Friday, September 25, 2009

From Greg Mankiw's blog

How much should one care about efficiency anyway?

-- some speculations --

Economists are so preoccupied with efficiency that they twist debates away from more important aspects. Often, big social questions do not need to be answered optimally at the margin, they need the right answer.

We economists argue based on what economics is: the subject concerned with the allocation of scarce resources - the subject that teaches efficiency.

But how much should (the rest of) society care about efficiency? How important is it compared to other goals?

ex 1: the equity debate. Economists usually retreat from any debate about unjust distributions of opportunities, wealth, factors and everything else by claiming that due to the "welfare theorems" of some mathematician from long ago by the name of Walras, they do not need to care: The cake should be as big as possible and the sharing can be done afterwards. In that sense, we claim, we do equity a favour - more is for sure better, isn't it?

We solve the social optimization problem and then say a perfect market can achieve it, given the right initial distribution of "x". But nobody says out loud that we can not impact the initial allocations practically. Lump sum transfers do not grow on trees on this planet… And, of course the initial distributions are not "right". So we do not get at the real issue - welfare loss due to wrong initial allocations. The welfare loss we can get at are efficiency losses from redistributive measures - that, we are good at. But that's often not the issue. And by the way, we have no social welfare function in first place, making the initial problem unsolvable (Arrow's impossibility theorem).

That is not to say that the entire subject of economics ignores the relationship betweeen inequities and efficiency - far from. Development economists point out causal effects running both ways. But when economists are not directly concerned with distributive effects, they rest their not so troubled minds on the pillow provided by Walras and go on arguing for efficiency as if there is no other game in town.

ex 2: the climate/technology debate. What might be efficient today is maybe not so good for future developments. If one wants to retain or increase living standards and avoid climate change, one needs new production and consumption technologies.

For sure, we economists do care about growth and have all kinds of growth models; Solow, Romer, Acemoglu, what have you. A caricature of standard models is to "throw money at the R&D sector and out will come growth". And of course there is research about climate change that accounts for technological development.

But what comes out of economics into the public are mainly static efficiency arguments - we should start cleaning up where it is cheapest. Buy into CDM projects in less developed countries. Put up a market and get the prices right and stuff.

The argument for doing the most expensive things today at home since that might mean more rapid technological advances rather than (or in addition to) installing scrubbers in some ancient coal plants in faraway does exist. So does the argument that this symbolizes political will and leadership, engages the public and other countries etc. But those are fuzzy and not as mathematically clear cut as the efficiency argument. Hence they tend to be mentioned as an aside rather than the real issue.

I'm not saying that this or that argument is right or wrong. I'm just saying the latter tends to be put aside because it isn't about what we are most preoccupied with - efficiency. And that means trouble if that's not what it's all about.

That doesn't get easier if economic results are sold as if they would contain more than just efficiency: everything. The implicit or explicit believe that all arguments can be wrapped into some fabric of supply and demand crosses leads to neglecting that there is more to a social issue that can be put into economic models. Again, economics is about allocation of scarce resources. A lot of the world can be looked at that way. But not everything.

Monday, September 21, 2009

Mathematical elegance and empirical relevance – part 2

The previous post raised two questions that need to be kept separate:

  1. “How could you convince a fellow die-hard-formalistic-member-of-the-economics-tribe that the profession has gone wrong because it fell in love with mathematical formalism?”
  2. Is it true? I.e., is it possible to empirically validate the hypothesis that economists believe stupid empirical claims on the basis of empirically irrelevant mathematical proof/beauty/elegance/norms/whatever?

These are two VERY different questions. The first one bites itself in the tail: If someone relies on irrelevant and false criteria when evaluating claims, then you would have to come up with an argument that succeeds given these false and irrelevant criteria in order to convince him. This sounds logically humorous, like convincing a Christian that God doesn’t exist by making him believe God told him so. Or, in our case, making a nice, formal model that fits the intuition of economists, coheres with all established formal protocol, elegantly extends the standard framework of microeconomic choice, and that results in some absurd and silly proof that rational agents can be rationally misguided by acting and judging in the way the economists see themselves as acting and judging.

Put differently, this is a question of how to manipulate a system (in this case the one determining the beliefs of economists), how to change someone’s beliefs. It is a question of pscyhology.

The second question is – I believe – more important. There is a clear distinction (at least conceptually) between things that influence choice/thought/behavior and things that a person him-/herself believes to influence his/her choice/thought/behavior. What we would like to know is what kind of evidence/arguments/proofs economists emphasize when judging empricial claims from other economists. I am writing as I’m thinking here – always a somewhat risky proposition – but some first thoughts:

  • Ask a sample of economists for the 5 strongest arguments that support some academically based empirical conclusion they support (“What are the 5 strongest reasons you know that support your view that technological shocks drive business cycle behavior?”)
  • Find some way of categorizing articles and their mix/share of theory/empirics and sum over the references used to support policy claims by economists
  • Ask what kinds of implications economists draw from some branch of theory – and then ask what kinds of evidence they believe to be relevant for judging these (an approach Hans Olav Melberg and I have attempted in an article we are currently finishing up)

Thursday, September 17, 2009

Mathematical elegance and empirical relevance

Economicst have been accused of confusing mathematical elegance with empirical relevance. I have tended to agree, but then it struck me: What is the basis for the belief? Is it just my accumulated subjective impressions or is it something that could be objectively quantified so as to convince someone who does not agree with me?

