Scott Sumner  

Questions for Keynesians (Show me the data)

Tsipras and electoral reform... Lawson's Economic Freedom Bet...

I'm seeing lots of comments to the effect that the 2013 austerity was not a big deal, and or the dreaded "fiscal cliff" never happened. OK, let's see some numbers. After all, if my critics just know that I am wrong, they must have the correct numbers. In an earlier post I estimated that the deficit fell by $500 billion in calendar 2013.

(It fell by $400 billion in fiscal 2013, but you want to use calendar 2013 because the serious austerity kicked in on January 1st, 2013. Indeed if that date were not important then the deficit reduction estimates would not vary by $100 billion merely by shifting the date by 3 months.)


1. Is my $500 billion deficit reduction estimate wrong? I used this source. If it's wrong, what source should I use for monthly deficits? What are the actual numbers for 2012 and 2013?

2. If the dreaded fiscal cliff had happened, what was the estimated deficit reduction that was expected to occur?

3. Is the difference between the answer to #1 and #2 large enough to account for the difference between a predicted recession and a speed up in economic growth (or at worst roughly no change, if you don't buy my speed up argument)?

My commenter Justin over at MoneyIllusion also has some questions:

Two questions for Keynesians in light of this discussion:

1) If the general consensus of the Keynesian camp is that $500 billion of deficit reduction (much of it exogenous and permanent) shouldn't be expected to put a noticeable dent in the time path of GDP or the unemployment rate, why credit the $800 billion stimulus (which only lasted for 2 years) with any meaningful effect whatsoever on the economy during 2009-2010?

2) Should we not try to generate a budget surplus for FY2016 if cutting the deficit by $500 billion isn't really austerity and therefore unlikely to produce obviously negative impacts on GDP and unemployment?

Comments and Sharing


COMMENTS (25 to date)
Ray Lopez writes:

Sumner: "...between a predicted recession "

My first objection: aside from the "Open Letter" by the liberal leaning economists, that indeed predicted a 'double-dip recession' if sequestration happened, who else predicted a recession? Not the CBO (which uses a Keynesian model to forecast). Not Krugman (who merely thought the USA might follow the trajectory of the UK). Then who? What am I missing?

Daniel Kuehn writes:

I agree with Ray. I've found this whole line of argument a little strange. It wasn't really a natural experiment in the first place (and I criticized Konczal and Krugman on that at the time). I don't understand how macroeconomists can get so boisterous over such bad counterfactuals.

Maybe there's something wrong with your data, I don't know. I've given you the benefit of the doubt on that - I assume you can look that up. It's not a data issue really, I just don't see much here as a test of anything.

Scott Sumner writes:

Hmmm, not a single Keynesian comes forward with any data to support the claim that my data is wrong. The silence is deafening. Or should I say the silence is defining?

Ray, That's right "only 350" Keynesian economists predicted recession, no big deal.

Daniel. I actually agree that it wasn't much of a natural experiment, nor were the similar UK and eurozone austerity episodes, for exactly the same reason. So Keynesians have essentially zero empirical evidence to justify abandoning their 2007 presumption that the multiplier is roughly zero due to monetary offset. Do you agree?

Or is there some evidence I missed?

And can I assume that you've never posted anything to the effect that UK or eurozone austerity was slowing the recovery?

TomM writes:

@ Daniel

The Keynesian thought process just seems completely political in my eyes. If the data seems to support there model, then they scream victory. If however, the data does not support there argument, the argument they turn to is "things would have been better". The entire argument is the one predicated on a counterfactual:

If country A pursues fiscal austerity and economic conditions worsen-> Keynesians claim a win

If country A pursues fiscal austerity and nothing changes-> Keynesians claim things would have be better without austerity

If country A pursues fiscal austerity and economic conditions improve-> Keynesians claim things would have improved more!

I think what frustrates professor Sumner and others is that no matter what the outcome, Keynesians claim this is what there model would have predicted.

P.S. It still is beyond me how Krugman and others can point to there "Scatterplot of Government Spending and Growth" as a defining piece of evidence in favor of fiscal stimulus and still be taken seriously.

Tracy W writes:

Daniel Kuehn: if I follow Scott Sumner correctly, he agrees that it wasn't a good natural experiment.
But, if it's not a good result against Keynesianism, then presumably similar experiments aren't good for Keynesianism, such as the argument that $800 billion of spending over two years could be credited with improving GDP.

Ray Lopez: you're dismissing 350 economists here. You're getting onto "but apart from the sanitation, the medicine, education, wine, public order, irrigation, roads, the fresh-water system, and public health, what have the Romans ever done for us?" territory.

