Group interaction in research: a general nesting spatial model with heteroskedastic shocks

Research output: Working paper

Author(s)

Department/unit(s)

Publication details

DateUnpublished - 2013
Number of pages35
Original languageEnglish

Abstract

This paper tests the feasibility and empirical implications of a heteroskedastic spatial econometric model with a full set of interaction effects applied to research productivity of over 2500 individual researchers in economics, business, and finance employed at 83 universities across Austria, Germany and German speaking Switzerland over the period 1999-2008. It is found that if the regressors satisfy mild conditions, identication is generally assured, that
heteroskedastic model versions outperform their homoskedastic counterparts, and that spatial econometric models with a limited number of spatial interaction effects, especially the well-known SAR, SEM, SLX and SAC models, demonstrably lead to incorrect inferences.
The question whether interaction effects among the dependent variable or among the disturbances are empirically more important remains difficult to answer straightforwardly. In view of the specific context of the empirical application, we favour the SDM against the SDEM specification and find that a 1% increase in the proportion of post-docs in a department raises the average research productivity of their colleagues by 0.086%, while a similar increase in the proportion of colleagues that do not publish in leading journals has a downward effect of 0.054%.

Discover related content

Find related publications, people, projects, datasets and more using interactive charts.

View graph of relations