Preventing early school-leaving can enable disadvantaged youth to break cycles of multi-dimensional disadvantage. This column analyses a sample of Australians with high rates of early school-leaving to gauge its effects on the most disadvantaged students. For men, leaving school before age 18 increases the likelihood of experiencing homeless, being incarcerated, using cannabis daily, and using illegal street drugs weekly, several years after leaving school. The effects for women are concentrated in the short term for those leaving very early.
Growing up in economic and social disadvantage has long-lasting consequences on people’s life trajectories, including for their children and future generations. Identifying pathways that enable young people to break cycles of multi-dimensional disadvantage and build a better life for themselves has long been an active area of research in social sciences. Schooling is often viewed as the main tool to lift the lives of disadvantaged children by providing them an opportunity to build the necessary skills to lead a productive and successful life.
The impact of additional schooling has been widely studied. Research consistently finds that extra years of education yield positive effects on earnings (Angrist and Krueger 1991) and nonpecuniary benefits such as improved health, stronger marriages, and greater success for children (Oreopoulos and Salvanes 2011). Among other benefits, schooling also reduces chronic illness (Shields et al. 2020), crime rates (Costa et al. 2018), and poverty (Hofmarcher 2021).
While these positive returns to education are well-documented, they are typically identified for marginal students – those who might have otherwise dropped out but remained in school due to specific circumstances, such as changes in compulsory schooling laws, birth timing, geographic proximity to a university, or even differences in educational attainment between identical twins.
However, the focus on marginal students provides limited insight into the most disadvantaged students – those whose educational trajectories differ significantly from the general population. Studies suggest that broad expansions in education often fail to deliver the greatest benefits to these students, leaving a gap in the literature regarding the returns to education for the most disadvantaged. Our study seeks to fill this gap by estimating the effects of additional years of schooling on the most disadvantaged students (Moschion and van Ours 2025).
Journeys Home
Our analysis is based on data from the Australian ‘Journeys Home’ survey (see Wooden et al. 2012 for details). The sample for Journeys Home was drawn from Australian administrative data and covers a broad sample of the most disadvantaged population. The sampling base was the universe of recipients of income support, which is managed by a single entity – Centrelink. In May 2011 (when the sample was drawn), more than one in five Australian residents were receiving some kind of income-support payment (such as childcare payments, rent assistance, disability benefits, or unemployment benefits). From this population subset, recipients who were homeless or at risk of homelessness were identified. Respondents are disadvantaged along all standard socioeconomic dimensions. Back-of-the-envelope calculations presented in an earlier paper of ours (Moschion and van Ours 2019) show that our sample corresponds to the lowest percentile of the Australian population in terms of multi-dimensional disadvantage.
We define early school-leaving as leaving school before 18 since most young people in Australia finish high school at 18 years of age (repetitions are extremely rare in Australia). To follow the prevalence of our outcomes of interest from age 12 to 25, we use the retrospective information collected in Journeys Home at different waves and reconstruct a balanced panel dataset. Specifically, we use the information collected about the age at which respondents first experienced homelessness, incarceration, daily cannabis use, weekly illegal street drug use, depression, and anxiety disorder. Figure 1 shows the lifetime prevalence of these outcomes between 12 and 25 years old. On the left-hand side, the figure for men shows that homelessness, cannabis use, and incarceration are the most common outcomes.
Figure 1 Lifetime prevalence of adverse life events by age
Note: Based on a sample of 573 males (8,022 observations) and 381 females (5,334 observations).
By age 25, 56% of men have experienced homelessness, 47% have used or are still using cannabis on a daily basis, and 42% have been incarcerated. The right-hand side figure shows a different pattern for women, with mental-health conditions being more common than for men. By age 25, one in three females in our sample has been diagnosed with depression, and one in five has been diagnosed with an anxiety disorder. Homelessness is also prevalent among females, with almost half of the sample having experienced it by age 25. Daily cannabis use, weekly use of illegal street drugs, and incarceration are less common among women. Importantly, the prevalence rates are not only high but also increase progressively, with onsets both before and after the age at which respondents leave school.
Early school-leaving and adverse life events
We adopt a difference-in-difference strategy to identify the causal effect of early school-leaving (the treatment) on a variety of adverse life outcomes. Respondents leaving school between ages 14 and 17 are ‘treated’ from the time they leave school, while those leaving from 18 onwards are assumed to be never treated.
Since school-leaving can occur at various ages, we are in a context of a staggered treatment. To address the issues that arise with a staggered treatment, several alternative estimators have been proposed in the literature. We follow Callaway and Sant’Anna (2021) and estimate average treatment effects on the treated, and weighted averages of the average treatment effects on the treated for various times and ages (for details, see Moschion and van Ours 2025). Our preferred estimates are presented in Figure 2.
Figure 2 Parameter estimates effects of early school-leaving on adverse life events
Note: Based on a sample of 573 males (8,022 observations) and 381 females (5,334 observations). Presented are the point estimates of the effect of early school-leaving and their 95% confidence intervals.
The left-hand side for men shows that, on average, leaving school early increases the risk of homelessness by 7 percentage points (borderline significant), incarceration by 13 percentage points, daily cannabis use by 10 percentage points, and weeklhy use of street drugs by 9 percentage points. It is interesting to note that even in a population that is at risk of becoming homeless, staying in school longer reduces the probability of becoming homeless at a young age. Homelessness in itself may negatively affect general health outcomes later in life. Meyer et al. (2014), for example, find that after accounting for demographic and geographic differences, the mortality risk of people who have experienced homelessness in the US is about 3.5 times higher than that of the housed population.
In our analysis, we find a clear gender gap. As shown in the right-hand side of Figure 2, the effects of early school-leaving for women are mostly insignificant, small in magnitude and inconsistent in sign (despite some uncertainty for the effect on homelessness). Effects of early school-leaving on depression and anxiety disorder are not significantly different from zero for either gender.
In one of the sensitivity analyses, we show that age of early school-leaving matters. Across the board, the effects of early school-leaving appear particularly large for young people who leave school at ages 14 or 15. For example, the largest effects for men suggest that leaving school at 14 increases the risk of homelessness by 19 percentage points. Leaving school at 15 increases the risks of incarceration by 15 percentage points, daily cannabis use by 14 percentage points, and weekly use of illegal street drugs by 11 percentage points. For omen, leaving school at 14 increases the risks of homelessness by 25 percentage points and incarceration by 13 percentage points. Even when they leave school very early, the risk of illegal drug use is not increased for women.
When young people leave school at ages 16 or 17, the effect of leaving early tends to be smaller in magnitude, but some effects remain significant. Indeed, leaving school at 16 still increases the risk of incarceration, daily cannabis use, and illegal street drugs for men. For those who leave at 17, only the risks of incarceration and illegal street drug use remain significant (11 and 6 percentage points, respectively). Leaving school from the age of 16 does not appear to affect any outcomes for women, or homelessness for men.
Conclusions
There is a clear gender gap in the effect of early school-leaving. Leaving school early substantially increases the risk of adverse life events for men, while the effects for women are concentrated in the short term for those leaving very early. These gender gaps may reflect different reasons why men and women quit school (e.g. females may leave if they expect a baby) and different sets of support programmes potentially offered to them as a result.
Overall, our findings reinforce other results in the education literature that men’s disadvantage in education is a critical policy issue. They suggest that if extremely disadvantaged men stayed in school longer, they would be less likely to become homeless, be incarcerated, or use illegal drugs.
Source : VOXeu
