1. Introduction.
In some sense, the world has become smaller. As it until half a century ago was commonly
accepted that people lived and worked in the same locality, it has become a widespread feature
to have distances between workplace and home. Tkocz and Kristensen (1994) report average
commuting distances for employees from a number of Danish cities ranging from 7 to 21
kilometres. Comparing this to the average size of Danish municipalities of 157 square kilometres,
i.e. an average cirkular radius of 7 kilometres, the prevalence of commuting becomes evident.
An increasing amount of commuting has been facilitated by technological conquests in the form
of refined infrastructure and faster and more comfortable modes of transport (Kristensen 1997).
Further, commuting has been necessitated by specific spatial patterns of industries which
contradict the spatial patterns in peoples preferences for living places. Many people - especially
when educated - prefer to live in recreative areas away from city centres (Graversen,
Hummelgaard and Nielsen 1997, Kristensen 1997). On the other hand, rural areas with low
wages are preferred locations for many industrial branches whereas other branches seem to prefer
urban areas (Kristensen 1997). Further, some cross-border studies (De Falleur and Vanderville
1994, Bacher, Kj0ller and Mohr 1995, Hansen and Schack 1997) indicate that people commute
from areas with low wages and high unemployment to areas with higher wages, lower
unemployment and more prosperous industrial structure. These studies, too, indicate that house
prices, rents for flats as well as amenities causes people to live away from urban areas where
these are relatively high.
Commuting emerges from contradictions between the industrial localization preferences and the
preference - and need driven choices of residences made by the workforce. To this point, the
literature has been able to provide empirical evidence on the living place choice and commuting
behaviour of the workforce, but no significant empirical findings has been provided on the nature
and impact of industrial localization. For the case of Denmark, this shortcoming is due to a lack
of data on a municipal base, as such data are mainly provided on an aggregated county level. The
present paper intends to fill this gap, using spatial autoregressive and spatial Durbin model
specifications. Specifically, the impact of workforce characteristics on in- and out-commuting
will be measured using a causal linear regression specification. In this specification, the impact
of spatial patterns outside the model - i.e. industrial clustering patterns - will be traced by adding
spatial autoregressive and spatial Durbin processes.
2. Model suggestions for commuting in Denmark.
The aim of the present study is to investigate the importance of various determinants for the
intermunicipal commuting variations between the 275 Danish municipalities. To facilitate a
detailed investigation of these determinants, the out-commuting and the in-commuting as well
as the net (in-) commuting are analyzed. Following suggestions from the literature, the in-
commuting and the net-commuting are expected to be influenced by the number of working
places, urbanization degree and personal incomes. Further, high unemployment in neighbourhood
municipalities should give rise to a higher in-commuting. For the out-commuting, determinant
factors are expected to be unemployment, a large workforce population share and low availability
of workplaces. Family structures are expected to influence the propensity to commute, as a low
average number of inhabitants per household dampens this propensity. The causal relationship
between commuting patterns and sociodemographic conditions will be investigated using the