Evaluating Maternity Leave Reforms When RD Fails

Clara von Bismarck-Osten; Andrei Zeleneev

AI summary · arXiv: 2610.41881v1 · econ.EM

Evaluating Maternity Leave Reforms When RD Fails

Clara von Bismarck-Osten, Andrei Zeleneev

Original paper Submitted to arXiv on 30 Oct. 2026 License: CC BY 4.0 Also in: fr

How to read this summary? Illustrated guide

The essentials

An imputation-based evaluation finds UK maternity-pay expansions improved children’s test scores while reducing maternal employment, whereas extending job protection alone produced no detectable child benefits.

Question

Did the UK’s 2000, 2003 and 2007 maternity-leave reforms improve children’s school performance and mothers’ labour-market attachment? How can their effects be identified when expected birth dates determine eligibility but outcome datasets record only actual birth dates?

Method

The authors combine England’s National Pupil Database, with roughly 600,000 children per academic-year cohort and assessments at ages 5, 7 and 11, with Quarterly Labour Force Survey samples of roughly 3,900–4,900 mothers per follow-up horizon. They use actual and expected birth dates for 18,818 Millennium Cohort Study children to estimate eligibility probabilities, impute untreated outcomes from a preceding cohort under parallel trends, and divide observed-minus-imputed outcomes by those probabilities. Children’s effects are estimated nonparametrically by birth date, reporting dates with estimated treatment probability at least 0.2, while maternal outcomes use regressions assuming linear birth-date trends and constant effects across birth dates.

Findings

Result The 2007 reform increased age-5 test scores by 0.045 standard deviations, with a smaller but statistically significant effect remaining at age 11.
See the passage in the paper
The 2007 reform, primarily an extension of pay, raised scores at age 5 by 0.045 standard deviations, equivalent to about 25 additional days of age at testing. At age 11, the estimated effect remains statistically significant and is equivalent to about 15 additional days of age at testing.
Result The 2003 combined reform produced a smaller estimated age-5 benefit, while the 2000 job-protection reform produced no measurable child benefit.
See the passage in the paper
The 2003 reform, which extended both pay and job protection, had an estimated age-5 effect less than half that of the 2007 reform, and the 2000 extension of job protection alone had no measurable effect.
Result All three reforms reduced maternal employment and working hours, with negative effects persisting up to six years after childbirth.
See the passage in the paper
All three reforms reduced maternal employment and hours worked, and although the point estimates become smaller, they remain negative up to six years after childbirth.
Result The imputation approach identifies effects among eligible children at birth dates with positive eligibility probability, even when birth-date measurement error is related to potential outcomes.
See the passage in the paper
Dividing the difference between observed and imputed average outcomes by that share identifies the effect on children who were eligible at each birth date with a positive treated share, rather than at the cutoff alone as in conventional RD. Unlike many existing correction methods, our approach remains valid when the gap between actual and expected birth dates is related to potential outcomes.
Authors’ interpretation The authors interpret the contrast between similar maternal labour-supply reductions and different child outcomes as suggesting a role for income support beyond additional maternal time at home.
See the passage in the paper
This pattern suggests that additional time at home alone may not be sufficient to improve children’s outcomes and is consistent with a role for income support during leave, potentially through reduced financial stress complementing maternal time investments.

Limitations

Parallel-trends diagnostics cover only birth dates with zero treatment probability and cannot directly validate the assumption for affected birth dates.
See the passage in the paper
This observable implication provides a specification check within the zero-treatment region, although it does not directly test the parallel-trends restriction at dates for which $p^{t}(x)>0$.
Applying eligibility probabilities from the validation sample to reform cohorts requires an untested transportability assumption.
See the passage in the paper
This comparison does not test transportability to the reform cohorts, which remains a maintained restriction.
The mechanism comparison cannot cleanly separate pay from protection because the 2007 reform also changed protection rights and employment after one year imperfectly measures maternal time at home.
See the passage in the paper
Two caveats apply: the 2007 reform also extended job protection to mothers with short tenure (who, as noted in Section 2, mostly did not qualify for Statutory Maternity Pay), and employment one year after childbirth is a coarse proxy for time spent at home.

Why it matters

The authors argue that maternity-pay expansions can yield lasting child-development gains when initial provision is low, without strengthening mothers’ long-run employment attachment. Methodologically, they offer an alternative to conventional RD when eligibility is latent and measurement error is economically meaningful, provided credible counterfactual imputation and validation data are available.

Written by AI from the paper. Passages are checked word for word against the text we analysed; the rest may contain errors.