Mr. Econotarian writes:

If you are debating Keynesians, you might want to address spending by all governments, not just the Federal one. suggests that 2013 total US government spending (federal, state, local) was $6.1 billion, and 2014 was $6.14 billion.

Steve J writes:

I agree with Mr. Econotarian that deficit reduction seems like the wrong number. I thought the fiscal cliff was a spending cliff not a taxing cliff. Did spending actually decrease from 2012 to 2013?

Daniel Kuehn writes:

Scott -

"So Keynesians have essentially zero empirical evidence to justify abandoning their 2007 presumption that the multiplier is roughly zero due to monetary offset. Do you agree?"

I certainly don't agree that the lack of a natural experiment implies zero evidence. I don't *think* I agree that there's no evidence, but I'm not especially fluent in the macro empirical literature. My impression from those who are is that there are plenty of studies that show a non-zero multiplier (including a guy paid by someone hoping to find a zero multiplier that concluded the multiplier's about 1.5 based on the best studies we have).

"And can I assume that you've never posted anything to the effect that UK or eurozone austerity was slowing the recovery?"

No, you can't assume that. I'm not entirely clear on why that should follow. But I don't think I've ever suggested that there's some kind of incontrovertible well-identified empirical evidence for that claim. That's quite different from saying that given the evidence that we do have we should expect that it slowed down growth relative to what it would have been.

Daniel Kuehn writes:

Tracy W.

"if I follow Scott Sumner correctly, he agrees that it wasn't a good natural experiment."

That's what I find so weird about Scott's recent posts. He lead the first post talking about the "spectacular implosion of Keynesian economics in 2013". If he sincerely thinks it wasn't a good natural experiment - and I share your impression that he has his doubts - then why is he so insistent that it has spectacularly imploded?

Doesn't that seem weird to you?

I agree, btw, with some of these data points - particularly the issue raised above that the federal government back-filling state cuts matters for how stimulative the whole thing is. But even if we get the data we want to look at agreed upon you still need to deal with causality and the counterfactual and I don't see how anyone has.

Daniel Kuehn writes:

TomM -
That seems like a thought process I see everywhere - not a Keynesian thought process. You probably notice it more with Keynesians.

Anyway, I obviously agree with you that it's wrong to latch onto bad arguments. That's precisely why I initially commented with the idea that the natural experiment was no such thing. I can't sympathize with your complaint about taking care to think about counterfactuals when things go south. I think that's precisely the right way to think and we should encourage more of that, not get frustrated with it.

More people need to nip it in the bud. When Konczal and Krugman initially did their post we should have had more people say "how will you disentangle what would have happened under business as usual from the observed outcome to isolate the impact of the policy change?".

james in london writes:

Just google "Fiscal cliff". There is more stuff out there than anyone could possibly want. Steve is right. It was roughly $400bn+ in tax increases, some Bush era expiring cuts from earlier years, the end of part of the payroll tax holiday. And $100bn+ in cuts in spending, mostly from defence.

A decent data set is the spreadsheet here:

Do a chart and on a trailing 12m average it looks like outlays fell $10-20bn per month, and maybe $20-30bn per month below trend. And receipts climbed $20-40bn a month as 2013 wore on. Perhaps receipts were on trend though.

The deficit certainly shrunk from around $100bn per month on the same moving average basis to $50bn by the end of 2013. So the $500-600bbn plan looks to have worked, potentially aided by tax receipts rising due to the recovery.

Matt McOsker writes:

I think Warren Mosler commented in your shell game post. He expected the fiscal changes to be a drag, but it was offset by private borrowing.

Mark Bahner writes:
I'm seeing lots of comments to the effect that the 2013 austerity was not a big deal,...

"Not a big deal!?" C'mon! (As Mike Wallace used to say.)

It was "unprecedented"!

Ya can't get bigger than "unprecedented." And it was "just awesomely destructive", to boot. And this judgment was rendered in *December 2013.* By Paul Krugman, who's a Keynesian god*, or something.

P.S. "A god, not the God." --Bill Murray, Groundhog Day.

Jeff writes:

Aside from casually dismissing the open letter, I don't understand Ray's claim that the CBO was not predicting a recession. In August of 2012, the CBO’s baseline forecast was for the 2013 deficit to be $641B and “Such fiscal tightening will lead to economic conditions in 2013 that will probably be considered a recession, with real GDP declining by 0.5 percent between the fourth quarter of 2012 and the fourth quarter of 2013 and the unemployment rate rising to about 9 percent in the second half of calendar year 2013.” CBO tallied the actual deficit to be $680B in 2013. Are we to infer that the extra $39B in spending is what made the difference between recession and no recession?

