In my new book entitled The Money Illusion, I argue that a tight money policy by the Fed in 2008-09 largely caused the Great Recession. I’d guess that 99% of economists don’t agree with me on that point. That makes me a contrarian.
But if I am a contrarian, it’s of a type that is quite common throughout history. Consider:
1. In the 1930s and 1940s, almost all economists believed that the Great Depression was not caused by a tight money policy at the Fed. In the 1960s, Milton Friedman and Anna Schwartz convinced many economists that an excessively tight money policy was largely to blame for the Depression. By 2002, even Fed officials like Ben Bernanke acknowledged the Fed’s guilt:
Let me end my talk by abusing slightly my status as an official representative of the Federal Reserve. I would like to say to Milton and Anna: Regarding the Great Depression. You’re right, we did it. We’re very sorry. But thanks to you, we won’t do it again.
2. In the 1960s and 1970s, most economists did not blame the Fed for the Great Inflation. A few decades later, most economists thought the Fed was to blame. (Ben Bernanke among them.)
3. In the 2010s, most economists did not blame the Fed for the Great Recession. Market monetarists did.
What do the first two cases have in common? In both cases, an economist focusing on interest rates would be unlikely to blame the Fed. Rates were very low in the 1930s, and hence money did not look tight. Rates rose sharply in the 1970s, and hence monetary policy did not look expansionary.
So why did economists change their mind? In both cases, NGDP signaled a problem. NGDP fell roughly in half during the early 1930s, which sure looks like tight money. NGDP growth averaged 11% during 1971-81, which sure looks like easy money.
This is why my contrarianism is so boringly conventional. I’m merely trying to do for the Great Recession what other economists have already done for the Great Depression and the Great Inflation. I’m attempting to get people to see that a “Great” economic problem, which didn’t look monetary in real time, actually was monetary. I am trying to get people to believe that money was tight in 2008, even though it didn’t look tight. I hope to convince people that the huge drop in NGDP growth during 2008-09 is prima facia evidence of an excessively tight monetary policy. I am trying to get people to see the post-Lehman banking crisis as being caused by falling NGDP, just as the 1930s banking crises were caused by falling NGDP.
In fact, my contrarian take on the Great Recession is so similar to previous reappraisals of the Great Depression and the Great Inflation that I’m tempted to say that it is I that is actually the conventional economist and all those who have not come around to my view are the true contrarians.
It’s a longstanding tradition for economists to initially blame “Great” problems on non-monetary factors, and then later see them as monetary policy failures. Why stop now? Keep the tradition alive!
READER COMMENTS
Rob Rawlings
Dec 1 2021 at 10:03pm
Great post.
I see there is now both a kindle and a audio book version of ‘The Money Illusion: Market Monetarism, the Great Recession, and the Future of Monetary Policy’. Will give it a listen!
https://www.amazon.com/Money-Illusion-Monetarism-Recession-Monetary-dp-B09KT4T4BK/dp/B09KT4T4BK/ref=mt_other?_encoding=UTF8&me=&qid=
Andrew_FL
Dec 1 2021 at 10:03pm
And when did it stop being tight? When exactly. Be specific, and show your work.
Matthias
Dec 3 2021 at 12:33am
According to the 5 year TIPS spread, inflation expectations were back to about to 2%-ish in perhaps 2011.
Alas, we don’t have such a good and deep prediction market for ngdp expectations.
Andrew_FL
Dec 3 2021 at 11:44am
I did not ask you, but if the reason Summer has not answered my challenge is a lack of NGDP prediction markets, then how, lacking them, does he know money was tight to begin with?
Jose Pablo
Dec 1 2021 at 11:16pm
Why economist that are so frequently wrong (to the point that claiming them to be wrong is “common throughout history”) keep being elected as FED officials?
Aren’t they supposed to be the best and brightest?
rsm
Dec 2 2021 at 1:12am
《NGDP growth averaged 11% during 1971-81, which sure looks like easy money.》
Why wasn’t Cost Of Living Adjustments the best answer? Did the S&P 500 return higher than inflation over that same time period, thus effectively indexing investor incomes to inflation? Why not extend the financial investor’s solution to inflation, to everyone?
robc
Dec 2 2021 at 9:15am
Short answer: no.
Long answer: At the start of 1971, the S&P500 was 91.15. It closed 1981 at 122.55. That is a 34% gain over 11 years, or 2.7% annual growth, if I did the math correctly. Over the same time period, the CPI went from 38.8 to 90.9, 134% growth, or 8.0% annually.
