Monday, September 22, 2008

Economics as a voodoo science, written under the influence of a whole bottle of Lambrusco



I was reading two books at the same time this week.

One was "The Great Influenza" by John Barry (http://www.amazon.com/Great-Influenza-deadliest-pandemic-history/dp/0143036491) which among other things lays the history of medicine's transformation into science as it gradually happened throughout the 19th century, from a trade largely practiced by uneducated, unlicensed doctors some of whom could barely read (in the early 1800s there were no prerequisites for medical school, not even college education, no entrance exams, and no laboratory training) into the true scientific discipline which used statistics, biochemistry, and extensive experimentation.

The second was "Ahead of the Curve" by Philip Broughton (http://www.amazon.com/Ahead-Curve-Harvard-Business-School/dp/1594201757), documenting the author's experience at the Harvard Business School. This one nicely superimposed itself over the week of the economic turmoil, and brought back my own MBA memories.

Being reminded of what an actual scientific method looks like while listening to the solemn bullshit pronouncements about the economy that is overflowing the news, I could not help but compare these pronouncements to the writings of the medieval alchemists.

Compare, for example, the following quote from "The Revelations of Hermes"...

"But it is a skillful, perfect equation of all the Elements, a right commingling of natural forces, a most particular union of spiritual virtues, an indissoluble uniting of body and soul. It is the purest and noblest substance of an indestructible body, which cannot be destroyed nor harmed by the Elements, and is produced by Art. With this Aristotle prepared an apple prolonging life by its scent, when he, fifteen days before his death, could neither eat nor drink on account of old age." (http://www.sacred-texts.com/alc/arr/arr27.htm)

...with the following citation from Alan Greenspan...

"We cannot be certain that this benign environment will persist and that there are not more deep-seated forces emerging as a consequence of prolonged monetary accommodation."

Or this amazing pearl from Greenspan circa 2000:

"This does not necessarily imply a decline in asset values - although that can happen at any time for any number of reasons - but these values will increase no faster than household income."

Wikipedia defined science as "the effort to discover, and increase human understanding of how the physical world works. Through controlled methods, scientists use observable physical evidence of natural phenomena to collect data, and analyze this information to explain what and how things work. Such methods include experimentation that tries to simulate natural phenomena under controlled conditions and thought experiments. Knowledge in science is gained through research."

A typical science - such as physics - starts with observing the natural phenomena, proceeds with deducing the basic laws that govern the system under examination, then uses these laws to create mathematical formulas to describe the behavior of the system, then use these formulas to predict some aspects of the behavior of the Universe.

A representative scientific formula might look like this...



...and lead to results that look like this:



Most importantly, the results are REPRODUCIBLE, can be used to produce NON-TRIVIAL predictions of system's behavior, and are considered TRUE by the scientific community within the limits of their applicability.

By contrast, let's examine a representative economic construct, a Laffer Curve.

Wikipedia says, "In economics, the Laffer curve is used to illustrate the idea that increases in the rate of taxation may sometimes decrease tax revenue. Since a 100 percent income tax will generate no revenue (as citizens will have no incentive to work), the optimal tax rate that maximizes government revenue must lie below 100 percent."



If you look closely at the picture above, you can notice several things. First, it contains only two numbers - 0 and 1. The Y axis is not marked at all. Its shape is deceptive - it resembles a parabola, but there is no mathematical proof that it really is one. Furthermore, the t0 point on the curve is deceptive - it appears to be exactly at the 0.5 coordinate (which might suggest a 50% income taxation as maximizing the government revenue), but there is no data that suggest that this is true.

If I were to produce a picture like this in any of my undergraduate classes, every one of my physics or math teachers would have given me an F on the spot. Doing it twice would have most likely resulted in premature termination of my college education.

Examining the "idea" that the Laffer curve illustrates, it becomes immediately obvious that it contains NO NON-TRIVIAL RESULTS. The claims that it makes is that at 0% and at 100% taxation rates the government income is 0. So what? The theory would only be USEFUL if it were to predict what the maximum is (or at least that there is only one).

What if the curve actually looks like this?


Or this?



Finally, did anybody prove that it is even a function?



There is no way of saying, of course - the theory behind the Laffer curve did not result from any statistical analysis - or any scientific analysis for that matter. It was based on a mental model, "common sense", and an ideology - much like Aristotle's derivations of the laws of nature. Only instead of 300BC, in economics this approach is practiced today.

Laffer curve happens to be a representative economic concept - a model that while sounding important can actually give one no useful information. Or, to use Greenspan's quote again...

"The fact that our econometric models at the Fed, the best in the world, have been wrong for 14 straight quarters, does not mean thay will not be right in the 15th quarter."

7 comments:

Serguei Michtchenko said...

Thanks for that post, never ceases to amaze me how quickly and badly one's (mine) perception of what it means to act as a scientist really ought to degrades.

Economic models are notoriously hard to construct, especially to model the human society. Forgetting the complexity, I think part of the difficulty comes from the tendency of creating feedback in that the prediction also affects behaviour of what (who) it models; media surely plays a big part. Intuitively it seems to me that a very good model would have to make provisions for such effects. Begs the question if such a model can exist at all, hard to do for sure.

"...would require a proof that the curve is contiguous..."
I presume you meant to say "continuous"?

Anonymous said...

But aren't you aware that the scientists at the Wall Street Journal editorial page have used actual data to reveal the true laffer curve:

http://www.brendan-nyhan.com/blog/2007/08/replicating-the.html

Anonymous said...

First, great post.
Second, more formal scientific criteria is Falsifiability.
http://en.wikipedia.org/wiki/Falsifiability — wikipedia article is worth reading.
Also it says «Aspects of economics have been accused of not being falsifiable …» (http://en.wikipedia.org/wiki/Falsifiability#Economics)

nathan3700 said...

I can see your frustration with economics as a science.
Obviously it is not a science in the sense that you can assure 100% reproducibility. So, by your strict definition of science economics can never be a science.

However, it doesn't mean that a study of aggregate human behavior isn't worthwhile. We can make some predictions and we're sometimes right about them.

I thought your citations of Greenspan weakened your argument. Greenspan had a tremendous affect on the economy based on what he said. He *had* to cloak his comments in ambiguous terms or else anything he said would be a self fulfilling prophecy.

I think your use of the Laffer Curve is a little bit of a strawman. I don't think that the Laffer principle ever asserted anything about the shape of the curve between 0 and 1. It is obvious that the curve is only meant to demonstrate extremes.

An economy is all about confidence and faith. I can assure you that matters of faith will always be outside the realm of pure science.

Sergey Solyanik said...

> An economy is all about confidence and faith. I can assure you that matters of faith will always be outside the realm of pure science.

:-)

Incidentally, I think alchemy was also all about confidence and faith.

On a separate note, imagine a world 200 years from now where human brain have been completely reverse-engineered... Where would be faith then?

nathan3700 said...

Sergey:
You seem to be confident we will someday completely reverse engineer the human brain.

That takes faith!
:)

Sergey Solyanik said...

Nathan,

Not faith - historical precedent.

Everything gets reverse-engineering eventually. Just ask the team that created SONY rootkit :-).