« Two more random arguments for microfoundations | Main | Living in a demand-side world »

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

did you see the recent release of "the end of value free economics" ? http://www.amazon.com/Value-Free-Economics-Routledge-Methodology-ebook/dp/B007M94R60

Nathan - no, I didn't - though I was thinking of Larry Boland (he taught me advanced undergrad micro theory) when I wrote the bit about "Providing a single counter-example is enough to prove that a statement is false".

I recently suggested, in a comment on Nick's teaching inflation expectations using IS-LM post, that it seemed to be a publishable idea. So does this, and I'd suggest the same two journals I suggested there (Journal of Economic Education and Perspectives on Economic Education Research); there is a host of other alternatives as well.

Nicely put, Frances. I'm sending this to my graduate class - they write their exam next week, so the timing is very good for me.

Linda - thanks, hope they find it helpful.

Donald - thank you! Doesn't JEE also have some kind of on-line thing?

Ok, are economists using the scientific method or are they engaging in something else? I would replace the use of "assumptions" in your post, Frances, with "Hypothesis". The question is what do you do when the hypothesis is proven to be empirically wrong. If you are a scientist, you re-examine the theory, find out what is wrong, generate a new hypothesis and try again. It is an iterative method. If you don't take the final step and close the loop, you are not using the scientific method, you are practising philosophy.

Here is an example. Wage stagnation has been debated on this board and here is the chairman of Morgan Stanley Asia discussing it. (http://americanreviewmag.com/stories/The-China-complex)

America faces many tough problems in the early 21st century. But none seem as painfully intractable as those bearing down on US workers. A productivity paradox is at the heart of the problem. Despite a doubling of trend productivity growth over the past 15 years—with 2.7 per cent average annual gains since 1995, representing a stunning turnaround from the anemic 1.4 per cent pace over the 1973 to 1995 period—worker pay has continued to lag. Gains in inflation-adjusted compensation per hour (wages plus benefits) have averaged only 1.6 per cent since 1995—fully one percentage point less than productivity growth over the same time frame.

This outcome violates one of the most basic tenets of economics—that workers are eventually rewarded in accordance with their marginal productivity contribution. The Great Recession of 2008–09 has added insult to injury—pushing the official unemployment rate up to 10 per cent and more comprehensive measures of labour market slack into the high teens. Little wonder that American workers are feeling more disenfranchised today than at any point in the post-World War II era.

If you are using the scientific method, you should recognize that the micro theory that says that wages should tend to converge on the marginal product of labour is wrong. Then you go about analyzing why. If your answer is "that's impossible by theory" then you aren't using the scientific method, in which case my next question is what is the value of a polemical position without evidence? That renders the conclusion of economics hollow. Is that really what you want?

Determinant: "The question is what do you do when the hypothesis is proven to be empirically wrong."

I'm not sure what your point is, but I'm hoping that you are perhaps trying to explore some of the consequences of the points that I made above about the nature of economics, and it being a deductive discipline.

The thing is, of course, that economists use all sorts of immunizing strategies to avoid being proven wrong - see some of the authors in the book that Nathan refers to in the first post. E.g. in the example you gave - what's the evidence that wages aren't converging to the marginal product of labour? When a person and a machine together do an activity, e.g. a cashier and a scanner together check out an item in a grocery store, how do we know what is being produced by labor and what is being produced by capital?

Now the wage=marginal productivity theory might be non-falsifiable, but that's another matter entirely.

Yes, I am exploring your assertion that economics is a deductive (or formal) science.

I thought my evidence was clear about the theory of the marginal product of labour.

America faces many tough problems in the early 21st century. But none seem as painfully intractable as those bearing down on US workers. A productivity paradox is at the heart of the problem. Despite a doubling of trend productivity growth over the past 15 years—with 2.7 per cent average annual gains since 1995, representing a stunning turnaround from the anemic 1.4 per cent pace over the 1973 to 1995 period—worker pay has continued to lag. Gains in inflation-adjusted compensation per hour (wages plus benefits) have averaged only 1.6 per cent since 1995—fully one percentage point less than productivity growth over the same time frame.

This outcome violates one of the most basic tenets of economics—that workers are eventually rewarded in accordance with their marginal productivity contribution.

