A defense of interwar macroeconomics
Gauti Eggertsson makes the following claim:
As a discipline, macroeconomics was born in response to the Great Depression, giving rise to Keynesianism; the rational-expectations revolution in macroeconomics was born in response to the great inflation on the 1970s.
I don’t wish to contest this claim, at least not directly. I agree that the Depression led to Keynesian economics and that the Great Inflation led to the rise of monetarist/Lucasian ideas. And there is a sense in which macro was born in the Depression. It became a separate field with highly specialized practitioners. But there’s also a sense in which that claim is misleading, and misleading in an interesting way.
Interwar (pre-General Theory) economists had all sorts of recognizable models of the macroeconomy. Today many look ad hoc, but I’d argue they were appropriately ad hoc. Here’s a couple models that form my vision of macro:
MB*V(i) = NGDP, where V is positively related to the 5 year bond yield.
I think it’s a pretty good model, even though other variables like marginal tax rates also impact V. And I think most of the better interwar economists had a model something like this in the back of their minds. Or take this model:
(Hours worked)/ (natural rate of hours worked) = f(W/NGDP), where high relative (not real!) wages reduce hours worked.
Again, many interwar economists assumed a similar sort of sticky-wage model. Fisher was the most advanced; he created a Phillips Curve model in the early 192os.
In my view lots of modern macroeconomists overlook these models, as their vision of a macro model uses a general equilibrium approach, where monetary policy works through the liquidity effect and sticky prices are more important than sticky wages. Something like IS-LM.
Because my preferred approach is similar to that of the interwar economists, I am naturally more likely to treat these early models with respect.
Evan Soltas makes the following claim about improvements in data:
In the beginning, there was no economics data. Greats like Adam Smith wrote treatises on political economy in the equivalent of near-total darkness. Later economists such as Vilfredo Pareto and Alfred Marshall introduced mathematical foundations, changing the direction of what had been a very qualitative philosophical endeavor. After the Great Depression, Paul Samuelson and John Hicks consolidated Keynes’ work into the modern field of macroeconomics — and they received critical (and I might argue significantly under-appreciated) support from econometricians and statisticians like Simon Kuznets.
Kuznets developed the United States’ program of national income accounting — from which the ubiquitous measure of GDP comes — and more broadly, he put heavy emphasis upon data collection. That enabled empirical analysis and complemented economics’ ever more quantitative bent.
Call Kuznets’ revolution the First Generation of economics data. Much of it was low-frequency, with figures released on yearly and quarterly bases. Only some data, largely from labor markets and prices, came out with greater frequency. In large part, data was supplied from government bureaus of statistics and industry groups — a highly centralized model of collection and distribution. And the supply of data was scarce, with each figure an expensive undertaking.
I think we are approaching a Second Generation of economics data. The model is changing, a trend driven by information technology.
I spent several decades immersed in interwar macroeconomic data. And what surprised me most was the high frequency nature of the data, and the fact that it was much more available in “real time” than modern data. I recall during the first quarter of 2011 and the first quarter of 2012 that there was a lot of uncertainty about the macroeconomy. Some data such as ISM numbers and employment numbers showed strong growth. But the actual real GDP numbers (released in late April) were quite disappointing (and later revised.) That wouldn’t have happened in the interwar period.
Christopher Hanes has shown that the interwar economy was very heavily dominated by commodities, and by manufactured goods that are not highly finished (things like steel.) It’s not at all difficult to ascertain the contemporaneous prices of these goods, as they are often traded in auction-style markets, or at the very least (in the case of steel) are relatively homogeneous, and hence the law of one price is approximately true. The WPI (forerunner of the PPI) was actually reported weekly, with very little lag. Admittedly it didn’t include all prices, but it did include the most volatile prices. Hence it did a good job of picking up surges of inflation or deflation.
The data for real output was dominated by agriculture and manufacturing. That’s a weakness in that it ignores services, however:
1. Agricultural and manufacturing comprised a far larger share of GDP during the interwar years.
2. Manufacturing (including mining and utilities) was by far the most volatile part of GDP.
Put those two together and the monthly industrial production numbers gave a pretty good read on the business cycle, far better than the modern IP numbers. And of course they were available much more frequently than our modern GDP figures. And there was even more data available at weekly frequencies, which were highly correlated with the macroeconomy. Steel production, and even more importantly rail shipments (at a time when most goods were shipped by rail.)
Today we are dominated by sectors like finance, consulting, health care, education, software, online journalism, law, accounting, etc, where I don’t even have a clue as to how we should measure “real output” and I don’t think anyone else does either. Under those circumstances the monthly employment data is probability our best cyclical indicator. Unfortunately we have two employment series, and they often diverge sharply.
When I was young I also had a sort of “Whig view of history.” Now that I am a grumpy old reactionary, I no longer think we are evolving toward the right model of the macroeconomy, or the perfect data set. Indeed in both areas I see us regressing, moving ever further away from the golden age of Calvin Coolidge.
The only thing that gives me hope is people like Evan Soltas and Yichuan Wang.