Bob Murphy writes:

Mark Bahner's comment above is unprecedentedly awesome.

Scott Sumner writes:

James, Finally, someone answers my question! Thanks. So it looks like the dreaded "fiscal cliff" that was averted was almost identical to the actual "austerity" of 2013. Just what I expected. BTW, it's true the recovery will naturally lead to some reduction in the deficit, but the reduction in 2013 was far, far greater than either before or since, and the recovery has been underway since late 2009.

And BTW, the deficit in calendar 2013 was only about $580 billion ($100 billion less than the fiscal year deficit), which shows how dramatic the policy switch on January 1 2013 really was.

So we can lay to rest two more myths:

1. The "it wasn't really much austerity" myth
2. The "not many Keynesians were predicting a significant slowdown in growth" myth.

If the Keynesians were intellectually honest they'd now be saying:

"Yes, the MMs were right about 2013, the dramatic austerity did not have the effect we expected, partly due to monetary offset. But there may be other occasions where fiscal policy is effective at the zero bound, because the Fed does not offset it."

If I had been wrong about 2013, I would have said:

"Yes, the Keynesians were right about 2013, the dramatic austerity did have the effect they expected. But there may be other occasions where fiscal policy is ineffective at the zero bound, because the Fed offsets it."

Everyone, About the question of whether this was a fair test. I regard the theory as wacky. The burden of proof is on those propounding the view that the conventional wisdom of 2007 (that fiscal stimulus is ineffective) was wrong. They needed a home run in 2013, not a single. Instead they struck out. There is a different burden of proof if you are trying to disprove a well established theory (say supply and demand) as compared to if you are offering up a new and contrarian theory (like the Laffer Curve.) Evidence too weak to disprove S&D, may well strong enough to lead people to reject a contrarian theory. Fiscal stimulus is much closer to the Laffer curve than to S&D. Indeed now you often read Keynesians claim that fiscal stimulus won't even increase the deficit, due to the growth effect. So modern Keynesianism has actually moved very close to Laffer Curve economics.

BTW, there is actually more evidence in favor of the Laffer Curve, (that cuts in very high marginal income tax rates lead the rich to pay more taxes) from the 1980s than there is for the fiscal multiplier from 2013. And yet most Keynesians say the 1980s "disproved" the Laffer Curve.

One other thing. I would expect fiscal stimulus in Greece to "work" due to lack of monetary offset. I use scare quotes, because I don't think Greece has a bright future, for supply side reasons. But cyclical recovery? That's very possible.

Ray Lopez writes:

Everybody has an ax to grind. Reading these comments, I find that Daniel Kuehn has the most balanced viewpoint.

I find this Sumner quote confusing:

If I had been wrong about 2013, I would have said:
"Yes, the Keynesians were right about 2013, the dramatic austerity did have the effect they expected. But there may be other occasions where fiscal policy is ineffective at the zero bound, because the Fed offsets it."

I think Sumner meant to say "right" not "wrong about 2013". And 'wrong' not 'right' in the next sentence following?

Anyway, economic cycles are largely random, so IMO neither monetary policy nor fiscal policy have any real long term effect on the economy, short of hyperinflation or unlimited reckless spending.

Nick writes:

I think you've nailed it! It's a knife's edge solution: monetary and fiscal policy have absolutely no effect until one of them 'goes too far' and then suddenly (or maybe even a handful of years later) it has a deleterious effect of random magnitude and duration.
The code has been cracked people. Go home.
Oh wait I misread you ... It's just no 'real long term' effect. Cool. Now what about the short term effect? You know, the thing this thread has always been about?

Ray Lopez writes:

@Nick - this blog is moderated, so it's not about what I think, but Sumner's post. But to answer your question, if this post goes through, the short-term is measured in days, hours, minutes. It's like a press release affecting the price of a stock. Most of the time the press release is good for a 'pop' of a small fraction in the price, then it's back to the 'steady state' solution. Monetary and fiscal policy, unless taken to the extreme, are largely irrelevant. Fischer Black showed this to be true, and traded on it, as has George Soros vis-a-vis the Bank of England, and indeed the history of nearly perpetual deficits in the USA has shown fiscal policy is a red herring, albeit a very wasteful rotten red herring.

Justin writes:

@Mr. Econotarian,

The overall story doesn't materially change using aggregate government spending rather than federal.

