An empirical analysis of airline safety

There are substantial differences across large and small carriers that previous airline safety studies have neglected. My inquiry controls for these differences in a test for consumer and airline reaction to accidents. I first identify and correct several weaknesses in a model introduced by Borenstein and Zimmerman (1988) that prevented the authors from deriving statistically robust results. I rank accidents by their level of severity to distinguish accidents with material losses from those with human losses and then complete estimations of the short-run effects of accidents on enplanement. I find that the aggregate impact of accidents on large carriers' enplanement is estimated at about 3% of their quarterly enplanement and equivalent to about 1.8 million passengers or {dollar}284 million in revenue losses in the quarter following the accident and no signs of recovery during the following four quarters. These losses are immediate and lasting. There is no statistically significant impact of accidents on small carriers.; Second, I test the short-run specification against two alternative specifications, a perfect recall model, that is whether consumers have perfect recollection about all accidents (old and new) and an imperfect recall model, that is whether consumers forget about accidents over time and update their beliefs toward carriers. While these models test intuitively interesting hypotheses, the short-run model provides the most statistically robust estimates suggesting that consumers may not view accidents as random events but rather that accidents may provide some information about the occurrence of future accidents.; Third, I test for consumer behavior across carriers and find that large carriers' accidents have no impact on enplanement for large rival carriers but lead to a generalized fear of flying on small rival carriers. Small carriers' accidents, on the other hand, have no statistically significant impact on (small and large) rival carriers' enplanement. Fourth, I test for airline pricing strategies in reaction to accidents and find that pooled estimations yield no statistically significant changes for large carriers and a 11.9% statistically significant decrease in fares for small carriers.; Finally, I test for airline pricing strategies across carriers and find that when major carriers are involved in accidents, there is no statistical evidence of changes in prices for large and small rival carriers. When small carriers are involved in accidents, large carriers do not vary their fares by much while small rival carriers raise them by 3.9%.