An engineer or physical scientist, confronted by the fact that wages and productivity gains have diverged for more than a decade, would start to re-examine the theory that workers are paid their for their marginal product contribution. Perhaps there is something else at work? That's how Modern Physics began in the late 1800's. Classical Physics started to fail for a certain number of significant experiments.

The history of Physics and Chemistry are replete with theories being discounted, rebuked and turned on their ear. Electromagnetism, Relativity, the entire controversy of Heisenberg's Uncertainty Principle and the development of Quantum Mechanics, new theories have supplanted the old.

If economists are not willing to treat your assumptions as a hypothesis and challenge them with experimental evidence, you are not engaged in science, you are doing philosophy. You're not even a social science. Where is the learning in this process? Science ultimately comes down to the real world. Lots of things *can* exist, but don't in the real world. That's what experiments are for.

It's almost as bad as Hayek's famous visit to Cambridge in the 1930's where he stated that a person buying a coat would increase unemployment (he said it would take a lot of math to prove, and his entire response was to a question after a seminar). He had nice theory but it didn't stand up to the reality of the Depression. And that is why Keynes became ascendant.

Is economics trying to learn as a discipline or is it just preaching? Is that why you use orthodoxy, heterodoxy and other terms usually confined to theology?

Determinant:

I'm not unsympathetic to your position - I get intensely frustrated by some of what I hear said about the relationship between productivity and wages.

Here's where I'm coming from:

A horse and a jockey together win a race.

What is the contribution of the horse to the victory? What is the contribution of the jockey to the victory?

It's a basically, fundamental, unanswerable question.

When people say "productivity has increased" they often mean output divided by the number of workers - or some slightly more sophisticated metric - has increased. That's totally different from the micro marginal productivity of labour concept, which is "what is the monetary value of what this person has actually produced."

Yes, it becomes somewhat like an exercise in logical philosophy - which means it is great intellectual training.

On the example you quote, there is also the argument that increased health care costs essentially ate up that entire productivity gain, which is why workers' direct, monetary compensation didn't increase.

Ok, but the solution, a partial one, is to try to isolate one actor and the other over an series of different but comparable results. Try the same jockey on five different horses, does he have a 10% better result on average? Try the same horse with five different jockeys, is the horse better?

With your health example, see where Canada, the UK, Australia and the US all stand on wages vs. output/worker growth. Let's avoid the word "productivity". All except the US have significant public health provision for working-age populations. In fact the NHS in the UK consumes the least amount of GDP (7%), Canada's public/private system (don't forget drugs!) consumes 9% and the US spends 14%. Are all countries exhibiting the same trend to lower wage gains than output/worker growth?

Or with my example, though I was using a quote for east of argument, the author states that he is comparing wages+benefits (health insurance) to output/growth, so the "health care ate my wage gain" explanation is out.

@ Determinant

The answer does not need to be “that’s impossible by theory” but instead “not all factors were taken into account which the theory requires”. I understand the chairman of Morgan Stanley to be some position of authority, but that does not mean he is infallible, nor the information provided to him complete.

Rather than challenging the foundation of micro theory, which your (Determinant) first comment suggests, Frances instead offers (through the above blog post) an explanation of economic relationships which are true by definition. Since you brought up the topic of philosophy, I think Frances above post is akin to an: “if p then q, p, therefore q” like illustration of economics.

Example: “Perfectly competitive markets are, by definition, markets with large numbers of buyers and sellers, in which firms sell homogeneous products. In such a market, no one seller will be able to raise his or her price above the market level - buyers would just go elsewhere. Hence firms in competitive markets must be price takers, and the answer is true.”

Correct me if I am wrong, but it seems to me that you are both talking past each other. One (Determinant) attacking the very foundation of economic theory (and the discipline more generally), the other offering support for logical economic deductions which necessarily follow from her particular definitions.

Am I missing something here?

Scott: I was hoping you'd read this post, because I thought it would fit nicely with your interests in econ and philosophy. In response to your comments:

"One (Determinant) attacking the very foundation of economic theory (and the discipline more generally), the other offering support for logical economic deductions which necessarily follow from her particular definitions."

As a summary of the debate, that's close. Just a couple of points to clarify:

The definitions in question are not mine ("her definitions") but rather standard ones that are taught each year by hundreds or thousands of econ profs.