Vivian Darkbloom
Dec 2 2021 at 1:28pm
The even longer answer is that the comparison needs to include the dividends paid by companies in the Index and not simply the change in the Index over that ten year period. The S&P companies paid significant dividends over that period of time and those dividends largely (but not quite) kept up with inflation as measured by the CPI. By my calculation, the S&P return including dividends returned around negative 1 percent annually if deflated by the CPI. That’s not great but it’s much better than what you calculated and I’d rather have held those stocks than 1970 10-year Treasury bonds for that period of time. Furthermore, if the CPI, as many think, overstates “inflation”…
https://www.multpl.com/s-p-500-dividend/table/by-year
robc
Dec 2 2021 at 4:57pm
Good point, I thought about looking it up with dividends included, but didnt think it would close THAT much of the gap.
Matthias
Dec 3 2021 at 12:38am
For the same reason that in general ngdp is a better indicator for these kinds of things than inflation:
Productivity shocks (good or bad) make inflation an unreliable indicator for the stance of monetary policy.
A recent example: during the pandemic the cost of living went up, because of supply issues. But that doesn’t mean that money was too loose.
John Hall
Dec 2 2021 at 8:33am
I’ve been reading around on the literature about structural VARs and identification of shocks. I think one difficulty with your approach is that you assume that an aggregate demand shock is basically the same thing as a decline to nominal GDP. I’m sure they are highly correlated, but other things could be going on too. For instance, if the population shrunk by -5% in a quarter, then you wouldn’t obviously be able to argue that the accompanying change in nominal GDP was due to monetary policy. Of course, population doesn’t change that much in a quarter. However, there may be other factors influencing aggregate supply, like productivity growth, that are difficult to measure but may explain a portion of the movement in nominal GDP. Maybe your argument then would be that productivity shouldn’t be more volatile in a recession so we should be able to associate the volatility in recessions to monetary policy mistakes? Regardless, I’m not saying that your underlying story is right or wrong. I’m just concerned that you reach a little too quickly for the argument that nominal GDP slowed or declined and therefore it is a monetary policy mistake that was the cause. It is possible for a real shock to have a short-term asymmetric impact on real GDP versus prices, which means that a real shock could have a nominal impact in the short-term, even if it doesn’t over the long-run.
Scott Sumner
Dec 2 2021 at 11:01am
I’m not claiming that falling NGDP causes less AD; I’m defining AD as NGDP.
John Hall
Dec 2 2021 at 12:27pm
I’m not sure that really clarifies things…
My point is that there are changes in nominal GDP that are driven by things that the central bank cannot control.
Don Geddis
Dec 2 2021 at 2:52pm
Your points are true — they are just irrelevant. Yes, real shocks could indeed have nominal effects. But that doesn’t change the analysis at all.
A central bank is a feedback control system. Imagine a house thermostat — but with a furnace and air conditioner of infinite power. The Fed has an inflation target. The Fed has the ability to make the quantity of money be anything at all that they want. If the resulting nominal variable is not on target, the problem is always that the Fed chose the wrong quantity of money (given current economic conditions).
It doesn’t matter at all what current economic conditions actually are. It doesn’t matter if they’ve changed recently. Monetary policy is not a fixed action; instead, it is a function that maps from current conditions to appropriate action (in order to reach target goals). If you aren’t experiencing your nominal variables being on target, the problem is bad monetary policy. By definition.
John Hall
Dec 4 2021 at 1:58pm
Interesting reply.
If real shocks have an effect on nominal GDP, then does it make sense for the central bank to respond to all changes in nominal GDP? Or, should they only respond to nominal shocks?
Suppose the central bank can freely create a nominal shock. If there is a negative real shock that causes real GDP to fall by more than prices rise, then it has a negative impact on nominal GDP. If the central bank creates a nominal shock large enough to stabilize nominal GDP, then it will basically boost both real GDP and inflation. This means the central bank acts when maybe they shouldn’t. Real GDP is not as low, while inflation is higher than it should otherwise be. Sounds familiar…
John Hall
Dec 5 2021 at 6:45am
I wrote up a response to this, but the site seems to have eaten it.
