Unravelling the relationship between employment,social transfers and income poverty: Policy and measurement

Abstract

Despite the rise in employment, consistently high EU-average poverty rates continue to generate debates about the factors that explain the level and changes in the relative poverty rate, both within and across countries. Assuming a strong negative correlation between poverty and employment, the article investigates the role of four mechanisms responsible for this blurred relationship. Using decomposition analysis and macro-level regression analysis, we investigate the extent to which (i) the distribution of employment across households with different levels of work intensity, (ii) the expansion of non-standard work, (iii) the change in the effectiveness of social welfare systems, and (iv) the change in median income and the corresponding shift in the poverty threshold have contributed to changes in relative income poverty in the last decades. We found that employment growth benefits poverty reduction, but this positive effect was partially offset by the precarious characteristics of some newly created jobs. If the distribution of jobs had favoured the jobless more in the pre-crisis period, the relative income poverty rate would have been lower. Although the share of persons in jobless households decreased during the recovery years, their risk of poverty increased due to the retrenchment of social transfers during and after the Great Recession. Furthermore, the use of a floating threshold, which is linked to changes in median income, underestimates the strength of the relationships between poverty, employment and social transfers: when the poverty threshold is kept fixed, not only do the dynamics of poverty look different, but the estimated coefficients are considerably larger.

Keywords

relative income poverty employment social transfers median income fixed-effects panel regression European Union

Introduction

The relationship between employment and poverty is of great importance in social policy research and policy debates. Some argue that employment is the key to poverty reduction and advocates strengthening the social investment state through activating social policies, while others emphasize the key role of benefits, stressing that employment alone is not enough to ensure prosperity for all and that even people in employment can be affected by in-work poverty.

The periods before and after the Great Recession were characterised by a standstill in poverty despite improved labour market conditions (Atkinson, 2010; Cantillon et al., 2019; Cantillon and Vandenbroucke, 2014). There are concerns regarding the declining poverty reduction capacity of welfare states, especially for vulnerable households (Cantillon, 2018). The literature on in-work poverty highlights the limitations for jobs in fully protecting against poverty (Crettaz, 2013; Hick and Marx, 2022; Lohmann and Marx, 2019), discussing the role of low wages (Marchal and Marx, 2018; Salverda, 2019), precarious forms of work (Eurofound, 2017; Horemans, 2019), long-term unemployment spells (Halleröd et al., 2015) and the number of workers in the household (Hick and Lanau, 2017; Tamayo and Popova, 2021). The social investment policy paradigm emphasises human capital development and enabling services for labour market transitions, while acknowledging the role of minimum incomes schemes as stabilizing buffers during economic downturns (Hemerijck, 2018). However, the efficiency of social investment policies has been questioned since the end of the Lisbon decade (Cantillon, 2011; Vandenbroucke and Vleminckx, 2011), and debates among social policy scholars continued on the effectiveness of social investment policies and of vertical redistribution policies in reducing poverty (Ronchi, 2018; Parolin and Van Lancker, 2021; Plavgo and Hemerijck, 2021).

Figure 1 illustrates the all-European employment and poverty trends in question, showing trends of poverty in the overall population as a benchmark, the poverty rate among the active-age population, and (to illustrate developments in redistribution) poverty rates before social benefits (excluding pensions).¹

The pre-crisis period

The period prior to the Great Recession was marked by a strong increase in employment rates across Europe, reaching historically high levels just before the crisis. However, this employment growth did not seem to have produced the desired relative poverty reduction in that period. Individual employment among the population aged 20–64 in the EU increased from 67.8% in 2005 to 69.8% in 2007, the average proportion of people at risk of income poverty remained unchanged over the same period (14.7%–14.9% for the active-age population).

The period of the Great Recession

While the EU-27 average individual employment rate fell from 69.8% to 68.4% between 2007 and 2012, poverty increased in the European Union as a whole, as well as in most of the member states. The EU-27 poverty rate among the active age population increased from 14.8% to 17.1% between 2009 and 2013.

The post-crisis period

The labour market performance improved again after the crisis: the EU employment rate of the 20–64 years old population rose sharply from 68.4 to 73.2% between 2012 and 2017. However, despite this, the EU-27 average at-risk-of-poverty (AROP) rate remained fairly stable, staying around 17% between 2013 and 2015 and it decreased only very slightly afterwards (16.4% in 2017). Overall, the descriptive trends indicate a rather weak relationship between employment and the relative poverty rates.

In what follows we examine the relationship between employment, social protection and relative income poverty rates in the EU countries among the working age population,² to get a better understanding of why the employment–relative poverty relationship looks weak at the macro level. We suggest that the strong negative correlation between employment and poverty persists, but it is overshadowed by (i) the distribution of employment change across households with different levels of work intensity; (ii) the expansion of non-standard work; (iii) the change in the effectiveness of redistribution; and (iv) the change in median income and the corresponding shift in the poverty threshold. As all four mechanisms are affected by changes in the macroeconomic context and social policy, as well as their interaction, we divide the analysed time period into pre-crisis, crisis and post-crisis recovery years as described above. We want to note that the EU level trends may mask differences in the association between employment and poverty across countries (see Figure A1 for country-level trajectories).

We find a negative relationship between (both individual and household level) employment and relative income poverty. The distribution of new jobs across households played a significant role, benefiting jobless households less during the pre-crisis period, but showing improvement during the recovery years. This latter contributed to an overall decrease in active-age poverty rates in most member states. The expansion of non-standard forms of work was positively associated with poverty. The analysis highlights weakened safety nets for jobless households and limited poverty reduction capacities of social transfers, particularly minimum income-type benefits, during and after the Great Recession. Furthermore, controlling for changes in median income strengthened the relationship between employment and poverty, also indicating a larger impact of social transfers on poverty reduction.

The article adds to the existing theoretical and empirical literature on the roles of employment and social safety nets as effective policy tools to reduce poverty, by jointly analysing their parts in times of economic recession and recovery. Our results may also feed into the discussion (Atkinson et al., 2017; Darvas, 2019; Jenkins, 2020; Marx et al., 2012) regarding the apparent failure to simultaneously fulfil the employment and social inclusion targets of the Europe 2020 strategy (European Commission, 2019) and the European Union’s minimum income initiative (European Council, 2023).

