Unexpected IPO deals

[Cong Gu]

Dear Statalists,
I am doing the research on some unexpected Initial Public Offering (IPO) events. Basically, with a sample of about 2000 firms, I first run the logit regression to identify the determinants of IPOs, where the dependent variable equals one if a sample firm is involved in IPO, and zero otherwise.

With the general logit regression results ready, I would like to investigate each specific IPO deal and to define whether it is an unexpected IPO or not. However, I feel a little bit confused on how to identify this by using the general logit regression results.

Therefore, it would be much appreciated if any kind suggestion could be provided.

Best wishes,
Cong

[Clyde Schechter]

Well, following the logit regression you can use -predict- (see -help predict- if you are not familiar with it) to get the logistic regression's predicted probability of being an IPO. Then you could call an observation an unexpected IPO if it is an IPO and the predicted probability is below some threshold. Just what threshold to use for this is a substantive issue in finance, not a statistical or programming issue, so you will have to figure that part out yourself, or in consultation with others in your field.

[Cong Gu]

Dear Clyde,

Many thanks for your kind reply, which is very helpful.

Best,
Cong