AI-generated keywords: Maternity leave; Maternity pay; Child educational outcomes; Maternal employment; Policy evaluation

Read further

Comprehensive Summary
Layman's Summary
Blog article
Original abstract

Clara von Bismarck-Osten and Andrei Zeleneev examine whether expanding maternity leave improves children’s educational outcomes and mothers’ employment when initial benefits are relatively limited. They study three UK reforms: a 2000 expansion of job-protected leave, a 2003 package extending both leave and pay, and a 2007 reform primarily extending maternity pay from 26 to 39 weeks. Their central question is whether maternity pay and job protection have different effects. Their methodological contribution addresses an obstacle to conventional regression discontinuity analysis. Eligibility depended on a child’s expected birth date, but the outcome datasets record only actual birth dates. Early and late deliveries therefore obscure both the eligibility variable and policy assignment. Treatment probabilities change gradually around the cutoff rather than abruptly, eliminating the discontinuity required for standard regression discontinuity estimation. Existing measurement-error corrections are problematic because premature delivery can itself affect outcomes. Excluding potentially misclassified births would require a wide exclusion window and substantial extrapolation. The authors instead combine an untreated birth cohort with separate validation data. They use the preceding academic-year cohort to predict outcomes in the absence of reform, assuming that untreated outcome profiles across actual birth dates differ between cohorts only by a constant shift. They estimate this shift using births sufficiently early to be considered unaffected. The Millennium Cohort Study, which records actual and expected birth dates for 18,818 children, provides estimated probabilities of assignment to the new policy regime. Dividing the difference between observed and predicted outcomes by these probabilities identifies conditional average effects among those assigned to the reform. These estimates capture the effects of policy-regime assignment, not actual benefit receipt. Children’s outcomes come from England’s National Pupil Database, covering roughly 600,000 pupils per academic-year cohort; maternal outcomes come from the Quarterly Labour Force Survey. The strongest evidence of improved child outcomes concerns the 2007 reform. It increased age-five test scores by 0.045 standard deviations, equivalent to the score difference associated with approximately 25 additional days of age at testing. Effects on maths and language were similar. By age eleven, the estimated gain remained statistically significant and corresponded to about 15 days of age, although the age-seven estimates were imprecise and statistically insignificant. Age-five effects were approximately 1.5 times larger in rural areas than in urban areas and twice as large in the most deprived fifth of areas as in the least deprived fifth. The 2003 reform produced a smaller age-five gain, equivalent to roughly ten days of age, but its age-eleven estimate was not statistically distinguishable from zero. The 2000 job-protection reform produced no detectable educational improvement, with estimates ruling out positive gains larger than approximately four days of age. All three reforms reduced maternal employment by around 9–10 percentage points one year after childbirth. They also reduced average weekly working hours, including zeros for non-employed mothers, by approximately 2.1–2.6 hours. Six years later, employment remained approximately 5.6–7.1 percentage points lower, while weekly working hours remained around 1.2–2.0 hours lower. The authors conclude that paid maternity leave can generate modest, lasting educational benefits where initial provision is limited, without strengthening mothers’ longer-term employment. Similar employment responses alongside different child outcomes suggest that income support matters beyond the additional time mothers spend at home, although the study does not directly establish this mechanism. Causal interpretation depends on parallel birth-date outcome profiles in the absence of reform and on the applicability of the birth-timing distribution across cohorts and populations. Placebo tests and alternative control cohorts support these assumptions but cannot prove them. Maternal estimates require additional functional-form restrictions, and the reforms’ overlapping components prevent a clean separation of the effects of pay and job protection.

Does Longer Maternity Leave Help Children? Why Pay May Matter More Than Protection

Maternity leave promises parents time to care for a new baby without losing their connection to work. But does expanding leave also improve children’s education? And does it help mothers remain employed over the longer term? In “Evaluating Maternity Leave Reforms When RD Fails,” Clara von Bismarck-Osten and Andrei Zeleneev examine three UK reforms to explore whether maternity pay and job protection deliver different benefits.

Their findings suggest that extending paid leave can produce modest, lasting educational gains when initial support is relatively limited. Extending job protection alone, however, produced no detectable improvement in children’s achievement. None of the reforms strengthened mothers’ longer-term employment.

Three Reforms, Different Kinds of Support

The researchers study policy changes that expanded maternity entitlements in different ways:

The 2000 reform expanded job-protected maternity leave.
The 2003 reform extended both leave and maternity pay.
The 2007 reform primarily extended maternity pay, from 26 to 39 weeks.

