Holger,

there is a paper on Small business survival and sample selection bias by Hanas Cader and John Leatherman, you will find Cader on Researchgate. The authors suggest to use the Cox Accelerated Failure Time approach.

HTH

Karin

Hi Karin, 

thank you very much. I will take a look at the paper. The problem, however is that we do not have longitudinal data.... :(

Best,

Holger

Holger Hi,

To the best of my knowledge selectivity bias can only be addressed with models with two equations. Usually a probit (the decision/state etc)  that controls for the selection and a linear/truncated etc equation(performance in your case). Since I have no clear picture on what you intend to investigate or even more importantly the available information I would refer you to a partial observability model  and a paper from Kounetas and Tsekouras, (2008) in Energy Economics which may be of help.

Just to note that firm survival kind of implies a dynamic setting by definition. David Audretsch and Steven Klepper have done a lot of work on firm survival from an economic perspective. 

Hope this helps

Best

Areti

If you have access to official industry classification codes and databases where you are able to search for basic economic data (turnover, number of employees, age, location, active or not, etc), you might be able to get a crude measure by comparing survivors with the overall population (or a sample of it). But it will of course be a crude measure and not work if the variable X is e.g. a question you included in a questionnaire.

• As mentioned by others Heckman selection models may be useful in treating the problem as one of missing variables.
• There is also a useful paper that examines growth under selection and replaces growth in the year of failure by -100%:
• The Growth and Failure of U. S. Manufacturing Plants

Timothy Dunne, Mark J. Roberts and Larry Samuelson

The Quarterly Journal of Economics

Vol. 104, No. 4 (Nov., 1989), pp. 671-698

• Hope this helps.

Robert Cressy