Sunday, 23 September 2012

Macroeconomists: Scientists or Engineers?

Some comments on my earlier posts discussing microfoundations have mentioned Mankiw’s well known paper on the macroeconomist as scientist or engineer? This is something I’ve also wondered about, so I went back to reread the paper. Mankiw writes:

“My premise is that the field has evolved through the efforts of two types of macroeconomist—those who understand the field as a type of engineering and those who would like it to be more of a science. Engineers are, first and foremost, problem-solvers. By contrast, the goal of scientists is to understand how the world works. The research emphasis of macroeconomists has varied over time between these two motives.”

Mankiw tries to relate this distinction to the debate between Keynesian and New Classical economists. When I first read the paper, and even more so now, I find this link ultimately unconvincing. The attempt to cast Keynesians as engineers and New Classicals as scientists requires too many qualifications: there is probably something there, but it does not seem to be the critical distinction. His argument that New Keynesian theory has had little influence on policy turned out to be premature, as he himself anticipated in his conclusions.

But suppose we use the engineer/scientist dichotomy and apply it to microfoundations, rather than particular schools of thought – will that work? Microfoundations macro is the science, while those pursuing different approaches are the engineers. Is the problem with academic macro at the moment that we have too many who think they have to be scientists, and too few who try to be engineers?

I really have not made my mind up about this. One of the motivations I give in my earlier post listing reasons for departing from microfoundations focuses on policy. We have some empirical finding that at present has no clear microfoundation, but policy cannot wait for theory to develop that microfoundation. We need to explore its macroeconomic implications now. I give price rigidity as a retrospective example – it took at least a decade to develop New Keynesian theories. So we can imagine those involved with policy as the engineers, busy looking at the implications of some empirical finding, while those exploring its microfoundations are the scientists.

Certainly I have found the distinction between scientist and engineer often has resonance with macroeconomists in policy making institutions. They feel the urgency of the problem, and probably have less concern about some of the seemingly more esoteric issues that might be raised in an academic seminar. My post suggesting that the core models used by central banks should not be DSGE models ties in with this. In that context I like the scientist/engineer dichotomy, because I don’t think central banks have the resources to develop alternative modelling frameworks alone. Although we find plenty of engineers outside academic departments, we also need engineers who are academics.

But then, would it make sense to have distinct departments, of pure and applied macroeconomics? Obviously not because there are not enough of us to go round, but isn’t there something more fundamental in that observation? I write papers that build and simulate DSGE models, but I have in the past built models that are not fully microfounded, and the only thing stopping me doing both at the same time is time itself and that a major part of my job is to publish in academic journals. More importantly, in which department would you put Michael Woodford (at least when he writes Jackson Hole papers)?

This isn’t really about people as about ideas. I certainly think that anyone building non-microfounded models needs to have a thorough knowledge of microfounded models to do it well. (That would go for heterodox economists as well.) But equally, I think it is difficult to build good microfounded models without having a good knowledge of the empirical evidence. This is not just a matter of selecting an appropriate puzzle: I think a good deal of the microfoundations game is about selecting particular ‘tricks’ that allow these models to get closer to the real world, as I suggested here. While I can think of some macroeconomists whose productivity is largely unrelated to what is happening in the real world, and others – particularly in policy institutions - who do good stuff without ever seriously thinking about microfoundations, I think the majority of academic macroeconomists need to do both. In other words, to be good scientists we need to be engineers, and to be good engineers we also need to be scientists. 


  1. Economists are neither Engineers nor Scientists, as each of these fields has a significant degree of precision in what they do, and test their hypotheses in a lab.

    The better choice for Economists are "Historian" or "Sociologists."

    The sooner the profession loses its "physics penis-envy", the better off we all will be . . .

  2. The way in which economists continually try and argue that what they are doing isn't a science seems crazy to me. Economists want to predict outcomes and make measurements to test hypothesis, sounds like science to me. Sure it's hard to make controlled experiments, but it seems to me that this is no harder than in biology (i.e. field work vs lab work).

