Empirical Estimation and Quantitative Analysis

Abstract

The translog variable cost function (equation 6.2) in our study is specified with one output (Y) revenue passenger miles, one quasi-fixed factor, the capital stock (K) of flight equipment that has been quality adjusted with technological parameters — payload, SFC, range, thrust and passenger capacity — and five variable inputs, labor (L), energy (E), materials (M), business services (S) and other expenses (O). A full description of the model can be found in Chapter 6. Efficient estimates of The Restricted Variable Cost Function (RVCF) over the period 1970-1992 (161 observations) for the seven major carriers were obtained using the Generalized Method of Moments (GMM)¹ estimation algorithm in the TSP (Time Series Processor) econometric software program.² All of the parameters in the RVCF model are identified by estimating a pooled time series cross-sectional translog variable cost function jointly with a revenue equation, and five input demand equations, instead of the value shares of inputs. The input demand equations are subject to the same linear homogeneity and symmetry restrictions as the share equations. Our regression coefficients are therefore in quantity terms rather than input value shares. The techniques of estimating translog cost functions with input demand quantities are described by McElroy (1987)³ and Norsworthy and Jang (1992)⁴. The fitted variable cost function satisfies at every sample point the regularity conditions that it be non-decreasing and concave in input prices.