Regression Discontinuity Estimates Research Articles

Getting away from the cutoff in regression discontinuity designs

Regression discontinuity (RD) designs are highly popular in economic research because of their strong internal validity and straightforward intuition. While RD estimates are local in nature, several recent articles propose methods that generalize RD estimates to units outside a small neighborhood of the cutoff. In this article, I introduce the getaway package, which implements the method proposed by Angrist and Rokkanen (2015, Journal of the American Statistical Association 110: 1331-1344) to extrapolate treatment-effect estimates “away from the cutoff”, relying on a classical unconfoundedness condition. Additionally, the package features a data-driven algorithm designed to identify a set of covariates that fulfills the unconfoundedness assumption. It also incorporates a toolkit intended for testing and visualization purposes.

Sorrow under prosperity: how downtown residence influences the welfare of disabled people in China

ABSTRACT While the welfare divergence between rural and urban disabled residents in China has attracted great scholarly attention, little to no research has examined the relevant disparities within the city. Due to the administrative-led resource allocation model in China, disabled individuals residing in the urban centre are expected to have a greater advantage in their daily lives compared to their counterparts in the suburbs. We use Tianjin, a Chinese municipality, as an example to explore the impact of disabled individuals’ downtown residence on their employment, education, marriage, and access to public services. Our spatial regression discontinuity estimates reveal that compared to suburban disabled residents, downtown residents have easier access to education but face difficulties in employment and marriage. A follow-up survey indicates that disabled residents in downtown areas face employment and marriage disadvantages due to implicit exclusion of low-skilled groups and high marriage burdens in urban centres. This paper sheds light on the welfare of disabled groups in cities and the latent impact of urban development on social minorities.

The Targeting and Impact of Partisan Gerrymandering: Evidence from a Legislative Discontinuity

Abstract A party’s influence over redistricting increases discontinuously when its seat share in the state legislature exceeds 50%. We apply bunching tests to show that, in the election before redistricting, parties systematically win narrow majorities in legislatures of states where they have lost recent U.S. House races. This trend of losses is reversed after redistricting despite no change in overall House vote shares in states near the cutoff. The pre-to-post-redistricting change in regression discontinuity estimates implies that the party that controls redistricting engineers an 11 percentage point increase in its probability of winning a House race.

Regression(s) discontinuity: Using bootstrap aggregation to yield estimates of RD treatment effects

Abstract Following Efron (2014), we propose an algorithm for estimating treatment effects for use by researchers employing a regression-discontinuity (RD) design. This algorithm generates a set of estimates of the treatment effect from bootstrapped samples, wherein the polynomial-selection algorithm developed by Pei, Lee, Card, and Weber (2021) is applied to each sample, the average of these RD treatment effect (RDTE) estimates is computed and serves as the overall estimate of the RDTE. Effectively, this procedure estimates a set of plausible RD estimates and weights the estimates by their likelihood of being the best estimate to form a weighted-average estimate. We discuss why this procedure may lower the estimate’s root mean squared error (RMSE). In simulation results, we show that this better performance is achieved, yielding up to a 5% reduction in RMSE relative to PLCW’s method and a 16% reduction in RMSE relative to Calonico, Cattaneo, and Titiunik’s (2014) method for bandwidth selection (with default settings).

Utilizing local likelihood in regression discontinuity design: Investigating the impact of antiretroviral therapy eligibility on retention in clinical HIV care in South Africa.

The regression discontinuity (RD) design is a widely utilized approach for assessing treatment effects. It involves assigning treatment based on the value of an observed covariate in relation to a fixed threshold. Although the RD design has been widely employed across various problems, its application to specific data types has received limited attention. For instance, there has been little research on utilizing the RD design when the outcome variable exhibits zero-inflation. This study introduces a novel RD estimator using local likelihood, which overcomes the limitations of the local linear regression model, a popular approach for estimating treatment effects in RD design, by considering the data type of the outcome variable. To determine the optimal bandwidth, we propose a modified Ludwig-Miller cross validation method. A set of simulations is carried out, involving binary, count, and zero-inflated outcome variables, to showcase the superior performance of the suggested method over local linear regression models. Subsequently, the proposed local likelihood model is employed on HIV care data, where antiretroviral therapy eligibility is determined by a CD4 count threshold. A comparison is made between the results obtained using the local likelihood model and those obtained using local linear regression.

