causality - exog

[josef...@gmail.com]

Interesting late night readings (while remembering when I was an economist)

a followup to targeted learning and treatment effects

 http://andrewgelman.com/2013/07/30/the-roy-causal-model/

http://chrisauld.com/2013/10/08/remarks-on-chen-and-pearl-on-causality-in-econometrics-textbooks/

ftp://ftp.cs.ucla.edu/pub/stat_ser/r395.pdf

brief comments:

I think people are not careful enough what the expectation operator `E` refers to, what's the relevant probability space. 

(I have problems reading formulas that nest several expectation operator without specifying what variable is integrated over.)

By default economists are causal, no need to spell it out. (But it might help.)

(We even have Lucas' critique for two levels of causality.)

statsmodels is causal, we have endog and exog and not y and X.

(But I still don't know what to call the explanatory variables in IV2SLS which is the main reason it's still in the sandbox.)

potential outcomes are implicitly defined by structural equations.

anecdotal observation: articles in econometric theory spend a lot more page space specifying assumptions than articles in statistics 

(noncausal population average)

Josef

PS: I'm structural but try to figure out what population averages refer to.

[josef...@gmail.com]

And I think I mentioned arguing schools of thought in economics

(found link in Skipper's tweet)

"The ascendance of the five core econometric tools – experiments, matching and regression methods, instrumental variables, differences-in-differences and regression discontinuity designs"

http://www.voxeu.org/article/mastering-metrics-teaching-econometrics

Economics as an experimental field with mostly a binary or discrete explanatory variable. ???

(I already said I like "Modern Econometrics")

Josef

Pinky: Gee, Brain, what do you want to do tonight?

Brain: The same thing we do every night, Pinky - try to take over the world!

 

[josef...@gmail.com]

Sorry if I'm spamming the mailing list too much. I'm trying to catch up with an unfamiliar branch of statistics and econometrics, expected to arrive in statmodels within a year.

Josef

[josef...@gmail.com]

For a different perspective on econometrics teaching  (macro econometrics instead of "treatment effects" econometrics)

Hendry, David F., and Grayham E. Mizon. 2016. “Improving the Teaching of Econometrics.” Cogent Economics & Finance 4 (1): 1170096. doi:10.1080/23322039.2016.1170096.

And then there are still several other schools of thought in econometrics where I haven't found a "teach our stuff" article yet.

Josef