Saturday, February 8, 2014

Keynes, Keynesians, and Deficits as Policy

“New Keynesianism” is a gross misnomer. The macroeconomics of people like Greg Mankiw and Paul Krugman has theoretically and methodologically a lot to do with Milton Friedman, Robert Lucas and Thomas Sargent — and very little, or next to nothing, to do with the founder of macroeconomics, John Maynard Keynes.

I have to talk about a Mike Kimel post from back in February 2011 -- his Critique of Tyler Cowen’s The Great Stagnation, by way of Alex Tabarrok’s Criticism of Keynesian Politics. You can see the post at Angry Bear or at Presimetrics. The Presimetrics link is missing Kimel's very important graph, but carries some comments by Mike Kimel that the AB post lacks.

I was introduced to Kimel's post almost three years ago by Jazzbumpa's Accidental Keynesians, and by comments on two other posts at Retirement Blues: Money Illusion Delusion and A Different Look At The Great Stagnation - Pt 2.

Back then I had two problems with Mike Kimel's post: First, his definition of the term "Keynesian". And second, his view that in the 1960s policy was (by Kimel's own definition) Keynesian.

Those problems persist.

Two Problems

Jazzbumpa's recent Republican State of Disunion: Taxes Edition links to Kimel's old post. That's why the subject comes up. Here are my problems with Kimel's post:

1. "What Keynes called for was deficits when the private sector cut back", Kimel says in a comment at Presimetrics, "and surpluses at other times". Yes, that is exactly what Keynes said. But that is not what the policymakers did, who called themselves "Keynesian" in the 1960s. The Keynesians did not think like Keynes.

2. "The fact that the government started running non-stop deficits means that by the very definition of Keynesian economics, it wasn’t following Keynesian policies," Kimel adds. He's right again -- by his definition of "Keynesian". But Kimel says the non-stop deficits started around 1967 "or 70 or 73 or thereabouts (I think 67 but I’m open to persuasion about what date the graph points to)".

That *is* what his graph shows: 1967 or '70 or '73 maybe. But as Basil points out in the first comment on the Presimetrics post, "since ~1960, the US has been running a deficit nearly nonstop."

Basil is right. Kimel disagrees, but Basil is right. The "Keynesian" policymakers of the 1960s used "full employment" budgeting, trying to push economic growth up to "potential" by stimulating the economy with deficits. The 1960s was a decade of deficits.

Not only does Mike Kimel overlook the difference between Keynes and the Keynesians. He also overlooks the policy of the 1960s. He shows a graph that says the US ran balanced budgets through most of that decade.

Hey, definitions are definitions. I can deal with that. But the discrepancy regarding the deficits of the 1960s has been eating at my brain for three years.

Mike Kimel says Tabarrok and Cowen are wrong -- "wrong that Keynesian economics hasn’t been tried, and wrong that it hasn’t worked."

I don't want to argue that point. But I did want to gripe about Kimel's definition of Keynesian economics. And I do want to reconsider Mike Kimel's 1960s.

From Kimel's post:
The following graph may look a bit odd, since it has no curve on it. But it shows something cool. If I did this correctly, the gray bars show periods when:

1.  real private sector spending hit a new high and the government ran a surplus

2.  real private sector spending fell below a previous high and the government ran a deficit

Keynesian governments will generally behave in that way.  The turquoise bars show non-Keynesian behavior:

1.  real private sector spending hit a new high and the government ran a deficit

2.  real private sector spending fell below a previous high and the government ran a surplus

Here’s what it looks like:

Figure 1

The graph is wrong.

On Mike Kimel's graph, the gray areas show times when policy was Keynesian -- or, not "Keynesian", but when policy was in accord with the ideas of Keynes himself. Let's call that "Keynes-like".

On Kimel's graph, the biggest gray area is a stretch that runs from 1955 to 1967. That means policy in those years was Keynes-like. By Mike Kimel's definition, it means the private sector was growing while the government ran a surplus... or the private sector was shrinking while the government ran a deficit.

In the 1960s the economy was growing while the government ran deficits. Kimel's graph is wrong.

One of the things I know without looking is that the 1960s was a time of economic growth. Before the 1980s came along, the 1960s were often held up as the longest period of uninterrupted expansion since the Second World War.

Any chance you remember that?

Kimel's data does show that the economy was growing in the 1960s:

Graph #2: The economy was Growing in the 1960s

And yes, Kimel's data also shows Federal revenue higher than Federal spending almost all through the 1960s:

Graph #3: Kimel's Data Shows a Budget for the Most Part in Surplus

Those source numbers produce the large gray, Keynes-like 1960s on Mike Kimel's Figure 1. But these numbers would have produced a large turquoise area:

Graph #4: The Current Version of Kimel's Data Shows
Expenditure Consistently Greater than Receipts through the 1960s.
In other words, Consistent Deficits.

Here's my problem. I know the economy was growing in the 1960s. I also know that in the 1960s, policymakers used "full employment" budgeting to stimulate further growth. They used deficits. It was policy.

