Regression Discontinuity Estimates Research Articles

The impact of IOM recommendations on gestational weight gain among US women: An analysis of birth records during 2011-2019.

The prevailing guidelines of the Institute of Medicine (IOM) of United States on gestational weight gain (GWG) are based on women's prepregnancy body mass index (BMI) categories. Previous research has shown that the guidelines issued in 1990 and revised in 2009 had no effect. We investigate the effectiveness of new guidelines issued in 2009 analyzing the records of all singleton births in the U.S. during 2011-2019 (34.0 million observations). We use the discontinuity in recommended guidelines at the threshold values of BMI categories in a regression discontinuity (RD) research design to investigate the effect of IOM guidelines on GWG. We also use an RD analysis in a difference in difference (DID) framework where we compare the effect on women who had any prenatal care to others who did not receive prenatal care. The naïve RD estimator predicts an effect in the expected direction at the threshold BMI values of 18.5 and 25.0 but not at 30.0. After the DID based correction, the RD analyses show that the GWG, measured in kg, drop at the BMI values of 18.5, 25.0 and 30.0 by 0.189 [CI: 0.341, 0.037], 0.085 [CI: 0.179, -0.009] and 0.200 [CI: 0.328, 0.072] respectively when the midpoint of the recommended range in kg drops by 1.5, 4.5 and 2.25. This implies a responsiveness of 12.6%, 1.9% and 8.9% respectively to changes in guidelines at these BMI values. The findings show that the national guidelines have induced some behavioral changes among US women during their pregnancy resulting in a change in GWG in the expected direction. However, the magnitude of the change has not been large compared to the expectations, implying that the existing mechanisms to implement these guidelines have not been sufficiently strong.

Summaries of Articles

Summaries of Articles

Impact of a guaranteed minimum income program on rural–urban migration in China

Abstract This article relies on a regression discontinuity (RD) design to estimate the impact of an unconditional cash transfer from Minimum Income Living Allowance (MLSA)—one of the largest basic income guarantee programs in the world—on the household decision to participate in rural–urban migration. The study is informed by novel survey data that provide the first and only representative information on China’s large, but understudied ethnic minority areas. Exploiting the income-based MLSA eligibility rule as an instrument, fuzzy RD estimates reveal that MLSA subsidy receipt significantly increases the likelihood that complier households participate in rural–urban migration, a finding that is robust to a batter of sensitivity tests and checks for robustness. In line with the idea that MLSA cash subsidies help to loosen household credit or risk constraints, I show that the observed positive effects are driven mainly by poor [ethnic minority] households that face relatively high perceived migration costs. Additional evidence shows that the MLSA program increases complier households’ disposable income and consumption despite having no significant effect on household investment behavior. The main findings suggest that a small cash infusion from a minimum income program like MLSA helps to promote migration-led urbanization and rural household wellbeing.

Association Between the Patient Driven Payment Model and Therapy Utilization and Patient Outcomes in US Skilled Nursing Facilities

In October 2019, Medicare changed its skilled nursing facility (SNF) reimbursement model to the Patient Driven Payment Model (PDPM), which has modified financial incentives for SNFs that may relate to therapy use and health outcomes. To assess whether implementation of the PDPM was associated with changes in therapy utilization or health outcomes. This cross-sectional study used a regression discontinuity (RD) approach among Medicare fee-for-service postacute-care patients admitted to a Medicare-certified SNF following hip fracture between January 2018 and March 2020. Skilled nursing facility admission after PDPM implementation. Main outcomes were individual and nonindividual (concurrent and group) therapy minutes per day, hospitalization within 40 days of SNF admission, SNF length of stay longer than 40 days, and discharge activities of daily living score. The study cohort included 201 084 postacute-care patients (mean [SD] age, 83.8 [8.3] years; 143 830 women [71.5%]; 185 854 White patients [92.4%]); 147 711 were admitted pre-PDPM, and 53 373 were admitted post-PDPM. A decrease in individual therapy (RD estimate: -15.9 minutes per day; 95% CI, -16.9 to -14.6) and an increase in nonindividual therapy (RD estimate: 3.6 minutes per day; 95% CI, 3.4 to 3.8) were observed. Total therapy use in the first week following admission was about 12 minutes per day (95% CI, -13.3 to -11.3) (approximately 13%) lower for residents admitted post-PDPM vs pre-PDPM. No consistent and statistically significant discontinuity in hospital readmission (0.31 percentage point increase; 95% CI, -1.46 to 2.09), SNF length of stay (2.7 percentage point decrease in likelihood of staying longer than 40 days; 95% CI, -4.83 to -0.54), or functional score at discharge (0.04 point increase in activities of daily living score; 95% CI, -0.19 to 0.26) was observed. Nonindividual therapy minutes were reduced to nearly zero in late March 2020, likely owing to COVID-19-related restrictions on communal activities in SNFs. In this cross-sectional study of SNF admission after PDPM implementation, a reduction of total therapy minutes was observed following the implementation of PDPM, even though PDPM was designed to be budget neutral. No significant changes in postacute outcomes were observed. Further study is needed to understand whether the PDPM is associated with successful discharge outcomes.

