2 a djustments the regression model described in the

This preview shows page 101 - 103 out of 139 pages.

2. A DJUSTMENTS The regression model described in the previous section relates traffic volumes between any arbitrary county-pair to socioeconomic parameters and the bilateral air services agreement. Variables (2)-(7) together define the incremental traffic growth resulting from total or partial liberalization of the air services agreement. The model demonstrates the importance of liberalization to the growth of international air traffic. However, the idealized model poses several challenges to any attempts to apply it to specific situations. The model uses a double-logarithm specification. This reflects the interdependency of the exogenous variables. However, such models tend to dampen any disparities between predicted and actual traffic observations. Even a small residual of such a model can result in a very large disparity when the coefficients appear as exponents. The coefficient corresponding to the Bermuda capacity control was not statistically significant. Although lower than the predetermined capacity coefficient, it is believed that the model may still overstate the traffic stimulus arising from the abolition of Bermuda controls. The predictive model therefore assumes that a Bermuda clause is half as detrimental as a capacity stipulated directly in the bilateral agreement. The raw model
75 also suggested that a very large stimulus of 75 percent or more would result from liberalizing price clauses (Variable 5). The results, when applied to several country-pairs, appeared excessive. A separate analysis, isolating this term alone, suggested a stimulus of 4.1 percent was more appropriate. This finding was superimposed on the predictive model, although it is fully recognized that this ad hoc measure is a methodological compromise. The predictive model reflects a view that restrictive bilateral agreements, whatever their economic justification, are the global status quo. It is usually thought that any unjustified change in the status quo is a more serious error than mistakenly retaining the current circumstances. This in turn means that an error in over-estimating the consequences of liberalization is more serious than any under-estimation. The model therefore tests each prediction. Should the predicted stimulus exceed a particular critical value, the stimulus is reduced to that particular value. No such correction applies if the model appears to under- predict the consequences of liberalizing a particular country pair. Furthermore, a “grand limit” capped the total growth resulting from a full liberalization. The limits were estimated by taking a sample of 600 country-pairs in various stages of liberalization. Each attribute of the relevant bilateral agreements was examined in turn, and subject to a step-by-step liberalization. The model calculated the conditional expectations of traffic resulting from each perturbation of the bilateral for each observation, generating a series of calculated stimuli. For each attribute in the bilateral, a

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture