Empirical economists using flexible functional forms often face the disturbing choice of drawing inferences from an approximation violating properties dictated by theory or imposing global restrictions that greatly restrict the flexibility of the functional form. Focusing on the cost function, this
A globally concave, monotone and flexible cost function: derivation and application
β Scribed by Asher Tishler; Stan Lipovetsky
- Publisher
- John Wiley and Sons
- Year
- 2000
- Tongue
- English
- Weight
- 166 KB
- Volume
- 16
- Category
- Article
- ISSN
- 1524-1904
No coin nor oath required. For personal study only.
β¦ Synopsis
Empirical analysis of demand often requires that the functional form of the cost function be speci"ed in advance. The form of the function has to be both consistent with economic theory and su$ciently #exible to accommodate the data. Recent research has indicated that #exible forms do not always generate empirically credible elasticity estimates, and often fail to satisfy concavity and/or monotonicity. A priori imposition of both global concavity and monotonicity is generally assured only with functional forms that are not #exible (the CES function, for example). However, such forms produce a poor statistical "t, and may be di$cult to estimate and interpret. Here we develop the CES}DCES cost function*which is globally concave and monotone at any possible vector of input prices. We prove that the CES}DCES is a #exible cost function when the number of inputs is two or three. We show how to extend the analysis to acommdodate more than three inputs. Using three data sets we estimate the parameters of the CES}DCES cost function for two and three inputs, and compare its performance to that of the CES}GBC function developed in Tishler and Lipovetsky (ΒΉhe Review of Economics and Statistics 1997; 638}646).
π SIMILAR VOLUMES
A new approach to visualizing spectral densities and analyzing NMR relaxation data has been developed. By plotting the spectral density function, J(Ο), as F(Ο) = 2ΟJ(Ο) on the log-log scale, the distribution of motional correlation times can be easily visualized. F(Ο) is calculated from experimental