Long‐term labour market and economic consequences of school exclusions in England .Pdf

(2022) - Joan Madia, Ingrid Obsuth, Ian Thompson, Harry Daniels, Aja Murray

Journal: British Journal of Educational Psychology
Link::
DOI::
Links::
Tags:: #paper #LabourMarket #Attainment #SocialClass
Cite Key:: [@Madia2022]

Abstract

Background. Previous research suggests that school exclusion during childhood is a precursor to social exclusion in adulthood. Past literature on the consequences of school exclusion is, however, scarce and mainly focused on short-term outcomes such as educational attainment, delinquency, and mental health in early adolescence. Moreover, this evidence is based primarily on descriptive and correlational analysis, whereas robust causal evidence is required to best inform policy. Aims. We aimed to estimate the mid-to-long-term impact of school exclusion on labour market and economic outcomes. Sample. The sample included 6,632 young people who at the age of 25/26 in the year 2015 participated in the Next Steps survey of whom 86 were expelled from school and 711 were suspended between the ages of 13/14 and 16/17. Method. Using high quality existing longitudinal data, we utilized four approaches to evaluate the impact of school exclusion: logistic regression-adjustment models, propensity score matching, school fixed-effects analysis, and inverse propensity weighting. The latter two counterfactual approaches were used to estimate causal effects. Results. We found that school exclusion increased the risk of becoming NEET at the age of 19/20, and then remaining economically inactive at the age of 25/26, as well as experiencing higher unemployment risk and earning lower wages also at the age of 25/26. Conclusion. School exclusion has pervasive negative effects into adulthood. Policy interventions should focus on both prevention and mitigating its negative effects. Interventions aimed at re-integrating excluded individuals into education or vocational training could be key in reducing the risk of poor socio-economic outcomes and social exclusion.

Notes

"Past literature on the consequences of school exclusion is, however, scarce and mainly focused on short-term outcomes such as educational attainment, delinquency, and mental health in early adolescence" ( :1)

"robust causal evidence is required to best inform policy." ( :1)

"ogistic regression-adjustment models, propensity score matching, school fixed-effects analysis, and inverse propensity weighting. The latter two counterfactual approaches were used to estimate causal effects" ( :1)

"We found that school exclusion increased the risk of becoming NEET at the age of 19/20, and then remaining economically inactive at the age of 25/26, as well as experiencing higher unemployment risk and earning lower wages also at the age of 25/26." ( :1)

"School exclusion has pervasive negative effects into adulthood" ( :1)

"England, the headteacher of a school can temporarily exclude students from school for a fixed period of time, typically from 1 to 5 days with a legal maximum of 45 days a year, or they can permanently exclude students from the school" ( :2)

"In recent years, school exclusion has steadily increased in England in contrast to the rest of the United Kingdom. In 2011, for example, the rate of permanent exclusion at primary and lower secondary education was 0.06% while in 2018 it increased to 0.12%, representing a doubling in only 7 years (Timpson, 2019)." ( :2)

"liable to worsen the situation of vulnerable students who are already at risk of poor educational and occupational attainment, or of social exclusion more broadly (Parsons, 2018)." ( :2)

"systemic pressures have led to a disproportionate number of school exclusions of students with social emotional and mental health special educational needs (Thompson, Tawell, & Daniels, 2021)." ( :2)

"empirical evidence on the impact of school exclusion remains rather scarce and largely limited to descriptive and correlational analyses" ( :2)

"Evidence from the United States suggests a direct association between school exclusion and school retention, dropout, anti-social behaviour, and delinquency (see for example, Costenbader & Markson, 1998; Skiba & Knesting, 2001; Perry & Morris, 2014)." ( :2)

"2 Joan E. Madia et al. Intheshortterm,thatis,oneyearfollowing the exclusion, excluded students were reported to have poor academic outcomes (DfE," ( :2)

"2011)" ( :3)

"They were also reported tobe atan increased risk for substanceuseand delinquency (Timpson, 2019) as well as engaging in self-harm (McAra & McVie, 2010; McCrystal, Percy, & Higgins, 2007)." ( :3)

"medium to longer term, young people who were excluded were reported to be less likely to be employed in early adulthood (DfE, 2011)." ( :3)

"Similarly, findings from Massey (2011), who followed a group of young people who had experienced exclusion, suggested that within two-three years after their exclusion, half of excluded students transitioned to a NEET status, compared to the UK average of around 13%." ( :3)