The article is organised as follows. In the second section, we overview the recent literature grouped under the four mechanisms we examine. The third section describes the data, variables and methods used. In the fourth section, we first present the results of the decomposition of country-specific poverty change before, during and after the crisis to gauge the contribution of distribution of work across households. Then, we present and discuss the results of country fixed-effect multivariate regressions to assess the employment elasticity of poverty throughout the three time periods, and to capture the average correlation between poverty, employment and social benefits in the EU. Finally, we discuss our results and their policy implications.

The relationship between employment, redistribution and poverty

The descriptive trends at EU level (Figure 1) indicate that despite positive economic and labour market developments, poverty outcomes did not improve as expected. However, economic theories establish a negative correlation between economic growth, increased income, and reduced unemployment.³ Empirical studies, both using macro- (Gábos et al., 2019; Moller et al., 2003; Tudorache, 2019) or micro-level data (Polin and Raitano, 2014; Valaavuo and Sirniö, 2022), are supportive. This relationship may not necessarily be stable, though, as changes in individual employment can affect household-level work intensity differently. For example, during crisis periods, the negative correlation was more pronounced due to the increase in the share of jobless households (Corluy and Vandenbroucke, 2014; Gábos et al., 2019; Marx et al., 2013). Less is known about developments during the recovery periods. Nevertheless, descriptive statistics show that the share of persons living in jobless households started to decline in most countries in 2013, but AROP rates continued to stagnate.⁴

Explanations for stagnant relative poverty rates in the EU despite rising employment can be grouped into four interrelated mechanisms explored in the literature. The first concerns how changes in individual employment translate to the household level, given that poverty is a household-level concept. Therefore, when employment growth is polarised across households, it may not result in overall poverty reduction. Empirical evidence on the pre-crisis period supports this explanation, indicating that employment growth benefited households already integrated into the labour force, while governments were less successful in activating people living in very low work intensity households (Corluy and Vandenbroucke, 2014; Gábos et al., 2019).

The second mechanism relates to non-standard work and low job quality. Structural changes in the labour market and deregulation of labour contracts have led to an increase in non-standard and precarious employment, such as part-time, fixed-contract or pseudo self-employment arrangements. The share of low-paid jobs increased slightly during the recession (from 16.7 to 17.2% between 2006 and 2014), but decreased afterwards (to 15.5% until 2018 in the EU) (Eurostat, 2018). These labour market risks may again distribute unevenly across households if an increase in low hourly wages and part-time employment contributes to labour market polarisation and a rise in in-work poverty (Alper et al., 2021; Brülle et al., 2019). Part-time employment reduces the chances of escaping poverty, while the impact of self-employment status is ambiguous (Valaavuo and Sirniö, 2022).

The level and trends of poverty also depend on the effectiveness of redistribution (Caminada et al., 2012; Moller et al., 2003; Marx et al., 2015; Nolan and Marx, 2009; Notten and Guio, 2019). The third mechanism is therefore related to the poverty reducing capacity of social protection for the working-age population. Income protection has become less adequate as social policies and social redistribution have become less pro-poor (Cantillon, 2011; Cantillon et al., 2020; Chen et al., 2018). The adequacy of minimum income schemes declined in many EU countries since before the Great Recession (Gábos and Tomka, 2022), and weakened safety nets contributed to the failure of poverty reduction during favourable employment trends in the pre-crisis period (Cantillon, 2011). Changes in the systems of redistribution since 2009, especially in unemployment benefit systems, added to the declining redistributive effect of social transfers (see Hermann, 2017 for an overview). This is also shown by a sharper increase in poverty during the crisis among the active-age population compared to the overall population (Figure 1).

Comparative research on minimum income schemes indicates an ongoing general retrenchment of their adequacy, but also highlights a stronger link between employment and social transfers in the recovery period through increased conditionality of benefits on taking up work, especially in Central-Eastern European member states (Knotz, 2018; Weishaupt, 2013). Higher public social spending is associated with lower income inequality and poverty, but tends to benefit the elderly more than the working-age population, especially in Western and Southern Europe (Chen et al., 2018; Chzhen et al., 2017; Jaumotte et al., 2013; McKnight et al., 2016).

The fourth explanation concerns the change in the median income and the consequent impact on the floating poverty threshold (Jenkins, 2020; Marx and Nolan, 2012; McKnight et al., 2016). When incomes near or below the poverty line do not keep pace with overall income growth, poverty stagnates or increases. Conversely, during recessions, a decreased threshold may result in reduced levels of income poverty.⁵ Various factors, including those discussed earlier, contribute to the unequal increase in incomes across the distribution. Job gains may increase the share of two-earner households, and short-term unemployed individuals may find a job more easily than long-term unemployed persons during recovery. Flexibilization of labour markets and lagging minimum wages in many member states, both during and after the recession, weakened the protection of low wage workers against poverty (Cantillon et al., 2020). The income gap between the jobless and the employed is likely to grow during recovery, even with unchanged redistribution. The upward shift in the poverty threshold hinders individuals in jobless households from crossing it. Overall, the link between the AROP threshold and median income may contribute to the perception of a weakened correlation between employment and poverty in recent decades. Consequently, poverty measurement is intensely debated, highlighting the need for alternative indicators to better capture the effects of anti-poverty policies (Darvas, 2019; Jenkins, 2020).

Based on the EU employment and poverty trends and the literature, we expect that a negative association between employment (individual and household level) and poverty exists. The distribution of jobs among households also matters, and we expect a negative relationship between the share of persons living in jobless households and the risk of income poverty. Furthermore, we expect that the quality of employment growth, particularly the prevalence of non-standard work, is a critical factor – with a higher share of part-time work arrangements (used as a proxy) associated with higher poverty rates. While the role of welfare transfers is important in reducing the level of pre-transfer income poverty, the effect of minimum income-type benefits is limited during periods of economic growth and increasing income levels. This limited capacity was further lowered by the retrenchment in these, particularly during the recovery period. Lastly, the use of the floating poverty threshold itself contributes to the weak employment–poverty relationship in descriptive statistics since employment growth typically leads to rising income levels and, consequently, to higher poverty thresholds.

