UPDATE: The video of this panel is here.
[On Friday, July 2, I am taking part in a panel organized by Economics for Inclusive Prosperity on “A new macroeconomics?” This is my contribution.]
Jón Steinsson wrote up some thoughts about the current state of macroeconomics. He begins:
There is a narrative within our field that macroeconomics has lost its way. While I have some sympathy with this narrative, I think it is a better description of the field 10 years ago than of the field today. Today, macroeconomics is in the process of regaining its footing. Because of this, in my view, the state of macroeconomics is actually better than it has been for quite some time.
I can’t help but be reminded of Olivier Blanchard’s 2008 article on the state of macroeconomics, which opened with a flat assertion that “the state of macro is good.” I am not convinced today’s positive assessment is going to hold up better than that one.
Where I do agree with Jón is that empirical work in macro is in better shape than theory. But I think theory is in much worse shape than he thinks. The problem is not some particular assumptions. It is the fundamental approach.
We need to be brutally honest: What is taught in today’s graduate programs as macroeconomics is entirely useless for the kinds of questions we are interested in.
I have in front of me the macro comp from a well-regarded mainstream economics PhD program. The comp starts with the familiar Euler equation with a representative agent maximizing their utility from consumption over an infinite future. Then we introduce various complications — instead of a single good we have a final and intermediate good, we allow firms to have some market power, we introduce random variation in the production technology or markup. The problem at each stage is to find what is the optimal path chosen by the representative household under the new set of constraints.
This is what macroeconomics education looks like in 2021. I submit that it provides no preparation whatsoever for thinking about the substantive questions we are interested in. It’s not that this or that assumption is unrealistic. It is that there is no point of contact between the world of these models and the real economies that we live in.
I don’t think that anyone in this conversation reasons this way when they are thinking about real economic questions. If you are asked how serious inflation is likely to be over the next year, or how much of a constraint public debt is on public spending, or how income distribution is likely to change based on labor market conditions, you will not base your answer on some kind of vaguely analogous questions about a world of rational households optimizing the tradeoff between labor and consumption over an infinite future. You will answer it based on your concrete institutional and historical knowledge of the world we live in today.
To be sure, once you have come up with a plausible answer to a real world question, you can go back and construct a microfounded model that supports it. But so what? Yes, with some ingenuity you can get a plausible Keynesian multiplier out of a microfounded model. But in terms of what we actually know about real economies, we don’t learn anything from the exercise that the simple Keynesian multiplier didn’t already tell us.
The heterogenous agent models that Jón talks about are to me symptoms of the problem, not signs of progress. You start with a fact about the world that we already knew, that consumption spending is sensitive to current income. Then you backfill a set of microfoundations that lead to that conclusion. The model doesn’t add anything, it just gets you back to your starting point, with a lot of time and effort that you could have been using elsewhere. Why not just start from the existence of a marginal propensity to consume well above zero, and go forward from there?
Then on the other hand, think about what is not included in macroeconomics education at the graduate level. Nothing about national accounting. Nothing about about policy. Nothing about history. Nothing about the concrete institutions that structure real labor and product markets.
My personal view is that we need to roll back the clock at least 40 years, and throw out the whole existing macroeconomics curriculum. It’s not going to happen tomorrow, of course. But if we want a macroeconomics that can contribute to public debates, that should be what we’re aiming for.
What should we be doing instead? There is no fully-fledged alternative to the mainstream, no heterodox theory that is ready to step in to replace the existing macro curriculum. Still, we don’t have to start from scratch. There are fragments, or building blocks, of a more scientific macroeconomics scattered around. We can find promising approaches in work from earlier generations, work in the margins of the profession, and work being done by people outside of economics, in the policy world, in finance, in other social sciences.
This work, it seems to me, shares a number of characteristics.
First, it is in close contact with broader public debates. Macroeconomics exists not to study “the economy” in the abstract — there isn’t any such thing — but to help us address concrete problems with the economies that we live in. The questions of what topics are important, what assumptions are reasonable, what considerations are relevant, can only be answered from a perspective outside of theory itself. A useful macroeconomic theory cannot be an axiomatic system developed from first principles. It needs to start with the conversations among policymakers, business people, journalists, and so on, and then generalize and systematize them.
A corollary of this is that we are looking not for a general model of the economy, but a lot of specialized models for particular questions.