The nominal deficit in calendar 2012 was $1,452.43 billion and it falls to $957.55 billion in calendar 2013 when defined as the gap between total receipts (line 31) and total expenditures (line 34). This occurs at the start of the year, too, as the deficit falls from $1,400.4 billion during 2012Q4 to $1,054.7 billion in 2013Q1 to $869.7 billion in 2013Q2.

The following numbers show the actual total government deficit as a % of GDP and a projection based on the 2010Q4-2012Q4 trend. This suggests an exogenous decline in the deficit of about 2% of GDP, which compares reasonably well with the various fiscal changes we know occurred during 2013 (expiration of the payroll tax cut, the sequester, the end of some of the Bush tax cuts, etc). Put anything like a Keynesian multiplier on that (1x or 1.5x) and the counterfactual is that 2013 GDP growth Q4/Q4 would have been +5% to +6% (i.e. the highest since 1984) rather than +3.1%. With 2014 GDP Q4/Q4 growth having slowed to +2.5% despite no additional fiscal tightening, it is difficult to accept that counterfactual as plausible.

Actual Deficit Projected Deficit
2010Q4 -11.3% -11.3%
2011Q1 -10.8% -10.8%
2011Q2 -11.2% -11.2%
2011Q3 -10.6% -10.6%
2011Q4 -10.4% -10.4%
2012Q1 -9.2% -9.2%
2012Q2 -9.2% -9.2%
2012Q3 -9.0% -9.0%
2012Q4 -8.6% -8.6%
2013Q1 -6.4% -8.2%
2013Q2 -5.2% -7.9%
2013Q3 -6.3% -7.5%
2013Q4 -4.9% -7.2%

Nick writes:

I love the way you use that Soros / Bank of England thing. I've seen you pull it out twice before in threads and you always deploy it to perfection: off-topic, and backward.
For fiscal impotence in the short term I think you may be on to something there...
As to "days, hours, minutes" in one obvious sense the shortest term measure of gdp we have is the quarterly gdp reports, since they literally are our gdp numbers and we only produce them every quarter.
You can try to simulate a shorter term look, but on the time scale of hours and days it's very hard to do. If only there were some useful method of producing extremely rapidly updating predictions the field of economics was aware of and could employ...

Justin writes:

I agree with the comments that there is no perfect natural experiment in macroeconomics - there are always going to be too many confounding variables.

However, consider this hypothetical situation.

Suppose you got Paul Krugman, Brad DeLong and other assorted Keynesians into a room during December 2012, after the November budget data had been released showing a $1,145.53 billion trailing 12 month federal deficit. You told them that you were going to force down the budget deficit to $559.51 billion during calendar 2013 and down again to $487.44 billion during calendar 2014. Furthermore, unemployment would fall from 7.7% in November 2012 to 6.7% in December 2013 and to 5.6% in December 2014. More payroll jobs would be created in 2013 than 2012. The S&P 500 would rise from 1,416.18 on 11/30/12 to 1,848.36 on 12/31/13 and 2,058.90 on 12/31/14. At the same time, headline inflation falls from 1.8% YoY in November 2012 to 1.5% YoY in December 2013 and 0.8% in December 2014. Most analysts at the end of 2014 will expect Fed rate hikes in the middle of 2015.

Can anyone honestly say Krugman, DeLong or any other Keynesian economist would have taken that forecast seriously? To have survived proposing it to them without not only skepticism, but outright scorn and ridicule?

Brad writes:

I don't have much to add other that Justin's post is just fantastic.

Would love to hear a Keynesian address it.

Scott Sumner writes:

Justin, Great comments.

Ray, Soros believes fiscal and monetary policy have no effect????

Ray Lopez writes:

@Justin - Sumner did a post on this once I believe, or something close to it. Indeed it's implausible the Keynesians would admit defeat, but economics is not (supposed to be) about personalities but economic science.

@Sumner - "Ray, Soros believes fiscal and monetary policy have no effect????" - no, I'm sure Soros, being liberal, believes in Keynesianism. But his conduct shows central banks are not omnipotent. I'm not sure why Nick thinks I have the history of how Soros broke the bank of England backwards. The Bank of England could have defended the exchange rate more, and indeed only 3.3B pounds were spent (see Wikipedia on Black Wednesday), a trifle for a de facto central bank. If you read the history behind Black Wednesday (not found on Wikipedia), you'll find the Major government of the UK lost their nerve way too easily and refused to fund the Bank of England to fight Soros. They could have defended longer and Soros and the others could have caved. But the larger point being: central banks follow the market much more than the public assumes with simplistic 'Don't Fight The Fed' dogma. The implications for your NGDPLT framework are obvious, but that's a post for another day.

Comments for this entry have been closed
Return to top