Monday, February 16, 2015

Why do Macroeconomists Think We Know Macroeconomics ?

This post is a comment on Noah Smith's brilliant post "Why do non-experts think they know about macroeconomics?" which is, itself, a reply to Scott Sumner's presumably brilliant post (which I haven't read and on which I am not commenting -- for one think I can't find a link in Noah Smith's post).

Noah Smith doesn't have a very high opinion or really existing macroeconomics. However, I think he is a bit too diplomatic. I think I will quote bits of his post and comment.

Sumner argued that macroeconomics uses familiar words (with unfamiliar meanings) so people think they understand it. Like Noah, I agree that this is part of what is going on. I also mostly agree with Noah that

there are some other reasons too: 1. Macroeconomics is relevant to most laypeople. String theory, to use Scott's example, is not. String theory is something you hear Brian Greene or Michio Kaku talk about, and you think "Wow, neato, the Universe is mysterious and funky!", and then you never think about it again. Macroeconomics is something related to our jobs and our investments. It affects us every day. Notice that laypeople do not often hold forth on game theory or decision theory. 2. Macroeconomics has political implications. Many people have political agendas. The "heterodox" people you meet in the blogosphere are almost all just leftists who see mainstream econ as a tool of the neoliberal oppressor, and since macro is by far the most visible branch of econ, they equate "econ" with "macro" and bash it. Or take the "Austrians", whose goal is actually to make econ into a tool of the neoliberal oppressor, for real. Then you have a whole bunch of people who aren't pushing political agendas, but who feel that macroeconomists are pushing agendas, and don't like that. In fact, some macroeconomists are pushing political agendas, though I think it's a clear minority (no, I won't name names). This is also why a lot of laypeople get involved in climate science debates. 3. There is the perception that macroeconomists don't understand their own subject. The Great Recession convinced a lot of people that macroeconomics hasn't solved any of the problems it was created to solve. Contrast that with physics or bio or chem, which have very obviously given us a lot of the awesome stuff that makes our society rich. In addition, you have very public and acrimonious debates between macroeconomists like Krugman, Cochrane, and Sumner. That convinces a lot of people that there is no consensus within macro, which in turn makes them suspect that macroeconomists haven't gotten any answers out of the Universe. If the experts don't understand anything, why can't the amateurs weigh in?
update: I see Mark Thoma found this post (this blog is for stuff I don't want many people to read update: I post stuff which I think is worth reading at -- I do put some effort into those posts end update). I wish to clarify and revise my remarks.

First, ordinary people think they know other technical subjects too. It is easy to get into debates about microeconomics (think pf discussing the incentive effects of welfare or the economic effects of international trade) climate science, food science, and military strategy. Clearly Noah's first two points are valid.

However, I think there is special disrespect for the opinions of macroeconomists. I see two quite different problems. The first (discussed vaguely below and discussed better by Noah) is that there isn't a macroeconomists' consensus on anything. There isn't even a consensus among academics (or for that matter Nobel memorial prize winners) but also wingnut welfare recipient hacks are presented to the public as prominent economists.

I think there is a main stream of contemporary macroeconomics -- New Keynesian DSGE. It matters that the implications of those models are dismissed by real business cycle theorists, austrians, market monetarists, Post-Keynesians, and modern monetary theorists (you can't imagine how much fun I had putting real business cycle theory in that list). But it also definitely matters that new Keynesian DSGE models don't explain patterns which would otherwise be puzzling and don't generate very good forecasts.

end update:

That [Noah's] is an excellent effort. I disagree with two points. First I think macroeconomists with a political agenda are a clear majority (I will name one name -- Robert Waldmann). Second, I don't really think that the Krugman, Cochrane, Sumner debate has a big effect on the opinion the general public has of macroeconomics. The reason is simply that most people don't know who Cochrane and Sumner are ( many times as many people know who Cochrane and Sumner are than know who Robert Waldmann is, but the number is still small). I think that radically different views on everything expressed by Nobel memorial prize winners does have that effect. I note (as a member of the public) that I haven't heard anything from Lucas or Prescott recently.

I realize that I have very little sense of what most people think macroeconomists say. I'm pretty sure that most people think of macroeconomics when they think of economics. I think it is a sign of the strength of the perception that we don't understand our subject that pollsters don't ask people what they think we think.