Impact of the Russia–Ukraine war on grain and fertilizer supply: Evidence from Kenya

AbstractThis paper examines the impact of the Russia–Ukraine war on grain and fertilizer supply to Kenya as the sixth largest economy in Africa using the Regression Discontinuity estimator on monthly data spanning from January 2017 to August 2022. The share of the quantity and value of cereals and fertilizer imports by Kenya from Russia and Ukraine in the total quantity and value of cereals and fertilizer imports by Kenya from world are utilized. The share of freight and insurance costs on grain and fertilizer imports by Kenya from Russia and Ukraine in total freight and insurance costs on grain and fertilizer from world have also been used. After controlling for sensitivity and incorporating depreciation of the exchange rate and a dummy on drought as covariates, the results indicate the war has had insignificant impact on imports of grain and fertilizer by Kenya though the impact is negative for grain but positive for fertilizer. The findings have policy implications touching on need for African countries to diversify import sources for grain and fertilizer, enhance productive capacities through investment on irrigation, strengthen diplomatic ties with existing import sources for grain and fertilizer, and invest in climate action to tackle drought incidences.

Translating Policy to Practice: An Association Between Medicare Access and Children's Health Insurance Program Reauthorization Act Implementation and Palliative Care Consultations and Perioperative Mortality in Critical Care.

To evaluate the 30-day postoperative mortality and palliative care consultations in patients that underwent surgical procedures in the United States before and after Medicare Access and Children's Health Insurance Program Reauthorization Act (MACRA) implementation. Retrospective, Observational cohort study. Secondary data were collected from the U.S. National Inpatient Sample, the largest hospital database in the country. The time span was from 2011 to 2019. Adult patients that electively underwent 1 of 19 major procedures. None. The primary outcome was cumulative postoperative mortality in two study cohorts. The secondary outcome was palliative care use. We identified 4,900,451 patients and categorized them into two study cohorts: PreM: 2011-2014 (n = 2,103,836) and PostM: 2016-2019 (n = 2,796,615). Regression discontinuity estimates and multivariate analysis were used. Across all procedures, 149,372 patients (7.1%) and 156,610 patients (5%) died within 30 days of their index procedures in the PreM and PostM cohorts, respectively. There was no statistically significant increase in mortality rates around postoperative day (POD) 30 (POD 26-30 vs 31-35) for both cohorts. More patients had inpatient palliative consultations during POD 31-60 compared with POD 1-30 in PreM (8,533 of 2,081,207 patients [0.4%] vs 1,118 of 22,629 patients [4.9%]) and PostM (18,915 of 2,791,712 patients [0.7%] vs 417 of 4,903 patients [8.5%]). Patients were more likely to receive palliative care consultations during POD 31-60 compared with POD 1-30 in both the PreM (odds ratio [OR] 5.31; 95% CI, 2.22-8.68; p < 0.001) and the PostM (OR 7.84; 95% CI, 4.83-9.10; p < 0.001) cohorts. We did not observe an increase in postoperative mortality after POD 30 before or after MACRA implementation. However, palliative care use markedly increased after POD 30. These findings should be considered hypothesis-generating because of several confounders.

Choice and consequence: Assessing mismatch at Chicago exam schools

The educational mismatch hypothesis asserts that students are hurt by affirmative action policies placing them in selective schools for which they wouldn’t otherwise qualify. We evaluate mismatch in Chicago’s selective public exam schools, which admit students using neighborhood-based diversity criteria as well as test scores. Regression discontinuity estimates for applicants favored by affirmative action indeed show no gains in reading and negative effects of exam school attendance on math scores and four-year college enrollment. But these results are similar for more- and less-selective schools and for applicants more and less likely to benefit from affirmative action, a pattern inconsistent with mismatch. We show that Chicago exam school effects are determined largely by the schools attended when not offered an exam school seat. In particular, apparent mismatch is explained by the fact that exam school admission diverts many applicants from high-performing Noble Network charter schools, where they would have done well. Consistent with these findings, exam schools reduce math scores for applicants applying from high-quality charter schools in another large urban district. Exam school applicants’ previous achievement, race, and other demographic characteristics that are sometimes said to mediate student-school matching play no role in this story.

