Conventional or Bootstrap or Jackknife Standard Errors in Between-Effects Estimator

[Prateek Bedi]

Hi,

I am trying to estimate the impact of directors' remuneration on firm performance. My unbalanced data set comprises 1696 firms and 16 time periods (in particular, years). I have 7 independent variables in total. For robustness purposes, I am using the between effects model (xtreg, be). Below is my model.

 
 xtreg Profitability4 Size2 Leverage1 CurrentRatio SalesGro CapitalExpenditure2 WPromoterSharesin1 AD_Totalremuneration, be

I checked for heteroskedasticity (using xttest3) and serial correlation (using xtserial) for my model and I found the presence of both. In addition, the between effects estimator in Stata (using xtreg, be) allows estimation of standard errors using three options: conventional, bootstrap and jackknife. Which method of estimating standard errors shall be most suitable for my between effects model?

Thanks!

[Felix Bittmann]

Most studies just report the regular SEs. However, if you want to be thorough, use bootstrap SEs in a second model and compare them to the regular SEs. If both results are similar you can be confident that your findings are robust.

[daniel klein]

How exactly do you interpret serial correlation in a model that uses only between unit/panel variation? That does not seem to make much sense to me.

[Prateek Bedi]

How exactly do you interpret serial correlation in a model that uses only between unit/panel variation? That does not seem to make much sense to me.

Absolutely, daniel klein. I agree that serial correlation does not make sense in between effects estimator. I just mentioned it because I am also using fixed effects estimator as my primary methodology. For robustness purposes, I am using between effects and I would like to know how should I choose among conventional, bootstrap and jackknife standard errors.