Scott Sumner  

The "other things"

PRINT
Why do we write too much?... Suderman on Republicans on Hea...

Russ Roberts did a recent post complaining about the way Paul Krugman uses data to support his arguments. Roberts pointed out that when things turn out as he expected, Krugman cites the events as proof of his (Keynesian) worldview. When they don't, he suggests that "other things" interfered, and that the model is still correct. Now Krugman has a new post that doesn't specifically mention Robert's post, but responds to this general criticism.

First let's think a bit more about fiscal policy and "other things." Suppose the authorities reduce federal spending on output. That's a contractionary fiscal policy. And suppose that GDP rises rather than falls. That seems inconsistent with the Keynesian model. Could this outcome have been due to other factors? Yes, it certainly would have been. By definition, either state and local spending rose, or investment spending rose, or consumption rose, or the trade balance improved. So yes, there definitely would have been "other factors." The question is how we think about the cause of these other factors.

Perhaps the most confusing of the other factors is state and local spending. Many economists wrongly assume that state and local (S&L) spending are a part of fiscal policy, because we attach the letter "G" to their output. But in the Keynesian model they are really no different from private investment. During recessions both private companies and state and local government are less likely to build things like new roads. These decisions are made by 1000s of entities, and should be viewed as endogenous in any model of federal spending as a tool of fiscal policy. For instance, if fiscal stimulus causes 100,000 new houses to be built in suburban Phoenix, then local governments in that area will endogenously add roads, sewer lines, schools, etc. Or as the economy improves revenues will rise, and states constrained by balanced budget rules will spend more.

In calendar 2013, fiscal policy in the US became dramatically more contractionary. This isn't even debatable. The budget deficit plunged by an astounding $500 billion in a single year. Now suppose your prediction of a slowdown doesn't turn out to be true. How would you try to minimize the amount of austerity that occurred in 2013? One technique would be to use fiscal years, as the big tax increases occurred on January 1st, not at the beginning of the fiscal year. Thus the fall in the budget deficit for fiscal 2013 was somewhat less than $500 billion, although still extremely large. (I have no evidence Krugman did this; his IMF data has no link.)

Another technique is to focus on government consumption, which is only one component of fiscal policy. This will make the changes look much smaller. For instance, in 2013 most of the austerity consisted of tax increases and cuts in transfer payments, not reductions in government consumption. And another technique would be to muddy the waters by citing figures for the change in both federal and S&L deficits.

In Krugman's posts he has two graphs of government consumption, which show that as federal consumption fell, state and local consumption rose. Interesting data, but it's not clear what it tells us about fiscal policy. Then he makes a different claim:

What we should have realized, but I didn't - at least not fully - was that the sequester just wasn't that big relative to the economy. The CBO put the first-year impact on the deficit at $68 billion, or 0.4 percent of GDP, in the first year with much smaller additional impacts thereafter - not trivial, but not huge either.
The sequester included government consumption and transfers, but not the big increase in government taxes. Recall that back in early 2013 Keynesian economists were extremely concerned about the dramatic shift toward austerity. And it wasn't just Krugman, there was a letter signed by 350 Keynesians warning that the austerity could lead to a recession. I think these guys were exactly right about one thing; there really was a lot of austerity occurring in 2013. (Yes, we didn't get 100% of the feared "fiscal cliff", but the $500 reduction in the deficit was only slightly less than that worst case, according to the estimates I've seen.) The estimates I saw suggested that the total austerity package would have been expected to reduce GDP by 1.5% to 2%. Can someone confirm that?

While I don't agree with Krugman's revisionist view that there really wasn't much austerity, I'd like to accept it as a working assumption for the rest of this post. And I'd like to analyze its implications.

1. First of all, it's worth pointing out that the same sort of mistake seems to have been made in Britain, for unrelated reasons. (S&L is less important over there.) At roughly the same time as the US misjudgment, many Keynesian economists were warning that austerity in Britain was a disastrous policy that would slow the recovery. But then Britain suddenly started doing much better, especially in the employment area. (Recall that British productivity was weak, but the Keynesian model says more AD will create jobs, it does not include a magic wand to make those new jobs more productive.) Once the strength of the labor market became apparent, many Keynesian economists decided that the austerity was actually much less than they had been assuming. Including during the period when they were loudly complaining about austerity.

So let's say there was nothing disingenuous about all these flip-flops. I've made similar mistakes at various times in my life. Even so, doesn't this call into question the effectiveness of fiscal policy? It's already a pretty blunt instrument; requiring politicians of different stripes to come together in a timely fashion to set the appropriate cyclically adjusted budget balance. If we are now to believe that even in real time it's extremely hard for the world's leading Keynesian economists to tell whether policy is contractionary or not, then how likely is it that this tool will be used effectively?

