Explaining the Unexpected Decline of Teenage Fertility in England: An Ecological Study
Ann Berrington, University of Southampton
Katie L Heap, University of Southampton
Roger Ingham, University of Southampton
Motivation
Historically, U.K. teenage fertility rates werehigher than many European countries (Darroch et al., 2001). Teenage fertilityfell slowly between 1998 and 2008 under-18 fertility rate from 27 to 19births per 1000. In the subsequent seven years, rates more than halved to 9births per 1000 in 2016, mirroring U.S declines (Kearney & Levine, 2015;Lindberg et al., 2016). Teenage fertility reduction was driven by plummetingteenage conception rates and consistent proportions of conceptions ending inabortion. Explanations for falling teenage conceptions/fertility include theten-year UK Governments Teenage Pregnancy Strategy (TPS), ending in 2010 (Hadleyet al., 2016; Wellings et al., 2016). The TPS provided young people withknowledge (e.g. Sex and Relationships Education) and sexual health services toreduce teenage pregnancy (Hadley et al., 2016).
This paper considers the potential TPS impact butsituates declines within wider 1998-2016 UK societal changes: risingpost-compulsory educational aspirations; altering teenage ethnic composition;housing affordability decline (postponement of permanently leaving the parentalhome); and increased deprivation associated with economic recession andGovernment austerity post-2008.
As noted by Wellings et al. (2016), vastgeographical differences exist amongst teenage fertility rates and theirdeclines. Insufficient attention has been placed on understanding thisvariation. This papers contribution involves an ecological approach toexamine, for the 326 Local Authority Districts of England, factors (includingthe TPS funding) associated with under-18 fertility rate decline by viewingunder-18 conception rates and abortion ratios.
Data and Methods
Our dependent variables are the annual under-18 conceptionrates and abortion ratios for all 326 LADs 1998 to 2016, available from theOffice for National Statistics. We attach time-varying contextual informationabout each LAD. Deprivation is measured by percentages of secondary schoolpupils in state-funded schools eligible for Free School Meals. The UK has agrowing second generation of Black and Asian teenagers, found to experiencedifferent fertility. Ethnic composition is identified by individual percentagesof Black and South Asian pupils within state-funded secondary schools.Educational enrolment post-16 is endogenous to teenage fertility. Therefore, wechoose age 11 attainment for LAD-level educational aspiration by percentages offemale pupils achieving level 5 in Key Stage 2 Mathematics assessments.Finally, housing affordability to young people was included by measuring socialhousing stock per adult and inflation-adjusted average house price at the lowestquartile. The rurality of the LAD is used as a control and London is modelled separately(not included) due to distinct fertility outcomes and LAD-level characteristics.
To investigate associations between LAD-levelcharacteristics and teenage fertility trends we employ panel regressions, withannual conception rates (or abortion ratios) clustered within LADs. Randomintercept linear regression is used to identify our dependent variablesvariability within and between LADs. Repeated measures capturetime-constant (e.g. region) and time-varying (e.g. ethnic composition)characteristics effects on teenage fertility trends 1998-2016. Interactionsbetween characteristics and year test whether effects alter throughout theperiod.
There is great variation in the levels ofteenage conceptions and the proportion of these ending in abortion throughoutEngland. Excluding London, conception rates were originally highest in theNorth East and lowest in the South East (Figure 1a). Inner London isdistinctive: under-18 conception initially higher than the English average whichdeclined rapidly. Londons abortion ratios are even more distinctive, with consistentlyhigher ratios but similar levels of growth to the rest of England (Figure 1b).Regions with higher conception rates generally have lower proportions of conceptionsending in abortion.
Multivariate Findings
Model 1 (Table 1) includes LAD-level covariatesrelationships with LAD-level teenage conception rates or abortion ratios in Model2. Local authorities with greater deprivation tended to have higher teenage conceptionsbut lower abortion ratios. Deprivations covariates reduced through 1998-2016in both models. The percentage of Black pupils were associated withdeprivation, so were removed, but the other ethnicity measure was not significantin predicting conceptions or abortions. Both housing affordability measureswere significantly associated with our conception and abortion measures and theserelationships covariates weakened between 1998-2016, apart from social housingfor abortion ratios which had a consistent negative relationship. Even whencontrolling for other LAD-level characteristics, rural LADs had significantlylower conception rates and higher abortion ratios than LADs which were in the urbanestcategory.
Conclusions and Next Steps
Area-level deprivation was traditionally positivelyassociated with teenage conceptions and negatively associated with abortions. Since1998, these relationships weakened. Factors consistently associated with teenageconceptions and abortions include LADs deprivation, average houseprice, and the stock of social housing per adult, but these relationships manifestdifferently for conceptions compared to abortions. We need to further explorethese factors and include additional contextual variables (e.g. each LADs TPSfunding). We will have also extended analyses by fixed effect modelling toexamine associations between within LAD change in teenage abortionsand conceptions and LADs key covariates.
Presented in Session 1168: Fertility