I hand the microphone back to Noah

Ryan Decker writes in response to my Reason #3 above: When I fire up my web browser I'm not bombarded with confident non-expert opinions about earthquakes, despite seismology's apparent inability to predict them.
True, but seismologists are pretty up front about this, including any seismologist in the press. I think there's a public perception that while seismologists realize their shortcomings, and are therefore probably "on the job" in terms of trying new stuff, macroeconomists might have declared premature victory.
I don't think the modesty of seismologists is the issue. For one thing I don't think that declaring premature victory deprives people of their influence. Prominent pundits and politicians all express extreme confidence that they understand the world. They are influential. I think one clear difference is that seismologists can predict lots of things, just not the thing that we care about for practical reasons. It is hard to doubt that their model of earthquakes is approximately correct. In any case, there is an overwhelming consensus -- when there is an earthquake siesmologists publicly agree about things that happened under the earth.

I think the problem here is that the analogy is much too kind to macroeconomists. It is true that macroeconomists can't predict recessions. It is also true that macroecomists almost all admit this. However, macroeconomists don't agree on the explanation of what happened. Also macroeconomic models have lots of implications which can be confronted with the data. However they don't fit the data as the implications of models of plate tectonics do.

Finally, I think point 2 it is critical. Seismology does not involve ideology (although it has gigantic implications such as get the hell out of Tokyo). Notably lots of non experts have strong opinions about global warming and evolution by natural selection.

A lot of macro people in the press express a lot of certitude about things. John Taylor expresses incredible confidence that the Taylor Rule (with coefficients of exactly 1.5 and 0.5!) is THE best monetary policy rule. Scott Sumner expresses incredible confidence that NGDP targeting is best. Paul Krugman expresses incredible confidence that fiscal stimulus is effective and that austerity is counterproductive. John Cochrane expresses incredible confidence that structural form - removing "sand in the gears" - is the best medicine for an economy in recession. Robert Lucas said that the "central problem of depression prevention has been solved." And so on, and so forth.

I think normal people realize that that certitude is basically never warranted. Yes, those economists often (but not always) have some evidence to back up their claims. But not the kind of evidence that people have in disciplines where data is more abundant, controlled, and replicable.

Here I object to the Ballance. I think there is overwhelming evidence that, when the economy is at the zero lower bound, stimulus is effective and austerity is counterproductive. I think the evidence collected in the 1930 was overwhelming. I think the evidence collected so far in the 21st century is overwhelming. I actuall challenge Noah Smith -- do you really think that Krugman's confidence in claims on those points is "incredible" ? Here's another one -- do you doubt that austerity in economies at the ZLB is counterproductive ?

On the others, Lucas said that long ago (I remember I was in the room) and wasn't widely quoted in the mass media. Again I ask how prominent Sumner and Cochrane are.

update 2: I don't know where to put this but here's some more Krugman fandom. Prominent pundits all express extreme confidence that they understand the world, but none of the others in a small sample of 26 were as successful forecasters as Krugman (pdf warning). Almost everyone must have been put off by his intellectual confidence and his eagerness to say I told you so. However, people who have been keeping track have to notice just how often he claims he told us so and how he always provides links to him telling us so. Noah and Decker agree that a problem for macroeconomists is that we have very little data and no experimental data. I think this is special pleading. Existing macroeconomics is based on the the decision that the 30s are irrelevant and Indonesia is irrelevant and Argentina is irrelevant. We have few data, and we ignore a substantial fraction of them, because they are inconsistent with our models. Even the most standard data set (post WWII US quarterly) doesn't fit the models at all. Yes it is hard to do well without decent data. But it is also hard to do as badly as macroeconomists have without decent data.

And this brings us to

Macroeconomists know more than a lot of people think they do. That doesn't mean they know a lot. And macro discussions in the public sphere tend to focus more on the contentious stuff - the stuff where no one really knows all that much. That's where normal people feel justified jumping in. If you tell them that investment is the most volatile component of GDP, they're not going to argue.
I note again that Noah is no admirer of current macroconomics, but this is pathetic update: not a convincing defence of current macroeconomics which convincing defence is not likely to come from harsh critic of current macroeconomics Noah Smith who wasn't trying to defend existing macroeconomics in that post anyway end update. The discovery is a simple measurement. People can guess that macroeconomists know the coefficient of variation of consumption and investment. People also have good reason not to bow down to the superior insight of someone who has done that calculation. A more interesting question is whether the pattern is convincingly explained by existing models (no) and whether it is easy to explain in many different reasonably convincing ways (yes).

