Solow Residuals without Capital Stock


For half a century, the Solow decomposition has served as the standard instrument
for measuring total factor productivity growth in economics and management.
Using synthetic data generated from a stochastic growth model, we show that measurement
error, in particular with respect to capital stocks, can seriously confound estimates
of TFP growth. We explore the quantitative extent of this measurement error when the
initial condition of capital is unknown to the analyst and when depreciation is stochastic.
We show that two alternative procedures which eliminate capital stocks from the
decomposition outperform the original Solow decomposition. This improvement is inversely related
to the length of the capital series and the distance from the steady state. We then
apply an example to the measurement of technical change in the ICT and retail trade
sectors as well as transition countries.

For half a century, the Solow decomposition has served as the standard instrument
for measuring total factor productivity growth in economics and management.
Using synthetic data generated from a stochastic growth model, we show that measurement
error, in particular with respect to capital stocks, can seriously confound estimates
of TFP growth. We explore the quantitative extent of this measurement error when the
initial condition of capital is unknown to the analyst and when depreciation is stochastic.
We show that two alternative procedures which eliminate capital stocks from the
decomposition outperform the original Solow decomposition. This improvement is inversely related
to the length of the capital series and the distance from the steady state. We then
apply an example to the measurement of technical change in the ICT and retail trade
sectors as well as transition countries.

Principal investigators
Severgnini, Battista (Details) (Economic Theory II)

Duration of project
Start date: 01/2008
End date: 12/2008

Last updated on 2022-07-09 at 19:05