Abstract
This paper investigates the relationship between default probability and value when there is a selection bias due to missing controls for firm heterogeneous likelihood to survive in the sample. Our model delivers the following implications for the conglomerate case: (a) the sample conglomerate value increases in their default probability, (b) the sample conglomerate discount falls together with their excess default probability with respect to focused companies, (c) both effects disappear when the analyst controls for survival probability. The data support the presence of a selection bias distorting downwards the relative value of sample firms with higher survival probability.
