How do I correct the standard error for cross-correlation among error terms in a cross-sectional regression?

I have a dataset of about 300 valuations of firm /financial securities for 200 days (unbalanced panel). I do not take the time series aspect seriously, and stack all the data up in a cross-sectional regression of these valuations on firm and market characteristics, that are observed at the same daily frequency. By clustering on firm/security, I can adjust the standard error for arbitrary within firm correlation over the 200 days. By using a White heteroskedasticity robust estimator, I am adjusting for heteroskedasticity (hopefully..any views?). But, how do I address the fact that there may be unobserved shocks on certain days that are uncorrelated with my independent variables, but affect all the firms (to different extents), thereby causing no bias in my coefficents but inducing cross correlation in error terms?


How do I take care of this in Stata?


Thanks,


Sid|||The white estimator ought to correct for hetero. The correction you need is Zellner's seemingly unrelated regression. Check sureg in Stata.