My point was that if central banks can only create nominal shocks. If they are responding to a real shock with a nominal shock, then they won’t be able to match the same forces. For instance, suppose there is a real shock that causes a decline in real GDP and a modest increase in inflation, on net there is a decline in nominal GDP. If the central bank responds and creates a nominal shock, then they will partially offset the decline in real GDP but then add more to inflation.
So the question is really, if real shocks have nominal effects, should the central bank be responding to them? I’m not sure, but I don’t know if it’s as simple as the way Scott normally talks about it.
Don Geddis
Dec 8 2021 at 2:59pm
Yes … you have asked a deep question about monetary policy: what is the actual purpose? What is the goal? What are you trying to achieve?
You are correct that nominal (monetary) stimulus cannot possibly undo a real crash caused by supply-side disruptions. Asteroid, tsunami, covid … there can be less real output for supply-side reasons, and there is nothing that demand-side stimulus can do about recovering that missing wealth.
But what should demand management be trying to do?
Sumner has been blogging about this since the 2008 recession. It’s too complex for me to convince you in a short comment, but the basic story is: nominal crashes cause additional economic pain. Because of (empirically observed) downward nominal sticky wages, a nominal crash is responded to by the economy with increased involuntary unemployment (and thus lower real output). Moreover, the key is nominal revenue, not the value of the currency. So inflation itself doesn’t matter nearly as much as NGDP. (Even though both usually move together, so it’s rare to have a situation where you can see the difference.)
So the basic answer to your scenario is: if a patient comes into the hospital with a knife wound, you don’t want to them give them an additional gunshot wound as well. You can’t immediately fix the knife wound, but you should avoid making the situation worse.
A real supply-side crash is going to reduce real output and real wealth. If you also allow a nominal crash, you are going to get additional – and unnecessary! – economic distress. (One more way to think about it: if you have an asteroid/tsunami with reduced real output and reduce national income, the labor incentives should be to work harder and use less leisure. So unemployment goes down with a real crash. But the characteristic symptom of insufficient aggregate demand is involuntary unemployment, where unemployment goes up. Covid may be an interesting exception, but in general a supply crash results in higher employment, as labor is substituted for missing capital.)
Alex S.
Dec 2 2021 at 3:04pm
@John Hall, maybe this will help clarify…or possibly confuse…hopefully the former, apologies in advance if the latter.
Mv = Py
M = money supply
v = velocity (money demand)
P = GDP deflator
y = RGDP
Y = Py (aka Agg Demand – try charting this in y-P space assuming Y = 1,000 – see what happens)
Y = NGDP
y = c + i + g + nx
Putting these together:
Y = Mv = P(c + i + g + nx)
your contention is…roughly…that c, i, g, or nx can also cause Agg Demand to change. That’s true but notice that if those change there must be a corresponding change to money supply or money demand.
I like to think the Fed affects M through open market operations and v through “open mouth operations” (e.g. forward guidance) crudely speaking.
Thus, if g falls, a proper central bank action would be to raise M or v to offset the tighter fiscal policy (assuming we started at full employment).
Ceteris paribus, “things the central bank cannot control” can affect Agg Demand ***but*** the central bank can control how it reacts to things it doesn’t control directly. The best approach is if the central bank acts to keep Agg Demand (NGDP) stable.
Hope this helps.
John Hall
Dec 4 2021 at 1:47pm
It seems you are trying to combine together accounting identities. I’m skeptical that this helps clarifying the underlying economic relationships better…
As I noted in my first comment, I came to my point when thinking about Structural Vector Autoregressions (SVAR). In an SVAR with real GDP and prices, we can identify two shocks (using a similar approach as Blanchard-Quah, 1989), one associated with long-run changes in both and one associated with long-run changes only on real GDP. We can interpret or label these as a nominal shock and a real shock, respectively. My point is that the real shock in this case can have an impact on nominal GDP. For instance, this approach does pretty well over the COVID-19 period in picking up that what we experienced in Q2 2020 was more of a real shock than a nominal shock. There were big declines in both nominal and real GDP over this period. Scott often says that this is a classic supply shock, but in the past he has said that any movement in nominal GDP is a nominal shock. My point is that a real shock can have a nominal effect.
Matthias
Dec 3 2021 at 12:40am
Productivity improvements are neutral for nominal GDP.
See George Selgin’s book Less than Zero for more. You can read it for free online.
And basically, changes in productivity are exactly why inflation is a less reliable indicator than nominal GDP.
