Sunday, 22 April 2012

Microfoundations and Evidence (2): Ideological bias

               Internal consistency rather than external consistency is the admissibility criteria for microfounded models. Which means in ordinary English that academic papers presenting macroeconomic models will be rejected if some parts are theoretically inconsistent with other parts, but not if some model property is inconsistent with the data. However the motivation for a paper will often be a ‘puzzle’, which is an empirical fact that cannot as yet by explained by a model. However the paper is not required to be consistent will all other relevant facts, so external consistency is not as important as internal consistency.
               In a previous post I expressed a concern that researchers might tend to choose puzzles that were relatively easy to solve, rather than puzzles that were really important. In this post I want to raise another problem, which is that some researchers might select facts on the basis of ideology. The example that I find most telling here is unemployment and Real Business Cycle models.
Why is a large part of macroeconomics all about understanding the booms and busts of the business cycle? The answer is obvious: the consequences of booms – rising inflation – and busts – rising unemployment – are large macroeconomic ‘bads’. No one disagrees about rising inflation being a serious problem. Almost no one disagrees about rising unemployment.  Except, it would appear, the large number of macroeconomists who use Real Business Cycle (RBC) models to study the business cycle.
In RBC models, all changes in unemployment are voluntary. If unemployment is rising, it is because more workers are choosing leisure rather than work. As a result, high unemployment in a recession is not a problem at all. It just so happens that (because of a temporary absence of new discoveries) real wages are relatively low, so workers choose to work less and enjoy more free time. As RBC models do not say much about inflation, then according to this theory the business cycle is not a problem at all.
If anyone is reading this who is not familiar with macroeconomics, you might guess that this rather counterintuitive theory is some very marginal and long forgotten macroeconomic idea. You would be very wrong. RBC models were dominant in the 1980s, and many macroeconomists still model business cycles this way. I have even seen textbooks where the only account of the business cycle is a basic RBC model.
But perhaps common sense here is wrong, and the RBC approach is right. Perhaps, despite appearances, high levels of unemployment in a recession are just people choosing to enjoy more leisure. Unfortunately not. One of the really robust findings revealed by happiness data (see here for a recent comprehensive survey) is that unemployment increases unhappiness. As Chris Dillow notes from some recent research, unemployment appears worse than divorce or widowhood, in the sense that the happiness of the unemployed does not adapt over time to their state. Given the future earnings loss implied by spells of unemployment documented here, this is not that surprising. It is also not surprising that quits (voluntary exits) from employment are negatively correlated with unemployment, which is also difficult to rationalise with the RBC approach.
Now the RBC literature is very empirically orientated. It is all about trying to get closer to the observed patterns of cyclical variation in key macro variables. Yet what seems like a rather important fact about business cycles, which is that changes in unemployment are involuntary, is largely ignored. (By involuntary I mean the unemployed are looking for work at the current real wage, which they would not be under RBC theory.) There would seem to be only one defence of this approach (apart from denying the fact), and that is that these models could be easily adapted to explain involuntary unemployment, without the rest of the model changing in any important way. If this was the case, you might expect papers that present RBC theory to say so, but they generally do not. New Keynesian models are RBC models plus sticky prices, but that plus bit is crucial. Not only does it allow involuntary unemployment, and therefore a role for policy to smooth the cycle, but it also changes other properties of the model.
What could account for this particular selective use of evidence? One explanation is ideological. The commonsense view of the business cycle, and the need to in some sense smooth this cycle, is that it involves a market failure that requires the intervention of a state institution in some form. If your ideological view is to deny market failure where possible, and therefore minimise a role for the state, then it is natural enough (although hardly scientific) to ignore inconvenient facts. For the record I think those on the left are as capable of ignoring inconvenient facts: however there is not a left wing equivalent of RBC theory which plays a central role in mainstream macroeconomics.
In this and a previous post I have looked at two biases that can arise in the puzzle selection that drives microfoundation model development. There may well be others. Do these biases matter? I think they do for two reasons. First from a purely academic point of view they distort the development of the discipline. As I keep stressing, I do think the microfoundations project is important and useful, but that means anything that distorts in energies is a problem. Second, policy does rely on academic macroeconomics, and both the examples of bias that I use in this post and the last could have been the source of important policy errors.
One way of reading these two posts is a way of exploring Krugman’s Mistaking Beauty for Truth essay. I know the reactions of colleagues, and bloggers, to this piece have been quite extreme: some endorsing it totally, while others taking strong exception to its perceived targets. My own reaction is very similar to Karl Smith here. I regard what has happened as a result of the scramble for austerity in 2010 to be in part a failure of academic macroeconomics. It would be easy to suggest that this was only the result of unfortunate technical errors, or political interference, and that otherwise the way we do macro is basically fine. I think Krugman was right to suggest otherwise. Given the conservative tendency in any group, an essay that said maybe there might just be an underlying problem here would have been ignored. The discipline needed a wake-up call from someone with authority who knew what they were talking about. Identifying exactly what those problems are, and what to do about them, seems to me an important endeavour that has only just begun.