In short: How could you convince a fellow die-hard-formalistic-member-of-the-economics-tribe that the profession has gone wrong because it fell in love with mathematical formalism? And is it true?

One could, or course, give examples. Rational addiction theory, real business cycles, rational expectations, the emphasis on microfoundations in New Keynesian economics. In addition to the problem that the die-hard-formalist will not agree with you on the particulars, there are two problems. First, picking examples do not give the whole picture. Second, it seems difficult to prove that the cause of the problem was an excessive love for formal models. It is certainly easy to get the impression that this is the case when reading these paper, but it is just an impression. I really do not know whether the author/editor was too much in love with formal models. All I can and should do with a paper - formal or not - is to point out what I think is right and wrong in it and perhaps I should just quit speculating about why they were wrong as long as it is impossible to test it more objectively?

Note the "perhaps." It may be possible to link the problems in a paper to its love for formalism, or more generally the problems of a profession, but I think we should dig deeper and not just accept that argument because we agree with it based on our own impressions. We need reasons and arguments that can convince the opposition as well!

Wednesday, September 16, 2009

Everyone and their cousins about macroeconomics

For a brief while there will continue to be a flurry of debate over the intellectual shortcomings of modern economics, centered on the perceived failure of modern macroeconomics in the wake of the financial crisis. A great list of recent contributions to the macroeconomics debate can be found here, on Mark Thoma’s blog “Economist’s View”.

Have fun while it lasts – but don’t think things will change. My prediction is that there may be a surface change in which models the profession believes and so on, but that the underlying issue of “as-if,” consistency with “standard assumptions,” over-mathematization of weird psychological assumptions, overreliance on “stylized facts” and quasi-scientific “rigor” and all the rest will remain more or less as it is. I believe the big problem is that the criteria we use to evaluate claims about the real world are severely lacking – and this is not really improved by exchanging the current models for models that a broader set of people feel comfortable with. If anything, this may impede progress as critical assessment of our subject drops as people become more comfortable with what we say.

Never underestimate the power of the dark side.

Thursday, September 3, 2009

Ketchup economists

Paul Krugman has an interesting column in the NYT with the title "How Did Economists Get It So Wrong?" (registration required). There are lots of quotable sentences (silly, absurd, crazy), but the short story is that many economists went wrong because they assumed that people always were rational and the market was efficient. He argues both elegantly and convincingly - and it is well worth a read - but it will not be earth shattering news to those who have followed the debate.

However, buried in the article are also some quotes about economists and their relationship with evidence that deserve to be repeated. First:

But there was something else going on: a general belief that bubbles just don’t happen. What’s striking, when you reread Greenspan’s assurances, is that they weren’t based on evidence — they were based on the a priori assertion that there simply can’t be a bubble in housing.


However, he also writes, more interestingly, that:

To be fair, finance theorists didn’t accept the efficient-market hypothesis merely because it was elegant, convenient and lucrative. They also produced a great deal of statistical evidence, which at first seemed strongly supportive. But this evidence was of an oddly limited form. Finance economists rarely asked the seemingly obvious (though not easily answered) question of whether asset prices made sense given real-world fundamentals like earnings. Instead, they asked only whether asset prices made sense given other asset prices. Larry Summers, now the top economic adviser in the Obama administration, once mocked finance professors with a parable about “ketchup economists” who “have shown that two-quart bottles of ketchup invariably sell for exactly twice as much as one-quart bottles of ketchup,” and conclude from this that the ketchup market is perfectly efficient.


Ketchup economists! The real problem was not that they did not use empirical evidence, but that only a certain type of evidence was admitted. This is different and it raises the question of why only some types of evidence were accepted? Ideology? Ease of quantification?

Tuesday, September 1, 2009

“Just so” stories in economics and biology

Two popular ways of making waves in Economics is by:

  • Revealing a seemingly smart thing to be dumb. Economists love unintended consequences. We love it if you can build up an argument that regulatory agencies will actually benefit the monopolistic industries they are set to serve (e.g., Stigler’s regulatory capture), that politicians are no more interested in the “public good” than business leaders (e.g., Buchanan’s Public Choice), that fiscal policy has no effects (e.g., Lucas and Barro), that minimum wages hurt the poor, etc.
  • Revealing a seemingly dumb thing to be smart. Economists also love it if you can prove that individual optimization and markets are smarter and better than non-economists believe. We love it it you can build up an argument that criminals rationally weigh costs and benefits, risks and penalties, that junkies getting hooked are implementing forward looking rational “taste-planning” (both Becker, who has a long long list of such arguments), that QWERTY-keyboards are actually better than any alternatives, that Betamax deserved to lose out to VHS, and so on.

Of course – any of these may be true. Sometimes truth happens to lie where we want it. My point here is more that I have a gut-feeling (that might well be wrong) that tells me these two types of narratives are welcomed more eagerly by economists than mechanisms or hypotheses that don’t fit the mould.

Sometimes I get the feeling that evolutionary biology has some of the same problem. This is related to the “feud” or argument between Dawkins and (now deceased) Stephen Jay Gould. Both fought for evolution and wrote excellent popular non-fiction, but Gould felt that Dawkins had too much of a “full optimization equilibrium” spin on reality around us.

An example of the latter is the paper discussed briefly here. This paper argues that depression – which seems like a dumb thing for an organism to fall into – is actually really really smart.

When one considers all the evidence, depression seems less like a disorder where the brain is operating in a haphazard way, or malfunctioning. Instead, depression seems more like the vertebrate
eye—an intricate, highly organized piece of  machinery that performs a specific function.