Data and methods

We use data covering years 2004 to 2017 to create country-level indicators of the EU-27 member states. We use the EU Statistics on Income and Living Conditions (EU-SILC, survey years 2005–2018, with income reference years 2004–2017), supplemented with individual-level employment and part-time employment rates from the EU Labour Force Survey (EU-LFS, Eurostat database). Details on the used indicators and data sources can be found in the supplemental material.

First, we turn to a decomposition analysis to assess the contribution of the distribution of employment growth (or decline) across households with different levels of work intensity (mechanism i) to the change in poverty rates separately before, during and after the crisis across countries. Following Corluy and Vandenbroucke (2014), in each period we decompose the change in poverty rate of the active-age population into three contributory factors:

∆ P (a) = \bar{{n j l}_{i}} ∆ P {(a)}_{n j l, i} + \bar{{j l}_{i}} ∆ {P (a)}_{j l, i} + (\bar{{P (a)}_{j l, i}} - \bar{{P (a)}_{n j l, i}}) ∆ {j l}_{i}

(1)

The change in the (group-specific) poverty risk ( $P)$ of non-jobless households ( $n j l$ ); the change in the (group-specific) poverty risk of jobless households ( $j l$ ); and the change in the proportion of jobless households $(∆ {j l}_{i})$ . Subscript $i$ stands for country i, $(a)$ for active age. Jobless households are defined as households in which none of the active-age members worked during the year (resp. with a work intensity of zero).

Afterwards, we use a time-series cross-sectional (TSCS) design for 27 EU member states between 2004 and 2017 to study the association between employment and poverty and the role of the mechanisms we identify. We run multivariate panel regressions at the macro level, where our unit of analysis is country i in year t. As we are interested in the role of employment indicators that vary over time within countries, fixed effects models that control for all other time-invariant differences between countries are suitable for our purposes, the use of which mitigates the risk of omitted variable bias compared to the random effects models.⁶

We use country-level fixed-effects-regressions and estimate the following fixed effects regression:

Y_{i t} = β_{0} + β_{1} E_{i t} + β_{2} {P T E}_{i t} + β_{3} T_{i t} + {β_{4} C o u n t r y}_{i} + β_{5} {P e r i o d}_{n} + β_{6} {{P e r i o d}_{n} E_{i t} + β_{7} {{P e r i o d}_{n} T}_{i t} + ε}_{i t}

(2)

where

Y_{i t}

is the dependent variable, the AROP(a) rate in country i in year t. We use the AROP rate in the active-age population (aged between 18 and 64 years), excluding students and pensioners (AROP(a), see Table A1). Active-age individuals who live in households with disposable incomes – including social transfers – below the poverty threshold set at 60% of the country- and year-specific median disposable household income, are defined as being at risk of poverty.

$β_{1}$ is the coefficient of our variables for employment on individual and household level, denoted by $E$ . We use the individual-level employment rate of people aged between 20 and 64 years and, to match the poverty indicator measured on household level, the share of the active-age population living in jobless households. It identifies households in which nobody of working age is employed and is therefore indicative of the distribution of jobs across households (mechanism i).

$β_{0}$ denotes the slope coefficient, $C o u n t r y$ stands for the country dummies, and $ε_{i t}$ is the error term. We extend the baseline model by including covariates to capture non-standard work (mechanism ii), and the effectiveness of social welfare systems (mechanism iii). Corresponding to our employment indicators, we include either an individual-level (part-time employment rate) or a household-level indicator (the share of the active-age population living in households with at least one part-time employed person), which capture the spread of part-time employment (denoted by $P T E)$ . These indicators are used as proxy variables for non-standard forms of work. By this, we aim to account for the possibility that governments or employers may use reductions in working time or other means of labour market flexibilization as a strategy to avoid employment loss or to foster employment growth, ultimately increasing the prevalence of non-standard work (mechanism ii).

To capture the role of social benefits (mechanism iii), we use the country- and year-specific mean of social transfers for the active-age population or those living in jobless households (denoted by $T$ ). Social transfers refer to the equivalised amount of all social benefits excluding pensions.

Furthermore, we differentiate the pre-crisis (2004–2007), crisis (2008–2012), and post-crisis (2013–2017) periods (denoted by $P e r i o d$ ) and include interaction terms with the employment and social transfers variables to assess whether their relationship with income poverty varied across periods of economic downturns and booms.

Finally, a second set of models uses an alternative outcome variable, the AROP indicator for the active-age population anchored in 2005, AROP(a)’05 (for descriptives, see Figure A2). Instead of the floating poverty threshold, it uses the anchored poverty threshold defined as 60% of the median total disposable household income after social transfers in 2005, adjusted over the years for inflation. On this basis, we can rule out that the poverty threshold by itself affects and blurs the association between employment and poverty (mechanism iv) and respond to debates on alternative poverty indicators to better capture the effect of anti-poverty policies.⁷

Results

Decomposition of poverty changes, 2005–2017

One of the mechanisms behind the relationship between employment and relative poverty is the distribution of employment (growth) across households with different work intensity levels. We assess this mechanism in the decomposition analysis of the AROP rate among the active-age population across countries, weighed up with the poverty risks among jobless and non-jobless households.⁸ The three components of the decomposition are changes in the poverty risks of non-jobless and jobless households, and their changing share in the population. The analysis covers pre-crisis (2004–2007), crisis (2007–2012), and recovery (2012–2016) periods, and illustrates how poverty changes when one component changes while holding the others constant (Figure 2).

During the pre-crisis period, poverty rate changes among the active-age population varied widely across countries. Decreases in poverty rates were often associated with a declining poverty risk of non-jobless households (Poland, Slovakia, Hungary), a declining share of persons living in jobless households (Estonia), or both (Czechia, Lithuania). Increases in poverty rates were generally associated with an increased poverty risk of persons living in non-jobless households, accompanied by a smaller increase in the poverty risk of persons in jobless households.