These differences offer a way to investigate whether financial support matters beyond the opportunity to spend more time at home. However, because the reforms contained overlapping components, they do not provide a perfectly clean separation of pay and job protection.

Why a Standard Research Method Fails Here

Policy eligibility cutoffs often allow researchers to compare people just eligible for a benefit with those just ineligible. This approach, known as regression discontinuity, depends on a sharp change in treatment at the cutoff.

Here, eligibility depended on a child’s expected birth date, while the datasets containing outcomes recorded only actual birth dates. Early and late deliveries blur the boundary: children born on the same date could fall under different policy regimes. Consequently, the probability of assignment changes gradually across actual birth dates rather than jumping abruptly.

Standard measurement-error corrections are also problematic because premature birth can itself influence later outcomes. Simply removing potentially misclassified births would require excluding a wide window and extrapolating substantially across it.

Building a Counterfactual From Another Cohort

The authors instead use the preceding academic-year birth cohort to predict outcomes without reform. Their central assumption is that, absent policy changes, outcome patterns across actual birth dates would differ between cohorts only by a constant shift. They estimate that shift using births sufficiently early to be considered unaffected.

They then use the Millennium Cohort Study, which records expected and actual birth dates for 18,818 children, to estimate probabilities of assignment to the new regime. Dividing the gap between observed and predicted outcomes by those probabilities identifies conditional average effects among children assigned to the reform. These are effects of policy-regime assignment, not actual benefit receipt.

Educational outcomes come from England’s National Pupil Database, covering roughly 600,000 pupils per academic-year cohort. Maternal employment outcomes come from the Quarterly Labour Force Survey.

The Clearest Educational Gains Followed the Pay Extension

The strongest evidence concerns the 2007 reform. At age five, children’s test scores increased by 0.045 standard deviations. To make that magnitude more intuitive, the researchers compare it with the score difference associated with being approximately 25 days older at testing. This is a benchmark, not an estimate of extra days of schooling.

Maths and language effects were similar. At age eleven, the estimated improvement remained statistically significant and corresponded to approximately 15 days of age. Age-seven estimates, however, were imprecise and not statistically significant.

The age-five benefits were unevenly distributed. Estimated effects were approximately 1.5 times larger in rural than urban areas, and twice as large in the most deprived fifth of areas as in the least deprived fifth.

The 2003 reform generated a smaller age-five gain, equivalent to roughly ten days of age, but its age-eleven estimate was not statistically distinguishable from zero. The 2000 job-protection expansion produced no detectable educational improvement; its estimates ruled out positive gains larger than approximately four days of age.

Employment Fell—and Stayed Lower

Despite their different educational effects, all three reforms generated similar employment responses. One year after childbirth, maternal employment was around 9–10 percentage points lower. Average weekly working hours, counting non-employed mothers as working zero hours, fell by approximately 2.1–2.6 hours.

Six years after childbirth, employment remained approximately 5.6–7.1 percentage points lower, while average weekly hours remained around 1.2–2.0 hours lower. These findings challenge the expectation that more generous maternity entitlements necessarily strengthen mothers’ longer-term employment.

What Policymakers Can Take Away

Similar employment responses alongside different educational outcomes suggest that income support may matter beyond additional time at home. The study does not directly establish that mechanism, however.

Causal interpretation also depends on comparable untreated birth-date outcome profiles and on birth-timing patterns transferring across cohorts and populations. Placebo tests and alternative control cohorts support these assumptions without proving them; maternal estimates require additional functional-form restrictions.

The central lesson is therefore measured: where initial provision is limited, paid maternity leave can support modest, lasting educational gains. But those gains should not be confused with stronger maternal employment, and leave duration alone does not tell the whole story.

Created on 02 Nov. 2026

Was this synthesis useful?

Score: 0

Ask questions about this paper to our AI assistant

How to chat with this paper? Illustrated guide

Compare with the PDF

You can also chat with multiple papers at once here.

AI assistant instructions?

Tools for this paper

Explain it differently 3 free rewrites left today

Save to a collection Group papers in collections to compare them, chat with them and export them. Create a free account Log in

Export BibTeX · RIS (Zotero, Mendeley, EndNote) · Obsidian (Markdown) · Notion (Markdown)

Advanced tools Premium

For researchers who follow a field closely: citation alerts, journal-club slides, version comparison and Zotero export. Discover Premium

Zotero

Citation alerts Citation alerts guide

Who works on it? The people and institutions who publish on its topic

Weekly letter Presented in "Economics and finance this week" of 2 November 2026. Economics and finance this week: When the measuring tool changes the answer

Journal club Premium+ Slides to present it Propose for the journal club

Versions What changed between versions

Talk about it in public Premium+ Press kit: release, questions, hype check Press kit guide

Papers around this one

See them on a map

Ranked by how close their title and abstract are to this paper’s. Select some to read them together.

