KOF working papers // KOF Swiss Economic Institute, ETH Zurich 278
This paper analyzes the labor productivity of Swiss university departments between 1995 and 2007. Using a parametric input distance function we estimate and decompose the Malmquist productivity indexes in line with Fuentes et al. (2001) and Atkinson et al. (2003). By contrast to those studies, this paper proposes a panel data specification to account for unobserved heterogeneity across production units. The adopted model is a mixed-effects model with department fixed effects as well as random coefficients for time variables. We also use an autoregressive stochastic term to model inefficiency shocks while allowing for gradual improvement of persistent inefficiencies. The results indicate a negative trend in overall productivity measured by Malmquist index, particularly after 2002, with an average productivity decline of about one percent per year. A major part of this productivity decline coincides with the recent developments in Switzerland's higher education system following the adoption of the Bologna agreement. However, the results do not provide any evidence of statistically significant relationship between productivity and reforms. Our decomposition analysis suggests that the observed productivity decline could be contributed to technical regress but also to a rising inefficiency with a relatively high level of persistence. The results also point to various patterns across different fields. In particular, economics and business departments and law schools show the lowest performance, whereas science departments stand out as an exception with productivity improvement.
Swiss Universities Parametric Distance Function Heterogeneity Malmquist Index Decomposition Autocorrelation