I'm also not offering "support for the logical economic deductions." Certain conclusions follow from certain definitions and assumptions, that is all: as you say, "If p then q". If a student thinks "not p" the best answer to choose is "uncertain." Demonstrate "if p then q" and then point out it's possible that "not p", in which case q may or may not hold.

Ideally, by the time a student finishes an undergrad degree, he or she will be able to take a basic economic model, derive basic predictions from it, and have some idea of how to go about exploring the model (testing might be a bit overoptimistic) using real world data.

Whatever the flaws of the standard micro way of viewing the world - and Nick is arguing in the post below that micro economists can learn from macro, that microeconomics needs macro foundations, i.e. micro models should be grounded in the assumptions of imperfect competition not perfect competition - it provides simple and useful analytical tools.

The point of an exam is for the student to demonstrate to the prof that he or she understands how to use those analytic tools.

The point of this post was to try to explain this to students and professors.

Your post is interesting. I have no doubt that it will be useful for students. It made me think about certain weaknesses of current practices in teaching economics.

You first speak about the “true by definition” type of questions. I hope there are not too many of them, because definitions by themselves have only the appearance of knowledge.

It is often said that we have to find in our tool-box the relevant model. You give precise tips about how to find the right model. That would be a good step for programming artificial intelligence for answering such questions. But a student who has to rely on such clues surely lacks familiarity with the models.

Then you answer the question « True, false, uncertain: "Natural gas price controls during the late 1970s hurt producers at the expense of consumers, but did improve economic efficiency." ». Your answer is standard, but I don’t like it. I don’t like the use of diagrams (and of economic jargon). I think that it obscures rather than illuminates the question. In one sense, it is an overkill since a verbal answer would have been sufficient. But in another sense it is less than a verbal answer since it is too specific. Implicit assumptions and loose ends are quite hidden in the diagram presentation, whereas a verbal answer can more easily be connected to common sense and public debate, and thus its weaknesses can more easily be spotted. What is the likelihood that a student will notice that perfect competition is assumed, and ask to either prove that this assumption was satisfied in the natural gas market in the late 70’s or prove that your point is still valid when competition is imperfect? What student will notice that the price in the diagram does not include some externalities (like pollution)? Another weakness of diagrams is that drawing them necessitates assuming some slopes, even if the conclusion does not depend on those slopes. We surely can hope that students will understand that your conclusion about the efficiency loss triangle is still valid with other slopes (of the same sign) and that other points (like whether the consumer surplus is larger or not thanks to price control) depend on the slopes. But if the conclusion does not depend on the slopes, it seems it would be more efficient that they do not enter the discussion. They are necessarily part of a diagram, but a verbal discussion tends more naturally to either avoid them or be explicit about the various possible cases.

It is said that a picture is worth a thousand words. But it seems to me that verbal answers(*) have certain advantages. I would even go so far as to argue that models (even a simple diagram) should be taught only when they are really necessary for answering a relevant question (just using a model for the sake of using it is not relevant).

(*) I give mine on my blog: http://unintendedoutcome.blogspot.com/2012/04/natural-gas-price-controls.html

Unintended - sometimes I wonder if I'm too rigid with "you must draw a diagram." These questions weren't taken from my own exams, and perhaps the person who set the question was actually looking for an answer more like the one you gave in your blog post. These days I've taken to adding things like "draw a diagram whenever you can" and "explain using an indifference curve/budget constraint diagram" because it makes life simpler all around when students are given clear expectations.

Let me point out some of the differences between the answer that you give in your blog and the answers that I see in a typical undergraduate verbal argument.

First, you make a really good point that I hadn't thought of, and that can't be shown in the standard one period supply and demand framework: price controls may be good for future generations. If I was grading your paper, I'd circle that. If you'd have done that plus the diagram, you'd have got 20/20 or 19/20 - basically anyone who comes up with something interesting that I haven't thought of gets an A (minimum) or A+. Typically that will be one or two students in a class of 70 - though this term I'm lucky and have a few really strong students.