Vocational training support and innovation at SMEs

Purpose While the value of human capital for technological innovation is well acknowledged, literature on the role of vocational training in corporate innovation is notably scarce. The purpose of this study is to assess the effect of government support for small and medium-sized enterprise (SME) competencies on Korean firms’ innovation. The author investigates SMEs’ patent applications (supported by the government to varying degrees) while accounting for firms’ market position, ownership and management structure, as well as prior changes in firms’ technologies, products, processes and other characteristics. Alternative hypotheses about management motivation – the “lazy manager”, “career concerns” and “special East Asian institutional constraints” hypotheses – are also evaluated. Design/methodology/approach Censored and count data analysis methods are used on a panel of 595 Korean firms covering 2005–2015 from the Korean Human Capital Corporate Survey, Intellectual Property Office and National Investment Commission. A regression discontinuity estimator accounts for potential endogeneity because of support for vocational training at firms. Findings Firms receiving training support are more innovative than firms without support, but latent effects may play a role. The regression-discontinuity model suggests that firms that succeeded only marginally in obtaining support had higher innovative output than non-recipients near the eligibility threshold. Originality/value The findings of this study establish that government support had the intended effect on SMEs’ technological capacity. This cannot be discounted as a simple crowding-out effect. The author also establishes that management–ownership separation within firms was conducive to innovation, that product competition had an inverse U-shaped effect and that management–ownership separation had a substitutable relationship with competition in overcoming managers’ effort avoidance. The findings support the “lazy manager” hypothesis over the “career concerns” and the “special East Asian institutional constraints” hypotheses.

Summertime and the drivin' is easy? Daylight saving time and vehicle accidents.

We investigate how exogenous variation in daylight caused by Daylight Saving Time affects road safety as measured by the count of vehicle crashes. We use administrative daily data from Greece covering the universe of all types of recorded vehicle accidents during the 2006-2016 period. Our regression discontinuity estimates support an ambient light mechanism that reduces the counts of serious vehicle accidents during the Spring transition and increases the count of minor ones during the Fall transition. The effects are driven from the hour intervals that are mostly affected from seasonal clock changes. We then discuss the potential cost implications of those seasonal transitions. In light of the talks about abolishing seasonal clock changes in the European Union (EU), our findings are policy relevant and can inform the public debate as empirical evidence for the block is scarce.

Formal Employment and Organised Crime: Regression Discontinuity Evidence from Colombia

Abstract Safety net programs, common in settings with high informality like Latin America, often use a means test to establish eligibility. We ask: in settings in which organised crime provides lucrative opportunities in the informal market, will discouraging formal employment via benefits eligibility criteria increase criminal enterprise activity? We link administrative socioeconomic microdata with the universe of arrests in Medellín over a decade, and exploit exogenous variation in formal-sector employment around a socioeconomic-score cutoff, below which individuals receive generous benefits if not formally employed. Regression discontinuity estimates confirm this policy reduced formal-sector employment and generated a corresponding increase in arrests associated with organised crime. We do not find increases for crimes unlikely to be associated with organised entities, such as crimes of impulse or opportunity. Effects on arrests are strongest in neighbourhoods where organised crime is most prevalent.

The Effect of Social Pension on Material Hardship among Older Adults in Korea: Regression Discontinuity Estimation

Abstract The goal of this study was to examine the effect of a social pension programme for older adults in South Korea, Basic Pension Scheme (BPS) on material hardship and subjective well-being. We apply a regression discontinuity design (RDD) to estimate the effect of the BPS on the material hardship and life satisfaction of older people between the ages of sixty-one and sixty-eight. Data come from Korea Welfare Panel Study (KOWEPS) wave 12 survey (2017, N = 3,932). The BPS benefit reduces the risks of housing hardship, bill payment delay and food insecurity. Interestingly, while the effect sizes of the BPS on mitigating the material hardship increase as income decreases, the lower-income groups were less satisfied with the pension provision than middle- and upper-income groups. This study contributes to the growing body of literature on material hardship for older adults in an Asian country facing persistent old-age poverty and immature public pensions.