    Also, why is there so much debate about micro-foundations vs macro theories. Surely every half decent economist knows they are working with what would be described as a "complex system" ( which has simple fundamental rules that give rise to phenomena on larger scales which has seemingly no relation to the underlying rules. Sound vaguely familiar?

  3. It seems to me that a clear distinction between scientists and engineers only exists for mature sciences. Engineers are perforce empiricists. If you build a bridge and it quickly falls down under ordinary use, you can't claim that you knew how to build the bridge. Modern science grew out of engineering, not philosophy, even though it was called natural philosophy at first. Newton may have been a theoretician, but, as Feynman pointed out in "QED", Newton ground his own lenses, and learned something about light from the experience. As our current quandary shows, economic theory has not progressed to the point that there is a consensus about how to get out of it. As with the Great Depression, we have to try different things and find out what works.

  4. In most other subjects where the terms scientist and engineer are used, there is a fairly clear distinction between the two. The scientist tries to understand a phenomenon. The engineer then uses the scientific knowledge to develop something that is useful to society.

    The dividing line is usually related to the education of non-experts in the scientific knowledge. The reason for this is that the engineer normally needs significant funding, and no-one is likely to provide that funding without an understanding of the scientific knowledge and the chances of success of the proposed engineering project. Even then, the initial engineering is normal at a small scale experimental level. Only after successful experiments, is full funding provided.

    To give an extreme example, a proposal to spend billions to send a man to the moon would not get off the ground if the physics profession was split on whether the theory of gravity was correct or not.

    Macroeconomics is an exception to this mostly sequential process. The reason for this is that while society has a choice on whether to send a man to the moon, it does not have a choice on whether to run an economy.

    This means that economists and central bankers must be trusted with billions of pounds of public money even though the economics profession is split on many fundamental issues.

    Society, macroeconomics and economists would all be better off if economists had to pass the education of the non-experts stage before being allowed to contribute to policy which involves spending billions of public money. This would require macroeconomists to debate much more robustly with the public, even if just to understand what the public would expect from this education. Note that if a scientist/engineer is looking for funding from a non-expert, it is the non-expert who asks the questions and determines whether the education is successful.

    A related point is that the split in the economics profession into multiple, incompatible schools of thought means that it is virtually impossible for a non-expert to educate himself in economics without first selecting a school of thought. Any such selection is likely to be arbitrary. This is one reason why so many voters choose to “believe” in one brand of economics and ignore the others. It also why Paul Krugman wrote recently in the NYT “let’s be realistic: the public doesn’t get Keynesian economics”.

  5. Agree with Ritzholtz.

    Also, see my response post/rant:

    1. I’ve just read Noah’s post. It gives an interesting history of the use of the terms “scientist” and “engineer” in the economic context.

      To a non-expert, Noah’s post documents a history of economists arguing with other economists to prove that one economic school of thought is better than another. That’s nothing new. However, the terms “scientist” and “engineer” are irrelevant in this debate, and are not used in a manner which any non-economist would recognise. Such debate is also insular.

      I agree with those who say that economics is not an exact science, so the terms “scientist” and “engineer” are not 100% fits. However, I still think that they can be of some value.

      ALL policy making is engineering. That’s what we elect politicians to do. They try to engineer a better society. Any economist who is involved in policy making is involved in engineering. Any policy proposal, such as quantitative easing, is engineering.

      If there is a useful distinction between the terms “scientist” and “engineer” then that limits the role of a scientist by excluding current policy making. An economic scientist would be involved in looking back at history to try to disentangle causes and effects, and to disentangle facts from political dogma. He might also carry out thought experiments or observe current “natural” experiments. This would still not be an exact science, but it would be an approximation.

      Based on these definitions, “microfoundations” is a theory (or a theory about what constitutes a good theory) employed by scientists. I know it’s a theory because I can name economists who don’t believe in this theory.

      When economists argue that belief in something like microfoundations is the essence of the difference between economics as science and economics as engineering, they are completely missing the point. They are using the terms “science” and “engineering” as propaganda.

