The evolution of the commuting network in Germany: Spatial and connectivity patterns
Network concepts have received a great deal of attention in spatial economics in recent decades. Examples are the well-known ideas of the network economy (Shapiro and Varian 1999) and the knowledge economy (Cooke 2001). Networks are based on the existence of interactions (which may occur on multiple levels) between agents operating in a network, giving rise to synergistic effects. The effects of these interactions are oĕen investigated and modeled by considering, amongst other things, network externalities or spillover effects (Yilmaz et al. 2002). The labor market literature is no exception to this trend: spatial job matching processes have been widely studied in a social network framework (Montgomery 1991), while work-induced mobility (commuting) has been investigated in both an urban and a regional network context (e.g. Russo etal. 2007; ăorsen etal. 1999; Van Nuffel and Saey 2005).
The directionality of commuting Ĕows has clear implications for urban form and for the development of regional networks of cities. Commuting has long been studied with these implications in mind, in particular concerning locational and developmental trends leading to either the monocentric (central) city or the polycentric city. Ʋ The latter perspective has been developed by observing the various deconcentration trends observed in many major cities (e.g. Bar-ElandParr2003;Fujitaetal.1999). Thesetrendsarenowincreasinglyevidentatlargerspatial scales leading, for example, to the idea of “network cities” (Batten 1995). In this context, horizontal relations between cities tend to emerge (Van der Laan 1998; Wiberg 1993). The emergence of network cities also results from improvements in transportation systems and accessibility, which diminish the importance of distance. Remarkably, Papanikolaou (2006) suggests that spatial structure alone does not strongly account for different commuting distances. Asaresultoftheongoingprocessdescribedabove,localhierarchies—originallyconsistentwith monocentrictheories—aresubjecttoconstantchangeandexhibitmoredecentralizedurbanregions; examplesare the Randstad areain the Netherlands (see Clarkand Kuijpers-Linde1994) or the emergence of edge urban areas (edge cities) (see Phelps and Parsons 2003).
In this framework, there have been many experiments with network-modeling approaches to the analysis of commuting Ĕows. ăorsen et al. (1999), for instance, examine the effects of transportation infrastructure and spatial structure on commuting Ĕows in a network of cities. Russoetal.(2007)usecommutingĔowsinGermanytoidentify“entrepreneurialcities”inGermany. Van derLaan (1998);Van Nuffeland Saey (2005) investigatetheemergenceoflocaland regional multi-nodality for the Netherlands and the Flanders area, respectively, on the basis of commuting Ĕows. In particular, van der Laan đnds that more horizontal (non-hierarchical) relations emerge for regions with modern manufacturing systems, while the (hierarchical) status quo is preserved for peripheral, less advanced regions.
On the basis of the aforementioned developments, the present paper investigates, for the case of Germany, the relevance in the đrst place, of the volume and distribution of the commuting Ĕows, and, in the second place, the connectivity and topology of the same network. In particular, we aim to assess whether the geographic commuting system and its hierarchies, in the years 1995 and 2005, are affected by network topology and its changes over time. In other words, we aim to investigate whether the most mobile districts are also the most connected. Our inspiration for studying the commuting network from a connectivity perspective is the idea that the network distribution of mobility can help explain other relevant economic phenomena, such as variations in key labor market indicators or production levels. The importance of spatial interaction (Niebuhr 2003), and primarily of commuting (Patacchini and Zenou 2007), for the development of regional labor markets has been stressed in the recent literature. Moreover, distance has already been shown to lead to greatly diminished labor market interactions, when over a certain threshold (e.g. Badingerand Url 2002), and accessibility is also seen as a possible source of spatial dependence (Anselin and Florax 1995). In this framework, the value added of network analysis is that its set of analytical tools supports an intuitive inspection of commuting-related topology and accessibility. Given these premises, our aim is to digdeeperintheconnectivityperspectiveinordertoimprove ourunderstandingofthespatialeconomic perspective. The paper is structured as follows: Section 2 brieĔy describes recent developments in network analysis, on which some of our empirical analyses are based. Section 3 illustrates a preliminary spatial analysis of commuting Ĕows in Germany, while Section 4presentstheresultsof thenetworkmodelingexperimentundertaken. Section5thenpresents a comparative multicriteria analysis that addresses the change in hierarchies in the main German districts. Finally, Section 6 concludes the paper with some đnal remarks and suggestions for future research.