Second, your answer is clear and logically structured - you systemically discuss the effects on consumers, the effects on producers, and the effects on overall efficiency, with paragraphs and headings. Wonderful! Ironically, this is where diagrams can be extremely helpful - they force the writer to systemically go through all of the various impacts shown in the diagram. They're a useful way of structuring and organizing a verbal argument.

Here's a question for you: you're teaching economics to someone right in the middle of the ability distribution. Odds are that English is their second language. For an average academic economist, which do you figure is is easier, to teach a person how to write a coherent, well-organized, essay, or how to draw and explain a diagram? I agree that being able to write well is an absolutely crucial skill, and probably I should have written more in the original post about the importance of explaining diagrams. But teaching people how to write coherently is really challenging - I don't think I can do it.

It is true that part of my love for diagrams comes from the fact that they are vastly easier to mark than a two or three page written answer that lacks paragraphs, structure, or any kind of organization (length itself is not an issue, but lack of structure is brutal). Another issue is this: how do you mark an answer about natural gas controls that says nothing more than anyone with a modicum of common sense would know without ever taking an economics course i.e. producers are hurt by low prices, while consumers may benefit?

It is clear that it is easier to teach and mark diagrams. But what is the benefit? I am not sure that most of these students will actually use these diagrams in their professional life. Even as students, they may not use them to understand economic news.

Why not build on common sense? The economics course should refine common sense. The student taking such a course should be able to do better than with only simple common sense. If not, what does that say about economics?

“Producers are hurt by low prices, while consumers may benefit” is not that bad an answer. The student who gives this answer should get points for understanding that the consumers “may” (but not necessarily do) benefit, but would get a low grade, because the question of whether the possible gain by the consumers (or other agents) could more than offset the producers’ loss should be discussed. This implies discussing in more detail why consumers may benefit. This said, the question could be framed differently “what are the advantages and disadvantages of natural gas price controls?”. This is a relevant question that a policy maker could ask economic advisors. And it is easier to mark by considering what points are mentioned and which are missing, and how well the points mentioned are explained.

Unintended - The great power of diagrams is
(a) that they provide a framework around which one can structure a coherent and articulate verbal analysis. They are also
(b) better than words for comparative statistics. It's really hard to do a comparative static analysis and keep track of all of the different confounding effects without using either a diagram or math. At least, perhaps some people can do it, but I can't. Even if I'm using words, I've got that diagram going in the back of my mind.

Now it is true that people can get seduced with diagrams. E.g., undergrads often think of elasticity as the "steepness" of the demand curve, which can be seriously misleading. Grounding elasticity with words - percentage change in quantity demanded with a given percentage change in price - is better.

The problems associated with marking verbal questions are not easily resolved - it's like judging synchronized swimming (or, sometimes, synchronized sinking). Sure you can give people marks for mentioning this point or that point. But then the answers that you get are just a list of points in bullet form, which involves even less analytic and logical thought than diagrams.

this is where computational analysis comes in. To me, economic modeling should migrate almost completely to computers (note that you can stay mainstream with computational general equilibrium models). asking students to model the question and provide their analytical results would be great. you can also make a student record a macro so you can see the steps the student went through to get their answer (also a good method against cheating since you literally have a personalized recording of them taking the test).

@ Frances - Just to clarify, I was not attempting to suggest that the above definitions were your own. Instead, I was assigning gender to the field of economics and speaking on 'her' behalf. Akin to, I think, when some speak on behalf of 'truth' or 'wisdom'. Perhaps it was too much of a poetic liberty…

@ Unintended Outcome - Are you serious? There is no way you actually believe graphs degrade analytical responses, do you?

I think Nick Rowe has put it best, if I can recall his Econ 1000 VOD quote correctly: "Economics is best understood as a combination of maths, words and graphs; you need each to succeed in its mastering"

What else I recall I paraphrase going forward: Without the use of each (of the above tools) you fail to grasp any economic idea in its entirety. If you master two, and neglect the other, you fail. If you master one and neglect the other two, you fail epically (Nick Rowe CUTV VOD ECON 1000).

I think Nick is just about right.

@@Unintended Outcome - I'd love for you to expand on what you meant by
"definitions by themselves have only the appearance of knowledge."

Are you trying to make some kind of epistemological argument here?

