Research project TA/00/02 (Research action TA)
With this project, we intend to build on previous research (Societal choice, structural poverty and social cost) in which we studied a number of social inclusion and exclusion processes, using the Panel Study of Belgian Households (De Blander et al., 2005). A dynamic simulation model was developed in order to evaluate the short-term and long-term effects of policy choices (and the associated social costs and benefits). By way of ‘test’, three concrete anti-poverty strategies were evaluated: improving the coverage of the guaranteed minimum income system, activation policy and, finally, reducing the outflow of unqualified people from education.
The main innovative features of this model, compared to existing Belgian microsimulation models are the following:
(1) the model is a joint Markov model for employment and poverty which takes into account the endogeneity of the initial conditions, including educational attainment. The explicit modelling of behavioural effects allows us to obtain pure estimates;
(2) our simulation model enables the long-term effects of programmes to be evaluated prospectively, rather than relying on purely retrospective (often much-delayed) assessment of past measures;
(3) poverty is modelled multidimensionally (i.e. taking into account the interrelationships between areas of life, such as learning, working, income, etc.), where we consider 3 poverty states: non-poor, minimum income and under-protection (i.e. income below the legal minimum threshold).
Our present proposal covers implementation of the following extensions:
- extension of the database to all waves of the PSBH;
- conceptual expansion of the model in order to take (additional) account of the multidimensionality of poverty. Apart from the standard definition of poverty in financial terms (income poverty), we intend to apply a multidimensional definition building on Dworkin’s concept of ‘resources’ (extended here to human, material, social and cultural capital) and Sen’s concept of ‘functionings’;
- refining the estimation model by using a “mixed GEV” model. A generalised extreme-values model (GEV) enables any correlation among poverty conditions to be modelled without having to rely on numerical integration or simulation. Mixed models allow for the use of the panel characteristics of the data;
- expansion of the model to longer time lags in order to model persistence more effectively;
- expansion of the number of anti-poverty measures, which will be examined closely. As far as the latter is concerned, the emphasis will be primarily on prevention;
- further analysis of social costs and benefits.
The determinants of poverty may include, for example, the effects of life events such as unemployment or family breakdown. In addition, the effects of (other than financial) resources such as schooling, housing, health, social capital, etc., will also be considered.
As far as the anti-poverty measures are concerned, special attention will be devoted firstly to labour market policy. This includes activation measures, ‘making work pay’-strategies, anti-discrimination measures, etc.. Secondly, we will examine social protection systems, in particular relating to unemployment insurance and guaranteed minimum income. Thirdly, attention will be devoted to family policy and, chiefly, measures in favour of single-parent families and the prevention of child poverty. As far as schooling is concerned, in addition to the initial level of education attained, we would also like to quantify the influence of adult learning on poverty transitions. Furthermore, simulations can also be envisaged with respect to housing policy, health policy, etc.
In determining priorities for simulations, a balance will be sought between feasibility and relevance to the National Action Plans for Social Inclusion.
