Poverty and social exclusion: two sides of the same coin or dynamically interrelated processes?

Abstract

There is growing interest in the analysis and measurement of social exclusion, to complement the static and dynamic literature on income poverty. On theoretical grounds, social exclusion and income poverty are seen as different processes, but with closely interrelated dynamics. However, our empirical understanding of the way these two processes dynamically interact at the individual level is still very limited. To shed some light on the issue, we use a dynamic bivariate probit model, controlling for unobserved heterogeneity and Wooldridge (2005)-type initial conditions. Both the first- and second-order Markov dynamics are examined. We estimate the model using the Italian sample of the European Community Household Panel (ECHP), waves 1–8, and find a sizable extent of state dependence in both poverty and social exclusion. Moreover, there are dynamic cross-effects implying that poverty and social exclusion are mutually reinforcing. Social policies aimed at eradicating poverty and avoiding individuals’ social and economic marginalization should take these interaction effects explicitly into account.

Notes

¹ We chose the age 60 threshold, rather than 64 as commonly in Labour Force Survey (LFS), in light of the low-average age of retirement, at about 58 during the 1990s, one of the lowest in Europe. Results were only marginally affected by the use of the higher threshold.

² We tested the robustness of the thresholds chosen re-performing the analysis using alternative definitions of the cut-off points: 40% and 60% of the median distribution. Our main results concerning poverty dynamics, and its relation to social exclusion, were essentially unaffected by the alternative cut-off points.

³ A number of papers have analysed poverty from a multidimensional perspective and have identified individual deprivation in terms of enforced lack of various durable goods and essential services (e.g. Pérez-Mayo, 2005; Borooah, 2007; D’Ambrosio et al., 2010). Note that deprivation so defined is part of the broader notion of social exclusion we focus in this article.

⁴ But, how many deprivations are necessary to define an individual as socially excluded? According to Sen (2000) we need simultaneous deprivations. The decision about how many deprivations (in our case ‘at least two’) is arbitrary, and partially depends on how many relevant functionings we consider in the analysis. We find that about 8% of the sample suffer at least two deprivations simultaneously over the period of analysis and slightly more than 1% is deprived in at least three dimensions. While in the article, we use deprivation in at least two functionings as our threshold we provide some robustness analysis in Section V.

⁵ For example, s/he can afford a durable good or has an indoor flushing toilet or does not have pollution in the area where s/he lives.

⁶ See Brandolini and D’Alessio (1998) for more details about the use of equal weights and alternative weighting structures.

⁷ Note that we tested the robustness of the thresholds chosen re-performing the analysis using alternative definitions of the cut-points: 40% and 60% of the median distribution and 40%, 50% and 60% of the mean distribution. Comparisons of the results suggest that the higher the cut-points, the higher social exclusion. However, the results about the dynamics of social exclusion are statistically equivalent for all the considered cut-points.

⁸ Note that the nonlinear Wald test is not invariant with respect to representation. In fact, an alternative representation of the null hypothesis is as follows: . If both representations lead to similar test statistics and p-values, one is able to accept or reject the hypothesis of proportionality of the coefficient of the two equations. The results of the tests are available upon request from the authors.

⁹ See Heitmueller and Michaud (2006) and Miranda (2007) for applications of the model of Alessie et al. (2004) in different contexts.

¹⁰ Heckman recommends the inclusion of exogenous instruments in the initial condition equations, whenever available. This is done in a few studies, for example Cappellari and Jenkins (2004) and Stewart (2007) use information on the respondent's parental background when the respondent was 14 as exogenous instruments. However, this pre-sample information is not available in the ECHP. Note that Alessie et al. (2004) too do not include exogenous instruments in their initial condition equations.

¹¹ For our simulated Maximum Likelihood Estimation (MLE) we used 2RN Halton draws (Train, 2003), where N is the number of individuals and we set the number of replications, R, equal to 60.

¹² The Alessie et al. model required approximately twice the computing time of the Wooldridge-type estimator. The programmes to implement the models are available from the authors upon request.

¹³ Note that time-constant variables (e.g. gender) may be included either in model 10 or in model 1–2. It is not possible to separately identify the effect of time-constant variables on unobserved heterogeneity and on the latent variables.

¹⁴ The estimates of these and further robustness checks are available from the authors upon request.

¹⁵ Note that the dynamic random-effects probit models and the pooled probit model involve different normalizations (Arulampalam, 1999) and, therefore, the estimated coefficients of the former need to be rescaled to be compared with the estimated coefficients of the latter. In our case the coefficients of the random-effects models need to be multiplied by an estimate of .

¹⁶ More generally, our econometric experimentation showed that estimates of true state dependence and the dynamic spillover effects are very much robust to the inclusion/exclusion of other controls in x_it .

¹⁷ Note that experiencing poverty and/or social exclusion in the past, even if not the year before, is nonetheless harmful: in Table 12 the probability of being poor (socially excluded) in t is about 0.2 (0.1) for those who experienced either poverty and social exclusion in t − 2 but not in t − 1.