Estimation usi a large set of firms from different countries

[Quant]

Hello there

I am trying to estimate a model using a large panel of firms (more than 500 with 4 years of available data) from different countries which probably has cross-section heteroscedasticity. In order to account for firm-specific heteroscedasticity should I estimate the model using GLS cross-section weights? Moreover, to account for any biases in the standard errors is bootstrap a good solution? Thanks in advance.