During the crisis period, poverty rates increased in all countries except Austria and Finland, and by more than 2.5 percentage points in 8 countries. This was driven by different components across countries, but the increase in the share of persons living in jobless households and the poverty risk of persons in non-jobless households were particularly prevalent. Overall, the findings for the pre-crisis and crisis periods confirm the results of previous work by Corluy and Vandenbroucke (2014) and Gábos et al. (2019).

In the recovery period, the sign of poverty rate change was mixed across countries, similarly to the pre-crisis period, though it decreased in the majority of the countries. The decreasing poverty rates were mostly reflecting a decline in the poverty risk of persons living in non-jobless households, coupled with a decrease in the share of jobless households. A striking difference compared to the pre-crisis period is that the poverty rate of persons in jobless households increased in almost all countries, contributing to rising poverty rates or counteracting a decrease in poverty overall. The deteriorating poverty trends among jobless households during the recovery years compared to the pre-crisis period may suggest a decline in the effectiveness of social welfare systems (mechanism iii), which will be further analysed in the regression analysis. The declining share of jobless households strongly contributed to decreasing poverty rates, while in countries with rising poverty rates, this component counteracted the trend.

Overall, the distribution of employment among households, illustrated by the change in the share of jobless households was a rather weak factor in the pre-crisis years but its increase during the crisis went hand in hand with rising poverty and it remained a significant contributor to poverty decline during the recovery, when much of the employment growth benefited households that were previously not in employment (as opposed to the pre-crisis period, when employment benefited the non-jobless more).

Regression analysis

To assess the relationship between employment and relative income poverty and its underlying mechanisms across EU countries between 2004 and 2017, we estimate two sets of regression models. The first set uses the AROP rate among the active-age population after social transfers (AROP(a)) as the dependent variable (Table 1), the other the anchored AROP(a)’05 rate (Table 2). In each table Models 1–5 use individual-level employment and part-time employment rate as explanatory variables, while Models 6–10 use household-level variables to capture employment, based on household work intensity and the presence of a part-time worker in the household.

Table 1.

The relationship between AROP(a) and employment indicators – country FE regressions.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
Variables	AROP(a)	AROP(a)	AROP(a)	AROP(a)	AROP(a)	AROP(a)	AROP(a)	AROP(a)	AROP(a)	AROP(a)
Employment rate (cent.)	−0.188** (0.0704)	−0.195*** (0.0438)	−0.182*** (0.0541)	−0.203*** (0.0413)	−0.184*** (0.0501)
Share in jobless households (cent.)						0.249** (0.0955)	0.292*** (0.0908)	0.190** (0.0789)	0.297*** (0.0832)	0.209** (0.0758)
Share in part-time employment (cent.)		0.814*** (0.102)	0.341*** (0.118)	0.771*** (0.104)	0.296** (0.116)
Share of households with part-time worker (cent.)							0.292*** (0.0737)	0.200*** (0.0632)	0.303*** (0.0762)	0.196*** (0.0626)
Mean of benefits to jobless households		−0.264* (0.130)	−0.177 (0.121)				0.130 (0.164)	0.0264 (0.0987)
Mean of benefits				−0.366 (0.371)	−0.354 (0.369)				−0.354 (0.394)	−0.430 (0.408)
Period dummy (reference: Crisis)
Period dummy: Pre-crisis			−0.833*** (0.257)		−0.956*** (0.260)			−0.576** (0.213)		−0.642*** (0.211)
Period dummy: Post-crisis			0.809*** (0.252)		0.763*** (0.263)			0.925*** (0.262)		0.911*** (0.262)
Interaction: Pre-crisis period # employment rate			−0.0214 (0.0469)		−0.0208 (0.0473)
Interaction: Post-crisis period # employment rate			−0.0761 (0.0610)		−0.0832 (0.0528)
Interaction: Pre-crisis period # share in jobless hh								0.247*** (0.0882)		0.246** (0.0894)
Interaction: Post-crisis period # share in jobless hh								0.170** (0.0688)		0.160** (0.0750)
Interaction: Pre-crisis period # mean of benefits to jobless hh			−0.0850 (0.0534)					−0.148** (0.0611)
Interaction: Post-crisis period # mean of benefits to jobless hh			0.00266 (0.0988)					−0.0468 (0.0888)
Interaction: Pre-crisis period # mean of benefits					−0.319* (0.172)					−0.430* (0.215)
Interaction: Post-crisis period # mean of benefits					0.0210 (0.294)					−0.124 (0.279)
Constant	14.65*** (0.00689)	14.58*** (0.00928)	14.58*** (0.111)	14.59*** (0.00936)	14.63*** (0.114)	14.64*** (8.72 × 10⁻⁵)	14.65*** (0.00296)	14.56*** (0.120)	14.65*** (0.00296)	14.57*** (0.123)
Number of countries	27	27	27	27	27	27	27	27	27	27
Observations	368	368	368	368	368	368	368	368	368	368
R-squared	0.120	0.365	0.478	0.354	0.478	0.113	0.236	0.448	0.236	0.453

Data source: Eurostat Database and own calculations based on EU-SILC (2018-1 19-09 release).

Note. Robust standard errors in parentheses, ***p < .01, **p < .05, *p < .1.

Table 2.

