is that GDP is also equal to national income. So any attempt to significantly boost GDP, if successful, would also boost national income. The group doing this would then earn a huge income. Unless I'm mistaken, to preserve their non-profit status they'd have to provide an enormous level of charitable services. That's costly.

Michael, I see many problems with your theory. First of all, if you were correct then this sort of thing should happen all the time. After all, there are hundreds of times in US history when growth prospects were just as uncertain as on October 19, 1987. For instance, in the 1930s things were far more volatile and uncertain.

It's important to remember that the 1987 crash was far, far bigger than any other stock crash in American history. There ought to have been some really dramatic news on October 19, 1987, but there wasn't. That's a puzzle.

That doesn't mean there was not some concern about a possible recession, just that it wasn't a big enough factor to drive stocks sharply lower. Recessions might depress earnings for a couple years, but I can't imagine they'd have a big impact on long run earnings. We have never gone more than 10 years without a recession. The stock market knew that even if a recession did not occur in 1988, it would occur in the near future. The previous recession was in 1982.

It's also not clear to me that the 2001 recession triggered the tech crash. The crash began well before the 2001 recession, and was partly due to industry specific problems.

Todd, "Program trading" doesn't actually explain anything. Humans create programs. If there were problems with computer trading, it's 100% human error.

In other words, there's really no difference between saying "Human's foolishly decided to sell stocks at Time=X", and "Humans foolishly instructed computers to sell stocks at Time=X"..

Scott Sumner writes:

Michael, Just to clarify, i accept your basic premise, that stock movements do predict the business cycle, to some extent. I used that hypothesis in my book on the Great Depression. The decline of 2008 was another example. I just think the October 19, 1987 decline was too large to be explained on that basis.

Matthew Waters writes:

"Make more assets eligible, or raise the trend rate of NGDP growth. Socialism or inflation."

This still just seems inherently problematic to me. Hard-money types will point to the fact that the Fed is buying non-government backed assets and that they will have influence over the private actors. They will not be wrong.

The NGDP future proposal meanwhile has an issue with the 10% margin in the proposal to not have counter-party risk. It's hard to see a true solution other than negative rates and restrictions on paper currency, as difficult as that is politically.

Michael Sandifer writes:

Scott,

You replied:

"Michael, I see many problems with your theory. First of all, if you were correct then this sort of thing should happen all the time. After all, there are hundreds of times in US history when growth prospects were just as uncertain as on October 19, 1987. For instance, in the 1930s things were far more volatile and uncertain."

I don't think uncertainty has anything to do with the '87 crash, but rather firm expectations of interest rate hikes.

You also replied:

"It's important to remember that the 1987 crash was far, far bigger than any other stock crash in American history. There ought to have been some really dramatic news on October 19, 1987, but there wasn't. That's a puzzle."

I don't think that's quite the way to frame it, for the following reasons. First, since 1987, US Stock exchanges have had "circuit breakers", which currently halt trading after S&P 500 declines of 7%, 13%, and 20%, respectively. The 7% circuit breaker was triggered 4 times in '08, and recall also that naked shorts were banned at times during that crisis.

https://en.wikipedia.org/wiki/List_of_largest_daily_changes_in_the_S%26P_500_Index

Also, the percentage drop in the S&P 500 during Black Monday and Tuesday of 1929 were of a similar magnitude to Black Monday 1987, when combined.

More importantly though, S&P 500 P/E's of around 20 have been relatively uncommon, historically.

http://www.multpl.com/

And, I don't think it's a coincidence that we've seen more stock market volatility since the Great Moderation began, along with somewhat higher P/E ratios, though peaks during the last two recessions were exaggerated during crashes, as recent earnings plunged.

So, I'll sum up by saying there's a reason Samuelson's joke about the stock market predicting "nine of the last five recessions" is often repeated. 1987 just happens to be an unusually large example, due to a higher than historically common P/E ratio, a new Fed Chairman making a mistake, and the resulting 1.4% shock to NGDP expectations being temporally acute. I don't see a huge difference between the '87 crash and the '29 crash, and you've stated before that the '29 crash certainly didn't cause nor foresee the depths of the Depression, unless I'm mistaken.


Michael Sandifer writes:

I meant to use the word "foretell", rather than "foresee" in that last line. And of course the '29 crash didn't foretell the Depression, since the S&P 500 index eventually fell a total of 83% at the depths of the Depression.

Todd Kreider writes:

Scott Wrote:

Todd, "Program trading" doesn't actually explain anything. Humans create programs. If there were problems with computer trading, it's 100% human error.

In other words, there's really no difference between saying "Human's foolishly decided to sell stocks at Time=X", and "Humans foolishly instructed computers to sell stocks at Time=X"..


There is a difference between the two since the latter takes the human out of the picture long enough for there to be a rapid avalanche effect.

“We haven’t agreed whether or not program trading was the primary culprit, but there’s been consensus it played a role,” says Paolo Pasquariello, assistant professor of finance at the University of Michigan’s Stephen M. Ross School of Business.

Timothy writes:

Scott, your point 6. is possibly the best 4-sentence statement about the crisis which I have seen.
"6. The lesson of the Great Recession is that macroeconomists need to spend less attention on the financial system. Bernanke attempted to integrate the financial system into macro in his famous 1983 AER article, which argued that banking problems directly depressed output during the early 1930s. That led Bernanke and others to wrongly conclude that fixing the banking problem was the way to stabilize the economy in late 2008. In fact, the Fed needed to use monetary policy to try to prevent a steep fall in NGDP."

Comments for this entry have been closed
Return to top