Second, it has national accounting at its center. Physical scientists spend an enormous amount of time refining and mastering their data collection tools. For macroeconomics, that means the national accounts, along with other sources of macro data. A major part of graduate education in economics should be gaining a deep understanding of existing accounting and data collection practices. If models are going to be relevant for policy or empirical work, they need to be built around the categories of macro data. One of the great vices of today’s macroeconomics is to treat a variable in a model as equivalent to a similarly-named item in the national accounts, even when they are defined quite differently.
Third, this work is fundamentally aggregative. The questions that macroeconomics asks involve aggregate variables like output, inflation, the wage share, the trade balance, etc. No matter how it is derived, the operational content of the theory is a set of causal relationships between these aggregate variables. You can certainly shed light on relationships between aggregates using micro data. But the questions we are asking always need to be posed in terms of observable aggregates. The disdain for “reduced form” models is something we have to rid ourselves of.
Fourth, it is historical. There are few if any general laws for how “an economy” operates; what there are, are patterns that are more or less consistent over a certain span of time and space. Macroeconomics is also historical in a second sense: It deals with developments that unfold in historical time. (This, among other reasons, is why the intertemporal approach is fundamentally unsuitable.) We need fewer models of “the” business cycle, and more narrative descriptions of individual cycles. This requires a sort of figure-ground reversal in our thinking — instead of seeing concrete developments as case studies or tests of models, we need to see models as embedded in concrete stories.
Fifth, it is monetary. The economies we live in are organized around money commitments and money flows, and most of the variables we are interested in are defined and measured in terms of money. These facts are not incidental. A model of a hypothetical non-monetary economy is not going to generate reliable intuitions about real economies. Of course it is sometimes useful to adjust money values for inflation, but it’s a bad habit to refer to the result quantities as “real” — it suggests that there is some objective quantity lying behind the monetary one, which is in no way the case.
In my ideal world, a macroeconomics education would proceed like this. First, here are the problems the external world is posing to us — the economic questions being asked by historians, policy makers, the business press. Second, here is the observable data relevant to those questions, here’s how the variables are defined and measured. Third, here are how those observables have evolved in some important historical cases. Fourth, here are some general patterns that seem to hold over a certain range — and just as important, here is the range where they don’t. Finally, here are some stories that might explain those patterns, that are plausible given what we know about how economic activity is organized.
Well, that’s my vision. Does it have anything to do with a plausible future of macroeconomics?
I certainly don’t expect established macroeconomists to throw out the work they’ve been doing their whole careers. Among younger economists, at least those whose interest in the economy is not strictly professional, I do think there is a fairly widespread recognition that macroeconomic theory is at an intellectual dead end. But the response is usually to do basically atheoretical empirical work, or go into a different field, like labor, where the constraints on theory are not so rigid. Then there is the heterodox community, which I come out of. I think there has been a great deal of interesting and valuable work within heterodox economics, and I’m glad to be associated with it. But as a project to change the views of the rest of the economics profession, it is clearly a failure.
As far as I can see, orthodox macroeconomic theory is basically unchallenged on its home ground. Nonetheless, I am moderately hopeful for the future, for two reasons.
First, academic macroeconomics has lost much of its hold on public debate. I have a fair amount of contact with policymakers, and in my experience, there is much less deference to mainstream economic theory than there used to be, and much more interest in alternative approaches. Strong deductive claims about the relationships between employment, inflation, wage growth, etc. are no longer taken seriously.
To be sure, there was always a gulf between macroeconomic theory and practical policymaking. But at one time, this could be papered over by a kind of folk wisdom — low unemployment leads to inflation, public deficits lead to higher interest rates, etc. — that both sides could accept. Under the pressure of the extraordinary developments of the past dozen years, the policy conversation has largely abandoned this folk wisdom — which, from my point of view, is real progress. At some point, I think, academic economics will recognize that it has lost contact with the policy conversation, and make a jump to catch up.
Keynes got a lot of things right, but one thing I think he got wrong was that “practical men are slaves to some defunct economist.” The relationship is more often the other way round. When practical people come to think about economy in new ways, economic theory eventually follows.
I think this is often true even of people who in their day job do theory in the approved style. They don’t think in terms of their models when they are answering real world questions. And this in turn makes our problem easier. We don’t need to create a new body of macroeconomic theory out of whole cloth. We just need to take the implicit models that we already use in conversations like this one, and bring them into scholarship.
