Multilevel Unbalanced Panel data

[James Kimber]

Hello

I'd like to explore the relationship between cumulative abnormal return and some explanatory factors. Attached is my excel data and I am going to explain the details to clarify my question.

Column 1: represents companies id. It means that the data includes 1331 companies.
Column 2: represents the industry ID. It means that there are eleven industries and their standard code ranges between 15 to 60.
Column 3: represents date of observations for each company withing a given industry. It means that Company 1 operating in industry ten has fourteen observations from 2009 to 2019.

The rest of the columns are explanatory variables which are not my main concern. My problem is as follows:

I want to study the variance among COMPANIES as one LEVEL and observe the variance among INDUSTRIES as the other LEVEL. When I study the first level, variance between companies, it is an unbalanced panel regression. However, level two would become an unbalanced panel with repeated measures based on time. Thus, I cannot proceed with the conventional panel regression.

My question is how can I study the variance among industries? I found several methods, but I am confused how to proceed.

**Please note that due to copyright, I replace my explanatory data with arbitrary numbers**

[James Kimber]

This is the clustering effect test, if I did it correctly:



Mixed-effects ML regression Number of obs = 22,433
Group variable: Industry Number of groups = 11
Obs per group:
min = 707
avg = 2,039.4
max = 4,079
Wald chi2(0) = .
Log likelihood = -133160.27 Prob > chi2 = .

------------------------------------------------------------------------------
CompanyID | Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
_cons | 1258.301 245.0403 5.14 0.000 778.0305 1738.571
------------------------------------------------------------------------------

------------------------------------------------------------------------------
Random-effects parameters | Estimate Std. err. [95% conf. interval]
-----------------------------+------------------------------------------------
Industry: Identity |
var(_cons) | 660486.7 281621.3 286368.3 1523362
-----------------------------+------------------------------------------------
var(Residual) | 8334.309 78.71317 8181.453 8490.02
------------------------------------------------------------------------------
LR test vs. linear model: chibar2(01) = 93922.47 Prob >= chibar2 = 0.0000

. estat icc

Intraclass correlation

------------------------------------------------------------------------------
Level | ICC Std. err. [95% conf. interval]
-----------------------------+------------------------------------------------
Industry | .9875388 .0052483 .971714 .9945599
------------------------------------------------------------------------------