top of page

Is there only one way out of In-Work Poverty? Gender & Race Differences in the US

Both great fluctuation and stability characterise the prevalence of in-work poverty in Europe and the United States across the last fifty years. A high level of in-work poverty is an especially worrisome phenomenon that can engender social exclusion as well as contribute to increasing social inequality and support for populist movements. The working poor are employed individuals who live in households with incomes below the poverty threshold. Similarly to poverty, long-lasting in-work poverty may have spill-over effects in other life domains. This is especially the case in the US, where the primary segment of the labour market is difficult to access for women and individuals from minority groups. Unstable and unprotected employment not only inhibits upward mobility chances, but also exposes these already marginalised groups to a higher risk of in-work poverty.

The standard approach to measure poverty in the US relies on an absolute threshold, which is a function of household composition adjusted yearly for inflation. For example, the federal poverty limit for a single adult in 2018 was $12,140 before taxes and government transfers and $23,900 for a household with three members. The working poor in the US are commonly defined as individuals over age 16 that have been employed for at least 27 weeks over the year and whose household income lies below the federal poverty line. This measure contrasts with the relative poverty approach, adopted for example by Eurostat, which considers individuals whose equivalized household income is below 60% of the national median as poor.

Figure 1 shows the proportion of employed men and women in absolute and relative poverty by race in the United States from 1971-2018. The proportion of employed white men and women below both the absolute poverty has remained remarkably stable (about 3 and 4 percent respectively). Relative in-work poverty rates are higher, but similarly stable for white men and women, roughly 9 and 10 percent, respectively. The share of employed Black and Hispanic men and women that are poor is considerably higher and more variable across time compared to white men and women. In fact, absolute in-work poverty rates for Blacks and Hispanics are more similar to the relative in-work poverty rates of white men and women. While the share of working Black and Hispanic men in poverty was similar during the 1970s, these trends began to diverge during the 1980s. In 1971, approximately 10 percent of employed Hispanic and Black men lived in absolute poverty. Following the passage of the Personal Responsibility and Work Opportunity Act in 1996, well over 13 percent of Hispanic working men lived in households under the federal poverty limit and over 30 percent lived in households under the relative poverty threshold, compared to 24 percent in 1971. In-work absolute and relative poverty rates decreased for Black men, reaching a low of 4 and 17 percent, respectively, in the year 2000.

Figure 1: Proportion of Employed Men and Women in Absolute and Relative Poverty by Race in the United States, 1971-2018

Source: 1971-2018 Current Population Surveys (CPS) and Annual Social and Economic Supplements (ASES); Employment defined as working 27 weeks or more; Proportions and 95% confidence intervals displayed; Absolute poverty (solid lines) defined as households with gross household incomes below US Census Bureau poverty thresholds; Relative poverty (dashed lines) defined as households with net equivalized household incomes below 60 percent of the median; Data weighted.

In contrast to men, in-work poverty rates for Black and Hispanic women converged by the mid-1990s. Before then, working Black women were more likely to be poor than Hispanic women during the 1970s and 1980s. In 1971, 16 and 31 percent of employed Black women lived in households under the absolute and relative poverty thresholds, compared to 7 and 18 percent for Hispanic women. By 1996, 10 percent of both Black and Hispanic women lived in households under the federal poverty limit and 28 percent under the relative poverty limit. Those rates have since changed only to a small degree.

Our results demonstrate that in-work poverty is not uniformly distributed across the US population, but also that large inequalities by gender and race exist. However, we still know little about how individuals experience in-work poverty. For some individuals, working and being poor may be long-lasting condition, while for others in-work poverty could be a short and sporadic event. Besides the duration of a single in-work poverty spell, some individuals who escape in-work poverty may never regress into it again, while others could experience recurring spells of in-work poverty and be unable to escape it.

