Solow Residuum und Kalman Filter

The State-Space models provide an accurate and innovative method for
estimating the rate and the biases of the technological change. The implementation
of this methodology requires initial conditions which are often not
possible to determine through the traditional econometric techniques and can
confound the productivity estimates, especially when the inputs are affected
by measurement errors. Applying the Kalman Filter to artificial data, we propose
a computation of the initial condition for productivity exploiting some
properties of the Malmquist index and the panel structure of the data. Moreover,
we compare our results in the Bayesian framework using Gibbs-sampling
and provide more robust results. The empirical application is to Danish industries
data.