Set it and forget it? Financing retirement in an age of defaults

Retirement savings abandonment is a rising concern connected to defined contribution systems and default enrollment. We use tax data on Individual Retirement Accounts (IRAs) to establish that for a recent cohort, 0.4% of retirement-age individuals abandoned an aggregate of $66 million, proxied by a failure to claim over ten years after a legal requirement to do so. Analysis of state unclaimed property databases suggests that workplace defined contribution plans are abandoned at a higher rate than IRAs. Finally, regression discontinuity estimates show that certain accounts created by default enrollment are at higher risk of abandonment by passive savers.

Student effort response to shifts in university admission policies

Policies which change the incentives students face are important levers for policymakers to improve learning. I study a statewide postsecondary admissions policy in North Carolina which introduces minimum admission standards for students wishing to attend college in-state. Regression discontinuity estimates show that high school students respond by increasing GPA and decreasing absences and suspensions, suggesting increased engagement. However, I find that the boost in GPA is driven by students substituting away from demanding coursework. Finally, I document stark heterogeneity of response across demographics, as minority and less affluent students have higher effort allocations but do not engage in strategic course-taking.

Political accountability for populist policies: Lessons from the world’s largest democracy

We know little about the electoral effects of policies with broad appeal that are implemented by popular leaders, but which have adverse economic effects. We analyze voter behavior following one such policy implemented in the world’s largest democracy – India’s 2016 ‘Demonetization,’ which unexpectedly made 86% of the currency-in-circulation redundant overnight, and led to severe cash shortages and economic hardship in subsequent months. Yet, the policy appealed to a majority of voters, and was framed as one that would combat corruption. We leverage a discontinuity in the number of bank branches arising from a nationwide, district-level bank expansion policy. Using the fact that districts with fewer banks had greater cash shortages, we identify the impacts of demonetization’s economic severity at the bank-expansion cutoff. Regression discontinuity estimates show that following demonetization, voters in places with more severe demonetization had less favorable views of the policy. Using a difference-in-discontinuity design, we find that the ruling party performed relatively worse in regions with more severe demonetization, receiving a 4.7 percentage point lower fraction of votes, and were relatively less likely to win seats in state legislatures. Areas that were historically strongly aligned with the ruling party were nearly unresponsive in voting behavior, despite having a less favorable view of the policy itself.

Does the early bird catch the worm? Evidence and interpretation on the long-term impact of school entry age in China

The long-term impact of children's age at primary school entry on educational attainment and labor market outcomes is one of the primary concerns to families, educators, and policymakers. Using a nationally representative survey of families and individuals, this paper is among the first to explore these effects in a causal sense in the Chinese context and understand the underlying mechanisms. We use a regression discontinuity (RD) design that employs the threshold date for primary school entry set by the 1986 Compulsory Education Law of China as a source of exogenous variation in the timing of school entry. We first document a salient and robust compliance rate of school entry requirement. RD estimates indicate that being born right after the cutoff date significantly increases years of schooling and annual earnings for non-agricultural jobs for the full sample. We also observe remarkable heterogeneous effects on labor market performance by gender. Being born right after the cutoff increases the probability of being in the labor force for men, but decreases that for women. We find that the decline in female labor force participation is mainly driven by women who come from economically and socially disadvantaged families. Further evidence suggests that this decline can be explained by supply-side factors including fertility decision and childcare provisions.

