More on Precise Mathematical Models

A little while ago, I discussed some of the issues I had with economists who look unfavorably upon economic theories that don’t use “precise mathematical models.”  In that post, I had discussed the inherent randomness free will injects into life, making precise calculations difficult.  Now, I’d like to discuss something more economic.

Economist Thomas Sowell once said that there are no solutions, only trade-offs.  He is absolutely right.  this is one of the first lessons an economics student learns.  He is taught to look for the unseen, to look for opportunity cost.  Opportunity cost is, to put it simply, foregone opportunities; what you had to give up to pursue another opportunity.

By its very nature, opportunity costs are imprecise.  They don’t occur, so they won’t show up in any statistic, but they are a very real thing.  For example, a store plans to open in an area and create 300 new jobs, but decides not to pursue the opportunity because taxes are too high.  Those lost 300 jobs will never show up in any statistic, but they are very real.  Good economists will try to estimate their impact, but these estimates are just that: estimates, subject to the researchers’ bias and inherently imprecise.

When people look down on economic theories or models because they are imprecise, because they rely more upon logic and reasoning rather than mathematics, they are making the mistake that Henry Hazlitt says separates the good economist from the poor economist: he is failing to look for the unseen.

To paraphrase Darth Vader, don’t be too proud of this mathematical terror you’ve created.  There is likely far more happening behind the scenes than you know.

One thought on “More on Precise Mathematical Models

  1. Logic is branch of mathematics. Rather a moot point to draw between the two when one is a subset of the set.

    I don’t think you get it Jon. For those you claim to look down upon models for not being precise, is not a mathematician. Nor in anyway worthy to be trusted to have an opinion on that subject. The data used, either it be from dildos sold over the course of a year, to the amount of Poles killed by Germans in World War II, to how many Irish died during the potato famine means little. Large quantifiable numbers isn’t in itself going to be a perfect result, which is why they have error values given in the data for p-values. The errors themselves are often done to a certain value of 95% to 99% certainty, that all depends upon your testing and what you’re doing.

    We know Jon. We know.


Comments are closed.