The relationship between AROP(a) anchored in 2005 and employment indicators – country FE regressions.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
Variables	AROP(a)’05	AROP(a)’05	AROP(a)’05	AROP(a)’05	AROP(a)’05	AROP(a)’05	AROP(a)’05	AROP(a)’05	AROP(a)’05	AROP(a)’05
Employment rate (cent.)	−0.879*** (0.243)	−0.921*** (0.206)	−0.941*** (0.202)	−1.002*** (0.197)	−0.996*** (0.197)
Share in jobless households (cent.)						0.790** (0.333)	0.846** (0.322)	0.810** (0.318)	1.097*** (0.331)	0.983*** (0.304)
Share in part-time employment (cent.)		1.211*** (0.426)	1.626*** (0.360)	1.215*** (0.386)	1.514*** (0.335)
Share of households with part-time worker (cent.)							0.725*** (0.252)	0.830*** (0.239)	0.747*** (0.245)	0.820*** (0.235)
Mean of benefits to jobless households		−0.918* (0.484)	−0.793* (0.412)				−0.208 (0.434)	0.0746 (0.469)
Mean of benefits				−4.571** (1.919)	−3.740** (1.614)				−5.023* (2.496)	−3.827* (2.215)
Period dummy (reference: Crisis)
Period dummy: Pre-crisis			2.474*** (0.804)		1.597* (0.852)			2.913** (1.251)		2.446** (1.154)
Period dummy: Post-crisis			0.971 (0.628)		0.468 (0.554)			1.524 (0.959)		1.360 (0.900)
Interaction: Pre-crisis period # employment rate			0.133 (0.156)		0.0837 (0.140)
Interaction: Post-crisis period # employment rate			−0.141 (0.0886)		−0.151* (0.0781)
Interaction: Pre-crisis period # share in jobless hh								0.532* (0.288)		0.544* (0.272)
Interaction: Post-crisis period # share in jobless hh								0.806** (0.301)		0.751***
Interaction: Pre-crisis period # mean of benefits to jobless hh			−0.814*** (0.225)					−0.772*** (0.257)
Interaction: Post-crisis period # mean of benefits to jobless hh			−0.342* (0.183)					−0.362 (0.237)
Interaction: Pre-crisis period # mean of benefits					−2.567*** (0.643)					−2.329*** (0.764)
Interaction: Post-crisis period # mean of benefits					−1.495** (0.564)					−1.428* (0.781)
Constant	14.43*** (0.0259)	14.25*** (0.0338)	13.23*** (0.255)	14.25*** (0.0285)	13.59*** (0.278)	14.27*** (0.000304)	14.29*** (0.0102)	13.23*** (0.471)	14.28*** (0.00938)	13.33*** (0.481)
Number of countries	27	27	27	27	27	27	27	27	27	27
Observations	368	368	368	368	368	368	368	368	368	368
R-squared	0.271	0.384	0.482	0.431	0.519	0.128	0.204	0.349	0.278	0.400

The estimations in Table 1 using AROP(a) show a strong negative association between employment and poverty independent of how employment is measured (on an individual or on a household level), which confirms our expectations. This can be already observed in Models 1 and 6, showing the pure association between employment and poverty. Model 2 introduces individual part-time employment rate to assess the role of non-standard employment (mechanism ii) and social transfers to jobless households to assess the role of the effectiveness of social welfare systems (mechanism iii).

In Model 3, we add the period dummies, as well as the interaction terms of the period dummies with the individual employment rate and the social transfers, respectively. Based on this model, a 10-percentage point increase in the individual employment rate was associated with a 1.8-percentage point decrease in the poverty rate over the whole time period. Also, the individual part-time employment rate is associated with higher AROP(a) rates on average across the EU, a 10-percentage point increase (a relatively sizeable change) in part-time employment is associated with a 3.4-percentage point increase in the poverty rate. The coefficient for the mean of social transfers to jobless households is negative, but not significant.

Models 4 and 5 use the mean of social transfers to all (and not only jobless) households. The covariate per se is not significant, but the interaction terms suggest that the association between social transfers and poverty was stronger before the crisis (an increase in €1000 in the average social transfer between 2004 and 2007 is associated with a 3.9 percentage point additional decline in the poverty rate compared to the recession period), which may be indicative of a retrenchment of social benefit systems. The coefficients of our main explanatory variable, employment rate, are only slightly smaller in the models that include period dummies, while the coefficients of part-time employment rate are much smaller, suggesting that the part-time employment variable captures other (i.e., labour market or welfare state related) factors during these periods that are associated with poverty. The period dummies indicate increasing poverty rates over time.

We see similar patterns when using the share of individuals in jobless households as our main explanatory variable (Models 6–10 in Table 1). Based on Model 8, a 10-percentage point increase in the share of jobless households is associated with 1.9 percentage point higher AROP rate on average. However, while the association between individual employment and poverty did not vary over time, the importance of the distribution of employment across households clearly differs across periods: in both the phase before and after the crisis, changes in the share of persons in jobless households had a stronger link to poverty dynamics than during the crisis (see Model 8). This could be due to compositional structures: the relative disadvantage of jobless households could be lower if the population in general was affected by the crisis and if the compensatory effects through the social systems had a stronger impact. This supports our finding from the decomposition, that the decrease in the share of jobless households was a particularly meaningful factor in the post-crisis period (mechanism i). Similar to the part-time employment rate, the association between the share of households with a part-time worker and poverty is strong and positive across the models (elasticity between 2 and 3). The coefficients of the main effect of social transfers are not significant, but the interaction terms indicate that the association between social transfers and poverty was stronger in the pre-crisis years, both when considering benefits for jobless and for all active-age households (Models 8 and 10). In that period a €1000 increase in the average of social benefits was associated with a 1.2 percentage point decline (the sum of 0.26 and −1.48) in the poverty rate in case of transfers to jobless households and with an 8.6 percentage point decline in case of transfers for all active-age households.

Note that these moderate effects (given that a €1000 increase in the average is a substantial increase) capture only the changes of benefits within countries over time. The time-invariant cross-country differences in the adequacy of income protection impacting poverty rate are captured by the country dummies in our models. To further assess how social transfers moderate the employment–poverty relationship, we also use the AROP indicator before social transfers (preAROP(a)) as a dependent variable (see Table A3 and its discussion in the supplemental material).

The estimations in Table 2 use the anchored AROP(a) rate as dependent variable, allowing us to control for the changes in median income and therefore to check for mechanism iv. These results, in line with our expectations, confirm the strong negative association between employment and poverty, again independent of whether employment is measured at individual or household level.

