Model Specifications
The new econometric research tested a number of different estimation methods and model specifications to provide explanations of economic phenomena within each given market. In all models, traffic was the dependent variable. The different specifications used were9:
• Ordinary Least Squares (OLS). OLS is a method relating passenger traffic to air travel prices,
income levels and other variables, while minimising the variance (randomness) of the estimates. The regression analysis allows the relationship between traffic and air travel prices to be isolated and quantified while controlling for other factors that may impact air travel, such as GDP, population levels, route distance and seasonality.
• Two-Stage Least Squares (2SLS). 2SLS is often used to improve the consistency of elasticity estimates when explanatory variables are believed to be correlated with the regression model’s error term.
• Autoregressive Distributed Lag (ARDL). An autoregressive ARDL model uses similar explanatory variables to the OLS model, but also uses lagged values of the traffic variable. The inclusion of lagged values accounts for the slow adjustment of supply (in the form of capacity) to changes in the explanatory variables.
Explanatory Variables
The econometric research estimated regression models that included a variety of explanatory variables to search for the best fit:
• Average Price. Average air travel price was used to measure the price of air travel, reflecting average route prices over the period reported. The average price variable appears in all model specifications.
• Gross Domestic Product. GDP is used to measure the effect of income on air travel. GDP estimates
are widely available, providing a variable that can
be consistently defined between regions and over time. Within the US, GDP estimates are available at the city pair level. Regression models using the IPS data set used UK national GDP. Regression analysis for all other regions used national GDP values,