"The difficulties faced by excluded students might be even larger at later stages of their life, with the accumulation of disadvantage. It is well known in the economic literature, youth unemployment and inactivity are linked to higher risk of unemployment, earning loses, and higher dependency on the welfare system in adulthood (Gregg, 2001; Gregg & Tominey, 2005)." ( :3)

"We focused on the likelihood of becoming NEET (i.e., not in education, employment, or training), unemployment and being in low paid jobs as a consequence of 3- 8- permanent exclusion 4 (age 19/20) and 9 years (age 24/25) following the exclusion. Testing this is challenging because different sources of endogeneity can bias results" ( :3)

"a range of individual and social background characteristics are likely to affect both the risk of school exclusion and socioeconomic outcomes in early adulthood. Therefore, to identify the casual effects of school exclusion, it is necessary to find a counterfactual group of students who have similar background characteristics of those excluded but have not experienced this event. To achieve this, we adopted several complementary approaches: regression adjustment, school fixed effects, propensity score matching, and inverse probability treatment weighting." ( :3)

"We focused primarily on the impact of permanent exclusions because the data related to suspensions was incomplete. Specifically, it did not include information about the length of suspensions. Given that in the United Kingdom it is possible to suspend students anywhere from a few hours/one day all the way to four weeks or more, not having this information rendered the definition of the latter category difficult." ( :4)

"classified students whose parents/guardians reported their exclusions from the age of 13/14 until the age of 16/17 into (a) those who never experienced a suspension or permanent exclusion, (b) those who were suspended, and then (c) those who were expelled from school." ( :4)

"also created a binary variable representing the treatment status of those who were expelled from school and the control status of those who never experienced any suspension or permanent exclusion" ( :4)

"counterfactual models, those suspended from school were removed from the analysis." ( :4)

"suspended students represent a different kind of treatment and thus they cannot be used as control units in the counterfactual analysis. In other words, if we include the suspended students in the reference category, our control group would be characterized by the presence of defiers, that is, units that do not behave in accordance with the hypothetical treatment assignment (Angrist, Imbens, & Rubin, 1996; Balke & Pearl, 1993)," ( :4)

"we focused on self-reported transitions to labour market, economic activities, and economic well-being" ( :4)

"we considered someone as such if they reported to be unemployed, not in education or training but also if they were not looking actively for a job (i.e., an economic inactive respondent)" ( :5)

"examined whether respondents had ever been employed; the probability of being unemployed at the time of the interview; and whether respondents were experiencing economic hardship" ( :5)

"examined the employment conditions of those who reported being employed at age 25/26, considering their gross monthly wages, the probability of being in a routine occupation, having a zero-hours contract, working full-time, and being in a job where some specific skills are required." ( :5)

"We controlled for a set of variables that past research suggested to be strong predictors of permanent exclusion from school (Strand & Fletcher, 2014) and are also associated in the literature with lower labour market prospects (Holmes, Murphy, & Mayhew, 2019" ( :5)

"sex, the number of siblings, and other adult members in the household (both continuous); an ethnicity dummy variable" ( :5)

"whether English was the main language of the household; mother's age which is typically positively associated with children's cognitive development; whether the respondent grew up in single parent household; whether parents were in contact with social services due to their child's behaviour at home or school; and whether respondents were identified as having Special Education Need (SEN) when they were attending mandatory education" ( :5)

"included two indices that capture family resources and contextual disadvantages in the neighbourhood area of residence. First, a 'Socio-Economic Status index' (SES)" ( :5)

"Second, the 'Income Deprivation Affecting Children index' (IDACI) which measures the proportion of children under the age of 16 who live in low-income households in the local area of residence." ( :5)

"Labour market consequences of school exclusion in England 5 We first ran logistic regressions for the binary outcomes and OLS" ( :5)

"regressions for the log of gross monthly wages applying three different specifications: (a) an unconditional regression (i.e., without any control variables) to describe the raw difference in means by school exclusion experiences; (b) conditional regressions included the full set of socio-demographic and student behaviour covariates described above to account for students' characteristics and background of origin (c) a school-fixed effect approach aimed at removing any (constant) unobserved heterogeneity at the school level and tackling self-selection into poor/rich schools" ( :6)

"reduce selection bias since vulnerable and disadvantaged students with challenging behaviour are more likely to attend poor schools with a lack of experienced teachers and resources to deal with disruptive situations in classes (Allen & Sims, 2018), increasing the likelihood of excluding these pupils from school as a less costly intervention for preserving the learning environment (Parsons, 2018" ( :6)