      I like reading Paul Krugman and sympathise with his views. However, in his posts he acts as "scientist", "engineer" and also “political propagandist”. He is a scientist when he looks back and uses charts to explain historical evidence e.g. the impact of past stimulus on unemployment; he is an “engineer” when he calls for more (or less) quantitative easing; and he is a “political propagandist” when he subjects one party’s proposals to detailed scrutiny and glosses over the failings of the other party.

      One of the problems for someone in Krugman’s position is that anyone who agrees with him sees his work as brilliant science and masterful ideas on engineering, while his detractors see him as a political propagandist. As a result, one of the key concepts missing from economics is “progress”. We are still debating the same things that were debated in the 1930s, with the same sides arguing the same cases.

      One of the most difficult tasks for economists is to help society disentangle “science” and “political propaganda”. Economic “engineering” can be carried out based on either science or political propaganda. “Progress” depends on science making headway against political propaganda.

  6. Ritholz has the correct view. The idea of being a scientist appeals to the vanity of Simon. But a macro economy is too complex an entity for us to model like a physical system. Have you not read Hayek's Pretence of Knowledge Mr Wren-Lewis ?

  7. Priests and engineers, as Sam Bowles(?) once said, rather than scientists. There is nothing scientific about non-falsifiability (even in principle) or a purely deductive axiomatic approach plus methodologically dubious things like calibration.

    Needless to add, I agree with Ritholtz and Noah.

    1. Correction:
      I partly *disagree* with Ritholtz.

      What economists appear to suffer from is not physics-envy but maths-envy. Open any graduate textbook on economics (eg the Mas-Colell micro text or any macro book by Tom Sargent) and compare it with a textbook on physics and another on mathematics.The essentially axiomatic development of the subject found in the economics text is similar to the development in the maths text. The physics text, on the other hand will be completely different as the whole approach is alien to physics.

      Of course, apart from mathematics, the axiomatic approach thrives in philosophy and theology. Hence Bowles's reference to economists as priests

  8. I'm an engineer by training (Chemical Engineering), and an autodidactive economist/computer programmer with about five years in banking and finance. When I'm classifying something as a science or a non-science I take da Vinci's approach--one must observe, then develop a logical model, then develop a mathematical model from theory, then test that model against new data. Economists take this approach. I would say that to improve as a science, economists need a deeper knowledge of mathematics and history than most currently have.

    Macroeconomics is much more like macrobiology and ecology than like physics or chemistry. An ecologist doesn't necessarily have much use for any one mathematical model. However, he or she needs to know the history and qualitative dynamics of the area under study, and be able to deploy the correct mathematical model for any situation. This is very different from laboratory science, and a lot more like macroeconomics than physics or chemistry.

    Economics is important in many arenas, such as risk analysis, pricing, financial projections, etc, etc. But the profession is terribly immature. Engineering, like medicine, is an ancient profession with plenty of bad actors and good actors in it's past. As a result, engineering, medicine, and their related professions are closely regulated and self-policed (we can include public accountants in this list). If I misuse my position as an engineer and put lives at risk through negligence or ignorance, I will lose my license for at least five years and possibly forever. If I sign off on fraudulent forms or botch a job for personal gain, I can serve time. If someone dies as a result of an ethical breach on my part, I will serve time, plus my family and I will lose everything we ever had.

    This is why I cannot regard economists as engineers. As an engineer, my first duty is to protect the health and well-being of the public, but there is no ethical code economists must follow. No one of them is sworn to tell the truth. Economists are routinely paid over the table and under it to make up garbage arguments in favor of their employer's approach. The profession has no shortage of liars and bad actors, and the worst punishment that will befall any of them is the possibility of appearing in a film like "Inside Job" or being interviewed by a poorly educated journalist. A disturbing irony for a pursuit that began as a branch of moral philosophy.

    So, in answer to the question. Economists are scientists. They are not like engineers, doctors, accountants, or even lawyers, because they have no code of ethics. If economics is ever to be regarded as a genuine, reliable profession, economists must do the hard and bloody work of scouring their shire and putting an effective sanctions regime in place. It may sound difficult right now, but it has been done before. Best of luck.

