The transmission of poverty across generations is a long-standing concern for national and EU-wide anti-poverty strategies. This column uses retrospective data for 30 European countries to compare the strength of the association between growing up in poverty and the risk of current poverty. Countries differ in the strength of this relationship: Scandinavian countries have low intergenerational poverty association, whereas Romania and Bulgaria have the highest. It also finds a positive relationship between intergenerational poverty persistence and parental poverty levels, which can exacerbate the reproduction of poverty across generations.
The transmission of poverty across generations and the role of intergenerational poverty traps is a long-standing concern of poverty research across economics, sociology, and social policy as well as national and EU-wide anti-poverty strategies. Despite this, most of the comparative empirical literature on intergenerational transmission in rich countries focuses on mobility in terms of earnings, income, education, or social class (e.g. Corak 2013, Bukodi et al. 2020, and recent Vox contributions from Doepke et al. 2022, Qureshi et al. 2022, Marrero et al. 2024) rather than poverty per se. The much-discussed relationship, labelled the ‘Great Gatsby curve’ by Alan Krueger, positing that intergenerational mobility is lower where cross-sectional inequality is higher thus relates to earnings or income, from which it is not possible to simply ‘read off’ implications for intergenerational poverty persistence (Nolan 2024).
One way to fill that gap relies on data from longitudinal surveys, as in the recent paper by Parolin et al. (2023), bringing together long-running panel data for five rich countries to measure poverty in income terms in a comparable fashion across generations and countries. Given the limited number of countries where these data exist, in a new paper (Bavaro et al. 2024) we take a different approach, instead using retrospective data for 30 European countries to compare the strength of association between growing up in poverty and the risk of current poverty across this substantial set of countries. Additionally, we assess whether a ‘Great Gatsby-type’ relationship between poverty transmission and poverty levels is present.
The data are from the ad hoc module on intergenerational transmission of disadvantages included in the 2019 EU statistics on income and living conditions (EU-SILC). This sought retrospective information from respondents on household circumstances when they were about 14 years of age such as whether the household could meet specific basic needs and experienced financial hardship, together with parental education and occupation, but not household income. The central analytical challenge is how best to employ this retrospective information to produce proxy measures of poverty for the parental household. We construct both a ‘narrow’ measure based solely on basic deprivation and financial hardship and a ‘broad’ measure also incorporating parental education and social class. Analogously, for current poverty we employ as a ‘narrow’ measure the conventional at-risk-of-income-poverty rate and as a ‘broad’ measure the risk of poverty and social exclusion as captured by the EU’s high-level social inclusion target (encompassing low income, material deprivation, and household worklessness).
Comparing what we find with more versus less restrictive poverty measures allows us to assess the robustness of the results, which are generally quite similar. Therefore, here we focus on those using the narrower measures. We employ logistic regressions to estimate the marginal effects of growing up in a poor rather than non-poor household on the predicted probability of being currently poor. In Figure 1 we show how countries rank in terms of this measure of intergenerational poverty association. Scandinavian countries have particularly low marginal effects whereas those with the highest are Bulgaria, Lithuania, Romania, and Slovakia. However, the other Baltic countries – Estonia and Latvia – are in the lowest one-third. Few central or Eastern European countries are towards the top, but Hungary, Poland, and Slovenia are around the middle. Southern/Mediterranean countries are in the bottom half but not consistently at the bottom as some previous mobility studies suggested. Among ‘continental’ countries, Switzerland and France have relatively low marginal effects, Germany and the Netherlands are closer to the middle, and Belgium has a relatively large marginal effect. These patterns thus do not fully align with conventional clusters of countries in geographical or welfare regime terms.
Figure 1 Intergenerational poverty association country rankings
Note: Marginal effects are estimated using a logistic regression between current and parental poverty. Countries are ranked from lowest to highest and include the 95% confidence interval. An estimate for the pooled sample labelled ‘EU’ for convenience is also included.
Source: Authors’ elaborations based on EU-SILC 2019.
The ‘Great Gatsby curve’ has played a prominent role in recent research and debate about intergenerational mobility, with the causal processes seen as underpinning such a relationship being many and various (e.g. Di Prete 2020, Durlauf et al. 2022). Many of these causal channels are highly salient in thinking about poverty, so it is helpful to be able to assess whether an analogous relationship between intergenerational poverty persistence and poverty levels is evident. Figure 2 plots intergenerational poverty persistence and parental poverty level by country, which does suggest a positive correlation between them. This suggests that intergenerational poverty traps relating to household circumstances in childhood and adolescence may indeed be even more salient where poverty at that point is higher, exacerbating the reproduction of poverty across generations.
Figure 2 Association between poverty persistence and parental poverty
Note: Vertical axis shows the marginal effects of parental poverty on current poverty, estimated via logistic regression, and horizontal axis shows the parental poverty rate. Linear fit and the Pearson correlation between them are also reported.
Source: Authors’ analysis of EU-SILC module 2019.
As certainly remains the case for the original Great Gatsby curve, there is still much to be unpacked about the drivers and determinants of this association. The methodological challenge of how best to capture parental poverty with retrospective data where long-run panel data are not available also merits much more investigation, to advance our understanding of what is distinctive about the transmission of poverty and where that fits in the broader picture of socio-economic mobility.
Source : VOXeu