2. And if we have not one but two major misdiagnosis of the stance of fiscal policy in key economies, in just the past few years, how many others have occurred that we don't know about? How many posts by Krugman can you find that say "I claimed fiscal austerity in country X would lead to recession, and recession did occur as I predicted? But new data shows that there really wasn't any fiscal austerity. I now think the recession was due to other stuff." Can someone point me to these posts? Indeed how likely is it that Paul Krugman would have done the post I am commenting on, if the US and Britain had fallen into a euro-style double-dip recession in 2013? "Oops, no austerity, my mistake." In other words, using a legal analogy, is Krugman more like an impartial judge, or an attorney that advocates for his client, and only feels a need to present evidence that supports his case?

Now before commenters start claiming that everyone does this, let me admit that this is normal behavior in the blogosphere (with a few exceptions such as Tyler Cowen.) So then are we just left with a "he said she said"? Just a series of anecdotes that are open to interpretation? I'd make a couple observations here:

1. We do have some systematic evidence. The early cross-sectional regressions supported the Keynesian model. Later work by Mark Sadowski and others found this result only applied to countries that lacked an independent monetary policy. (And I'd add that there are some tricky causality issues for even those countries.) As of this moment, I know of no systematic evidence for the effectiveness of monetary fiscal policy under the current inflation-targeting regime used by most countries. BTW, I've acknowledged that fiscal stimulus is effective under other regimes, and have cited the boost to GDP provided by war spending in the 1940s. That spending did not boost welfare (consumption fell) but it undeniably reduced the unemployment rate. And certain types of fiscal stimulus can be effective even under inflation targeting; such as employer-side payroll tax cuts and VAT cuts.

2. In a sense this entire debate is an artifact of a flawed stabilization policy regime, with unclear lines of authority. As I just indicated, fiscal policy would be effective (i.e. effect GDP) in certain types of clearly defined monetary regimes. And Paul Krugman has agreed that fiscal policy would not be effective if interest rates were above zero. Indeed he has advocated a 4% inflation target precisely because that would eliminate the need for fiscal stimulus. If the precise role of the Fed at the zero bound were made explicit, then this debate would go away.

There are very few debates in economics that could be resolved by Congress, but this is one of them. In this post I explain how.


Comments and Sharing






COMMENTS (20 to date)
meets writes:

"I know of no systematic evidence for the effectiveness of monetary policy under the current inflation-targeting regime used by most countries."

Did you mean to say fiscal policy?

Lewis writes:

Definitely agree about state and local spending, although the stimulus consisted in large part of aid to state govts.

I also wonder, based on politics in my region, whether asset prices have become critical to state/local spending. The reason is that many pensions are far less than 100% funded. It's plausible that, when QE props up asset prices as you have shown it to do, agencies make lower contributions to the pension funds and hire more people. Not that that's a good idea, but I think it happens. The effect could be fairly pronounced because it is precisely the labor-intensive/likely-to-hire agencies that are most exposed to pension problems.

There is also the potential effect on property tax collections, but I do not know how sensitive home prices are to fed policy.

E. Harding writes:

Yeah, there was no doubt 2013 was highly fiscally contractionary:
http://research.stlouisfed.org/fred2/graph/?g=14Kr

Brendan Riske writes:

[Comment removed for name-calling. Email the webmaster@econlib.org to request restoring your comment privileges.--Econlib Ed.]

Steve Fritzinger writes:

Krugman has repeatedly claimed that FDR's 1937 "austerity"caused the "Recession in the Depression". Those cuts were roughly the same size as the 0.4% GDP cuts he know dismisses as not big enough to matter.

There must have been some "other stuff" going on.

The Original CC writes:

Lewis wrote:

Definitely agree about state and local spending, although the stimulus consisted in large part of aid to state govts.

This would have shown up in federal spending/transfers. In other words, all else equal, it would increase the deficit, so it's already taken into account by looking at the federal side of things. (Please correct me if I'm missing something obvious.)

Frenchguy writes:

"And if we have not one but two major misdiagnosis of the stance of fiscal policy in key economies, in just the past few years, how many others have occurred that we don't know about?"

A bit early but Japan seems to be going in this direction. In late 2014, Krugman was all over it, claiming the VAT increase had been a huge mistake. GDP figures are very preliminary but GDP looks like it's rebounding nicely. More importantly, unemployment continues to drop and has barely register the VAT increase. True, inflation is not rising but it wasn't rising before (FX effect were doing most of the work).

David R. Henderson writes:

@Steve Fritzinger,
Krugman has repeatedly claimed that FDR's 1937 "austerity"caused the "Recession in the Depression". Those cuts were roughly the same size as the 0.4% GDP cuts he know dismisses as not big enough to matter.
There must have been some "other stuff" going on.

Interesting. And, by the way, there was another thing on: the doubling of reserve requirements. See Gene Smiley, “The Great Depression."

E. Harding writes:

The 1937 recession was weird: it was the only mainly aggregate-demand driven (unlike the 2008 recession) U.S. recession for which interest rates were very close to the ZLB for the entire recession and which did not seem to have any obvious causes (e.g., the end of postwar inflation).

bill writes:

@original CC,

Basically, you are correct. Payments to states are a transfer. However, in Krugman's post "2013 and all that", he chooses to link to a FRED graph that uses federal consumption which excludes transfers. I find that choice of his misleading because it would mean that unemployment benefits (a transfer) don't count as stimulus. I find that bizarre. I think that's how he excludes the 2 point increase in the Soc Sec tax on Jan 1, 2013. It's a pity. He used to be a more honest voice.

