One thing people has in common is overconfidence in their own ability to reason and predict. Ian Ayres in his book "Super Crunchers" described couple of examples, where so called experts were easily beaten by simple statistical method called linear regression. The method is very simple. The whole idea is to chose some data, that we believe (or tested before using different methods) is correlated with the numbers we are trying to predict and then find some polynomial, that has most predictive qualities, what is generally almost the same as solving a set of equations. I won't be describing details about measuring probability of accuracy or overfitting, or trying to find a prediction with a non-polynomial function.
Statistics are not some mambojumbo to tell us who is likely to win next elections. It is mathematical vehicle, that allows us to deal with uncertainties in a scientific way. Science is problematic in many areas. While in math and computer science everything is abstract and everything can be proven or disproved using logical rules, almost all science is dealing with a real world, where nothing can be really proved, so the only thing it can do is to test theories and go on with those, that were not disproved by empirical facts. In physics, chemistry, biology theories need to be always accurate and every contradiction disproves theory, but there are also economy, psychology, sociology, actually medicine is here too, where theories may be thought valid even if there are cases, where they didn't predict the outcome. Yet still certain predictions can be made and certain theories can be formulated, but to test them, we need mathematical tool, which is statistics. How science look in these areas? There is barely any theory prior the research, more often only a hunch, that something maybe interesting there. We conduct our experiments and then apply our statistical tools to see if something valid can be made out of the outcome. If there is something, then we formulate theory and look for another experiments, that may disprove it in a statistical sense.
One problem many economists face, when doing their research is how to test their theories. Their theories deal with whole nations and you can barely conduct an experiment with whole nation. I mean you can't just create different scenarios and execute them on different nations, or can you? Of course there a ways to deal with this obstacle, since life sometimes offer us natural experiments, that maybe treated in a scientific way. Sometimes we don't care about scientific content of a theory and we just make a prediction without really testing the theory. That's how many financial experts work, but at least they use statistics, so it has at least some merit (ok, not all of them, but let's leave it here).
How then our governments introduce new laws? And when? The timing is really important. New law is brought under consideration, not when someone comes up with a new theory, but when some people see a problem with a current law system. Of course sometimes the problem is "obvious" to all of us, like for example during crisis. Sometimes is "obvious" only too a certain group or company, that is struggling under the current law. What happens next? Problem is "obvious" and often so is the solution, at least generally. The details are then discussed among experts and the new law is drafted. Then comes political machine and bargaining between different parties and their lobbyists. They believe the experts knowledge and intuition about how our economy should work. There may be some value in the opinion of experts, but regardless of that, what they actually do is that, they create a huge and yet scientifically invalid experiment, that is conducted upon whole nation. I find it kind of scary. Yet at the same time, no one is even thinking about conducting really useful experiments, that may bring us some knowledge, not only about how our society works, but also what is the best for us.
Concluding above I would like to bring up some theories, that were never tested and won't be in a near future, yet they may give us an idea of better behaving system. Let's start with a more reasonable one. In USA corn farming is largely subsidized by a government. Theory is, that subsidizing a certain crop will be more often used as a food for a live stock, regardless how healthy it is for animals and/or meat consumers. Patent law has negative effect on creativity, since every discovery published will be comprehended by others and may be use it to come up with new ideas, even though it will bring down profits from an idea. Another one: lack of intellectual property law won't make authors go bust, but only distribution companies, even though they may earn less money from their previous work, so it would raise a quality of our entertainment.
We may argue about how valid these theories are, but the point is, that many laws introduced may not help a nation's economy, but only switch a power to a one side of the market, what translates to a better profits for some companies and people. Even though the experiments with laws are conducted on larger scale, when different countries create different laws, but in our times of global economy we can observe a tendencies to uniformity. What's more if you could valuate laws not only by its contribution to GDP, but also how it affects the lives of simple people, than maybe some laws we would find less profitable, yet still more attractive, because it increases the quality of our lives. There are many candidates for such assessment.
No comments:
Post a Comment