Closest papers

Related Broader context
Workplace Breastfeeding Legislation and Labor Market Outcomes
Julia Hatamyar · 2022 · General Economics · Not summarized yet

Workplace Breastfeeding Legislation and Labor Market Outcomes estimates how workplace breastfeeding laws affect women’s labor-force participation, a neighboring maternal employment policy question.

Abstract
This paper studies the effects of legislation mandating workplace breastfeeding amenities on various labor market outcomes. Using the Panel Study of Income Dynamics, I implement both a traditional fixed-effects event study framework and the Interaction-Weighted event study technique proposed by Sun & Abraham (2020). I find that workplace breastfeeding legislation increases the likelihood of female labor force participation by 4.2 percentage points in the two years directly following implementation. Female labor force participation remains higher than before implementation in subsequent years, but the significant effect does not persist. Using the CDC's Infant Feeding Practices Survey II, I t…
Related Broader context
Bounds on Distributional Treatment Effect Parameters using Panel Data with an Application on Job Displacement
Brantly Callaway · 2020 · Econometrics · cited 12 times · AI summary available

Bounds on Distributional Treatment Effect Parameters develops panel-data methods for treatment-effect identification and applies them to job displacement, rather than maternity-leave reforms.

Abstract
This paper develops new techniques to bound distributional treatment effect parameters that depend on the joint distribution of potential outcomes -- an object not identified by standard identifying assumptions such as selection on observables or even when treatment is randomly assigned. I show that panel data and an additional assumption on the dependence between untreated potential outcomes for the treated group over time (i) provide more identifying power for distributional treatment effect parameters than existing bounds and (ii) provide a more plausible set of conditions than existing methods that obtain point identification. I apply these bounds to study heterogeneity in the effect of …
Related Another approach to the problem
Labor Market Outcomes and Early Schooling: Evidence from School Entry Policies Using Exact Date of Birth
Pedro Cavalcante Oliveira, Daniel Duque · 2019 · General Economics · preprint, not peer reviewed · AI summary available

Labor Market Outcomes and Early Schooling uses an exact-date-of-birth regression discontinuity to estimate schooling’s effects on adult labor outcomes, unlike the paper’s imputation approach when RD fails.

Abstract
We use a rich, census-like Brazilian dataset containing information on spatial mobility, schooling, and income in which we can link children to parents to assess the impact of early education on several labor market outcomes. Brazilian public primary schools admit children up to one year younger than the national minimum age to enter school if their birthday is before an arbitrary threshold, causing an exogenous variation in schooling at adulthood. Using a Regression Discontinuity Design, we estimate one additional year of schooling increases labor income in 25.8% - almost twice as large as estimated using mincerian models. Around this cutoff there is also a gap of 9.6% on the probability of…
Related Broader context
Welfare Effects of Labor Income Tax Changes on Married Couples: A Sufficient Statistics Approach
Egor Malkov · 2021 · General Economics · cited 2 times · AI summary available

Welfare Effects of Labor Income Tax Changes assesses labor-supply and welfare responses to tax reforms for married couples, not maternity-leave effects.

Abstract
This paper develops a framework for assessing the welfare effects of labor income tax changes on married couples. I build a static model of couples' labor supply that features both intensive and extensive margins and derive a tractable expression that delivers a transparent understanding of how labor supply responses, policy parameters, and income distribution affect the reform-induced welfare gains. Using this formula, I conduct a comparative welfare analysis of four tax reforms implemented in the United States over the last four decades, namely the Tax Reform Act of 1986, the Omnibus Budget Reconciliation Act of 1993, the Economic Growth and Tax Relief Reconciliation Act of 2001, and the T…
Related Broader context
Skills and Liquidity Barriers to Youth Employment: Medium-term Evidence from a Cash Benchmarking Experiment in Rwanda
Craig McIntosh, Andrew Zeitlin · 2022 · General Economics · cited 2 times · Not summarized yet

Skills and Liquidity Barriers to Youth Employment evaluates training and cash-transfer effects on young adults’ work and entrepreneurship in Rwanda.