It was Keynesian policy. It was not Keynes-like. There were budget deficits in the 1960s, lots of budget deficits. It was policy in the 1960s to stimulate growth by means of deficits.

There is nothing iffy about this. It was policy.

I have to say this again. The economy grew in the 1960s. Policymakers at the time thought their policy of using deficits to boost growth was working. If the economy was growing, and there were deficits, then Kimel's graph should show turquoise in the 1960s, not gray. There is a problem with his graph. I can't get around it.

Graph #5: The Federal budget "surplus or deficit" was barely above zero in 1960.
It was not in balance again until 1969 under Nixon.

Graph #6: Both the gross Federal debt, and the credit market portion of it,
increase throughout the 1960s. There had to be deficits in those years.

I have to look at Kimel's spreadsheet.


Mike Kimel writes:

I would request that you immediately put up some sort of postscript indicating that a) I made the spreadsheet available to you as soon as you requested it and b) you could duplicate my results using older data, and please notify me when it is done.

He's absolutely right.

If you visit here often, you may know that I tend to split up a post into several sections. I didn't even think about it. Plus, I wrote some make-nice stuff that later got deleted. So I left a bad impression, and I apologize for that.

Not my intent, Mike. Sorry.

Even though his post is three years old, Kimel found the relevant spreadsheet and sent it to me right away. And I love the spreadsheet. It was done right, so that it was easy for me to plug in the current data. See tomorrow's post: "Kimel's spreadsheet is a beautiful thing".

Oh, and yes: I looked at the older "current expenditures" and "current receipts" at ALFRED and found data that show what Kimel showed -- budget surpluses in the 1960s. This graph:

Graph #7: Federal Budget in Surplus, except 1967 and the last bit.
I looked into that a little more. But that's my post for 11 Feb. Outcome: Kimel's numbers were good in 2011 when he wrote the post, and remained good until the "comprehensive revision" of mid-2013.

One more thing. It's not the changed data that pesters my mind. It is that Kimel used data that showed balanced Federal budgets through most of the 1960s. But the Federal debt grew in the 1960s, so I know there were deficits. Everybody knows there were deficits, I think. So I don't understand why Kimel used the data he used.

Mike, as I wrote to you this morning:

What bothers me is that I know perfectly well the Federal budget was in deficit almost without letup in the 1960s. But you selected data that (until the revision) showed the Federal budget mostly in surplus. I don't understand your choice of data.

I didn't ask for a response, because you said you are busy. Now I'm asking.

It was policy in the 1960s to use "full employment" budgets -- to use deficit spending to stimulate economic growth beyond the growth we already had at that time. As a result of that policy, there were Federal budget deficits in the 1960s.

Mike, please explain to me why you chose to use data that did not show the deficits.


Jazzbumpa said...

As I recall, Mike got his data from BEA, not FRED. I think FRED gets this data from BEA also. If there is a before - after revision situation, then that explains the discrepancy.

I doubt that it was a choice of data, but rather the data that was available at the time it was accessed.

If the revision blows up the concept of Mike's graph, then so be it.


The Arthurian said...

"As I recall, Mike got his data from BEA, not FRED."


"I think FRED gets this data from BEA also."

Probably. I couldn't tell. Sometimes I can find clear source info for FRED series, but not for these. Also, FRED seems to have several "current receipts" series and several "current expenditures" series, just to complicate things.

"If there is a before - after revision situation, then that explains the discrepancy."

No it doesn't. It's too bad, I should have done this post six months BEFORE the "comprehensive revision" instead of six months after. Perhaps then it would have been clear that the POLICY of JFK and LBJ was FULL EMPLOYMENT BUDGETS and DEFICIT SPENDING to boost growth. They were using deficits to boost growth, so the 1961-1966 period was not "Keynesian", not the way Kimel defines it. On his graph, 1961-1966 should be turquoise, nit gray.

"I doubt that it was a choice of data, but rather the data that was available at the time it was accessed."

Listen to what you are saying, Jazz. Listen to what I'm saying. The policy of the 1960s was to deficit spend.

"If the revision blows up the concept of Mike's graph, then so be it."

So be it. But then, don't keep linking to Kimel's post to support your arguments!

The Arthurian said...

...turquoise, NOT gray.

Jazzbumpa said...

First off, I wasn't aware of the comprehensive revision.

I wasn't citing Mike's post to make my argument. Nor did I rely on it in any way. In fact what I said was pretty close to a throw away comment. Mentally delete it from the post, and nothing changes.

And if the revision accounts for the discrepancy, then how can you say no it doesn't?

Your point is that there was deficit spending in the 60's. That's what the current data shows. If that is right, then Mike's graph is wrong.

Your focus on Mike'e 3-year-old graph is getting bogged down in minutia. Let's just write off the graph as either obsolete or an error, and move on. We're all human.


Mike Kimel said...

