st: Which test are applicable for dummy independant variables

[Max Tymoszuk <maxtymoszuk@gmail.com>]

Hey Guys,

I am writting my thesis at the moment and I face some difficulties
with STATA or the econometrical/statistical analysis. I am a complete
STATA rookie and this code is the best that I could produce ;)

I constructed a regression where the dependent variable is annual
stock return and the independent variables are all dummy variables
with (1 and 0).

I try to test for heteroskedasticity, autocorrelation,
multicollinearity and omitted variables. The sample consists of 961
observations.

However I am not quite sure if my approach is correct.

1) To test for heteroskedasticity, I use:

- Corrgram

- Varsoc

- Estat bgodfrey

= Does it make sense to use all of those tests or can I just
concentrate on one? Which one?

= Can I use these tests although I only have dummy independent variables?

= I am not quite sure about the lag. I used varsoc and corrgram to
indicate which lag to choose. But I have no autocorrelation in the
sample. For the sake of completeness I tested the following lags:

- Corrgram RETURN, lags(25)

- Varsoc RETURN, maxlag(25)

- Varsoc RETURN, maxlag(12)

- Varsoc RETURN, maxlag(6)

- estat bgodfrey, lag(1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
20 21 22 23 24 25)

- estat bgodfrey, lag(1 2 3 4 5 6 7 8 9 10 11 12)

- estat bgodfrey, lag(1 2 3 4 5 6)

Shall I keep using all of those or is it unnecessary as lag 25 already
tests for lag 6 (if it's the case) etc.?

2) In general, I use the following tests:

a. Autocorrelation (see above)

b. Hettest

c. As I have heteroskedasticity, I use the "robust" regression
(Huber-White standard errors)

d. Vif

e. Ovtest

= Can I use these tests although I just have dummy independent
variables and my dependent variable is continuous?

Thanks sooooo much for your help!!!!

Max

For the sake of completeness, the whole code:

**Import File

import excel "\\...........FinalSample.xlsx", sheet("Total") firstrow

**Sort by Date

sort Date Obs

**Create Dummy Variables

tabulate Payment, gen(Payment)

tabulate Status, gen(Status)

tabulate Country, gen(Country)

tabulate Industry, gen(Industry)

tabulate Listing, gen(Listing)

tabulate MB, gen(MB)

**Prepare Timeseries for Autocorrelation Check

****I have overlapping time data so I created a new time variable

gen time = _n

tsset time

**Plot BHAR over time to show no trend over time (Serial
Correlation/Autocorrelation)

twoway (tsline RETURN, recast(scatter))

**Regression MB2 MB3

regress RETURN Payment1 Status1 Country1 Industry1 Listing1 MB2 MB3

**Autocorrelation Check

corrgram RETURN, lags(25)

varsoc RETURN, maxlag(25)

varsoc RETURN, maxlag(12)

varsoc RETURN, maxlag(6)

estat bgodfrey, lag(1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
21 22 23 24 25)

estat bgodfrey, lag(1 2 3 4 5 6 7 8 9 10 11 12)

estat bgodfrey, lag(1 2 3 4 5 6)

**Heteroskedasticity

hettest Payment1 Status1 Country1 Industry1 Listing1 MB2 MB3

**Regression with Huber-White standard errors

regress RETURN Payment1 Status1 Country1 Industry1 Listing1 MB2 MB3, robust

**Check for Multicollinearity. VIF-Test

vif

**Check for Omitted Variables. OVTest

ovtest

**Regression MB1 MB3

regress RETURN Payment1 Status1 Country1 Industry1 Listing1 MB1 MB3

**Heteroskedasticity

hettest Payment1 Status1 Country1 Industry1 Listing1 MB1 MB3

**Regression with Huber-White standard errors with MB1 & MB3

regress RETURN Payment1 Status1 Country1 Industry1 Listing1 MB1 MB3, robust

**Check for Multicollinearity. VIF-Test

vif

**Check for Omitted Variables. OVTest

ovtest

** Summary Stat

estat summarize
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/