Individualisation and Poverty Over Time: The Italian Case (1985–2011)

ABSTRACT

In the last few years, much sociological debate has focused on individualisation theory, especially on Beck's risk society version. According to this theory, contemporary social change can be interpreted as the progressive weakening of the influence of social structures on individual behaviour. Individualisation theory has been adopted in many fields of study (voting behaviour, consumption behaviour, etc.). Although much of the debate has a theoretical character, there have been some attempts to empirically assess individualisation theory. As far as poverty is concerned, scholars supporting individualisation theory, as well as scholars opposing it, have adopted one of the following methodological strategies: highlighting the role played by individual variables (especially by life course variables) rather than structural variables; controlling for individual rather than structural variables. Both these approaches focus on short observation windows; however, it is necessary to consider long periods in order to assess the core of individualisation theory, i.e. the decreasing influence of social structures. Our approach assesses the change (rather than the stability) of the individual-level relationship between structures (occupational classes, education, etc.) and poverty over a long time period. This changing-parameter model is implemented through multilevel modelling with families at level one and years at level two. The analysis focuses on the Italian case and it is based on data from the Family Expenditure Survey (Indagine sui Consumi delle Famiglie) that was collected by the Italian Statistical Institute (ISTAT). It covers the period from 1985 to 2011. The results seem to indicate that there is stability in the relationship between structures and poverty.

Key words:

• individualisation
• poverty
• longitudinal data
• multilevel analysis
• Italy

Notes

1. Translated by the authors.

2. The different poverty patterns were derived from a latent class analysis of the complete ECHP.

3. In all these papers, the authors have analysed how structural variables mediate the effects of life course variables, but it is also possible to find in the literature examples of the reverse approach, i.e. controlling for individual variables (Layte and Whelan 2002: 228–230).

4. The main advantage of the diary technique of data collection is that it furnishes reliable information on low-level expenses or daily consumption activities (Fortini 2000).

5. See also http://www.oecd.org/redirect/dataoecd/61/52/35411111.pdf. Used in Italy is another scale of equivalence, called Carbonaro (1985). In this alternative formula, the weights are based only on the size of household. For 1 member the weight is 0.6, for 2 members it is 1, for 3, 1.33, for 4, 1.63, for 5, 1.9, for 6, 2.16 and for 7 or more it is 2.4. We also considered this scale to test the robustness of analysis. The results can be obtained from the authors.

6. The reference person is the respondent of the familial section of the questionnaire in the surveys (he/she corresponds to the traditional concept of head of household).

7. In preliminary analysis we considered other variables to describe family life course more accurately (children 0–5 years old in the household, children 0–14 years old in the household, individuals 65 or more years old in the household, individuals 75 or more years old in the household) but their effects were mostly captured by other variables already in the models (age of the reference person, size of households) due to collinearity problems.

8. Southern regions include Sicily and Sardinia. We tested a geographical categorization in five areas, but north-western, north-eastern, and central Italy showed similar coefficients. Likewise southern Italy and islands (Sardinia and Sicily) showed similar coefficients in the models.

9. While in some other countries it is usual to aggregate primary and lower secondary education, here we preferred to aggregate upper secondary and tertiary levels, given the low percentage of graduates that characterises Italy (OECD 2012).

10. We considered only the households in which at least one member was employed. We attributed to the household the social class of the individual with the higher employment status (following the usual hierarchy: bourgeoisie, white collars, petty bourgeoisie and manual workers).

11. This dummy variable varies only at the second level.

12. We used MLWIN software.

13. If the number of units in the first level is large and contexts are homogenous in size, and the dependent variable is not too asymmetric, bias in estimates are not relevant (Goldstein and Rasbash 1996).

14. Similar results were reached by Albertini (2013: 34) on the basis of different methods (index decomposition and quantile regression analysis) and focusing on economic inequality: ‘Descriptive statistics, therefore, signal that in the last three decades the relation between individuals’ social class and economic situation has not weakened’.

Additional information

Notes on contributors

Ferruccio Biolcati-Rinaldi

Ferruccio Biolcati-Rinaldi is Assistant Professor of Sociology at the University of Milan where he teaches undergraduate and doctoral courses in methodology of social research and data analysis. His research interests include poverty and income support policies and programmes evaluation.

Simone Sarti

Simone Sarti is Assistant Professor at the University of Milan where he teaches undergraduate courses in methodology of social research and society and social change. His studies include social stratification, social vulnerabilities and inequalities in health.