Update: welcome noahpinioners I really really want to extend and revise this paragraph. First, it is important that Noah Smith is a very harsh critic of macroeconomics. In the post to which I link, he mentions that there are non controversial claims in macroeconomics, but one just can't expect him to be the one who comes up with a convincing defence of current macroeconmics. I didn't mean to say that no example of a statement which is accepted by almost all macroeconomists and isn't obvious -- say just a measurement exists. I just mentioned that he (unsurprisingly ) didn't present such a statement. Oh now I am in a bit of trouble, as a commenter might ask me if I, Robert Waldmann, can come up with a non obvious statement accepted by almost all macroeconomists, and especially one not accepted by almost everyone in general. I will try 1. This recession isn't the end of the world. It's mostly cyclical. We are well into a recession so growth is likely to be unusually high averaged over the next 5 years. I think every reccession a lot of people decide that this time its different and we will go into a second great depression. Macroeconomists generally believe in mean reversion. This is, in fact, a measurement with atheoretical VARs implying hump shaped impulse response functions. I don't think it is controversial anymore and it is not obvious. Now maybe it should be controversial as the current troubles seem to have lasted for a long long time. but they haven't lead to total collapse (except in Greece -- massive foreign denominated debt makes things different and macroeconomists know this).

2.Government spending and deficits do not cause low output quickly and in the short run. There is a wide range of guesses of fiscal multipliers but none is well below zero. I think the range is zero (Fama Lucas) to 2.3 (Brad DeLong of course). I think this view is not shared by lots of ordinary people and also by a signficant number of policy makers. Macroeconomists also generally guess that deficits and unproductive government spending probably reduce the trend rate of growth.

I am, if anything, more harshly critical of existing macroeconomics than Noah, but that was a good faith effort. end update.

There are statements which aren't contentious in macroeconomics because almost all macroeconomists assume they are true, when doing academic work, and almost all macroeconomists agree that they are false (but maybe models based on those assumptions might be OK because who knows anything might be). This is not the sort of consensus which wins a field respect.

I think the post is excellent. I would put a lot of stress on point 3. I'd say that, when I think as a citizen and try to guess what policies are effective, I rely on Keynes and ignore the other stuff which was written since 1937 (date of the QJE article not The General Theory). I think I am not so unusual (and I am being paid as a macroeconomist). When I think of pure economic science, I think the same way. This is very unusual, but why ?


greg said...

Well, for one thing I think you guys are all, at least to some degree, and especially the politicized ones, pretty deep in an echo chamber. The fact that the deleterious effects of bad distributions of wealth and income are just coming to the awareness of the field, and are still contentious, really reflects badly on the field. This is something that (I think) should have been well studied by now and anticipated, and instead macro is really behind the curve on it. We should already know what might happen if inequality gets worse. But we don't, and this excuses policy that will likely exacerbate it. Indeed, the whole notion that the distributions of wealth and income might be important seems to have been overlooked.

Change in the distribution of demand was an important factor in the Great Depression, and should have been noticed. Well, Keynes did notice. But why was it allowed to be forgotten? Were the alternative explanations so coherent? Or does some sort of authoritarian mind set dominate the field, quelling intellectual initiative?

Greg said...

I just want to say something about Scott Sumner comparing people opining on macro econ with him opining on string theory.

First, string theory is not trying to do anything which will materially change anyones day to day life. If string theory were to be proven, nothing would change in anyones day to day life other than a nobel prize would likely go to the physicist or mathematician providing the proof.

String theory really revolves around the grand unification of Einsteins relativity, and quantum mechanics. Einstein was doing the macro, what happens to matter as it accelerates and approaches the speed of light, while quantum mechanics is doing the micro, how do the elemental forces and elemental particles interact. Learning the fundamental nature of matter is very interesting, and Im sure some new technologies could arise using the knowledge, but its mostly an academic exercise. None of the problems humans face today require understanding these things to solve. Our problems are all a result of how we have decided to organize our societies....... period. So opining on string theory is more or less trying to decide how many angels can fit on the head of a pin, relative to the issues that macroeconomics deals with.

Secondly, macroeconomists cannot even agree on a few basic things like what is the difference between fiscal and monetary policy, what is the nature of money, how do we measure productivity, how do we measure NGDP in real time, what is a bank doing when it extends a loan, what is unemployment, while physicists do not ever quibble about what is a proton, what is a negative charge, what is the speed at which light travels.

