Geoscience Reference
In-Depth Information
activity: this large gap between the level of modeling (households) and the level of
questioning (economic sectors) was then an innovation in social sciences. In the
1980s, the concepts of time-geography developed by Hägerstrand in his advocating
lecture “What about people in Regional Science?”,
5
are introduced in
microsimulation by geographers [HOL 89, HOL 07a, MOE 03] who develop
dynamic spatial microsimulation models. These models, centered on the individuals'
representations and decisions, allow us to endogenize the origin of changes in the
demographic, social and economic structures of a society (the aging of individuals
leads to more pensioners, for example).
Microsimulation consists of working on an artificial population. The initial
situation corresponds to the state of a population observed at a given time
6
or a
construction in the case where individual data are not accessible. In the latter case,
an artificial population is constructed by cross-sectioning statistical tables of
demographic, social and economic data at the aggregate level of spatial units
[CLA 87, WIL 98, STR 05]. The objective is to have a complete representation of
the population [CAL 00], and to ensure that this representation is consistent from the
point of view of the interdependencies between individuals' spatio-temporal
trajectories in terms of family and generational compositions
7
[HOL 07b].
Individuals are described by a set of attributes corresponding to the data
provided by a census or a large survey (age, sex, level of education, income,
occupation, place of residence and place of work). Transition rules define how these
attributes vary over time. The transition from the state
t
to the state
t +
1
is
formalized in the following way:
At each iteration of the model, each individual is successively examined in order to
determine the changes in his/her state. Some changes simply reflect the effects of the
passing of time [DAV 01], for example the increase in the age. Others are more
complex and relate to either quantitative changes of certain attributes (for example,
5 Lecture delivered at the European Regional Science Association (ERSA) in 1969.
6 Such is the case of the applications of Holm
et al.
's team [HOL 02] in Sweden, where
researchers have had access to a complete database of the Swedish population at
the individual level, with a localization provided at the 100 m cell level. This case is rare;
complete individual data is seldom accessible.
7 In order to reproduce the process of family formation in a consistent way, empirical work
about family planning and the decision to have children were used. Studies about the
interactions between the members of a family were also mobilized, notably about the
constraints imposed by the choice of one of the members on those of the others (a parent
who migrates to change job takes with him, in general, the other members of the family)
[HOL 07b].