Until now, the temporal dynamics of in-work poverty that reflect individuals’ experiences have not been thoroughly investigated. Therefore, we adopted a longitudinal approach to in-work poverty that went beyond a binary definition of entering and exiting in-work poverty. Rather, out conceptualisation of in-work poverty captures different routes out of in-work poverty by combining employment status and economic vulnerability. For example, individuals that exit in-work poverty by leaving the labour market find themselves in a state of social exclusion, while those that exit in-work poverty by crossing the poverty threshold continue to find themselves in a state of vulnerability. By following individuals after they exit in-work poverty, we can gain a better understanding of the various pathways they follow and how more and less advantageous pathways are distributed across the population.

To investigate gender and race differences in pathways out of in-work poverty in the US, we combine data from three high quality data sources – the 1979 and 1997 National Longitudinal Surveys of Youth (NLSY79 and NLSY97) as well as the Panel Study of Income Dynamics (PSID). Our combined sample consists of 37,925 individuals, whose life courses could be followed from age 18 up to age 50 maximum. Instead of considering the probability of exiting or entering in-work poverty at a specific point in time, we construct a trajectory for each individual, where each year of their life is defined according to a state identifying their position in the labour market status and the poverty status of the respondent’s household. In any given observation period, a respondent can either be “working and not poor”, “working and at-risk of poverty”, “working and poor”, “not working and not poor”, “not working and at-risk of poverty”, or “not working and poor”. We use the US Census Bureau poverty thresholds to define the absolute poverty thresholds and the Eurostat definition of relative poverty to delineate being at-risk of poverty. We used those trajectories to identify typical pathways that individuals experience in the 5 years after exiting an episode of in-work poverty.

Our findings demonstrate that there are five pathways out of in-work poverty, which are characterised by varying degrees of labour market attachment, economic vulnerability, social exclusion, and volatility. Each plot of Figure 2 displays 100 trajectories out of in work poverty, where each sequence begins in in-work poverty. One fifth of the exits from in-work poverty lead to an “immediate recovery”, where the following 4 years are spent in stable employment outside of poverty. Compared to all other pathways, this is the most advantageous one. However, the probability of experiencing this pathway is highly stratified by gender and race: compared to white men, all other gender-race groups are less likely to “immediately recover” from in-work poverty. These very same groups are instead more likely to experience “cyclical in-work poverty” or “impoverished non-employment”, which altogether account for the 30% of the trajectories out of in-work poverty. In fact, within four years, they regress back into in-work poverty or fall below the absolute poverty line outside of the labour market.

Figure 2: Relative Frequency Sequence Plots of Pathways Out of In-Work Poverty Clusters

Source: NLSY79, NLSY97, PSID, authors’ calculations.

Women and non-white men are more likely than white men to escape in-work poverty by exiting the labour market. These individuals are also more likely to remain employed and exit absolute poverty, but continue to live in households that are at a constant risk of poverty. White men are the least likely to experience this pathway of “continuous vulnerability”. Black women are also disadvantaged when it comes to the probability of experiencing at least a “progressive recovery”, namely a less smooth transition towards non-poor employment passing through being at-risk of poverty and unemployment. Our findings further show that these gender and race inequalities in the pathways out of in-work poverty are not attributable to differences in labour market related characteristics, such as educational attainment, work experience, and occupational group. Moreover, differences in family demographic behaviour, such as entering into parenthood, marriage, or divorce, cannot account for these inequalities.

Our study has important policy implications for anti-poverty measures in the US and beyond. Specifically, our results demonstrate that anti-poverty strategies that concentrate on employment alone will be unable to reduce in-work poverty. In fact, rather than decreasing in-work poverty rates, pushing individuals into low-paying positions will increase the proportion of the working population that live in impoverished households. Work-related anti-poverty strategies must be coupled with adequately high wages, for example regulated through minimum wage schemes, and employment protection legislation to effectively and continuously raise working households out of poverty.


Struffolino E., Van Winkle Z. (2019) Is there only one way out of in-work poverty? Difference by gender and race in the US. WZB Discussion Paper SP I 2019–601.

Van Winkle Z., Struffolino E. (2018). When Working Isn’t Enough: Family Demographic Process and In-Work Poverty across the Life Course in the United States. Demographic Research, Vol. 39(12), pp. 365–380.

84 views0 comments


bottom of page