  1. No one disagrees about rising inflation being a serious problem.

    That makes no sense. I know you know the difference between level and rate (and this is clearly about rate.) Many economists today are promoting more inflation in Europe and the US these days, I know you're reading them (aren't you one of them?) And I am sure you know that rising inflation in, say, a period of deflation is not a serious problem, quite to the contrary.

    Now if you mean hyperinflation... but that's about level and not what you're saying.

    1. It's not only wrong, it's a lie. Perhaps 'Mainly Macro' could start with Krugman, and work from there.

    2. The post does make sense.

      You seem to have interpreted it as a claim that rising inflation is always bad whatever the context, and I don't think this is what Simon meant.

      What the post says is not at all inconsistent with promoting inflationary policy during a recession in an effort to reduce unemployment. To say that rising inflation is problematic is not at all to say that the Phillips curve trade-off should never be exploited.

      I also partially disagree with Nick Rowe. I think central banks are very much concerned with inflation, and so are business cycle economists. I do agree in the sense that there is less of an appreciation of the real effects of monetary policy amongst the general public.

  2. Suppose you took an RBC model, and bolted on (say) an efficiency wage model of involuntary unemployment? Would it really change the other results much?

    In other words, is the empirical problem the *existence* of involuntary unemployment, or the procyclical *fluctuations* in involuntary unemployment?

    I had the same reaction as Anonymous above. Inflation seems to be seen as a bigger problem by the public than by economists. Just last year I had to argue with a blogger who said that a looser monetary policy would be bad, even if it cured the recession, because it would increase inflation and reduce real wages.

  3. Thanks for the summary of the current state of play -- how seriously RBC is taken, how it is typically presented, etc. This is a lot of work to figure out from outside the profession.

    Given your description, I have to say that unfortunately I can't muster any respect for the RBC theorists or for the disciplinary forums that give them a respectful hearing. Something is terribly wrong with this discourse.

    Counter-intuitive results are often interesting and useful. But ignoring (or it sounds like actually rejecting) a central fact -- that employment in recessions is largely involuntary -- is self-indulgent and intellectually dishonest. Professions that want to maintain credibility need to limit that kind of behavior.

    Given comments by RBC advocates that I've seen elsewhere I doubt they'll pipe up with a claim that they *do* accept that a lot of unemployment is involuntary. But if they did I'd be interested.

    Barring something like that, economists who want to maintain credibility as scientists will have to publicly reject RBC and similar violations of scientific ethics. It might be useful to look at debates in physics (for example about string theory) in which physicists have taken very strong positions, motivated by much less serious ethical issues.

    Given what you say about the state of RBC, do you really feel that an economist who cares about empirical validity can honestly support it? If so, how? If not, then I think explicitly recognizing that the supporters are dishonest would be a good first step.

  4. I'm not an economist (preface over) and it's fascinating (though depressing) to hear that a theory such as RBC is taken seriously as a policy instrument. It sounds like precisely the sort of 'science-mimicking' method of abstraction criticized by Marxists and perhaps others.
    @jed. I wonder whether the 'violation of scientific ethics' you describe is really a violation at all. Abstraction is, to my limited udnerstanding, a key part of scientific method. Couldn't it be within the application of this 'science-like' method to social policy that the 'unethical' kernel lies? (That's if ethics is your bag).

  5. A monumental, integral -- and tragic -- example of libertarian bias is how in economics the concern, the goal, is almost always Pareto optimality rather than total societal utils optimality.

    A common reason given is that it’s hard to measure total societal utils. Well, it’s hard to measure lots of things, so is it then optimal to completely ignore them, and just go 100% with random luck for our strategy in optimizing things that are of great importance to us? We can’t improve on that? When you choose a city and a home, do you say it’s hard to estimate the expected utility of a home, so I’ll just throw a dart at a map of the world and decide that way? I’ll put no thought into it whatsoever, and completely ignore the great deal of important information that I do have, and can get? Of course not.

