![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
Abstract
Teen Parent Units (TPUs) provide education and support for high school students who are pregnant or parents in New Zealand. They provide childcare, links to health and other social services, guidance and mentoring. Because this programme is only available in some schools, evaluation is possible using teen mothers and schools in other geographic areas as controls. Using administrative data, this study evaluates the impact of TPUs on school attendance and completion outcomes amongst nearly all teen mothers born between 1991 and 1994 in New Zealand. We find that young women who had access to TPUs were less likely to dropout of school and more likely to complete school qualifications. Among all teen mothers, access to a TPU at or prior to conception significantly increased the probability of school enrolment after giving birth. Among teen mothers enrolled in school post-birth, TPU access substantially increased the probabilities of completing formal high school qualifications.
Acknowledgements
The authors gratefully acknowledge the assistance and advice provided by the many people who helped at different stages of the project. Particular thanks are due to Associate Professor Tue Gørgens (Australian National University), Dr Dean Hyslop (Motu Economic and Public Policy Research), and Professors Steven Durlauf and Christopher Taber (University of Wisconsin). We gratefully acknowledge the support provided in New Zealand by Rissa Ota (MSD), Sarah Tumen (Treasury) and Ashlee Cuneen (Ministry of Education).
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 While teen fathers are also able to enrol, in practice few attend these units (Families Commission, Citation2011).
2 See Figure and Table in Appendix for details on the locations of these units.
3 Young women who were home schooled or in casual, unregistered schools beginning in 2008 are not included in our dataset. Because of data limitations, we can’t remove individuals in these same circumstances prior to 2008.
4 We drop women whose first enrolment records were after 1 January 2008, because we assume that these young women were immigrants and we therefore cannot tell if they had given birth prior to entering the country.
5 New Zealand does not have a unique person identifier that applies across all government databases. All data were de-identified prior to our analysis and accessed by the research team through the secure Statistics New Zealand Datalab and a secure server at the Ministry of Social Development. More details on the data linkage history and protocols can be found at http://www.msd.govt.nz/about-msd-and-our-work/publications-resources/evaluation/family-start-outcomes-study/index.html.
6 School deciles are used to target funding at disadvantaged schools in New Zealand. Schools are allocated to deciles based on the socio-economic status of the communities from which their students are drawn.
7 For more information on the NCEA system see: http://www.nzqa.govt.nz/qualifications-standards/qualifications/ncea/understanding-ncea.
8 Because this is a nonlinear regression model, we cannot interpret the coefficients as the change in the probability of this outcome for a change in the value of this dummy independent variable. This requires the computation of partial derivatives of this cumulative standard normal. We report the sample mean of these estimated marginal effects in all subsequent tables.
9 It’s important to emphasise that for young women who were not enrolled in school at the estimated time of conception, we know the identities of the last schools in which they were enrolled. Therefore, we know whether or not these were TPU schools, and the distances to the nearest TPU schools if they were most recently enrolled in non-TPU schools.
10 We used a ‘macro’ that Google Searches the origin and destination and returns the corresponding distance between the TPU governing school and conception school locations in kilometres. Where the physical address of the governing school was different to the physical address of the TPU, we use the physical address of the TPU.
11 We do not know if a young woman was attending TPU classes, only that she was enrolled in a TPU governing school. However, it would be very rare for a teen mother not to attend a TPU programme if it was available.
12 We recalculated these rates for those young women who conceived in 2008 or later, and we can confirm that the results in Table are not due to the censoring problem that we discussed earlier.
13 We are unable to estimate attainment prior to the exact date of first birth because we only have information about the calendar year in which credits were awarded.
14 The set of controls are only a subset of the controls used in the previous regressions. We have richer set of controls for young women who gave birth compared to others because we’re able to draw on maternity data for young women who gave birth.
15 The location and availability of abortion services could also play a role in this teen birth probability. This information would be difficult to access because we don’t know the earlier residential location of these young women (just the location of their most recent school attended). Advice, referrals and subsidies to the nearest abortion clinic are universally available, and the nearest clinics varied in the types of abortions that could be performed. If we think of this omitted variable in our analysis, the estimated coefficient on TPU enrolment at age 14 would be biased upward if abortion service access has a negative effect of birth probabilities and TPU schools are located in areas further away from abortion clinics. Despite this possible upward bias, we find no measurable effect of being enrolled in a TPU school at age 14 on this teen birth outcome.
16 Instrumental variables must be correlated with the regressor, but uncorrelated with the disturbance term. A valid instrument would be something that both influences the propensity to enrol in a TPU school post-birth and is directly unrelated to the propensity to complete NCEA qualifications. We argue that distance to the nearest TPU governing school prior to conception meets these criteria. This information is captured in two variables: enrolment in a TPU school and distance to the nearest TPU school prior to or at conception. Because the underlying information is taken from a single information source, no tests for the exogeneity of these instruments was conducted.
17 We should be cautious in interpreting these IV results. It is well known that this IV estimation produces Local Average Treatment Effects (LATEs) where greater weight is placed on individuals whose treatment is most influenced by these instrumental variables. For this reason, LATEs could easily exceed Average Treatment Effects for the general population.