Cobb-Douglas Production Function Revisited, VAR and VECM Analysis and a Note on Fischer/Cobb-Douglass Paradox
22 Pages Posted: 15 Feb 2012 Last revised: 27 Nov 2012
Date Written: September 20, 2011
Abstract
Cobb-Douglas production function is a basic function in growth models. The modeling in this paper showed that VAR is stable; KPSS test showed that output, capital and labor are not trend stationary. Johansen’s co-integration test showed that a requirement for Fischer/Cobb-Douglass paradox to work is met at 3 lags, there factor shares are I(0). The Fisher/Cobb-Douglas Paradox is based on constant factor shares. (In terms of time-series analysis, such constancy is equivalent to factor shares being I(0). The Fisher/Cobb-Douglas Paradox is thus why the estimated σ equals unity independent of the underlying production technologies generating the simulated data. At 4 lags however these variables are I(1) variables i.e. Cobb-Douglass is not CES function anymore. ADF test for factors of production showed that natural logarithm of capital is stationary variable, while log of labor is not-stationary except at 10% level of significance. Adjustment parameters showed that labour responds more/faster than loutput (log of GDP) and lcapital on if there is change/shock in the system. VECM model failed the stability eingevalues test.
Keywords: Fisher/Cobb-Douglas Paradox, cointegration, VAR, VECM, ADF test, unit root
JEL Classification: 040, B23
Suggested Citation: Suggested Citation