The impact of dismantling state monopoly on market integration: Evidence from the edible salt reform in China

AbstractState monopoly hinders market integration through rising interregional trade barriers. This article provides an account of the market reform on dismantling state monopoly as a natural experiment to investigate the extent to which the reform affects domestic market integration. The state monopoly on China's edible salt market was dismantled by implementing a market reform on the first of January 2017 that terminated the 2000‐year state monopoly on edible salt. Using a set of unique retailing price datasets on edible salt across cities over 10‐day periods every month, we take advantage of the regional and temporal variances of edible salt retailing price between origins of production and other cities to demonstrate market integration nationwide in China. The regression discontinuity estimation is leveraged based on an apparently sharp discontinuity in inter‐city price differences. The results suggest that the dissolution of state monopoly on edible salt market leads to a 3.14 percent decrease in price differences of edible salt between origins and other cities, thus ultimately resulting in the promoted market integration. The results also suggest that the varied iodine contents in edible salt, the market power of state‐owned enterprises and the government‐firm relationships exert promising roles in the influence.

Dual Identification? The Effects of English Learner (EL) Status on Subsequent Special Education (SPED) Placement in an Equity-Focused District

This study examines the effects of English Learner (EL) status on subsequent Special Education (SPED) placement. Through a research-practice partnership, we link student demographic data and initial English proficiency assessment data across seven cohorts of test takers and observe EL and SPED programmatic participation for these students over 7 years. Our regression discontinuity (RD) estimates at the English proficiency margin consistently differ substantively from positive associations generated through regression analyses. RD evidence indicates that EL status had no effect on SPED placement at the English proficiency threshold. Grade-by-grade and subgroup RD analyses at this margin suggest that ELs were modestly underidentified for SPED during Grade 5 and that ELs whose primary language was Spanish were underidentified for SPED.

Effects of universal oral healthcare coverage in an adult population: A long-term nationwide natural experiment.

A large and long-term natural experiment occurred in Finland from the late 1980s-2000, when adults' entitlement to subsidized oral healthcare was strongly dependent on the arbitrary classification based on their year of birth: people born in 1956 or later were entitled to subsidized care, while people born before 1956 were not. The aim of this study was to investigate the effect of this expanded universal oral healthcare coverage on service use and oral health outcomes. Data from annual nationally representative cross-sectional postal surveys among 15-64-year-olds between 1990 and 2014 were used. For this study, the following outcome variables were formed: experiencing toothache during the past month (yes/no), the number of missing teeth with three different thresholds (over 10, over 5 or at least 1 missing tooth), brushing more than once a day and the number of visits to the dentist. Regression discontinuity plots and bias-corrected local polynomial regression discontinuity estimators measuring the effect of the extended universal coverage on the outcomes at the year-of-birth cut-off of 1956 were generated separately from the data from 1990 to 2000 and from 2002 to 2014. Between 1990 and 2000, the number of visits to the dentist (0.2 visits, 95% CI, confidence intervals: -0.03; 0.43) and the proportion of those who visited the dentist during the past 12 months (4.2%, 95% CI: 0.1%; 8.3%) increased at the year-of-birth cut-off of 1956. There were minor drops (1.5%-1.9%) in the number of missing teeth across all thresholds (over 10, over 5, or at least 1 missing teeth) at the cut-off. Analyses with the data from the surveys from 2002 to 2014 showed that there were no discontinuities in these outcomes at the cut-off of 1956. Regression discontinuity estimates related to toothache experience and toothbrushing frequency were inconclusive due to high variability in the underlying data and the likely small effect of the more universal coverage on these outcomes. The current study provided evidence of the beneficial effects of universal oral healthcare coverage on the oral healthcare service use and teeth preservation from a large and long-term natural experiment occurred in Finland from the late 1980s to 2000.

Regression Discontinuity Designs

The regression discontinuity (RD) design is one of the most widely used nonexperimental methods for causal inference and program evaluation. Over the last two decades, statistical and econometric methods for RD analysis have expanded and matured, and there is now a large number of methodological results for RD identification, estimation, inference, and validation. We offer a curated review of this methodological literature organized around the two most popular frameworks for the analysis and interpretation of RD designs: the continuity framework and the local randomization framework. For each framework, we discuss three main topics: ( a) designs and parameters, focusing on different types of RD settings and treatment effects of interest; ( b) estimation and inference, presenting the most popular methods based on local polynomial regression and methods for the analysis of experiments, as well as refinements, extensions, and alternatives; and ( c) validation and falsification, summarizing an array of mostly empirical approaches to support the validity of RD designs in practice.