About PersiPov

Project supported by the Luxembourg Fonds National de la Recherche (contract C10/LM/783502) and by core funding for CEPS/INSTEAD by the Ministry of Culture, Higher Education and Research of Luxembourg

Project output

The importance of taking into account longitudinal aspects when analysing poverty has long been acknowledged. Rather than analysing who is poor, the focus of the analysis switches to questions such as who becomes poor and who remains poor. Little was known up to now about the dynamics of poverty in Luxembourg. With the support of the CORE programme of the Luxembourg Fonds National de la Recherche, from 2011 to 2014, this project aimed at shedding light on Luxembourg longitudinal poverty. We report here the main conclusions of PersiPov.

Furthering our understanding of low income dynamics in Luxembourg

Between 2003 and 2008, the poverty rate in Luxembourg remained stable between 12% and 14%. This apparent cross-sectional stability is silent about possible turnovers in the poor population. Indeed, out of this six-years window, close to one individual out of four has been poor at least once, highlighting substantial movement in and out of poverty, while slightly more than 2% have been permanently poor. One objective of PersiPov was to analyse the determinants of these poverty transitions, defined as movements across a low-income threshold, exploiting the “Panel Socio-Economique ‘Liewen zu Lëtzebuerg’” (PSELL3) and taking into account various econometric challenges. Particular attention was paid in disentangling among two set of determinants: individual heterogeneity and genuine state dependence. Persistence in low income may be due to the fact that the poor possess adverse characteristics that increase their risk of being poor and of remaining poor; in this case, persistence into low income may be due to the persistence of adverse characteristics and policies aiming at changing these individual characteristics, like training policies, might be favoured. Persistence in low income may also be the result of a behavioural effect from current poverty on future poverty – what is called genuine state dependence. The mechanisms through which experiencing low income in the current period may increase per se the risk of being low income in future period are, for example, the effect that poverty may have on the depreciation of human capital or on the quality of the social network the poor are confronted to, which may lead to difficulty in finding good quality jobs, reducing in turn the probability of exiting poverty. In this case, policies aiming at reducing poverty, and at preventing individuals from falling into poverty, are likely to have long-term effect.

Past experience of low income account for a large proportion of persistence in low income

In Luxembourg, the risk of remaining poor for individuals that were initially poor is around 70%, while the risk of entering poverty for individuals initially not poor is around 4% - more than 65 percentage point difference. One result of the project suggest that more than half of the level of state dependence (60%) can be ascribed to past experience of poverty while 40% can be attributed to individual heterogeneity. Regarding the latter, we found that employment protects from both entering poverty remaining poor and remaining poor. These results were obtained when taking into account two possible factors that may bias the results: non-random attrition (the poor are usually more likely to leave the panel than the non poor) and initial conditions (individuals more likely to remain poor may be overrepresented in the initial group of poor people). Our analyses confirmed the importance of taking these element into account in the analysis. These results, which are similar to those obtained in other countries, suggest that both policies aiming at reducing state dependence and at changing the adverse characteristics that make some individuals more prone to reproduce the state of poverty may be needed.

Household size and poverty

Another aspect of the project consisted in focusing on the link between household size and poverty. We analysed the extent to which the effect of household on poverty still holds when we take into account two potential econometric problems, namely possible correlation between unobserved factors and observed characteristics and different forms of state dependence in low income. Using correlated random effects models, we found that the number of children of different age groups significantly increases the probability to be in a situation of low income. However, the magnitude of the effect varies across specification.

Can perceptions of financial difficulties affect individuals’ risk of being objectively poor?

Individuals' economic welfare can be assessed both objectively, through the standard income poverty approach, and subjectively on the basis of individuals' perceptions about their financial difficulties. In addition to be distinct but complementary concepts highlighting different dimensions of disadvantage, income poverty and perceived financial difficulties are also likely to be dynamically interrelated. For example, current negative perceptions may be affected by the lasting effects of previous income poverty. Conversely, past perceived financial difficulties may affect an individual income-generating ability (for example, through its impact on individuals’ cognitive abilities or executive control) and thereby current poverty status. Empirical knowledge about the extent to which these concepts are dynamically interrelated is limited and the project aimed at filling this gap. Our results highlight the existence of a positive feedback effect from past perceived financial difficulties on current income poverty suggesting that subjective perceptions can have objective effects on individuals’ behaviour and outcomes.

Taking another perspective on the measurement of poverty

As argued by many scholars, the perceptions of individuals regarding their economic situation constitute a relevant approach to analyse the well-being of residents in rich countries with high living costs. Therefore, PersiPov also aimed at analysing the determinants of transitions in perceived financial difficulties in Luxembourg and the extent of state dependence in this subjective outcome. As for low income transitions, employment and a good education protect from both entering into remaining in and remaining in entering into perceived financial difficulties. Some differences were found in the socio-economic correlates affecting persistence in perceived difficulties and in low income suggesting that perceptions of financial difficulties capture dimensions of disadvantage not covered by low income. A sizeable proportion of genuine state dependence was also found in the case of perceived financial difficulties, highlighting the importance of appropriately taking into account dynamic issues when modelling subjective variables.

The impact of macro factors on poverty dynamics in Europe

Finally, moving to a cross-national analysis, the final objective was to analyse the effect of macroeconomic factors, and especially family policy, in mediating the relationship between the number of children of different age groups in the household and poverty transitions. Using a recently proposed methodology and controlling for the depth of poverty (by including variables measuring the distance between equivalent income and the poverty line), our results suggest that the number of children has a significant effect on the poverty transitions only in some countries and for few distances. This underlines the importance of controlling for the depth of poverty when looking for the determinants of poverty transitions. Moreover, when the mediating effect of macro variables on the role of the number of children is significant, it gives an interesting insight as for its moderating role on poverty transitions.

A fertile ground for future research

The research done in PersiPov has been successful in carrying out analysis describing various aspect of longitudinal poverty in Luxembourg. This provides a good reference point on this topic which can evolve in different directions. One of them is for example the analysis of the channels explaining the issue of state dependence. Indeed, PersiPov has contributed to the recent literature highlighting the existence of state dependence in socio-economic by providing evidence on the Luxembourg case. While this knowledge is important per se, the mechanisms explaining the way state dependence occurs are less well known and need to be understood for efficient policy-making. In the framework of PersiPov we have investigated this line of research through our inquiry into the behavioural-economics approach to poverty which is now expanding and which aims at uncovering how poverty affects the set of options available to individuals. For example, poverty has been found to affect the cognitive functioning of individuals, which leads some authors to propose ways of integrating this finding in policy response formulation. No doubt that these elements might be fruitful venues for future research.