"also employed Propensity Score Matching (PSM) (Rosenbaum & Rubin, 1983) and Inverse Probability of Treatment Weighting (IPTW) (Robins, Hern n, & Brumback, 2000) as a second approach following a counterfactual logic. PSM has the advantage of balancing the treatment and control groups in a more flexible manner than common regression adjustment methods. It also restricts the estimation to the group of observation within an area of data in which there is a 'common support' between observations rather than extrapolating across the sample (Guo & Fraser, 2015" ( :6)

"standard propensity score methods usually have larger standard errors due to the fact that the propensity score, used for matching individuals, is itself estimated prior to the treatment-control groups comparison, thus affecting the large sample distribution of PSM estimator" ( :6)

"counterfactual models were complemented with a series of sensitivity analyses to assess the extent to which results could be negated due to Mantelomitted variables. For this robustness check we employed the Haenszel bounds for our nonlinear models (Becker & Caliendo, 2007) and the Rosenbaum bounds (DiPrete & Gangl, 2004) for the earnings outcome" ( :7)

"To deal with non-random attrition, the data release provides attrition and non-response weights, which we used in our empirical analysis" ( :7)

"There was substantial variation in the probability of being expelled from school by social background. First, this group of students were more likely to be boys from disadvantaged backgrounds (40% of the total were girls compared to 58% girls in the never excluded group). Their parents had a lower SES, the household was usually single parent, and they were more likely to live in deprived neighbourhoods. In terms of ethnicity, 13% were from a black Caribbean backgroundcompared to 5% in the suspended group and 3% in the never excluded group." ( :7)

"expelled students were also more likely to be found among those identified as SEN (38% vs only 17% among those who never experience a suspension or permanent exclusion and 33% among suspended)." ( :7)

"Taken together, results point to the conclusion that both suspended and expelled students have experienced several disadvantages in early adulthood, even among those who find employment" ( :9)

"based on the school-fixed effects specification, respondents who experienced suspension and permanent exclusions from school are had a higher risk of being NEET at age 19/20 (0.10 for suspended and 0.17 for expelled) and of remaining NEET later on at age 25/26 (0.08 and 0.12, respectively) or unemployed as well (0.05 and 0.10, respectively)." ( :9)

"Figure 2 displays the Average Treatment Effects on Treated (ATT) across three different model specifications: (a) the PSM with one-to-one matching, (b) PSM with one-to-three matching, and (c) the IPTW" ( :10)

"Our counterfactual models confirm the majority of the findings based on regression analyses presented in Figure 1. However, point estimates from the counterfactual models were higher than those obtained by standard regression adjustment, most likely because the regression adjustment was downwardly biased due to sample selection bias (i.e., more disadvantaged respondents with higher probability of school exclusion and poorer outcomes are likely to dropout)" ( :10)

"other words, even those who found a job still experienced several disadvantages in the labour market. However, we may have not detected any significant effect in these outcomes due to potential power issues since the sample of those employed was considerably smaller." ( :11)

"In summary, our counterfactual analyses have shown that school exclusions have pervasive negative effects over the life course and can, unintentionally, exacerbate inequality and social exclusion in society through unemployment. Specifically, we found that school exclusion increases the risk of being NEET at the age 19 and, then, remaining NEET (economically inactive) and suffering more often unemployment at the age of 25. Further, these disadvantages do not end there but they are also present once these people start to work. Excluded respondents at the age of 25 are also at risk of remaining trapped in precarious jobs (with zero-hours contracts) and earn much lower wages than similar counterparts that only differ because they have not been excluded from school." ( :11)

"School exclusions, therefore, represent a precursor to exclusion from society in adulthood." ( :12)

"Notably, our findings are consistent with findings presented in correlational studies related to suspension and permanent exclusion from school (see e.g., Massey, 2011; Spielhofer et al., 2009)" ( :12)

"While in this study our focus was on the individual-level consequences of school exclusion, the effects we identified are likely to have society-level costs too; societal costs that are often not discussed in the policy agenda. First, from a pure economic perspective, a loss of human capital reduces productivity and economic growth in society" ( :13)

"First, this study focused on comparing never excluded versus temporarily suspended and expelled students, but we were not able to distinguish between early and late exclusions." ( :13)

"Second, we also lacked disaggregated information on the length, number of exclusions, and the reasons for permanently excluding students from scho" ( :13)

"Third, information on school exclusions were self-reported by parents which might lead to measurement error issues" ( :13)

"Fourth, our sample size was relatively small to further investigate potential heterogeneities by gender and ethnicity" ( :13)