A nice way to teach economics would be to prove Nick’s point empirically to each student. At the beginning, the student doesn’t know the diagrams and the maths, but has words, common sense and hopefully follows economic news. So let’s begin with everyday language and go as far as we can with that. When necessary we will introduce technical words, diagrams and maths (my guess is that we will go a long way without needing those). Thus, the student will personally experience the usefulness of diagrams and maths. Most importantly, the student will learn when and how to use these tools, and how to integrate them into a verbal discussion afterwards (in the end the economist usually has to give a verbal explanation to the decision-maker).

I definitely believe that models (even simple diagrams) may (but do not necessarily) worsen analytical responses.

Consider the opportunity cost of teaching models: could scarce student time not have been better used to understand the economy? Instead of teaching models and jargon, one could refine common sense. This may be more useful. Or are there any reasons to believe that the current curriculum is optimal? The usefulness of models tends to be overestimated because we use them even when they are not necessary (I gave an answer to the natural-gas price-controls question without diagrams, and I do not see anything missing there that the diagram would have provided ).

What happens in situations where no model in the tool-box applies? Maybe students who have simply upgraded their common sense (learning for example how to reduce the risk of making false analogies) will fare better than those who learned all those models that are useless in that situation. Maybe those who have learned the models have lost some of their common sense (or at least some of their skills for thinking without models) in the process and would even fare worse in that situation than if they had learned nothing. Even worse, they may not notice that none of the models in their tool-box applies and be unconsciously side-tracked by the unrealistic implicit assumptions of the model, applying to the situation a model that is not appropriate.

Paul Krugman wrote a nice book (“Development, geography and economic theory”) where he shows with concrete examples that “A successful model enhances our vision, but it also create blind spots, at least at first […] during the process of model-building, there is a narrowing of vision”. Yes, he is a fan of models (his book is in fact a plea for modeling), but he keeps an excellent peripheral vision. I do not want to overstate my case. It is not “keep away from models” it is “use models only when necessary (try first without models)”.

Something similar also applies to technical words. If some specific concept recurrently occurs in relevant economic discussions, it may be useful to give it a technical name. The teacher should not teach a technical word before its usefulness is clear. The danger is that students will only learn a jargon to speak in a complicated way about things that are either irrelevant or obvious, without noticing that they are not acquiring real knowledge about the economy.

Your diagram doesn't correspond to the question. The question implies the entire real economy. It doesn't say just the natural gas market. It doesn't say presume the real economy is Pareto optimal (which it isn't).

Now if the question were "what answer would an economics teacher expect you to give...", that would be different.

What follows is, to my mind worse, since logical truth that isn't necessarily always true, is a concept that I have a bit of trouble with.

"A more constructive strategy recognizes the fundamental nature of economics: it is a deductive study. Economics typically begin with a set of assumptions, and use these to derive a set of conclusions."

This makes economics something like plane geometry. This would be fine if it were true, but economics as usually taught is either provably untrue as a deductive science, or is true under extremely restrictive assumptions (you as an economist should be able to list them), quite unlike plain geometry (and no, the necessary restrictions DON'T resemble postulates).

The impression I always get is that economists believe economics as usually taught and the economy stand in the same relationship as thermodynamics and statistical mechanics. It would be nice, but it ain't true.


Unintended - "Thus, the student will personally experience the usefulness of diagrams and maths. Most importantly, the student will learn when and how to use these tools, and how to integrate them into a verbal discussion afterwards (in the end the economist usually has to give a verbal explanation to the decision-maker)."

If you have been successful able to teach students these skills, please contact me off-line, and send me your course outlines, exams, and any other materials you are willing to share.

Peter N - how would you recommend that an undergrad student answer such a question?

Or would you test knowledge of the effects of the natural gas price controls in the 1970s in a totally different way?

@Frances Woolley

It depends upon what you are trying to test for and what you've taught. For instance did you deal with that actual episode in class or homework? It isn't a good idea to ask questions to which you know the answer is actually quite complicated and just expect that students will apply the simplifications you intended.

I know this is difficult, but you shouldn't let teaching of simplifications get too far ahead of discussions of their limitations. Most practicing economists don't hear about the domains of validity until graduate school, after which they forget or ignore them. Economics isn't like thermodynamics. You shouldn't teach it as if it were. The development of thermodynamics from statistical mechanics is elegant and clear. Likewise the application of quantum mechanics to statistical mechanics. When a technique is applicable is always important. Perturbation methods are a good example.