LK Beland writes:

It looks like Krugman's claim, i.e "FY 2013 austerity was nothing special" crucially depends on cyclical ajustements:

Primary Balance as a % of GDP (IMF):
Year Value (%) || Difference_with_prev_year
2009 -11.6 || -6.6
2010 -9.2 || 2.4
2011 -7.6 || 1.6
2012 -6.3 || 1.3
2013 -3.6 || 2.7
2014 -3.2 || 0.4

Cyclically adjusted Primary Balance as a % of potential GDP (IMF):
Year Value (%) || Difference_with_prev_year
2009 -8.3 || -3.4
2010 -8.0 || 0.3
2011 -6.2 || 1.8
2012 -4.6 || 1.6
2013 -3.2 || 1.4
2014 -2.4 || 0.8

Using the raw data, 2013 looks like the most austere year in a very long time. The cyclical adjustment "filters" it out.

I would be very interested to understand what made the cyclical adjustments so large in 2013.

ThomasH writes:

The behavior of state and local governments should not de different in kind, although they will be different in degree from the behavior of a national government. A recession will under most monetary policy regimes lead to lower borrowing costs for the US Government and so would normally lead to increased expenditures as more expenditures will pass the positive NPV test. The lower interest rates would lead them to increase expenditures as it should the USG, but the spreads against what the USG can borrow at may increase and so moderate or even reverse this tendency.

Of course when politicians start to make taxing and spending decisions based on the size or changes in the size or the rate of change in the size of the "deficit" all bets are off.

Scott Sumner writes:

Meets, Yikes, That was a bad mistake. Yes, I meant fiscal.

Lewis, Good points.

E. Harding. Thanks for the link.

Steve, Good point. Of course transfers count as Federal spending but not consumption. I'd add that the 1937 austerity was mostly taxes and transfers, just like 2013. And 1937 even included a 2% point rise in payroll taxes, just like 2013.

E. Harding. There were two main causes of the 1937-38 depression:

1. A big wage shock, from strong unionization after the Wagner act.

2. Deflation caused by a global bout of gold hoarding (of which the higher reserve requirements were only a modest cause.)

Frenchguy, Yes, I've done some posts on Japan, including one quite recent one. The Japanese tax increase was a big success.

Bill, Yes, that's part of what I'm complaining about in this post.

Thomas, You said:

"A recession will under most monetary policy regimes lead to lower borrowing costs for the US Government and so would normally lead to increased expenditures as more expenditures will pass the positive NPV test."

That's equally true of the investment portion of GDP, but no one would call an investment offset a part of fiscal policy.

E. Harding writes:

@Scott Sumner
-I'm still surprised at the gold hoarding not leading to significant growth in nominal short-term rates and there being only a pause in the growth of the money supply. On the other hand, as you've pointed out, there were only moderate gains in short-term rates in Europe, 2011.

James in London writes:

PK: "my fallibility says nothing about how the economy works", I love that. Even when I'm wrong, I'm right.

James in London writes:

More seriously. The Keynesians claim "G" in the C+G+I+(X-M) model didn't fall much, therefore no "austerity". However, am I right in thinking that the impact of the increase in taxes and cuts in transfers should have had it's impact via "C" but didn't. Thanks to monetary offset on "C" as a whole.

Scott Sumner writes:

LK, Very good observation. That adjustment would only make sense if growth accelerated sharply in 2013. But Keynesians have been insisting that growth did not accelerate in 2013. So which is it.

E. Harding it was very similar to Europe in 2008 and 2011, small increases caused recessions.

James, Yes you are right, and it's even worse because fiscal policy only includes Federal G, not state and local G (which is endogenous.)

James in London writes:

Are state and local even in G, normally or formally?

o. nate writes:

Very interesting and informative post. It definitely shows how hard it is to draw firm conclusions from macro data. I agree that by the standard "textbook" model of Keynesian stimulus, ie., one based solely on net government surplus or deficit, you are correct. I would argue that perhaps what is needed is a refinement of that model, perhaps one that argues that government spending financed by taxes that fall primarily on the wealthy is not especially different from that financed by borrowing that also is (ultimately) financed primarily by the same individuals.

Brian E. writes:

I guess some of this depends on what you plan to use your model for, but if your model says the economy is going to behave a certain way and it does the opposite, it's probably not a very good model regardless of the reasons behind its failure. You can make all the excuses you want about external factors or hidden variables that you couldn't take into account, but the model doesn't exist for its own sake, it's there to provide a better means of understanding the actual system and if it doesn't provide that, it's a failure as a model.

For hundreds of years, Newton's equations for motion were the standard model for how the universe works, but it failed to explain the behavior of the system at the extremes. Then Einstein came along and developed a better model and people used that to better explain the things Newton's model couldn't explain. You can still see Newton's original equations in any freshman-level Physics class, but when it comes to doing the heavy lifting of understanding the universe, the people doing the work are using Einstein's model.

Comments for this entry have been closed
Return to top