Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Pierre Nguimkeu

Second Advisor

Andrew Feltenstein

Third Advisor

Givi Melkadze

Fourth Advisor

Augustine Denteh


This dissertation consists of three essays on aviation. The first chapter investigates the relationship between competition and airline safety. The U.S. airline industry transports almost a billion passengers a year. Accidents still cause much apprehension and angst among the public; the two 737 Max crashes are the latest examples. Despite massive improvements in safety over the last decades which are largely due to technological improvements, there is still a lack of research done on the effect of market conditions. We provide a simple theoretical model to explain the relationship between safety and competition. Using data from 1995 to 2018, we employ a negative binomial regression to evaluate the impact of competition through the Inverse Herfindahl Hirschman Index, Concentration Ratio 4 and 8, on injury outcome of passengers involved in an airline accident. Since accidents are costly both directly and indirectly through brand image deterioration, a competitive market may incentive airlines to reduce risks of accidents. Meanwhile, since safety is unobserved by passengers, under a highly competitive environment where profit margins are thinner, airlines cut costs, some of which are related to safety. Our result suggests that a less competitive industry has a positive impact on safety, suggesting that excess profits are in part reinvested into safety measures.

The second chapter tests the hypothesis that monopsony power is an important determinant of wages and employment in the U.S. pilot labor market. We estimate the labor supply curve of the U.S. airline industry using firm-level employment and accident data from 1995 to 2018. Utilizing a labor demand instrument, the prevalence of aircraft accidents, allows us to directly measure monopsony power. We also investigate the effects of competition, as measured by the Inverse Herfindahl-Hirschman Index, on the labor supply elasticity. We estimate a labor supply elasticity of 2.56, indicating that airlines have substantial monopsony power in pilot hiring, resulting in a labor shortage and wages 28.11% below the marginal revenue products. We also find that as market competition rises, airlines slightly lose market power in hiring: wages increase as competition increases. The source of monopsony power lies elsewhere, mostly in the training and career structure, which we address and provide policy recommendations.

The third chapter proposes a new method to calculate the p-values of a treatment variable in a cross-sectional small sample. Causal evaluation is becoming increasingly popular in industry and government. In small sample scenarios inference is more difficult. This often occur for several reasons such as budget constraints or noncompliance, but also in phenomena with low frequency. Small samples complicate causal evaluations for at least three reasons: (i) they are associated with greater sampling error, (ii) p-values based on standard tests are not trustworthy and the statistical power of these tests can be too low to detect significant program effects, (iii) the validity of parameter inference strongly depends on distributional assumptions. This paper proposes a simple approximation for the p-values to use in the regression analysis of treatment effects models with normal or nonnormal error distributions. The approximation is derived from recent developments in likelihood analysis and has a third-order distributional accuracy. Thus, for very small or medium-sized samples, the proposed method has remarkably higher accuracy compared to traditional ones that usually rely on normality or large samples. The method is then applied to aviation data to evaluate the impact of accidents on airfares, which is relevant to both airlines and insurance companies.