In-work poverty in Western Europe. A longitudinal perspective

ABSTRACT

This study investigates levels and determinants of in-work poverty (IWP) in Western Europe using EU-SILC longitudinal data 2004–2019. We compared IWP risk and their dynamics across fourteen countries by examining individual labor market positions, household total labor supplies, and employment patterns. We further explored the social class gradient in exposure to IWP, as well as drivers and patterns of longitudinal accumulation of poverty. Relying on a single (standard) earner is often not enough to keep families out of poverty, confirming the importance of dual-earner household arrangements, even if they entail non-standard employment conditions for one partner. This holds particularly true for countries with high levels of IWP and for less privileged social and occupational groups across all contexts. Analyzing IWP inertia, we examined the interplay between genuine state dependence (GSD) and unobserved heterogeneity in the accumulation of economic disadvantage over time. Previous experiences with IWP can lead to future IWP for some, yet this causal effect appears rather small. Our findings have clear implications for the social stratification of risk and policies designed to combat poverty accumulation.

KEYWORDS:

• In-work poverty
• low pay
• household employment patterns
• genuine state dependence
• EU-SILC

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 https://ec.europa.eu/eurostat/web/products-eurostat-news/-/DDN-20180316-1 [Eurostat 2018, last Accessed 18 Jan 2023].

2 See: https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:At-risk-of-poverty_rate.

3 IWP might even decline in times of crisis, as those with lower incomes lose their jobs (and thus exit from the count of IWP) disproportionately more often.

4 A related yet different question would be to focus on duration effects, that is, the idea that the longer one experiences poverty, the more difficult it is to exit from it. However, a meaningful contribution in this vein would require a longer observation window.

5 For data preparation, see Borst and Wirth (2022). For replication files refer to: https://osf.io/cbyv6/ doi:10.17605/OSF.IO/CBYV6.

6 We controlled for self-employment to deal with underreported incomes (Hurst et al. 2014).

7 More in detail on the fulfilment of the employment condition in the longitudinal setting: an individual that at time t and t + 2 satisfies the employment criterion but not at t + 1 (e.g. due to prolonged inactivity or unemployment), is consider to be at risk of IWP and thus enters the sample solely at time t and t + 2. This follows from applying the (standard) IWP definition – which’s shortcomings we discuss in the text.

8 Due to data features, for individuals for whom working condition could not be identified by means of self-reported monthly main activity – which is the case for non-primary respondents in those countries that base EU-SILC on administrative/register-based data (DK, FI, NL, SE) – we considered as workers all the individuals with a market income higher than the 25° percentile of country and year specific market income distributions, computed separately for self-employed and dependent employees. Further, in these countries it was not possible to properly identify non-standard working conditions, such as temporary, part-time and (to some extent) low-wage job spells for other than the primary respondent. Consequently, for register-based countries accumulation of non-standard working conditions (NSE) among household members is underestimated. Therefore, some of the households with one (the main respondent) non-standard job might in fact have more than one member in NSE. This fact is arguably associated with a “conservative” interpretation of the findings, since IWP risks associated with non-standard working conditions are, if anything, upward biased.

9 Main results do not change significantly in case of alternative analytical options, such as selecting households with no more than two adult workers, or, conversely, households with at least two adults in working age.

10 Our interest was to evaluate the role of ‘secondary labor market’ employment positions over accounting for possible heterogeneities across countries in the composition of the NSE group or the associations of IWP risks with each of the three specific labor market conditions.

11 OECD (2014) defines ‘non-standard employment’ (NSE) as comprising all forms of employment that do not benefit from the same degree of protection of open-end contracts against contract termination. See also Bentolila et al. (2019) for a similar definition of NSE and a discussion of the dualization of (Western) European labor markets.

12 To ensure our focus on the outcomes of various types of employment activation, we have omitted the complete combination of earners and NSE and the first two categories are not differentiated according to NSE. Further, despite their structural inability to activate additional income, singles have been included in the first category. See more on this below.

13 Data for Germany contained only ISCO main groups, which did not allow to construct ESeG.

14 Figures reporting the effects of social class and household employment patterns are included in the Appendix: Figure A-1.

15 We did not distinguish between single-earner households with multiple potential earners and single-adult/single-parent households, but poverty was defined based on incomes adjusted for the equivalence scale and household numerosity was accounted for in the model. Single-adult households, by definition, cannot increase their labor supply, and are, therefore, a category particularly at risk of IWP, requiring specific attention and policy interventions. For a more thorough discussion of single parent poverty exposure in European societies, see Niuewenhius and Maldonado (2018). Even if looking at the role of specific households' composition is somehow beyond the scope of this contribution, we ran several alternative model specifications allowing for the inclusion of family-related dummy variables, such as one person households, single headed households, absence of partner, and no-cohabiting partner. The overall pattern of findings remains substantially stable. Results can be found in the supplementary material, Table A2.

16 In Fig. 2 and Fig. 3 results for Germany are not presented due its insufficient level of detail in the ISCO measure. Figure A-2 in the Appendix reports the effects of household employment patterns only.

17 Figure 4 must be interpreted considering that it plots three different IWP situations (yes/no) at three different timepoints: the time of the interview (t), one year before (t – 1) and at first observation (t0 or initial condition). The supplementary material contains robustness checks.

18 In terms of implemented policies, the minimum wage has been the most important policy for tackling IWP in most EU member states, followed by taxation policies and reductions in social contributions by individuals at the lowest income levels and in-work benefits (Garnero et al. 2023). Alongside these targeted policies, minimum income schemes, active labor market policies (ALMPs) (including activation and training policies) and policies aimed at tackling labor market segmentation have been deemed by the European Commission as necessary and effective measures against IWP (Peña-Casas et al. 2019).