7.
Commuting distance in the concentrated model of employment
The following three sections present the results of the theoretical models for each of the
urban regions and the comparison with actual commuter flows. First, commuting distance
related to the basic model with full employment concentration will be looked at. Next, the
outcomes of the deconcentrated model will be considered. The possible reduction in
commuting distance will be calculated, assuming the deconcentrated distribution of
employment locations. By commuting distances in the case of full concentration, and the
reduction realized in the case of deconcentration of employment, the minimal commuting
distance is calculated as well as the degree of ‘wasteful commuting’. Finally, the cross-traffic
model is analysed in a similar way.
Starting with the basic model with full concentration of employment (see figure 2),
the region of Amsterdam serves as an example for calculating the commuting distances. With
its total area of 2941.32 square kilometres Amsterdam is the largest of the four regions. The
first step was the calculation of the distance of the city-centre (x=1) towards the edge of the
urban region (xmax) with means of equation (2). For Amsterdam this showed up to be 30.6 km.
The next step was the potential population density function. The potential labour force (P)
in the Amsterdam region is 1,701,104 persons. The constant (C) and the coefficient (τ) are
calculated by a loglinear regression.
C = ln 10,816,546 = 49839
τ = -0.28
Substituting this in the potential population density function results in:
P(x) = 49839 x-0.28 R2 ≈ 0.07
(21.76) (1.94)
(t-values in brackets)
R2 is the expected exponential relationship between the actual and the expected dispersion of
the residential locations of the potential labour force. Starting from the potential population
density function, together with distance x and the total potential labour force, the average
distance of commuting in case of a concentrated pattern of employment is calculated (see
equation (2) with limits x=1 and xmax).
The results for Amsterdam and the three other regions are presented in table 3, which
shows that in The Hague the average distance is the lowest and in Utrecht the highest. The
level of explanation of the concentrated model for The Hague reaches a level of 0.65.
Amsterdam scores very low and clearly distinct from that of the others. Actual commuting in
the region of Amsterdam differs substantially from the suppositions of the concentrated
model. Moreover, the table shows that if C is higher, τ becomes higher too: the gradient
becomes more negative. Particular τ can be related to the level of suburbanization of the
various urban regions. An urban region is more suburbanized as τ lessens with the extreme case
of τ= 0, in which density is uniform (see Mills 1992). The table shows that Amsterdam has
the most flat overall pattern of the population density. This parallels other studies in which
the Amsterdam region, followed by Utrecht, is characterised as more ‘advanced’ in the
deconcentration process than Rotterdam and The Hague (see Van der Laan, 1998).
16