  9. As a engineering student, I fear there is some confusion about the nature of engineering here. Engineering is but one kind of something more generally called wizardry, or magic, that is the art of making wishes come true. In the case of engineering, these wishes are often very simple and engineers design magical artifacts that do the work whenever the proper ritual is accomplished (mostly by pushing the right button. There are more complex ritual though. The ritual to teleport mail from one place to another is quite elaborate for example). Policy is another kind of magic and many other fields. So instead of using the word engineer, it would be more accurate to use the words sorcerer or wizard for problem-solvers.
    The difference between science and engineering is the same as between economics and economic policy. It is not about being a scientist or an engineer, it is about what you do. Building a model is not wizardry, but deriving normative implication from the same model is. Engineering was institutionalized to keep scientists from confusing science and magic, and thus keep them focused on explaining the world as it is.

  10. The best quote from Mankiw's paper: "The sad truth is that the macroeconomic research of the past three decades [the 'scientific' micro-founded stuff] has had only minor impact on the practical analysis of monetary or fiscal policy. The explanation is not that economists in the policy arena are ignorant of recent developments. Quite the contrary: The staff of the Federal Reserve includes some of the best young Ph.D.’s, and the Council of Economic Advisers under both Democratic and Republican administrations draws talent from the nation’s top research universities. The fact that modern macroeconomic research is not widely used in practical policymaking is prima facie evidence that it is of little use for this purpose. The research may have been successful as a matter of science, but it has not contributed significantly to macroeconomic engineering." Modern macro is not used in policy!

  11. Historians and sociologists, they are not, but they should be.
    It's not that maths is irrelevant to economics, but it should be the icing on the cake, the part that allows for synthetic description and quantitative testing, the part that comes at the end of the reflection.

    Stop me if I'm wrong, but Adam Smith, Keynes or Schumpeter did not use maths. They just made observations and applied a logical reasoning to explain what they saw. Other economists translated that into equations. But it is not those economists doing the maths who made economics progress, it is those who had just been thinking.

    1. I'll agree with you. You don't have to use maths in order to be a scientist. Historians, sociologists, political analysts, some economists etc don't use maths (they may use statistics, like time series analysis or symbolic logic, though).
      The key element that transforms a theory into a science, is the method you use in order to create knowledge relative to an earthbound phenomenon. Actually, in my opinion there should be more discussion about all these "microfoundation" stuff and if this micro-theory is scientific (at least the marginal approach).

  12. Slightly off-topic.
    Does not SCIENCE need evidence/proof etc... before it becomes SCIENCE? Then why there are so many SCHOOL OF THOUGHTS in ECONOMICS! Does BELIEF matters in SCIENCE instead of EVIDENCE/PROOF?

    1. The main reason behind all these schools of thoughts is because there is no experiment in economics. As a result, you have to use logic and logical arguments to establish your theory. Empirical analysis is only a tool. You can prove almost whatever you want to prove. It's not easy to provide evidence and proof for every single theory. You can not use economics only based on empirical analysis. That's why it's necessary to have a theoretical framework. And that's why there are so many schools of thought.

  13. Some further thoughts.

    Science is simply the pursuit of the scientific method. Scientists make observations, record facts, make up theories out of thin air, test the theories against the facts, and then, crucially, keep those theories which fit the facts and discard those theories which don’t fit the facts.

    Any observable phenomenon can be studied using the scientific method, even the economy.

    When people argue that economics is more like history than science, I would re-phrase this and say that the study of economics uses historical facts – but so do astronomy and geology. Why are astronomy and geology sciences, whereas economics is not?

    When people argue that economics is more like magic (or similar), I would re-phrase this and say that we haven’t studies economics for long enough to understand it properly, so we need to do more science. Astronomy was still a science, albeit a primitive one, when people thought that Mars orbited the earth, but couldn’t understand why sometimes it moved forwards and sometimes backwards. At that stage, the scientists’ observations and data could not be explained by theory, but so what? That was their reality.