That brings me to my second reason for optimism. Once people realize you don’t have to have microfoundations, that you don’t need to base your models on optimization by anyone, I think they will find that profoundly liberating. If you are wondering about, say, the effect of corporate taxation on productivity growth, there is absolutely no reason you need to model the labor supply decision of the representative household as some kind of intertemporal optimization. You can just, not do that. Whatever the story you’re telling, a simple aggregate relationship will capture it.
The microfounded approach is not helping people answer the questions they’re interested in. It’s just a hoop they have to jump through if they want other people in the profession to take their work seriously. As Jón suggests, a lot of what people see as essential in theory, is really just sociological conventions within the discipline. These sorts of professional norms can be powerful, but they are also brittle. The strongest prop of the current orthodoxy is that it is the orthodoxy. Once people realize they don’t have to do theory this way, it’s going to open up enormous space for asking substantive questions about the real world.
I think that once that dam breaks, it is going to sweep away most of what is now taught as macroeconomics. I hope that we’ll see something quite different in its place.
Once we stop chasing the will-o-wisp of general equilibrium, we can focus on developing a toolkit of models addressed to particular questions. I hope in the years ahead we’ll see a more modest but useful body of theory, one that is oriented to the concrete questions that motivate public debates; that embeds its formal models in a historical narrative; that starts from the economy as we observe it, rather than a set of abstract first principles; that dispenses with utility and other unobservables; and that is ready to learn from historians and other social scientists.
Hopefully you can extend these ideas into a longer, more detailed post after the talk
Not likely I’m afraid. There are a lot of things I should be writing, and “what’s wrong with economics” posts are pretty far down the list. What I might manage to do is point to a few recent papers that are examples of the direction I’d like to head in.
Please! It would be great to know more about these -recent- example papers.
That was an interesting panel but people seem far too optimistic, in particular about the extent to which more/better empirical work will solve the problems in macro. First of all, economists do not agree on when a discrepancy between an empirical fact and a theory is grounds to reject the theory. Suppose for the sake of argument it is an empirical fact that when you interview CFOs, they tell you they do not decide their capex budgets by solving an intertemporal optimization problem. Economists would certainly not all agree that this is grounds to reject models where investment is determined as the solution to such an optimization problem.
Conversely, if you define “model” broadly, there are very many models equally consistent with the empirical facts that are commonly used to discriminate between models. These facts only select one model as the winner if one has already narrowed down the set of models under consideration (usually to models that are as close as possible to the Arrow-Debreu/RBC core). So if the universe of models is restricted to {RBC model, NK model}, the “empirical fact” that monetary policy has real effects selects the NK model. A decade later another empirical fact might select a NK model with a behavioral twist, etc. But this process only moves us away from the core one step at a time, and will never move us very far, unless radically different models are considered a priori and/or more discriminating “facts” are treated as grounds to reject.
I agree that empirical results;ts in the sense that this is usually understood, are not going to overthrow the reigning orthodoxy. Two important caveats however.
1. Economists typically use “empirics” in a narrow sense, meaning formally estimating the parameters of a model or testing the consistency of the data with a posited model. But there is a much broader set of ways to engage with observable reality, some of which may be much more powerful. I think in particular the ability to give a compelling narrative account of some important historical development can shift people’s views in a way that econometrics never will. (Larry Summers’ old article on “The Scientific Illusion in Empirical Macroeconomics is quite good on this.)
2. Regardless of the specific content of their work, the turn toward empirical macroeconomics means that there are an increasing number of younger macroeconomists in prestigious positions. Even if these people are not themselves going to develop an alternative theory, they may be more receptive to it if it comes form the outside, or at least less invested in gatekeeping.
Economic sociology and political economy (in political science) are the ideal macro that you are dreaming of.
Or, I dunno, read _Human Action_.
Isn’t macroeconomics dead simple – if you want more employment/output, spend more.
How can microeconomics have anything to say about the macroeconomy? Microeconomics is about optimal resource allocation under constraint. What has that to do with the level of spending?
*laughs in Austrian*
The curious task of economics is to demonstrate to men how how little they really know about they imagine they can design.
This is a wonderful commentary, Josh. I agree with it pretty much down the line, and you explain the points much more clearly than they’re usually presented.