It is often said that physics isn't about solving the problem you have, but rather transforming the problem you have into one you know how to solve. Without a clear understanding of limits, such a program can never work.

Knowledgeable physicists and engineers say, based on analogous problems in their disciplines, that economics is quite difficult. Economists think the difficulties are generally manageable by generally accepted methods, and politicians think economics is either common sense, a religion, or both.

Who do you think is right?

Peter N: "You shouldn't teach it as if it were."

To be clear: none of these questions are from my own exams.

What I'm doing in this post is, in part, exposing the methodological and other assumptions behind such questions. Students are routinely given - and expected to answer - t/f/u questions without any explanation or discussion of what economists mean by "true", or what these questions are actually designed to test.

Don't shoot the messenger.

B.t.w. if you're interested in knowing what I do on my own exams, here are a couple of questions from my most recent midterm:

2. Right now Canada’s GST (goods and services tax) taxes clothing, but not food. Professor Jack Mintz [sorry Michael] has recently proposed that food should be taxed under the GST. Use words and an indifference curve/budget constraint diagram to compare the efficiency of a tax on clothing and a tax on both food and clothing. Drawing from the five characteristics of a good tax system, are there other arguments for/against taxing food?

3. True, false or uncertain: “Lower taxes mean people work harder.” Your answer should include an indifference curve/budget constraint diagram and a clear explanation.

5. Why is tax evasion a problem?

The answer to 3 is uncertain, incidentally, because it depends upon the relative magnitude of income and substitution effects - something that it would be hard to explain without diagrams.

These are much better, though in 2 you seem to have implicitly assumed homothetic preferences, which is problematic.

3 is incredibly hard once you get past the superficial level. I hope you gave them a lot of time for it.

@Peter N: Okay, I'll bite: why is 3 "incredibly hard once you get past the superficial level"? Seems to me that the point of this question is that without making further assumptions on preferences, if leisure is a normal good than the effect of lower taxes on work effort - again, assuming "work" is synonomous with paid labour market work - is ambiguous. Why does going deeper into the issue remove the ambiguity (if this is what you are syaing)?

Sorry: the last word is "saying".

It was a typo on my part, I meant 5. I think the missing 4 had something to do with it. Still, 3 is fairly hard. It depends greatly on what you presume about the country in question.

In the US, the elimination of the tax avoidance industry would have a huge effect.

It matters which taxes you propose to lower - corporate taxes ("end double taxation"+-) capital gains taxes ("these aren't income and shouldn't be taxed"+-), payroll taxes ("they're regressive"+-), income taxes (whose?).

Then, how are you proposing to aggregate the result? Mean or median extra hours worked? Extra real output?

The marginal value of income varies with income (but then why do we have billionaires working 70 hour weeks?).

My gut tells me that the social effects of how taxes are collected and from whom is at least as important as level of taxation. Do people in the US work more hours because their taxes are lower. I doubt it.

You have to consider cost shifting. Say there were no taxes, then we'd have to replace the government services, where we would often be much less efficient providers.


Peter N - all of these points would be perfectly legitimate ways of supporting a (correct) answer of 'uncertain'.

I have no problem with that answer, or with the question. What did the students do with it?

On the value of assumptions:

As a timely example, when the Titanic sank in 1912, the "rescue model" the shipping industry employed was totally different from today. Ships did not have to carry lifeboats for all aboard; given the high traffic flow in the North Atlantic it was assumed that a ship could call for help when in distress and be confident of help arriving in a timely fashion. Radio and rockets (both of which the Titanic used) were depended upon to save the day.

That model failed, spectacularly. The Titanic was too far away from help (the Californian and the Carpathia), isolated by ice and couldn't be rescued.

The assumption was wrong. Thereafter SOLAS came into effect mandating that there be a lifeboat for every person on board.

Assumptions are dangerous things. Justifying design assumptions through empirical validation is a large part of what engineers do. Failure is an empirical contravention of design assumptions.

The comments to this entry are closed.

Search this site

  • Google

    WWW
    worthwhile.typepad.com
Blog powered by Typepad