First, thank you for posting the update. As to your question: "
What bothers me is that I know perfectly well the Federal budget was in deficit almost without letup in the 1960s. But you selected data that (until the revision) showed the Federal budget mostly in surplus. I don't understand your choice of data."

When I used to blog regularly, I had a habit of trying to a) be as transparent as possible so nobody had to take my word for what I claimed (one chapter in Presimetrics, the book I co-authored with Michael Kannell, was entitled "Please Try This at Home") and b) using, as much as possible, the original source of data.

For GDP and it's components, the original source is the BEA, and there's a whole of nice data compiled neatly in the National Income and Product Accounts Tables (NIPA). For deficit data, the best source would normally be the Executive Branch, so the White House's historical tables should be it. Unfortunately, the former is based on a Calendar Year, and the latter on a Fiscal Year. Making matters worse, the start of the FY for the Federal government changed in 1976, making it a pain to use.

So I made a habit of using the NIPA tables, which the BEA converted from FY to CY. (hence, table 3.2.) That still leaves the question - do you use the "total receipts" and "total expenditures" or the "current receipts" and "current expenditures." (The former showed deficits for some years in the 1960s that the latter did not for data prior to the revision in the data).

I went back and forth, but eventually settled on using the "current" figures rater than the "total" figures. (Note - they generally were directionally the same.) Reasons:

1. Even with the revised data, Total Expenditures aren't reported until 1960. You'll notice that for the analysis we're discussing, as was usually my habit, I went as far back as there was official data, namely 1929.
2. I'm working off of memory here, but as I recall I had a few reasons for thinking "current" was more relevant than "total." The main one, I believe, was that total expenditures include "gross government investment" and current expenditures did not, and I believe I satisfied myself at some point, while digging through the numbers, that some of the gross government investment did not actually involve spending any money in the year that they are booked and/or spent. Again, it's been a while, and I don't have the time right now to verify.

Mike Kimel said...

One more thing - as per our exchange of email:

"... I hesitate to mention it because it smacks of moving the goalpost, but I wonder if the definition I used was inaccurate. By that I mean the following: very few of the cuts in marginal tax rates that we've observed were for Keynesian purposes. The 64 tax cuts were to make the economy grow faster. While the economy wasn't doing well in Reagan's early rounds of tax cuts, it was doing OK by the time of the later cuts. GW shfited his rationale for tax cuts so many times that he too was clearly was not doing it for Keynesian reasons. All three felt cutting taxes would make the economy grow faster regardless of whether the economy was doing well or poorly. Besides, even if you think they would help, you cannot push through cuts in marginal tax rates quickly enough to help in your typical recession. (Of course, you could instruct the IRS to be lenient, and possibly work with Congress to reduce payroll taxes, etc., but even that takes time.)

More to the point, Keynes felt the government was (my phrase, I believe, no his) the demander of last resort. That means: the spender of last resort. So perhaps what is needed is a measure of how much spending - and by this one clearly means domestic spending, which eliminates at a minimum a lot of military dollars - jumps relative to other years. In a Keynesian environment, private spending downturns are compensated for by the public spending upturn. I'd have to think about the "right" way to measure that issue. It could be just be a simple matter of "if private spending down by X (or X%), does public domestic spending go up by X (or X%)."

I know this smacks of the recent debate between "yes there was austerity" v. "no there wasn't austerity" in USA/Great Britain/Iceland/Ireland/Latvia/wherever, but it is important to pin down good definitions of terms like "austerity" and "Keynesian" and so forth. And sometimes, as more information becomes available, it points to holes in the definition one is using. Whether a "true" definition" of Keynesian economics would show something like the graph to in my original post or not now, I cannot say, as I haven't rooted out a better definition than I used at the time, but I think the definition I used at the time was probably inadequate.

The Arthurian said...

Thank you, Mike.

The Arthurian said...

Two years on...

Looking up federal surplus or deficit at FRED, I came upon Federal Surplus or Deficit, High Employment Budget for the United States (Q15018USQ027SNBR).

The what???

I googled high employment budget for the united states and came to a PDF:
Estimates of the High-Employment Budget: 1947-1967 at the St. Louis Fed -- and apparently from 1967 or later.

"The high-employment budget is an estimate of the national income accounts budget at some arbitrarily defined high-employment level of economic activity...
Originally the high-employment budget was developed in terms of a target for Government fiscal operations, i.e., it was suggested that budget policy be formulated in such a way as to produce a balanced national income accounts budget at high employment. More recently, the high-employment budget concept has served as a general tool of economic analysis ..."
(emphasis added)

In the early and mid-1960s, the high-employment budget was a target for fiscal operations. It told the Council of Economic Advisers how much of a deficit they should design in to the Federal budget. The goal was to have deficits that would bring GDP up and unemployment down.

The Arthurian said...

Christina D. Romer, MACROECONOMIC POLICY IN THE 1960S, page 1:

The 1960s introduced Americans to the phenomenon of persistent peacetime deficits.

The Arthurian said...

for the Romer PDF try