The bankruptcy of the general equilibrium/representative decision-maker/intertemporal optimization framework has been in my mind recently. I was just updating the manuscript for my climate change book (finally going into production after years in purgatory) and addressing the recent literature on the distributional effects of carbon pricing. There’s meaningful research that takes the empirical structure of employment and consumption as a starting point, and then there’s the stuff the economic profession regards as state of the art, which begins with a representative household choosing leisure and employment to maximize its utility function. How will we slay this beast?
Thanks!
How will we slay this beast?
Well, probably we won’t, it will just die of old age.
I think to the extent one is going to criticize it, it’s important to do so on the grounds that it has no content as a description of real economies and no implications for policy. Not, as heterodox people have tended to, on the grounds of its lack of realism, which is very easy for them to deflect. If I write a followup to this, that’s what I will emphasize.
I do have to say, I’ve been surprised at how positive the responses to my comments have been. EFIP is hardly a radical organization and they asked me to do this and have asked me to write up a longer version for them to publish.
“Well, probably we won’t, it will just die of old age.”
Yes there’s the usual maxim about new paradigms advancing one funeral at a time. However, as is evident in the conference video the younger members have been socialized into and are not inclined to toss out the mainsteam approach. So you might be waiting a long time.
For people inside the mainstream, it might be hard to imagine what *theory* looks like without optimization and equilibrium (everyone can imagine reduced form empirics). So ideally, I think you would advance the dialectic one step by presenting/popularizing examples where non-eqm models are used to answer substantive questions. (This is already a lot of work and probably has to be a collective endeavor, maybe heterodox folks somewhere have already done all the necessary work, idk.) Once you do that, I expect the negative reactions will be ‘Lucas critique’ and ‘isn’t this just a retrogression to bad 1970s large-scale models’. At that point, you can put your example side by side with an orthodox way of answering the same substantive question and ask people why they think the orthodox way is more reliable (e.g. why the `structural’ features of the eqm model are any more likely to be policy-invariant than the non-eqm model).
Josh,
Having listened to the EFIP panel discussion, if the distribution of opinions expressed are representative of the profession at large then I think your leaning towards optimism is misplaced.
Those pushing the empirics line are still steeped in the GET paradigm and dealing with frictions.
And like you I would say the emphasis has to be not on empirics but on a basic theoretical foundation.
And good ol’ Joe raised the question of the part ideology plays, which was dismissed by the younger participants.
Well, the critique of mainstream macro is spot on. But the implicit message for structuralist/heterodox macro doesn’t work for me, so here is some comradely criticism. First, we’re not on a mission to “change the views of the economics profession” so no failure there. They are instituionally and structurally resistant (aka intellectually bigoted), and have grown more so over my career. Second, the future you envision in your last paragraph is already here. Its modern structuralist/hetero macro, which does all those things. It may not do them as well as we could, but that’s where to start building. Your recipe for reconstructing macro, starting with “data” and conversations about policy strikes me as an empiricist retreat from the method Marx laid out in the Grundrisse?
Tom,
Can you explain where in the Grundrisse Marx’s method was explained?
Yes, sorry. Its actually reprinted in the Preface to a Contribution to a Critique of Political Economy or see this link: https://www.marxists.org/archive/marx/works/1857/grundrisse/ch01.htm#loc3.
Basic idea is that investigation starts with deconstructing (to use modern terms) concepts, and working back from the abstract to the concrete. Not by starting with concrete ‘facts’. In this context, it means taking critical and comparative economic theory as one starting point.
Are you by any means the same Thomas Michl as the one who wrote the book “Macroeconomic Theory, a short course”? If so, thank you so much for writing the book. The book is probably the reason I pass Macroeconomic undergraduate this year.
Greg Mankiw was ruining my life with his books, thankfully I found your book and that has made my life much easier. The problem with Mankiw is that he over-simplifies, sometimes ignoring equations entirely. This may help in understanding the concepts, but it stumps any chance of actually solving problems, which is the way they test you. Your book avoids this conundrum by going straight to the models themselves.
Again, thank you for writing the book, I tried looking for your email on the internet but couldn’t find it, so I am offering my thanks here.
Also, the problems in the book were very nice. Are there any resources where I can find questions with their solutions similar to yours?
