topic Re: Testing Coefficient difference in SAS Programming

Testing Coefficient difference

abdulla — Sat, 17 Apr 2021 17:37:01 GMT

The dependent variable is investment. So, Lead_Dir affects investment differently when FRQ is low and high.
It has control variables, industry FE and Year FE and they cluster the SE at Firm level.
What I am looking for is to calculate coef. comp, p-value.

Would anyone please help me?

Re: Testing Coefficient difference

PaigeMiller — Sat, 17 Apr 2021 18:48:19 GMT

Is this a SAS question? What SAS PROC produced these results?

Re: Testing Coefficient difference

abdulla — Sat, 17 Apr 2021 19:04:15 GMT

Hi PaigeMiller,
This is from a published paper. I do not know what proc the author used. What I can see here is that there are two regressions for two groups and then the author compared the coefficient IV. Running regression is not a problem but I don't know how the author tested the difference between two coefficient and found the p-value of the difference

Re: Testing Coefficient difference

PaigeMiller — Sat, 17 Apr 2021 19:22:55 GMT

I don't know how the author did it either.

Can you describe the experiment/data collection and analysis done? What is IV?

Re: Testing Coefficient difference

PGStats — Sat, 17 Apr 2021 19:20:20 GMT

What does the Methods section of the article say about statistical analysis methods?

Re: Testing Coefficient difference

abdulla — Sat, 17 Apr 2021 22:41:38 GMT

Hi PaigeMiller,
The IV is Lead_Dir and the dependent variable is Investment. It has control variables, industry FE and Year FE and they cluster the SE at Firm level. The author runs OLS.
The author divided the sample into two based on FRQ. So, it is the same OLS. The only difference is the FRQ high and Low group. After running OLS, the author compared the two coefficient difference of IV (Lead_Dir) and calculated p-value of the differene.

Re: Testing Coefficient difference

abdulla — Sat, 17 Apr 2021 22:42:41 GMT

Hi PGStats,
Could you please check the previous explanation?

Re: Testing Coefficient difference

PaigeMiller — Sun, 18 Apr 2021 18:07:50 GMT

Typically, comparing slopes or coefficients from two groups is done via putting an interaction into the model, and seeing if the interaction is statistically significant.

So, for OLS, it might look something like this:

proc glm data=have;
     class frq;
     model investment = frq lead_dir frq * lead_dir;
run; quit;

The interaction between frq and lead_dir is what you want to look at. If it is statistically significant, then the coefficient of lead_dir changes due to frq. This doesn't completely match the example you showed, but its the best I can do since I don't have the original paper.