    Plus, you think it's hard to rate the utils people get from goods, but it's not hard to rank their preferences for any and all bundles? You're not ok with a function that estimates their utils, but you are ok with one that ranks any and all bundles of goods? You're ok with saying a function like lnX does this, THAT'S Ok?

    Clearly, it’s hard to get a more pro-libertarian and anti-utilitarian bias than saying I’ll almost always make Pareto optimality a central focus and almost always completely ignore total societal utils; you're virtually coming right out and saying I'll always be a libertarian in my analyses and never a utilitarian – or anything in between.

  6. "some researchers might select facts on the basis of ideology"

    Isn't this, I don't know, a little limited? Don't Kuhn and Feyerabend and Merton do a pretty solid job of showing that ALL researchers select "facts" on the basis of ideology? Not that they deliberately exclude other facts, of course. The claim isn't so dire as that. Rather, their ideology conditions what counts as a fact and what doesn't.

  7. "Counter-intuitive results are often interesting and useful. But ignoring (or it sounds like actually rejecting) a central fact -- that employment in recessions is largely involuntary -- is self-indulgent and intellectually dishonest. Professions that want to maintain credibility need to limit that kind of behavior. "

    Really, it's 'The Bell Curve' level of both ethics and science. One blames racism on the minorities being discriminated against; the other blames unemployment on the ones suffering from a bad economy.

  8. Involuntary unemployment is like saying you are involuntarily unmarried. You could marry the first person you meet, if they will have you, but few would say that is wise.

    Most Keynesian economists are convinced that something exists which deserves to be called involuntary unemployment and, especially that people can be unemployed through no fault of their own. See for Involuntary Unemployment: the Elusive Quest for a Theory for a good round-up

    In the late 1970s, Modigliani dismissed the 1969 Lucas-Rapping paper on the U.S. great depression as the great vacation theory of depressions where the voluntary unemployment of the 1930s is explained, in Modigliani words, as a congenital attack of laziness.

    An increased preference for leisure is another name for voluntary unemployment.

    As Prescott pointed out, the USA in the Great Depression and France since the 1970s were both experienced 30% drops in hours worked per adult

    Blanchard attribute lower labour force participation in the EU to their greater preference for leisure

    An unusual left-right unity ticket has emerged to explain depressions in the 1930s and the depressed EU economies: the great vacation theory.

    In reality, time use studies conclude that lower hours of market work in Europe is entirely offset by higher hours of home production, implying that Europeans do not enjoy more leisure than Americans despite the widespread impression that they do.

  9. Instead of reading papers that deal with RBC models, I think it is far more advantageous to work out the 2nd edition of a book like Macroeconomics in Emerging Markets, by Peter Montiel (CUP, 2011).

    I am neither the author nor the publisher nor someone marketing the textbook. I am just a person that read the book and found it to be illuminating.

  10. John Maynard Keynes teaches about involuntary unemployment.

  11. Nice article, thanks for the information.

  12. I think the criticism that 1980s era RBC models do not address unemployment is completely valid. I also think times have moved on since the 1980s.

    In particular, I find it a bit odd that you decided not to mention the popular integration of labour market matching frictions into the RBC framework. It's now very easy to have involuntary unemployment in an RBC model - and has been for a while, since at least the mid-90s.

    The RBC-matching literature is so huge that I think it's only slightly inaccurate to say that unemployment is pretty much a standard feature of modern RBC models. The same goes for New Keynesian models.

    The criticism that RBC theory ignores unemployment is outdated and is no longer true.

    There's been an explosion of research on business cycle models that try to explain unemployment fluctuations and job creation. I'd be interested to hear what you think about it and how it relates to the points in your post.

    I also think that you exaggerate the difference in substance between modern RBC and NK models. The fact that NK models are just RBC models with sticky prices shows how flexible the framework is. Putting ideology aside, the technical differences between the models are minimal. What this means is that a criticism of one framework often applies to the other. Sticky prices can yield an output gap, but the NK models which are the corollary to the RBC models you describe also lack labour market imperfections and consequently an extrinsic measure of job search. That is, everyone who wants a job in a basic NK model has one, just as in a basic RBC model. The difference is not in the labour markets.

    Sticky prices can have important implications, depending on what you are trying to explain. But do you think that NK models are any better at quantifying the welfare effects of business cycle fluctuations than RBC models are? The standard representative agent model does not predict large welfare consequences of business cycle fluctuations - regardless of the introduction of involuntary unemployment.

    To me, this is the bigger issue about modern business cycle research; the standard model does not predict large welfare effects, which is a different issue than whether or not unemployment is voluntary.

  13. This comment has been removed by a blog administrator.