The U.S. Department of Justice is interested in how competition impacts airfares, merging airlines can be stopped from combining because the government thinks that decreased competition would not increase social welfare. Airlines are the typical example of a market organized as an oligopoly, controlled by a handful of players who must think strategically about their pricing decisions. Triggering a price war is hurts all airlines, but this even occurs often. These intricacies makes understanding the airline market/industry an interesting and policy relevant example for this post. Competitive markets tend to push prices down and increase the quantity of a product/service produced, and one would think that the airline industry is no exception. This post will analyze how market concentration impacts airfares using advanced panel data methods. I find that a 10% increase in the market concentration on a given route will cause a 2.16% increase in airfares, after controlling for distance and unobserved fixed factors on a given route.
Using data from 1,149 routes over a period of 4 years I estimate an OLS model, OLS with heteroskedasticity robust standard errors, a fixed effects, and random effects regression models to arrive at the impact that market concentration has on airfares. In order to estimate this relationship, the distance traveled per route will be used as a control variable. Using fixed effects regression will account for specific time invariant differences in prices per route, maybe the route is to a vacation destination or from an economically depressed city, this will help to mitigate some of the unobserved fixed heterogeneity present in panel data.
Fixed Effects vs. Random Effects Regression
A formal theoretical treatment of fixed effects and random effects regressions, including assumptions, will be conducted in another post. The intuition is that if the unobserved fixed heterogeneity is uncorrelated with the explanatory variables. If we think that the unobserved heterogeneity is correlated with any explanatory variables, then using first differencing of fixed effects is required. The regression estimates for both models will be estimated and the Hausman test will be used to see which model, RE or FE, is the best model to use.
The fixed effects regression above calculates a statistically significant positive relationship between market concentration and airfare. A 10% increase in the market concentration of the largest carrier on a route would raise prices by about 2%. Save the results…
The RE estimate suggest that a 10% increase in the market concentration of the largest carrier per route would increase airfares by 2.16%. Which model is the best model? Here is the Hausman test for differences in coefficients to help answer this question.
The rejection of the hypothesis that the difference in coefficients is not systematic, strongly suggest that FE model is the better model to use to estimate the affect of market concentration on airfares. The final conclusion is that there is overwhelming evidence that increased market concentration increases prices, and that a 10% increase in market concentration would account for a 2.16% increase in airfare.