Article ID: | iaor20107378 |
Volume: | 58 |
Issue: | 5 |
Start Page Number: | 1287 |
End Page Number: | 1302 |
Publication Date: | Sep 2010 |
Journal: | Operations Research |
Authors: | Raghunathan Srinivasan, Cavusoglu Huseyin, Koh Byungwan |
Keywords: | security, terrorism |
The proponents of airline passenger profiling claim that profiling will reduce the cost of security, improve the detection of attackers, increase the reliability of signals from screening devices, and reduce the inconvenience to normal passengers. In this paper we show that if the Transportation Security Administration (TSA) manually inspects all those passengers classified as likely attackers and sends others through a screening system, as it did when it deployed the Computer Assisted Passenger Prescreening System (CAPPS), then it is superior to no profiling on all four performance measures if and only if the quality of the profiler vis-à-vis that of the screening system is sufficiently high. If the quality of the screening device is sufficiently high, profiling could be detrimental on all four performance measures. On the other hand, if the TSA deploys two screening devices along with the profiler–each screening device optimally configured for each of the two groups of passengers–then profiling improves the reliability of screening device signals, reduces the inconvenience caused to normal passengers, and improves the social welfare even when quality of the screening device is high. One of the implications of our findings is that the security architecture used by the TSA when it deployed CAPPS could provide a strong support to the arguments by some against the use of profiling; however, if the TSA deploys a two-screening device architecture, it might not only blunt the criticism that profiling is discriminatory but also benefit normal passengers and overall society economically.