    People also argue that economics is difficult because economists can’t do controlled experiments. However, that is also true of astronomers and geologists. These subjects are more difficult than, say, chemistry, from that perspective, but so what?

    Economics is science if, and only if, economists follow the scientific method.

    In an earlier post, “Arguments for ending the microfoundations hegemony” (August 30), Simon listed a number of reasons in support of his argument. The first of these reasons was “empirical evidence”. Later in the post, after making a strong argument for his proposition, Simon wrote “if this all seems very reasonable to you, then you are probably not writing research papers in the macroeconomics mainstream”.

    In terms of the scientific method, Simon says that microfoundations is a theory which does not fit the facts. The scientific method would say that it should be discarded. However, Simon says that economists are unwilling to discard the theory because it would prevent them from getting papers published.

    Assuming that Simon is correct then, as another poster pointed out, this debate is really about ethics. Economists can choose to retain a theory which does not fit the facts and get their papers published, or they can follow the scientific method at the risk of not getting their papers published. It’s their choice, but it’s that choice which will determine whether economics is, or is not, science.

    1. Some fifty years ago people thought that there was a scientific method, aka the hypothetico-deductive method, by which hypotheses were proposed and tested. Earlier, induction was considered **the** scientific method. Now, people do not consider that there is a single scientific method. Each science has its own challenges, and develops its own methods.

  14. Interestingly the commenters here seem generally committed to the proposition that replicable understanding is achieved by way of either inductive or deductive reasoning.

    It appears safe to say that the vast bulk of "conventional uncertainty reduction" is accomplished via engineering professional practice and application of the scientific method to extend understanding outward from the basis of current knowledge.

    In the domain of conventional science (in Kuhn's sense) or standards-based engineering, ethical conventions emerge and accountability systems such as licensing follow to sort the bulk of good actors from the few bad ones.

    But all of this is about solving "well-formed problems" - by definition problems are circumstances to which hypothetical "advances" can be conceived and subject to either design verification or scientific falsification.

    But where then to dilemmas and predicaments fit into this schema of challenging circumstances. Predicaments are the inherently ill-formed brethren of problems - they are complex, not simply complicated situations that call for intentional action.

    Predicaments involve massively parallel inputs and, to some significant extent, unpredictable outcomes - predicaments must be navigated - generally on the basis of guiding principles and relevant experience - decision-making will be contingent on the event that just proceeds the next moment of risk-reckoning.

    The macro-economy always presents predicament circumstances to policy decision-makers. For globally engaged firms, nations, and other very large enterprises (e.g. the IMF, Fed, UN etc.) there is the matter of co-evolution between their enterprise policy intentionality and the responses of other entities plus the natural phenomenon (e.g. a global influenza epidemic or AIDs crisis, or extreme climate variations).

    There is a third form of inference about risk insights in predicament circumstances - abductive ethnography. This is the work of broad pattern discernment and coarse model building that seeks not so much to be predictive as to better inform the large-scale capital interventions with which we seek to moderate the variability (i.e. regulate rather than control) as many relatively local predicaments are interacting contingently with each other.

    The scientific basis for a professional practice of systemizing in complexity, and for the contribution of abductive ethnography resides in the principles of complementarity and indeterminacy - the insights of post Newtonian science that speak to how limits emerge in know-ability.

    It might be helpful for those bold abductive hypothesizers (the Guild of Approximation Model Makers) to distinguish their field from conventional economics with a hybrid-meme such as Econogy that acknowledges the inherent complex adaptive character of the work.

    As offensive as it might seem such practice has much in common with the practice of alchemy at the time of Newton - a case could certainly be made that throwing differential equations together in the hope that the result can prove useful often betrays no more advanced understanding.

  15. Alternatives to the neoliberal mumbo jumbo of Greg Mankiw Prof Lars Pålsson Syll hhtp://

  16. Engineering has been one of the most exciting fields to enter for several decades now. Perhaps more than any other industry, field or time in history, during this time in computer hardware engineering there has been more advancement and evolution than ever before or anywhere else. Visit our website for more details.

  17. It seems to me macroeconomics is naturally a combination of both engineering and science, no?