I am doing my thesis at the moment and I needed to check the BLUE assumptions for my data set. I am doing my thesis on trying to explain IPO underpricing and my data set consists of 100 companies who did an IPO. I have found second order correlation via the Breusch- Godfrey test in stata and I was wondering if anyone here knows how to fix this. The research I have done so far tells me that is it extremely rare for a data set like this to have serial correlation, but it happened. As newey-west standard errors only work for time series and I don't have that I have absolutely no clue on how to fix this.
Do you have the time points when the IPO was conducted included in your dataset? Otherwise how did you apply the Breusch-Godfrey test since it is a test for time-series data? It would strongly depend on the ordering of your data, which should be i.i.d. in your cross-sectional data set.
But it is thinkable that the volume of the IPOs depend somehow on the overall development of the stock market. You could control for that if you include an appropriate stock market index, or in your case of cross-sectional data you rather have to "deflate" your data to some basis year.
Just some thoughts, good luck with your thesis.