The fact that he likens what he is doing to what physicists are doing is unbelievable really. In a sense his is more important than string theory yet he takes far less care in being accurate and is far less oblivious to real issues what happens to people when you obsessively worry about govt debt and cut their financial lifelines in the name of saving SS.

MaxSpeak said...

I read Noah all the time and agree he is very smart. But I wish he would shut up about heterodoxy since he seems to know nothing about it except for some blogs.

Anonymous said...

The "heterodox" people you meet in the blogosphere are almost all just leftists who see mainstream econ as a tool of the neoliberal oppressor, and since macro is by far the most visible branch of econ, they equate "econ" with "macro" and bash it....

-- Noah Smith

[ This is stereotypical nonsense, or just hippie bashing. ]

Nathanael said...

What the first "greg" said. It is pretty obvious to nearly everyone that distributional issues are *vital*, and the failure of macroeconomists to seriously study them for about 100 years really makes the field look pathetic.

Heck, when I'm discussing the political economy of distributional issues, I go straight to Veblen, which is over 100 years ago.

Because from then until Stiglitz / Saez / Piketty in the last few years, there's nobody much studying & writing about distribution except Jamie Galbraith -- I guess there's the Joan Robinson school of economists in England as well.

That's just *pathetic* for one of the most important topics in macroeconomics. What's happened is that the subjects of study have been chosen for political reasons, and have excluded vital ones. Why should we respect that? It's as if big money had paid biologists to study ANYTHING BUT EVOLUTION -- biology would be deeply disrespected.

EliRabett said...

"Seismology does not involve ideology"

Eli refers you to current studies on earthquakes in Oklahoma and Pennsylvania associated with fracking, and the mess in Italy over earthquake non-prediction.

reason said...

I guess you won't read this, but on the outside chance that you do, I'd like to ask you a question. Somewhere recently, you left a long post on a very old blog post from Mark Thoma's blog. I replied to it, and can't find my reply, but it regards a puzzle I have that has now come up several times.

In building a toy model to make a point you used the phrase "the (and I should write THE) rate of time preference". I have a great deal of problem with this concept - especially when it is used as a constant. It makes no sense at all to me.

At any point in time I am confronted with the choice for marginal income of spending it or saving it. In terms of the rate of time preference, I will save it if the (safe?) interest rate is high enough to compensate me for the lost immediate utility I could have gained from consumption given my (MARGINAL) rate of time preference. This cannot be a constant, or at any given time I would either save all my income or spend it according to the current interest rate. Do you actually know anyone who does this (OK, OK, permanent income makes some difference with this - but depending on how long you expect to live it is marginal)? Given these thoughts, I can't see how "the rate of time preference" is somehow different than "the rate of interest". What is wrong here?

Editor of the Fabius Maximus website said...

This is great. If revised might be among your best (jumping a high bar)!

Still, I doubt that anything compated to the mess in climate science. Some arrogant behavior by scientists plus strong political-based opposition has created a base of remarkably ignorant laypeople who despise large numbers of climate scientists (these people exist on both Left and Rughtly, although more on Right).

Also, this disrespect for scientists easily takes root in America, probably due to our long-standing anti-intellectualism. It does not take much to arouse people against the pointy-headed effete snobs in the ivory towers.

invhand said...

Thank you for the content and tone of the posts. Dr. W, Noah etc.

One difference between macro and, say, climate change is that most everyone who is shouting has to make macro decisions - How much to put in a 401K and whether to put it in stocks or bonds or cash; whether to move up or stand pat or cash out of the housing market; whether to borrow for college? What will U or i or w look like in a year? In 4 years? Everyone explicitly or implicitly makes these decisions. Most of us look for some information sources. Some 'sound knowledgeable' - which means in part that they share our assumptions and use language in a way that impresses us.

So they know.

When things don't work out? That is because EVIL has shouted down virtue.

Better to believe than to know we are alone.

Anonymous said...

I don't understand why you consistency comment, "...this blog is for stuff i don't want many people to read.)"

to me, this sounds like a euphemism that you want to write, but that you can't really be bothered to meet the standards of such other economists like professors krugman and wren-lewis.

these latter two bloggers receive a high level of confidence exactly because they are both thoughtful, but also very careful in what and how they right.

their blogs are not full of "oops, i need to update this and that."

right now, your blog "seems neither fish nor fowl".

so, in future, i would urge you take the time and effort to write to the level that you are obviously capable of.

on the other hand, if you can't be bothered, my suggestions is either to scrap the blog or to advise mark thoma not to reference to your blog.

that'll save those of us who do link over, the time and frustration that is generated from reading through your blogs.

thanks for your consideration of this matter.