Yup, that was me. Glad you found it useful. I do have answers to the odd numbered problems. You can find me at the Colgate Univ. website I think.
Professor Mason,
Your analysis is a constructive critique about the situation of macroeconomics.
I agree with you that macroeconomics is dead for a long time. And there are certain reasons behind this situation. I will, by connecting your points, try to explain step by step.
Firstly, economy as you also point it out is monetary. And monetary economy creates constantly changing disequilibrium positions. Therefore, variables such as interest rate, unemployment, growth and so on evolve to ‘new normals’ continuously. The primary inference we can get from this state is that the uncertainty. And subsequently, uncertainty creates so much knowledge in which some part of it might contradict to similar knowledge that is derived in a different time period. Therefore, macroeconomists must adapt themselves to constantly changing paradigms. This is a compelling task we should acknowledge.
Secondly, uncertainty brings specialization as a byproduct. Macroeconomists today are, I assume, hyper specialized to work on certain variables. In a discipline that is characterized by uncertainty, constructing all the variables in one model is an excessively difficult task. Therefore, this trait leads to narrowly focused models instead of aggregative approach. Economists focus on certain variables today that might give them particular results compared to past. John Maynard Keynes created macro by investigating all the variables in economy by connecting them together. So, this is the point you made which is macro economy is aggregative. Or in other words, what is macroeconomics? It is the explanation of aggregate supply and aggregate demand. This is the fundamental character of macro. So it must be examined in this way.
Thirdly, Micro foundational models would not work in an economic system where there is no equilibrium. Micro modeling implicitly assumes equilibrium position. However, this is not the case in the real world. Therefore, micro foundational macro is ineffective to explain macroeconomic problems.
Fourthly, I don’t think more empirical work will help to shape a better macroeconomics. As I pointed out earlier that there is too much knowledge created by modern economists and since there is fundamental problem in theory, augmenting useless knowledge will not help. The people who created concrete hypotheses were the ones that focused on fundamentals not the ones who insisted on details.
I think the change will come sooner or later by heterodox economics.
Excellent piece. Having followed your blog for some time – and greatly enjoyed it – it does not surprise me you would come up with something like this. Some devil’s advocate thoughts:
– the orthodoxy you oppose is due to the Lucas revolution. But that revolution came from the (alleged) inability of the “hydraulic” reduced-form old Keynesian models. There is nothing in your piece about this. The few pieces I have read in heterodox macro never mention the role of expectations which was a key component of the revolution (I am glad to be shown wrong on this and pointed to the relevant papers). Once you think that expectations at least sometimes matter (e.g. for sudden stops), then the micro aspect may rear its head back.
– the standard micro based macro model allows welfare calculations. I am skeptical of many of those calculations for many reasons (e.g. there isn’t a representative agent etc.) but at least these models make the relevant trade-offs explicit and allow some cost-benefit analysis. Without these tools, a non orthodox macro model may tell us that policy X1 increases Y more than policy X2 but it may be hard to compare the relative costs of the two policies. We can give up welfare calculations altogether but there may be babies thrown away with the dirty water: e.g. I find useful that most trade models show that further trade liberalization will increase GDP by a relatively small amount while the gains from liberalizing immigration are one order of magnitude higher.
In any event, I agree with many of your points – definitely on national accounting.
the orthodoxy you oppose is due to the Lucas revolution. But that revolution came from the (alleged) inability of the “hydraulic” reduced-form old Keynesian models. There is nothing in your piece about this. The few pieces I have read in heterodox macro never mention the role of expectations which was a key component of the revolution
I am not sure about this. In settings where prediction matters, macro still uses old-fashioned Keynesian models. Look at how much discussion revolves around multipliers. Where the Lucas revolution succeeded was on the terrain of pure theory.
As far as expectations goes, what’s striking about rational expectations (or more precisely, model-consistent expectations) is that it *rules out* expectations as an object of inquiry. Whatever your model says about the path of x, it also must say about expectations of x. There is no possible role for people’s beliefs about the future as an independent factor.
the standard micro based macro model allows welfare calculations.
This is an important point that I will need to talk about if I write a followup to this piece. In my opinion, it is almost the only coherent defense of the micro founded approach. Personally, I do not think “welfare” is a topic for economics (or positive science) should be trying to answer. But defenders of the mainstream approach are right that if you want to answer it, you need to explicitly represent the choices of individuals.