However, the association between employment (both the individual- and household level measure) and poverty is substantially stronger when shifts in the poverty threshold (due to changes in the median income) are cancelled out. A 10-percentage point increase in the individual employment rate was associated with an 8.8 percentage point decrease in the poverty rate, compared to 1.9 in the models based on the floating poverty line. Similarly, a same increase in the share of jobless households is associated with a 7.9-percentage point increase in the poverty rate, compared to only 2.5 (Table 1). This confirms that a shift in the poverty threshold itself could contribute to the impression that the relationship between employment and poverty has been weakened (mechanism iv). Introducing part-time employment (mechanism ii) and social transfers (mechanism iii), points to similar results as in the case of AROP(a) (see Table 1), with only marginal increases in the association between employment and poverty. However, most of the coefficients for social transfers are significant, especially for all households, suggesting that the alternative measurement using the anchored poverty threshold also captures in a better way their relationship with income poverty. Most strikingly, the introduction of the interaction terms shows that the association between social transfers and poverty was considerably stronger before the crisis, the association being twice as strong compared to the crisis period. This underpins the findings based on AROP(a), which already pointed to weakening social safety nets. Changes in the share of jobless had a stronger impact on poverty dynamics in pre-crisis and post-crisis periods than during the crisis, confirming the prior results. Individual employment has a stronger link to poverty in the years following the crisis, too. Part-time employment again is an important factor, and its coefficient varies less across models than in the results using the floating poverty line. This suggests that in earlier models, indicators for part-time employment captured trends connected to the distribution of income.

Robustness checks omitting one country at a time yield similar estimates, suggesting that our results are not driven by any specific country. Random effects models also produce comparable results.

Discussion and concluding remarks

In this article, we analysed the relationship between poverty, employment and social benefits among the active age population. Our aim was to contribute to the existing literature through a longer period of analysis from 2004 to 2017 and better understand why the relationship between employment and poverty appears weak at macro level. We proposed four mechanisms that could explain why employment growth, both before and after the Great Recession, did not result in favourable developments in poverty.

Our findings confirm a negative relationship between employment both at individual and household levels and relative income poverty (AROP(a)). This aligns with economic theories and previous empirical findings (Gábos et al., 2019; Valaavuo and Sirniö, 2022). Our regression results indicate that a 10-percentage point increase in individual employment rate is associated with a 1.8-percentage point decrease in AROP(a) in Europe on average between 2004 and 2017. The estimated elasticity is 1.9–2.1-percentage points when employment is measured at household level. These relationships remain strong and robust across various specifications. While the role of individual employment remains consistent over time, the effect of the share of jobless households is more pronounced outside of the crisis period, particularly during the recovery phase. These results suggest that poverty trends should have been more favourable in recent decades compared to what has been observed. Furthermore, they somewhat challenge previous findings on the limited impact of employment growth on poverty reduction (Chen et al., 2018; McKnight, 2016) and concerns of social policy experts regarding the potential of employment and social investment policies in improving poverty outcomes (Cantillon, 2018). Why don’t EU (and many country)-level descriptive statistics reflect this relationship after all? To address this, we suggested four mechanisms that blur the relationship, and our analysis provides empirical evidence in support of each of these mechanisms.

The decomposition analysis showed that the distribution of new jobs across households played an important role, but this role varied notably between the two recent periods of economic upturn. In the pre-crisis period, jobless households did not benefit (or benefited less than non-jobless households) from employment growth, so that the uneven distribution of jobs hindered poverty reduction. However, during the recovery period, the share of persons living in jobless households declined significantly, leading to an improvement in the overall active-age poverty rate in most member states.

The same decomposition results revealed an increase in the poverty risk of the jobless in the majority of countries, indicating weakened safety nets for these groups (Cantillon, 2018; Gábos and Tomka, 2022). This deterioration in poverty trends during the recession and recovery is further supported by the regression results. Our estimates show that while overall changes in the level of social transfers are (mostly) negatively associated with changes in AROP(a), the estimates are significant only for the pre-crisis period. This suggests that the poverty reduction capacity of social transfers, particularly of minimum income-type benefits, seems to be limited overall, and was further diminished by the retrenchment during and after the Great Recession. To improve the accuracy of estimates, the use of institutional variables, which are independent of the composition of jobless households, could be taken into account. However, the available data did not meet the requirements for inclusion in our regression model. In addition, the financial incentive capacity of social transfers should be also considered, as welfare states face a ‘social trilemma’ where they must balance trade-offs between controlling social spending, guaranteeing decent incomes for the poor and maintaining work incentives (Cantillon et al., 2020; Cantillon and Vandenbroucke, 2014; Collado et al., 2019; Marx et al., 2015).

The poverty trends would have also been more encouraging during economic upturns if standard forms of work had prevailed to a greater extent among new jobs. The expansion of non-standard forms of work was positively associated with poverty in our regressions, emphasizing the importance of job type and quality when evaluating the poverty-reducing potential of employment growth. We used part-time employment on individual and household levels to proxy this mechanism, as previous analyses have suggested that it may be the most relevant form of non-standard work to consider in poverty research (Valaavuo and Sirniö, 2022). However, future research should put more emphasis on other job characteristics linked to in-work poverty, especially low wages.

Our results provide compelling empirical evidence supporting the existence of these three mechanisms. Accounting for them improves our understanding of the relationship between employment and relative income poverty. However, the most important factor in this regard proves to be the change in the median incomes. When using the anchored AROP(a) rate as the dependent variable, the association between employment and poverty becomes substantially stronger, up to four times the strength of the association with the AROP(a) indicator based on the floating poverty line. Not only does the strength of the employment–poverty relationship change when using the anchored poverty rate as the dependent variable, but these models also capture a more robust relationship between social transfers and changes in poverty. Consequently, the effectiveness of minimum income schemes can be more accurately assessed when they are evaluated independently of changes in the income distribution. In this respect, our results can enrich the ongoing debate on the European Union’s social target indicator (Darvas, 2019; Jenkins, 2020) and provide further insights into the developments of alternative poverty thresholds (Goedemé et al., 2019, 2022).

