Tuesday, February 3, 2009

Falsifiability and macroeconomics

Karl Popper's philosophy of science is probably the most influential philosophy among practicing scientists. Popper stressed doubt, and the importance of ideas improving through falsification. Falsifiability has become an important concept in the twentieth century and beyond for determining what can properly be called "science."

The concept of "science" has evolved over the centuries, in Ancient and Medieval times just meaning conclusions built upon rigorous logic and self-evident principles. Then, the importance of observation was added with the rise of the scientific revolution, and "science" began to mean more an empirical exercise, though still relying on self-evident assumptions and logic. But the old concept persisted in philosophical circles and was used by systematic philosophers like Kant, Hegel and Marx, who all viewed their highly speculative and abstract philosophizing as "scientific" (German for "Science"=Wissenschaft, which is different from Naturwissenschaft, "natural science" which includes things like Biology, Chemistry, Physics, etc)

Macroeconomics also has made this claim to being "scientific," but again it really is not the same as we mean nowadays. Physicist Frank J. Tipler has an article from a few days ago comparing Macroeconomics to Astrology, and I think the comparison is very apt. One amazing thing about Astrology, as it is practiced is how "scientific" it can be, in one sense of the term. Many practicing astrologers are out there collecting data about what the alignment of planets and stars was when a person was born and how things turned out for that person and what type of person they are. They are using this research to try to refine the predictive power and improve the results derived from Astrology. Now, when astrologers do this sort of research, most of us find it a little absurd because we know that there's no way you can predict a person's fate based on the stars and this whole endeavor is fruitless. But much very sophisticated research and many very respectable journals are devoted to publishing research in Macroeconomics that's very similar in nature.

The reason why both Macroeconomics and Astrology are not "scientific" as we usually mean it nowadays is that they lack falsifiability. They are operating according to these more antiquated notions of science: of concepts built upon sound logic and self-evident principles, possibly with observation also thrown in. Macroeconomics admittedly has done a lot better than Astrology, in terms of accuracy, but its predictive is still very weak. Macroeconomics logically is an incredibly rigorous discipline, with some of the most sophisticated math thrown in, but when it is forced to make predictions it fails, and yet the failure of these predictions don't quite doom the theories that propounded.

There was a recent piece on NPR about how Keynesianism is being given its first real world test with the upcoming stimulus. The immediate response to such a statement by many economists is, "No, Keynesianism has been tried many times and in many different ways and has failed." But the authors, Davidson and Blumberg reply that it hasn't really been tried. For example, what FDR was doing with his New Deal, though Keynesian in nature, wasn't Keynsianism proper because FDR didn't spend enough money. In fact, Keynes was quite critical of the New Deal. So, though it was sort of a failed Keynesian experiment, it can't disprove Keynesian, since it wasn't proper Keynesianism. This argument can always be made, and not just for Keynesianism, but for other macroeconomic perspectives. It makes these theories unfalsifiable and thereby not "science" as we mean it nowadays. Thus economics fails at being "science" according to the most rigorous standard.

Even the empirical side of Economics can be a little bit dicey. Empirical research in economics is notoriously slippery, because in economics we can't do experiments, with controlled conditions where we can isolate variables. Some new researchers in economics have figured out ways to find natural experiments, where they have huge data sets where in which many variables are held constant, so they can observe statistical fluctuation among the variables, and though these may hold promise, its hard to find many of these natural experiments, so its utility is limited. Ultimately, empirical research in economics is a difficult and highly speculative business. This mostly leaves economics with just being "scientific" in the sense of using rigorous logic and self-evident principles.

The failure of empirical research led economists in the Austrian tradition to reject "economic positivism," which is a concept that economics can be built up on empirical grounds. Hayek especially, who was a good friend and was influenced by and influenced Popper, saw the fatal conceit of economists and policy makers who thought that they understood how the economy works and even worse, thought they could manipulate and make the economy do what they wanted. Experience and history has shown that this is a dangerous delusion, and that things never work out the way economists think they should. Partly because the economic is too complex also because it is filled with living breathing humans who adapt and react to given circumstances. Economics suffers the same difficulty that a weatherman faces, of dealing with a highly complex system, but its even more difficult because every particle in the economic system is alive and is making decisions on their own.

Hayek wanted us to recognize that economics is a theoretical discipline and that fancy mathematics and statistical research cannot change the fact that economics is quite speculative and fairly conservative. All we need nowadays are for more policy makers and economists to recognize their limitations and come face to face with their ignorance and stop trying to think they have the magical formula to fix the economy.


Anonymous said...

Just imagine what it would be like if macroeconomics WERE falsifiable. We could study wealth the same way we study asteroids and organelles, and then, we could apply that knowledge through governance to make our society prosper beyond belief! That's a sweet dream. Of course, I don't expect this to ever happen, but for the sake of open-mindedness, I'll keep my small shred of hope. One mustn't ignore that economics is a social science and a holistic understanding of economics can only come after a proper understanding of the unit of society: the conscious being that is the individual.

Tom Brown said...

I don't see the consciousness of economic participants as an undo burden. Neuroscience and medical science progress and their subjects of study are often conscious beings. I don't see anything special at all about conscious beings in fact: the evidence is starting to demonstrate that we have no free will whatsoever. Freewill appears to be an illusion. Supposing that hypothesis is true, then humans are just like any other clumps of matter (see biologist Jerry Coyne on this, or the research by Soon el al), and there's no reason we can't study them. Now it might be true that the problems of applying scientific techniques to macro economics are just too great for us to make any progress, but I'm skeptical of that conclusion, and I see no harm in trying.

One new idea I've seen is to abandon trying to frame economic forces in terms of human motivations, and simply model humans as though they were little gas molecules randomly bouncing about. What's interesting, is that the opposite (thinking in terms of utility maximization, like an economist) can be used to study some cell biological funcions.

Here's a blog that uses a version of information theory to essentially remove the humans (and trying to understand their motivations and incentives) from macro economic theory:
The author of that blog has a PhD in physics and he has made numerous falsifiable macro predictions, and has additionally said that he's not interested in patching up his models to match the data after the fact by adding a lot of "epicycles." I've never in my life heard a professional macro-economist put their theory on the line like that!

Here's a particular post on that blog showing how utility maximization can be used to analyze systems with NO so-called "conscious" beings in them:

Tom Brown said...

... the field of evolutionary psychology also seems at first to be amenable to the scientific method (and it really is in theory), but once you delve a little deeper the obstacles are so high that it becomes questionable if progress can really be made.

However, I don't see that as a reason to turn to some method other than science (such as arm chair philosophy divorced from all empiricism, which is what I typically associate with Austrian economic concepts). Although granted: Hayek is far from being the poster boy for the worst of Austrian excesses in that regard.