Abstract
We present results of an experiment benchmarking a workforce training program against cash transfers for underemployed young adults in Rwanda. 3.5 years after treatment, the training program enhances productive time use and asset investment, while the cash transfers drive productive assets, livestock values, savings, and subjective well-being. Both interventions have powerful effects on entrepreneurship. But while labor, sales, and profits all go up, the implied wage rate in these businesses is low. Our results suggest that credit is a major barrier to self-employment, but deeper reforms may be required to enable entrepreneurship to provide a transformative pathway out of poverty.
Related Broader context
The Structure and Incentives of a COVID related Emergency Wage Subsidy
Jules Linden, Cathal O'Donoghue, Denisa M. Sologon · 2021 · General Economics · cited 2 times · AI summary available

The Structure and Incentives of a COVID related Emergency Wage Subsidy simulates alternative wage-subsidy designs and their labor-market incentives.

Abstract
During recent crisis, wage subsidies played a major role in sheltering firms and households from economic shocks. During COVID-19, most workers were affected and many liberal welfare states introduced new temporary wage subsidies to protected workers' earnings and employment (OECD, 2021). New wage subsidies marked a departure from the structure of traditional income support payments and required reform. This paper uses simulated datasets to assess the structure and incentives of the Irish COVID-19 wage subsidy scheme (CWS) under five designs. We use a nowcasting approach to update 2017 microdata, producing a near real time picture of the labour market at the peak of the crisis. Using microsi…
Related Broader context
Effect of breastfeeding on gastrointestinal infection in infants: A targeted maximum likelihood approach for clustered longitudinal data
Mireille E. Schnitzer, Mark J. van der Laan, Erica E. M. Moodie et al. · 2014 · Annals of Applied Statistics 2014, Vol. 8, No. 2, 703-725 · cited 63 times · AI summary available

Effect of breastfeeding on gastrointestinal infection estimates infant health effects of breastfeeding duration, rather than effects of maternity-leave policy.

Abstract
The PROmotion of Breastfeeding Intervention Trial (PROBIT) cluster-randomized a program encouraging breastfeeding to new mothers in hospital centers. The original studies indicated that this intervention successfully increased duration of breastfeeding and lowered rates of gastrointestinal tract infections in newborns. Additional scientific and popular interest lies in determining the causal effect of longer breastfeeding on gastrointestinal infection. In this study, we estimate the expected infection count under various lengths of breastfeeding in order to estimate the effect of breastfeeding duration on infection. Due to the presence of baseline and time-dependent confounding, specialized …
Related Broader context
Nonparametric Identification and Estimation of Earnings Dynamics using a Hidden Markov Model: Evidence from the PSID
Tong Zhou · 2023 · Applications · cited 1 time · AI summary available

Nonparametric Identification and Estimation of Earnings Dynamics uses nonparametric methods to model earnings processes in the PSID, not policy treatment effects.

Abstract
This paper presents a hidden Markov model designed to investigate the complex nature of earnings persistence. The proposed model assumes that the residuals of log-earnings consist of a persistent component and a transitory component, both following general Markov processes. Nonparametric identification is achieved through spectral decomposition of linear operators, and a modified stochastic EM algorithm is introduced for model estimation. Applying the framework to the Panel Study of Income Dynamics (PSID) dataset, we find that the earnings process displays nonlinear persistence, conditional skewness, and conditional kurtosis. Additionally, the transitory component is found to possess non-Gau…

Go further

How are these papers found?

We compare the title and abstract of this paper with those of every paper summarized on this site, and with the papers Semantic Scholar recommends, those it cites and those citing it (for a much-cited paper, its most recent citations and, from OpenAlex, the most cited ones). Among the prior work and the follow-ups, the papers that use each other in depth, a method it uses and the much-cited follow-ups come a little higher. The levels (very close, close, related) come from the similarity of the texts, not from a judgment of quality; * marks a paper without an abstract, compared by its title. The sentence under a paper is written by an AI from the two abstracts: check it in the papers. A paper marked not relevant by three readers leaves the list.

Updated on 2 November 2026.

Disclaimer: The AI-based summarization tool and virtual assistant provided on this website may not always provide accurate and complete summaries or responses. We encourage you to carefully review and evaluate the generated content to ensure its quality and relevance to your needs.

Evaluating Maternity Leave Reforms When RD Fails

The essentials

Question

Method

Findings

Limitations

Why it matters

Read further

Definitions

Does Longer Maternity Leave Help Children? Why Pay May Matter More Than Protection

Three Reforms, Different Kinds of Support

Why a Standard Research Method Fails Here

Building a Counterfactual From Another Cohort

The Clearest Educational Gains Followed the Pay Extension

Employment Fell—and Stayed Lower

What Policymakers Can Take Away

Ask questions about this paper to our AI assistant

Tools for this paper

Advanced tools Premium

Papers around this one

Closest papers

Go further

Cite this paper (BibTeX)