Hacker News Reader

Comment by KaiP | Hacker News Reader

This article is a pretty clear demonstration of what I dislike about popular economics (liberal and conservative) compared to hard science. Conclusions are drawn up based on ideological assumptions, then evidence is gathered, then public policy is changed. In any given case, several explanations are possible, but rarely is empirical testing done to prove that one is more feasible than the other. Instead they coexist as equals, gaining support based on random political futures. Perhaps this is a fundamental weakness in economics and cannot be overcome, but it seems to me that not enough pressure is being put on the field to make the ideas fully testable. As a engineer and numbers guy, I want to see my tax dollars spent (or not spent) only on ideas that have been rigorously examined and tested to be the most productive for society.

View on HN

The fundamental problem with economics is physics envy. Its nickname is the dismal science. The goal of economics has been the predictive power of the natural sciences. Well it aint going to happen. First, it is very difficult to run repeatable experiments. You cant have a double blind test to see if HSA's are better than single payer insurance. Friedman looked at the two and concluded that HSA's were better. I looked at his article and my experience with an HSA and concluded the opposite. Second, it appears to me that economics has a lot more in common with chaos theory than Newtonian physics. There are so many variables that it is probably impossible to determine what the initial conditions are. Third, economics is not rational. It has way more in common with psychology and politics than it does with physics, even if money is a nice quantum measurement.

So what's to be done. Look at other attempts and see what works and what doesnt, England and Canada for example. Be conservative. Even though out system is a train wreak, transform it, dont throw it completely away. Make a series of transformations. Assume people are going to game the system in ways that you dont anticipate and make adjustments along the way. It is impossible to design a perfect system; dont even bother to try. Throw away anything that doesnt work.

Fair enough, but how would you rigorously test economic principles without ruining the lives of millions?

First use historical statistics and if possible, historical examples of the ideas being implemented. What the article presented was not much more than circumstantial (costs are going up, we have these programs, therefore these programs are causing costs to go up). If the evidence seems sound by an independent advisory board, have a pilot program put in place at one hospital, one city, or one county and carefully monitor the results. If proven successful in those trials, expand further.

I see some possibilities to test hypotheses like "$MEASURE will achieve $GOAL"

a) Gather historical data where $MEASURE was attempted and see whether $GOAL was reached. This can give results like "$MEASURE is positively correlated with achieving $GOAL".

b) Test it on a smaller scale. Instead of reforming the entire health-care system, search for something small but similar to reform.

c) simulate. Build a simulation of economic subjects and, before simulating the Friedman health-care reform, test it on past scenarios whose outcome is known.

These usually just result in people picking holes in the dataset/experiment/simulation -- and rightly so; circumstances change too quickly for historical data to be much use, simulations necessarily have some severe assumptions, and small-scale tests don't simulate macroeconomic effects very well.

This is the reason that I've long classified macroeconomics in with religion: they're both very influential ideas, but neither are really based on any evidence.

Hmm yes... fairly quickly, one runs out of large rich countries at the beginnings of twenty-first centuries to draw parallels from.

I liked A Mind Forever Voyaging, though, which is about an attempt to do c) properly.

There is certainly an epistemological flaw in predictive economics. On the one hand, economic history gives us plenty of data on the behavior of economic systems. Unfortunately, each new economic system is essentially novel. That's why I don't give much credence to economic predictions that are overly precise, and I am skeptical of lessons drawn from long-ago events applied to today's issues.

This is where I think Austrian Economics has diagnosed the problem correctly, although its solution (an escape into extreme deductive logic) doesn't appeal to me as a reliable method to discover truths about the world.

Reply on news.ycombinator.com