Non-Random Survival and Long-Run Firm Performance

Firm survival is not random: it depends on firm performance. This fact has been neglected in prior studies that assess firms’ long-run abnormal performance. I show that, because of non-random survival, a legitimate long-run abnormal performance measure (1) should assign equal weights to each period rather than equal weights to each firm; and (2) should not use firms with different likelihoods of survival as a benchmark. A measure that infringes these principles will misstate long-run abnormal performance. Applying this rationale to IPO firms, I show that the average IPO firm’s abnormal return relative to matched seasoned firms understates IPO firm performance because this measure infringes the two principles: it assigns equal weights to IPO firms, and it uses seasoned firms, which have survived much longer than IPO firms, as a benchmark. The rationale predicts that IPO abnormal performance will concentrate in IPO firms delisted shortly after their IPO. Indeed, I find that only early-delisted IPO firms earn low-er returns than matched seasoned firms; IPO firms that survive long enough to turn into sea-soned firms earn the same returns as matched seasoned firms. This study sheds light on the debate over how to evaluate long-run firm performance and how to conduct long-run event studies