i am doing an assignment on bank mergers and want to see if there is a significant difference in the the mean of return on average assets 1 year before the merger announcement and 3 years after. I am using a panel dataset of 16 banks, have used xtset = company to set out my data as panel.
I am unsure as to what is the best approach to this; i thought a ttest would be at first, but after further research developed the following code and output with xtreg;
. gen byte era = 0 if time == 0
. replace era = 1 if time == 16
(15 real changes made)
. xtreg returnonavgassetsroaa i.era, fe
Fixed-effects (within) regression Number of obs = 31
Group variable: companyid Number of groups = 16
R-sq: Obs per group:
within = 0.0428 min = 1
between = 0.1344 avg = 1.9
overall = 0.0323 max = 2
F(1,14) = 0.63
corr(u_i, Xb) = 0.0418 Prob > F = 0.4418
------------------------------------------------------------------------------
returnonav~a | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.era | -.1293333 .1633765 -0.79 0.442 -.479741 .2210744
_cons | .9167742 .1127258 8.13 0.000 .6750014 1.158547
-------------+----------------------------------------------------------------
sigma_u | .30860528
sigma_e | .4474249
rho | .32237222 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(15, 14) = 0.90 Prob > F = 0.5785
can anyone please let me know if this is a correct way to assess this, and if possible help me with interpretting the tables. i have some idea but have limited experience in this field.
any help much appreciated.
Related Posts with Testing if the mean of a variable has changed significantly over time
How to set 3 way panel and run triple difference regression?Hi everyone, I've got a dataset which varies across time, across country and across products (trade …
Concatenating variables as a function of other variablesDear all, I have a database comprised of an id variable (siteid), two variables that indicate the l…
A Statalist glossary This is a revision of a glossary that appeared on 1 January 2011. Striking (to me any way) that mos…
Negative margins value with penalized maximumlikelihood estimatesHello, I've tried the logit with a binary variable but I found out that the sample size is only 95. …
Variance function regression using HAPC (age-period-cohort) Hi Statalist, I am attempting to run a model on stata, specifically, a Variance function regressio…
Subscribe to:
Post Comments (Atom)
0 Response to Testing if the mean of a variable has changed significantly over time
Post a Comment