Tags:
30. July 2012 at 06:06
This is WHY our bet is potentially such a big hairy deal:
“As a discipline, macroeconomics was born in response to the Great Depression, giving rise to Keynesianism; the rational-expectations revolution in macroeconomics was born in response to the great inflation on the 1970s.”
Because IF I win, than we must establish a newer form of macro thinking that ACCEPTS as true the hegemony.
Realpolitik Macro as it were.
IF I’m right, and Obama loses, then the right + CB have a macroeconomic policy grip on shrinking the wealth transfer side of government, pushing us further to free market by forcing liberals to re-think HOW they achieve their ends at the very core.
It isn’t enough for Sumner to admit he was blind-sided by the Tea Party.
IF 100% GDP debt + less than 2% inflation = Private Corporate power INCREASES as govt. crumbles under its own weight, forcing ALL Dem Presidents to either cut deficits or lose elections, which in turn dis-enfranchises the hand-out side of the electorate.
THEN we have a cultural experience as great as the Depression and the 1970’s to internalize.
BECAUSE then we have a trend like this:
1. GD – very liberal response
2. 1970’s Moderate-Conservative Response
3. Nov 2012 Obama loses – RADICAL CONSERVATIVE response
This would connote a trend that all future crisis should ASSUME as true, or their predictive value is worthless.
If I win, Scott credits me in his book.
But if I win, the entire field of Macro MUST see what hegemony means to Monetary Policy.
If Scott wins you all can go back to insisting MT and Money etc. is like GOVT. a Democratic good.
It is a great bet.
30. July 2012 at 06:15
I can think of a couple of good reasons why we are not getting some services data in real time. Private insurers often wait six monts or more, to get billing specifics sorted out from healthcare practitioners. Patients are constantly put in a bind when they need to present medical expenses for compensation in specified timelines for Medicare and Medicaid, because it takes the billing months to show up from the providers.
30. July 2012 at 06:37
Why and to what extent does your current view of history (people becoming more sensible, utilitarian progress) differ from the Whig view?
It’s really hard for people to reverse the trend you’re seeing in Macro, because it’s really hard for most people (and most economists) to believe that the Fed isn’t doing all it can to unambiguously improve the economy. The problem must lie elsewhere.
30. July 2012 at 06:40
Becky, Good point.
Saturos, I do believe in the Whig view of history, just not in macro.
30. July 2012 at 06:50
Silent Cal. Greatest Vermonter ever.
30. July 2012 at 07:47
[…] Sumner notes that the data is becoming more complicated, and therefore more difficult to dissect and understanding. He links to a post by Evan Soltas, […]
30. July 2012 at 08:01
What element of Keynes didn’t exist prior to 1930?
“I agree that the Depression led to Keynesian economics.”
Foster & Catchings work on the inadequacy of aggregate consumer demand was all the rage — in the 1920s. Hoover & Sen. Wagner of New York were huge fans, and every macroeconomist in the world was well aware of the work, and all of them were utterly unable to find a flaw in its argument.
Keynes was making Keynesian arguments for government spending and government inflation from the mid 1920s — when the Great Depression began in Britain — and he had his idea of the inadequacy of aggregate demand in his writings from at least 1930, before America’s downturn had become The Great Depression.
And Alvin Hansen was pushing Keynesian secular stagnation arguments from an early period, and similar ideas were popular in Chicago.
See David Laidler on _Fabricating the Keynesian Revolution_ for more on this topic.
There is hardly an economist under the age of 60 in America who knows any history of economic thought.
Economists need to stop pretending that they do.
What economists take to be “Keynesian” economics was created in the 1940s by Lerner, Kaldor, Samuelson, Hansen and others.
It wasn’t a product of the Great Depression, it was a product of post-war big government math economists.
30. July 2012 at 08:24
Scott, do you assume that V varies inversely with marginal tax rates?
30. July 2012 at 08:55
When the early data sets were collected their audience was small. As more poeple began to dicuss the economic data it became political. It is rumored the LBJ would regurlarly send his economists away because the facts didn’t fit his version of the truth. I doubt that Nixon was much better.
Then in the great inflation, goverment spending and union contracts became indexed for inflation. This gave CPI futher political importance. If CPI was overstated or underastated this was not just an acadmeic exercise. Real dollars were at risk.
Do we have a problem of Heisenberg uncertanty? Merely looking at the numbers will cause them to change?
Regarding the establishment vs. hosehold payroll surveys. The establishment survey has a margin of error of 100k and the establisment survey has a margin of error of 400k, however a 20k miss on will cause no end of excitment among the market makers, fed watchers, academic and journalist communities.
30. July 2012 at 08:58
One more problem I have with the economic data is the endless revisions. Some of the data can be revised years after the fact.
The old data is thrown down the memory hole. At the very least those distributors of economic data should store a the data series as orriginally reported side-by-side with the revised data.
30. July 2012 at 09:18
After observing mainstream economists for quite some time, I think I am justified in declaring that our age can be characterized by a prevailing fundamental misconception regarding the role of theory (economics) and history within the social sciences.
