Spatial diffusion of demographic behaviour

The aim of the communication is to analyse the onset of extra-marital births in each district of Switzerland between 1970 and 2000. The starting time of the spread of extra-marital birth differed according to regions. It has, for example, first increased in the French speaking area of Switzerland and in urbanized areas. We suspect a spatial diffusion process. Data we use is the level of out-of-wedlock births for each district of Switzerland at four moments, 1970, 1980, 1990 and 2000. We define the time of the onset of out-of-wedlock births for each district following a method suggested by Bocquet-Appel and Jakobi (1998). We then estimated an event history model of this time in which we introduce as a characteristic a social distance between districts which already experiment an increase of extra-marital births and those which did not. This social distance is considered as a proxy of the nearness between districts. Provisory results show a significant effect of this distance on the hazard rate. Lesser is the distance and higher is the probability of the onset. This result confirms our hypothesis of a diffusion process on the Swiss territory.

The purpose of this study is to estimate the effect of spatial diffusion on fertility behaviour changes in Belgium during the First Demographic Transition at a fine-scale geographic level. The analyses lie in a new data series from 1886 to 1913 providing yearly fertility indicators at the municipality level. Firstly, this data series will be used in a preliminary analysis of the evolution of fertility behaviour in order to produce a first overview of spatial patterns, diffusion effects and barriers to diffusion. This preliminary analysis is expected to reveal local particularities of fertility changes that are undetectable at higher geographic levels. Secondly, the yearly data series will be used in event-history models in order to estimate neighbourhood effects among municipalities. These models will test whether the presence of contiguous municipalities j that experienced a fertility change at time t−1 has an effect on the probability, for a municipality i, of experiencing the same change at time t. These models will also attempt to estimate the barrier effect of the Belgian linguistic border.

The emergence and diffusion of fertility control strategies as part of the demographic transition are usually not occurring randomly in space and time. Next to individual-level characteristics also prevailing socio-economic contextual conditions as well as geographic characteristics such as distance to centers of the decline seem to be relevant. However, most existing studies on the fertility decline focus either on macro-level trends or on micro-level studies with limited geographic scope. With this paper we aim to contribute to close this existing research gap. We use 100% individual-level samples of the Swedish censuses in the years 1880, 1890 and 1900 which include detailed information on socio-economic status. Multi-level models are applied to link these individuals to contextual information on the local parishes they were living in. Our main research question is whether in this initial phase of the fertility decline in Sweden social distance or spatial distance were more relevant as constraints for the diffusion process. Our preliminary results suggest that in this initial phase of the decline social class differences were putting higher constraints on the diffusion of the fertility decline compared to spatial distances. This is in line with theoretical considerations by Szreter (1996) on “communication communities”.

Within three decades Algerian fertility dramatically fell, from 8 children per woman in the early 1970s to 2.2 in the late 1990s, before climbing up after 2000. As usual, the decline started in cities. But it could not have been so rapid at the national level without diffusing in rural world quickly. We check here such assumption by analysing data available at regional and local levels. Local data are rare and hardly accessible, except those from the 1998 census now published at the “commune” level. A special treatment of 1987 census data for the central part of the country allowed us to analyse changes occurred in the last period of sharp decline at the commune level. It is also very interesting to look at the geographical changes in age at marriage, the rise of which was the first mean of fertility reduction, and to discuss the relation between the two behaviour changes.