Eur. Phys. J. B 84, 521-533 (2011)
https://doi.org/10.1140/epjb/e2011-10986-3

Regular Article

Identification of a core-periphery structure among participants of a business climate survey

An investigation based on the ZEW survey data

¹ Department of Economics, University of Kiel, Olshausenstr. 40, 24118 Kiel, Germany
² Kiel Institute for the World Economy, Hindenburgufer 66, 24105 Kiel, Germany

^a e-mail: lux@bwl.uni-kiel.de

Received: 17 December 2010
Received in final form: 10 March 2011
Published online: 4 May 2011

Abstract

Processes of social opinion formation might be dominated by a set of closely connected agents who constitute the cohesive ‘core’ of a network and have a higher influence on the overall outcome of the process than those agents in the more sparsely connected ‘periphery’. Here we explore whether such a perspective could shed light on the dynamics of a well known economic sentiment index. To this end, we hypothesize that the respondents of the survey under investigation form a core-periphery network, and we identify those agents that define the core (in a discrete setting) or the proximity of each agent to the core (in a continuous setting). As it turns out, there is significant correlation between the so identified cores of different survey questions. Both the discrete and the continuous cores allow an almost perfect replication of the original series with a reduced data set of core members or weighted entries according to core proximity. Using a monthly time series on industrial production in Germany, we also compared experts’ predictions with the real economic development. The core members identified in the discrete setting showed significantly better prediction capabilities than those agents assigned to the periphery of the network.