A main limitation of our article is that our results can only provide limited insights at the individual country level through the decomposition analysis. Although the association between gaining employment and exiting poverty varies substantially among EU member states (Valaavuo and Sirniö, 2022), conducting a country level analysis was beyond the scope of the current study. However, this limitation serves as a motivation for further research, as exploring the country-level dynamics would enable us to use the community of EU countries as a large ‘policy lab’. In this context, different policy combinations at the national and European levels could yield diverse outcomes, from which member states can learn from and inform their policy decisions.

Supplemental Material

Supplemental Material - Unravelling the relationship between employment, social transfers and income poverty: Policy and measurement

Supplemental Material for Unravelling the relationship between employment, social transfers and income poverty: Policy and measurement by András Gábos, Barbara Binder, Réka Branyiczki and István György Tóth in Journal of European Social Policy.

Footnotes

Acknowledgements

We thank the participants of the ‘7th European User Conference for EU-Microdata’ in 2021 and the Spring Conference of the DGS Section ‘Social Inequality and Social Stratification’ in 2023 organized by GESIS Leibniz Institute for the Social Sciences, participants of the ‘Work in Progress’ research seminar in TÁRKI Social Research Institute in March 2022, Tímea Laura Molnár and other participants of the ‘Graduate Student Conference on Inequality and Poverty’ in May 2022, organized by Central European University (in Vienna, Austria).

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Hungarian National Research, Development and Innovation Office (K 132883).

ORCID iDs

András Gábos

Barbara Binder

Réka Branyiczki

István György Tóth

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Alper

Huber

Stephens

(2021) Poverty and social rights: among the working age population in post-industrial democracies. Social Forces 99: 1710–1744.

Atkinson

(2010) Poverty and the EU: The New Decade (Macerata Lecture on European Economic Policy). Working Paper 24. Macerata, Italy: Università Degli Studi di Macerata.

Atkinson

Guio

A-C

Marlier

(2017) Monitoring social inclusion in Europe. In: Atkinson

Guio

A-C

Marlier

(eds) Monitoring Social Inclusion in Europe. Luxembourg: Eurostat.

Brülle

Gangl

Levanon

, et al. (2019) Changing labour market risks in the service economy: low wages, part-time employment and the trend in working poverty risks in Germany. Journal of European Social Policy 29(1): 115–129.

Caminada

Goudswaard

Koster

(2012) Social income transfers and poverty: a cross-country analysis for OECD countries. International Journal of Social Welfare 21(2): 115–126.

Cantillon

(2011) The paradox of the social investment state: growth, employment and poverty in the Lisbon era. Journal of European Social Policy 21(5): 432–449.

Cantillon

(2018) Social Security and Poverty Reduction in Rich Welfare States: Cracks in the Post War Policy Paradigm, Avenues for the Future. Working Paper 1817. Antwerp: Herman Deleeck Centre for Social Policy, University of Antwerp.

Cantillon

Vandenbroucke

(2014) Reconciling Work and Poverty Reduction. How Successful Are European Welfare States? Oxford: Oxford University Press.

Cantillon

Goedemé

Hills

(eds) (2019) Decent Incomes for All: Improving Policies in Europe. New York and Oxford: Oxford University Press.

10.

Cantillon

Parolin

Collado

(2020) A glass ceiling on poverty reduction? An empirical investigation into the structural constraints on minimum income protections. Journal of European Social Policy 30(2): 129–143.

11.

Chen

Hallaert

J-J

Pitt

, et al. (2018) Inequality and Poverty Across Generations in the European Union. Washington, DC: International Monetary Fund.

12.

Chzhen

Nolan

Cantillon

(2017) Impact of the economic crisis on children in rich countries. In: Cantillon

Chzhen

Handa

, et al. (eds) Children of Austerity: Impact of the Great Recession on Child Poverty in Rich Countries. Oxford: Oxford University Press, 8–29.

13.

Collado

Cantillon

Van den Bosch

, et al. (2019) The end of the cheap talk about poverty reduction: the cost of closing the poverty gap while maintaining work incentives. In: Cantillon

Goedemé

Hills

(eds) Decent Incomes for All: Improving Policies in Europe. New York and Oxford: Oxford University Press, 223–244.

14.

Corluy

Vandenbroucke

(2014) Individual employment, household employment, and risk of poverty in the European Union: a decomposition analysis. In: Cantillon

Vandenbroucke

(eds) Reconciling Work and Poverty Reduction: How Successful Are European Welfare States? Oxford: Oxford University Press, 94–130.

15.

Crettaz

(2013) A state-of-the-art review of working poverty in advanced economies: theoretical, models, measurement issues and risk groups. Journal of European Social Policy 23(4): 347–362.

16.

Darvas

(2019) Why is it so hard to reach the EU’s poverty target? Social Indicators Research 141(3): 1081–1105.

17.

Davis

Sanchez-Martinez

(2015) A Review of the Economic Theories of Poverty. York: Joseph Rowntree Foundation. https://www.jrf.org.uk/file/47487/download?token=mD0hhMRh&filetype=full-report.

18.

Eurofound (2017) Work Poverty in Europe. Luxembourg: Publications Office of the European Union.

19.

European Commission (2019) Employment and Social Developments in Europe. Sustainable Growth for All: Choices for the Future of Social Europe. Luxembourg: European Commission, Directorate-General for Employment, Social Affairs and Inclusion.

20.

European Commission (2020) Employment and Social Developments in Europe. Leaving No One Behind and Striving for More: Fairness and Solidarity in the European Social Market Economy. Luxembourg: Directorate-General of Employment, Social Affairs and Inclusion.

21.

European Council (2023) Recommendation on adequate minimum income ensuring active inclusion, 3 February. Official Journal of the European Union 66(C41): 1–12.

22.

Eurostat (2018) Low-wage Earners as a Proportion of All Employees (Excluding Apprentices) by Educational Attainment Level. Luxembourg: Eurostat. https://ec.europa.eu/eurostat/databrowser/view/earn_ses_pub1i/default/table?lang=en.

23.

Gábos

Tomka

(2022) Developments in minimum income levels in Europe. EUROSHIP Working Paper No. 11. Oslo: Oslo Metropolitan University.

24.