John Hall
Dec 4 2021 at 1:35pm
Not sure what you mean by neutral here. All else equal, an increase in productivity will boost real GDP and thus nominal GDP. The productivity norm is a specific monetary policy rule, so we wouldn’t be holding everything equal.
Spencer Bradley Hall
Dec 2 2021 at 9:13am
Sumner doesn’t go far enough. There were only 3 periods during which the rate-of-change in monetary flows, volume times transactions’ velocity, AD, fell below zero (became negative) since the FED was established. And the GFC recorded the second longest time for negative AD growth.
Dale Doback
Dec 2 2021 at 11:46am
I’d like to think that a majority of the 99% of economists who don’t agree on the causes of the Great Recession would at least acknowledge that monetary policy at the time was inappropriate considering that the Fed essentially admitted this when announcing AIT.
Kevin
Dec 2 2021 at 2:00pm
As someone who barely knows the basics of economics, what then should have the Fed done during the Great Recession in order to allow for easier money. Is it more QE?
Kevin
Dec 2 2021 at 2:05pm
To be clear…I’m learning a lot from reading this blog and am just hoping to find out more.
Scott Sumner
Dec 2 2021 at 5:11pm
They probably could have prevented a severe recession even without QE, if policy had been more expansionary in 2008—say cutting the target interest rate more aggressively.
Kevin
Dec 2 2021 at 6:19pm
Thanks, that makes sense.
Kevin
Dec 2 2021 at 2:08pm
What about lowering IOR? Again, forgive my ignorance, I’m just hoping to get a crude understanding of this kind of thing.
bill
Dec 5 2021 at 9:45pm
The funny thing is that the Fed paid zero IOR for 95 years. For some bizarre reason, they started paying IOR in October 2008.
Don Geddis
Dec 2 2021 at 3:04pm
Set expectations: clear statement of nominal targets that will be achieved by “whatever it takes”
Zero Interest on Excess Reserves (IOER)
”Infinite” QE: the Fed will continue to purchase bonds (exponentially doubling every month?) until nominal inflation (or NGDP) rises into the Fed’s target range
A better monetary policy framework (than “inflation targeting”), e.g. NGDPLT: GDP instead of inflation, nominal instead of real, level target instead of growth target
Kevin
Dec 2 2021 at 4:43pm
Thanks for the explanation.
Matthias
Dec 3 2021 at 12:41am
Refrain from interest on excess reserves, for a start.
Jose Pablo
Dec 4 2021 at 7:14pm
If what the FED should have done is so clear, the inevitable question is what prevented them for doing the right thing to do?
Is it a question of technical capabilities (or the lack thereof)? is there a problem with the “institutional design”? with political interference?
Or is it a “every master has his own trick” kind of problem?
Michael Rutland
Dec 3 2021 at 5:22pm
Michael D Sandifer
Dec 4 2021 at 12:43pm
Scott,
Yes, you and the market monetarists generally, have been contrarian by understanding and adhering to macroeconomic fundamentals better than most other economists. More importantly, your models are shown to be better, but they’re not without contradictions.
To address your implicit model, you have long stated that most of the slowdown in US real GDP in recent years, along with lower real rates, has been due to real factors, such as changing demographics and a shift toward lower productivity growth sectors, such as services. This implies that wage adjustment aftre the Great Recession took roughly 6 or fewer years to occur. This seems to be very consistent with the opinions of most economists. But, there are serious problems with this perspective, especially from a market monetarist viewpoint.
First, as I’ve pointed out previously, if we’re to trust Treasury market forecasts, why didn’t this market see lower real rates coming prior to the Great Recession, and why does the market continue to predict real rates will be higher 10 years in the future?
https://fred.stlouisfed.org/graph/?g=Jw7m
Second, the stock market gives us exactly the same message, but focuses more on the issue of how long it actually takes an economy to heal on its own during a tight money recovery. The S&P 500 fell a bit more than 50% during the Great Recesssion, indicating that average earnings expectations to infinity fell by the same amount. But, of course, the temporal distribution of the earnings collapse was front-loaded, with earnings falling more than 90% during the recession. If you do the calculation, just using the standard present value of discounted earnings model, the stock market was forecasting that the economy would be fully healed in about 14.5 years. If you play with the numbers in the equation, you can plausibly even get a longer predicted recovery period, but you can’t get one that’s much shorter. I can show the mathematics, if you like.
Also, the magnitude of the differences between forecasts and reality for both markets is roughly the same.
Comments are closed.