As regards the natural sciences, theory and history are treated as equivalent. This is no accident. It is due to an inference of constancy in the operations of causes as such, by way of being an actor. Being an actor constrains our view of the natural world in this way.
But this inference of constancy cannot be redirected back in on the actor themselves. For in order to learn, the actor necessarily presupposed non-constancy in action.
Thus action, which is the subject matter of economics, strictly delineates theory and history into two realms of inquiry. On the one hand, we have history, namely, what people thought and did in the past. This is what is called “data”. On the other hand, we have theory, which is our mental tools used to understand historical data.
The lack of grasping the difference between theory and history in the social sciences, has led to the belief that central planning can be improved upon if only the central planners had access to more “data”. But no amount of historical data can ever reveal constancies in human action, since the very learning of such “improving” history presupposes non-constancy in the actor who studies it.
30. July 2012 at 10:31
Ah yes, there is no such thing as the hegemony:
http://www.gallup.com/poll/156347/Americans-Next-President-Prioritize-Jobs-Corruption.aspx
30. July 2012 at 16:41
Greg, I’ve read Laidler’s book on Keynes, and he believes there really was a Keynesian revolution, as is obvious from the title.
Tommy, Yes, base velocity is definitely inversely related to MTRs. I did my PhD dissertation on that subject.
Doug, Yup, modern economic data isn’t very good.
30. July 2012 at 20:00
I agree with your last sentence, Scott Sumner. Young and promising people like Evan Soltas and Yi-Chuan Wang give me hope for the economics profession.
31. July 2012 at 09:19
Here’s a video that captures the problem with statistics;
http://www.youtube.com/watch?v=V1Ze_wpS_o0
31. July 2012 at 09:55
Wow. So why do people want to hold more cash when MTRs go up?
31. July 2012 at 10:07
Scott,
You are far more familiar with the interwar period and the Depression than I am, obviously. I suppose my point, better phrased, would be that the Keynesian revolution put new emphasis on macroeconomic management enabled by Kuznets et al.’s stats.
But I don’t see how you can downplay the potential of Big Data in the economy and in the academic discipline of economics. I’m absolutely convinced that this is coming — see this post I wrote a while ago about personal analytics and data, which you should really check out. (http://bit.ly/OzhNKt)
31. July 2012 at 12:10
Right — a revolution which took place in the work of Lerner, Hansen, Samuelson & Kaldor in the 1940s, and which was in large measure constructed out of ideas & ways of thinking that were in currency already in the 1920s.
Note well, the title of Laidler’s book is _FABRICATING the Keynesian Revolution_.
“Greg, I’ve read Laidler’s book on Keynes, and he believes there really was a Keynesian revolution, as is obvious from the title.”
What we have is a bogus mythology that there was no macroeconomics prior to 1936 — which is non-sense and patently false, and a false claim that the “Keynesian” revolution came fully formed via Miraculous Conception from the heavens in 1936, which is equal non-sense.
31. July 2012 at 12:14
Can we agree that the standard mythology of the history of macroeconomics is just that? A massive false mythology constructed for the narrative purposes and convenience of current professors, and not for the purpose of telling the truth and getting the history right?
31. July 2012 at 17:28
Saturos, To evade taxes. Cash is anonymous.
Evan, That’s an interesting post, although the vision has a dystopian “1984” feel to someone of my generation. (Loss of privacy, etc.)
I still feel the current data is much worse that the interwar period, but I can’t deny that it might get much better in the future. The real challenge is separating NGDP into prices and RGDP. Economists don’t have good models for doing so, and I still think we are a long way from figuring out “utility.”
Greg, I don’t think Laidler would agree with your characterization of Keynes. Obviously we all three agree that most modern economists know nothing about the interwar period.
1. August 2012 at 23:37
Thomas Humphrey wrote an excellent 250 year literature survey of the rules versus discretion debate in the 1998 Richmond Fed Quarterly.
He wanted to know if macroeconomics was a progressive science in the sense that superior new ideas relentlessly supplanted inferior old ones.
Humphrey found that:
– Keynesian ideas about a lack of demand and their great many antecedents gain currency when unemployment was the main concern.
– Monetarist ideas tended to reign when price stability was the main problem.
The policy debate keeps recycling because
1. people forget the lessons of the past and
2. politicians and the public have tended to believe that central banks, the focus of his studies, have the power to boost output, employment, and growth permanently.
Humphrey showed that stable rules are popular in good times to contain inflation, and when unemployment was rising, discretionary monetary policies returned to policy vogue.
Humphrey concluded that doctrinal historian knows that much of what passes for novelty and originality in monetary theory and policy are ancient teachings dressed up in modern guises.
Haberler, Jacob Viner and Frank Knight pointed out in their reviews of the General Theory that it lacked originality, and presented old ideas, often incorrectly in confusing new vocabularies that made the book difficult to read.
To be fair, under Stigler’s rule of scientific priority, Keynes deserves credit because he made sure that the old scattered ideas stayed discovered.