Gábos

Binder

Branyiczki

, et al. (2019) Employment and poverty dynamics before, during, and after the crisis. In: Cantillon

Goedemé

Hills

(eds) Decent Incomes for All: Improving Policies in Europe. New York and Oxford: Oxford University Press, 34–55.

25.

Goedemé

Penne

Swedrup

, et al. (2019) Is there common ground for defining a decent social minimum in Europe? In: Kotkas

Leijten

Pennings

(eds) Specifying and Securing a Social Minimum in the Battle against Poverty. Oxford: Hart, 93–109.

26.

Goedemé

Decerf

Van den Bosch

(2022) A new poverty indicator for Europe: the extended headcount ratio. Journal of European Social Policy 32(3): 287–301.

27.

Halleröd

Ebkrand

Bengtsson

(2015) In-work poverty and labour market trajectories: poverty risks among the working population in 22 European countries. Journal of European Social Policy 25(5): 1–17.

28.

Hemerijck

(2018) Social investment as a policy paradigm. Journal of European Public Policy 25(6): 810–827.

29.

Hermann

(2017) Crisis, structural reform and the dismantling of the European social model(s). Economic and Industrial Democracy 38(1): 51–68.

30.

Hick

Lanau

(2017) In-work Poverty in the UK: Problem, Policy Analysis and Platform for Action. Cardiff: Cardiff University.

31.

Hick

Marx

(2022) Poor Workers in Rich Democracies: On the Nature of In-Work Poverty and Its Relationship to Labour Market Policies. IZA Discussion Papers 15163. Bonn: Institute of Labor Economics (IZA).

32.

Horemans

(2019) Atypical work and in-work poverty. In: Lohmann

Marx

(eds) Handbook on In-Work Poverty. Cheltenham: Edward Elgar, 146–170.

33.

Jaumotte

Lall

Papageorgiou

(2013) Rising income inequality: technology, or trade and financial globalization? IMF Economic Review 61(2): 271–309.

34.

Jenkins

(2020) Perspectives of poverty in Europe: following in Tony Atkinson’s footsteps. Italian Economic Journal 6(1): 129–155.

35.

Knotz

(2018) A rising workfare state? Unemployment benefit conditionality in 21 OECD countries. Journal of International and Comparative Social Policy 34(2): 91–108.

36.

Lohmann

Marx

(eds) (2019) Handbook on In-Work Poverty. Cheltenham: Edward Elgar.

37.

Marchal

Marx

(2018) Stemming the tide: what have European Union countries done to support low-wage workers in an era of downward wage pressures? Journal of European Social Policy 28(1): 18–33.

38.

Marx

Nolan

(2012) In-work poverty. GINI Discussion Paper No 51. Amsterdam: Amsterdam Institute for Advanced Labor Studies (AIAS).

39.

Marx

Vandenbroucke

Verbist

(2012) Can higher employment levels bring down relative income poverty in the EU? regression-based simulations of the Europe 2020 target. Journal of European Social Policy 22: 472–486.

40.

Marx

Horemans

Marchal

, et al. (2013) Towards a better marriage between job growth and poverty reduction. GINI Policy Paper No. 5. Amsterdam: Amsterdam Institute for Advanced Labor Studies (AIAS).

41.

Marx

Nolan

Olivera

(2015) The welfare state and anti-poverty policy in rich countries. In: Atkinson

Bourguignon

(eds) Handbook of Income Distribution. Amsterdam: Elsevier, Vol. 2B.

42.

McKnight

Stewart

Mohun Himmelweit

, et al. (2016) Low Pay and In-Work Poverty: Preventative Measures and Preventative Approaches. Luxembourg: European Commission, Directorate-General for Employment, Social Affairs and Inclusion.

43.

Moller

Huber

Stephens

, et al. (2003) Determinants of relative poverty in advanced capitalist democracies. American Sociological Review 68(1): 22–51.

44.

Nolan

Marx

(2009) Economic inequality, poverty and social exclusion. In: Salverda

Nolan

Smeeding

(eds) Oxford Handbook of Economic Inequality. Oxford: Oxford University Press.

45.

Notten

Guio

(2019) The impact of social transfers on income poverty and material deprivation. In: Cantillon

Goedemé

Hills

(eds) Decent Incomes for All: Improving Policies in Europe. New York and Oxford: Oxford University Press, 85–107.

46.

Parolin

Van Lancker

(2021) What a social investment ‘litmus test’ must address: a response to Plavgo and Hemerijck. Journal of European Social Policy 31(3): 297–308.

47.

Plavgo

Hemerijck

(2021) The social investment litmus test: family formation, employment and poverty. Journal of European Social Policy 31(3): 282–296.

48.

Polin

Raitano

(2014) Poverty transitions and trigger events across EU groups of countries: evidence from EU-SILC. Journal of Social Policy 43(4): 745–772.

49.

Ronchi

(2018) Which roads (if any) to social investment? The recalibration of EU welfare states at the crisis crossroads (2000–2014). Journal of Social Policy 47(3): 459–478.

50.

Salverda

(2019) Low earnings and their drivers in relation to in-work poverty. In: Lohmann

Marx

(eds) Handbook on In-Work Poverty. Cheltenham: Edward Elgar, 26–49.

51.

Jara Tamayo

Popova

(2021) Second earners and in-work poverty in Europe. Journal of Social Policy 50(3): 470–492.

52.

Tudorache

M-D

(2019) Poverty rate determinants in the Central and Eastern Europe member states. Theoretical and Applied Economics 26(4): 163–180.

53.

Valaavuo

Sirniö

(2022) Jobs against poverty: a fixed-effects analysis on the link between gaining employment and exiting poverty in Europe. European Societies 24(3): 1–32.

54.

Vandenbroucke

Vleminckx

(2011) Disappointing poverty trends: is the social investment state to blame? Journal of European Social Policy 21(5): 450–471.

55.

Weishaupt

(2013) Origin and genesis of activation policies in ‘old’ Europe: towards a balanced approach? In: Marx

Nelson

(eds) Minimum Income Protection in Flux. Basingstoke: Palgrave Macmillan, 190–216.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.97 MB