A data-driven approach for estimating airport efficiency under endogeneity: An application to New Zealand airports

Abstract

Airport efficiency is commonly estimated via data envelopment analysis (DEA). This data-driven non-parametric approach is more flexible than the parametric approach, because it does not require an a priori function for the frontier or a very big sample size. However, DEA loses its discriminatory power when the number of observed airports is small compared with the number of airport inputs and outputs examined, particularly when examining airports within a country. In addition, the endogeneity problem may exist if one attempts to examine the determinants of airport efficiency. This study used the Slack-Based Measure (SBM) DEA-Window Analysis model to address the small sample issue during the first-stage analysis and used an instrumental variable (IV) in the Tobit model to solve the endogeneity issue during the second-stage analysis. Data from a sample of 11 New Zealand airports between 2006 and 2017 were used for analysis. The key findings showed the positive impact of tourism, regional economic development, an airport's domestic networks, airport privatisation, low-cost carrier services and the Christchurch earthquakes on New Zealand airports' performance and efficiency, whereas an airport's international networks has a negative impact. The literature contribution and policy implications are also discussed.