A Review of “efficiency and Returns to Scale in World Airlines 1998-2005”
IntroductionThe main purpose of the article, “Efficiency and Returns to Scale in World Airlines 1998-2005”[1], was to measure airline efficiency and economies of scale. The paper takes a heterogeneous set of data from 18 airlines over the period 1998-2005 and examines the relative efficiency and returns to scale. Through the use of data envelopment analysis[2] and stepwise linear regression analysis[3], the relative efficiency and returns to scale were measured. Results published in this article suggest that the constant returns to scale version of the Cobb-Douglas model of an airline is robust over a range of times, locations, and both regulatory and exogenous shocks. SummaryEconomist in the United States have long studied the efficiency and returns to scale of airlines. International regulatory developments and various events within the airline marketplace predicated the author’s examination of airline efficiency using contemporaneous data on a sample of world airlines of various types. In the study, 18 airlines, both legacy carriers and recent entrants geographically dispersed were examined over the period 1998-2005. This period bracketed the shocks of the September 11th terrorist attacks on the World Trade Center and the global SAARS outbreak. The issues of relative efficiency and returns to scale were considered through the use of data envelopment analysis and additional operational and network effects were analyzed through the use of a stepwise linear regression.
The quantitative analysis, Data Envelopment Analysis (DEA), is a powerful method widely used in benchmarking in this area of research (airline efficiency). It compares service units considering all resources used and services provided, and identifies the most efficient units (branches, departments, individuals) and the inefficient units in which real efficiency enhancements are achievable. The other empirical analysis employed in the study and discussed in the article was stepwise linear regressions. The focus of a stepwise regression is identifying the best combination of independent (predictor) variables to predict the dependent (predicted) variable. In stepwise regression not all independent (predictor) variables ultimately end up in the equation. The predictor variables are entered into the regression equation one at a time based upon statistical criteria. At each step in the analysis the predictor variable that contributes the most to the prediction equation in terms of increasing the multiple correlations is entered first. This process is repeated only if additional variables add anything statistically to the regression equation. When additional predictor variables cease to add anything statistically meaningful to the regression equation, the analysis stops.