Robert said...

Anonymous. I post the stuff which I think might be worth reading at This blog is for the leftovers (it has about 2% as much traffic). Some of my posts at ttp:// are the product of days of work.

Reason: Aiiiiiii A co-author and I are trying to deal exactly with that issue. By "the rate of time preference" I mean the rate of pure time preference. It isn't the marginal rate of substitution (that is the interest rate). I can only define it with an equation. In discrete time it is d in

1+d = 1+r(u'(c_(t+1))/u'(c_t)

this from the first order condition for consuming a little more now and less (with interest) later

0 = u'(c_t) - (1+r)u'(c_(t+1))

So you can have constant d and variable r with the ratio of marginal utilities of consumption making people spend more than zero but less than they could.

The reason I screamed is that I am trying to procrastinate and not revise a manuscript based on analysis of a survey in which they tried to elicit d by asking questions (in Dutch). The issue of contamination by (u'(c_(t+1))/u'(c_t) is a nightmare for me and my long suffering co-author.

reason said...

O.K. Thanks, but it only makes my puzzle worse because your definition makes it directly proportional to the rate of interest and negatively correlated with current consumption which accord to my priors. And it COULD be maintained as a constant if people adjust their consumption and expected future consumption in a particular way. But given that it is directly proportional to the rate of interest, how can people claim (as I have seen claimed) that it determines the long run rate of interest? It seems to me the determination goes the other way around.

reason said...

P.S. I still think talking about the THE (PURE) rate of time preference is nonsense. I cannot believe it is the same for everyone.

Robert said...

In the most standard macro models it is very definitely assumed that everyone has the same pure rate of time preference. I agree this is clearly not true. There is very strong evidence that, to the extent the concept even makes sense, some people have a higher rate of pure time preference than others.

This is also clearly extremely important. In fact the few models in which some people are more impatient than others are totally unlike standard models. The impatient people borrow as much as they can.

This isn't totally heterodox, but it makes a huge difference and recognizing diversity in impatience is not standard.

reason said...

Another puzzle given how you have defined it, it is unknowable isn't it (since the function u is unknown). I suppose we can make assumptions about the shape of u and fit that to point estimates, but it still seems very individual to me.

I guess my intuition is that technological and external factors (especially terms of trade) are much more important in determining (equilibrium in some sense) interest rates than time preference, because time preference can be changed by changing current consumption depending on circumstance (and more vaguely because income may move between people with different time preferences). Holding it a constant just seems to be ducking the actually interesting questions (i.e. assuming a simple answer and ensuring you get a simple but untrustworthy result).

Anonymous said...

I have been reading economics blogs since just before the financial crisis. To my great surprise I have been left with the impression that macroeconomics is a lot like scholasticism. Maybe that is inevitable, given the slow accretion of data. I expect that 500 years from now, and maybe earlier, it will be a science.

You say one thing that I am curious about, that macroeconomists believe in mean reversion. That seems likely to me, since long lived systems have strong negative feedback mechanisms. However, it is easy to confuse reversion to the mean with regression to the mean, which is a statistical artifact. Are there any studies which tease these two apart? Thanks. :)

Robert said...

There is, sad to say, a gigantic literature which attempts to tease apart true mean reversian and the statistical artifact of regression to the mean.

The number of papers written about, say mean reversion of US real GDP is vastly greater than the number of observations.

The literature is technical favoring the phrase Unit root. I think the best contribution is the abstract (which I quote in full) of

"Unit Root in GDP: Do we know and Should we Care"

Abstract: "No and probably not."

One can't ever tell if something is permanent -- will last forever. It takes forever to accumulate enough data. But, with the benefit of hindsight, it seems crazy to even try having only hundreds of observations (less than 1 kilobyte of data).

reason said...

sorry to bug again. But it occurred to me that nobody would care tuppence about the pure rate of time preference it is wasn't for Austrian economics or DSGE models, both of which I regard as bonkers or at best useless.

Anonymous said...

Dear Robert,

Thank you very much for your response. I will check out the Unit Root. :)