[marcosportillo]

Mr. Herbener,

I’m just about finishing up an intro to econometrics class. It was interesting but at times seemed like there was quite a bit of assumptions and aggregations being made. What’s your personal opinion on econometrics and do you ever use it in your research?

[jmherbener]

I agree with Mises’s criticisms of econometrics. He makes two main points. First, it isn’t possible to formulate economic propositions as mathematical functions because there are no constants in human action. Everything is a variable. So it is a mistake to postulate C = a + bY for example. Second, human action is subject to case probability, which has no numeric expression, and not class probability.

It seems to me that the burden of proof is on those who claim that econometrics renders either theoretical or historical truth.

Here is a recent example of insightful historical analysis without econometrics:

http://mises.org/journals/qjae/pdf/qjae15_1_1.pdf

[marcosportillo]

Thank you for your input and the journal article! Are you saying there is no place for mathematical analysis of statistical data to derive correlations? How should statistical data be analyzed? By just observation and deductive reasoning?

[jmherbener]

I’m saying that the burden of proof is on the proponent of statistical analysis. He must demonstrate that Mises’s criticisms are incorrect and that statistical analysis of the data of human action renders better understanding of historical events than what Mises called specific understanding. On the latter, look at Mises’s Theory and History:

http://library.mises.org/books/Ludwig%20von%20Mises/Theory%20and%20History%20An%20Interpretation%20of%20Social%20and%20Economic%20Evolution.pdf

[marcosportillo]

I think I understand. Statistical data can be better understood through what Mises called “specific understanding” rather than some mathematical computation that spits out an “understanding” or relationship.

[Author]

Posts