The Effect of Job Loss and Unemployment Insurance on Crime in Brazil

We investigate the impact of job loss on crime and the mitigating role of unemployment benefits, exploiting detailed individual‐level data linking employment careers, criminal records, and welfare registries for the universe of male workers in Brazil. The probability of committing crimes increases on average by 23% for workers displaced by mass layoffs, and by slightly less for their cohabiting sons. Using causal forests, we show that the effect is entirely driven by young and low‐tenure workers, while there is no heterogeneity by education and income. Regression discontinuity estimates indicate that unemployment benefit eligibility completely offsets potential crime increases upon job loss, but this effect vanishes completely immediately after benefit expiration. Our findings point to liquidity constraints and psychological stress as the main drivers of criminal behavior upon job loss, while substitution between time on the job and leisure does not seem to play an important role.

Minimum legal drinking age and the social gradient in binge drinking

Low minimum legal drinking ages (MLDAs), as prevalent in many European countries, are severely understudied. We use rich survey and administrative data to estimate the impact of the Austrian MLDA of 16 on teenage drinking behavior and morbidity. Regression discontinuity estimates show that legal access to alcohol increases the frequency and intensity of drinking, which results in more hospital admissions due to alcohol intoxication. The effects are stronger for boys and teenagers with low socioeconomic background. Evidence suggests that the policy’s impact cannot be fully explained by access. Data from an annual large-scale field study show that about 25 percent of retailers sell even hard liquor to underage customers. More generally, perceived access to alcohol is very high and hardly changes at the MLDA. However, teenagers consider binge drinking at weekends to be less harmful after gaining legal access.

Penalty-point system, deterrence and road safety: A quasi-experimental analysis

We use three complementary quasi-experimental approaches to study the causal effect of introducing a penalty-point system (PPS) in Spain in July 1 2006, on drivers' behavior. According to Regression-Discontinuity (RD) estimates, the PPS decreased the number of traffic offenders by 14%, and this deterrence effect was directly related to the severity of the penalty. Concerns that RD estimates may over- or under-state the longer-run effects of the policy change are addressed by presenting the Difference-in-Differences (DiD) estimates, which reveal that the deterrence effect of this reform increased over time. Crucially, the reform also curbed accidents by 14%, injuries by 16%, and fatalities by 14%, and these effects persisted over time. Difference-in-RD estimates are consistent with the other two approaches. Altogether, the reform represented net benefits of over 946 million EUR per year (or 0.09% of the Spanish GDP) during the first three years after implementation.

Evaluation of an intervention targeted with predictive analytics to prevent readmissions in an integrated health system: observational study

ObjectivesTo determine the associations between a care coordination intervention (the Transitions Program) targeted to patients after hospital discharge and 30 day readmission and mortality in a large, integrated healthcare system.DesignObservational...

Effects of school attendance on child labour and literacy: regression discontinuity evidence from Ghana

PurposeThe purpose of this paper is to examine the effects of school attendance on learning and child labour in Ghana.Design/methodology/approachThe paper uses a nationally representative sample of household and individual data in 2005/06 and 2011/12 for the analysis. Regression discontinuity, the capitation grant in 2005 as exogenous, is used to estimate the impact of school attendance on child labour and learning outcomes.FindingsThe study found that children who were exposed to the capitation grant spent more hours in school and were more likely to enrol in primary school. School attendance was found to increase the likelihood to read and write a standardised test in English. Also, the improvement in children’s school attendance was found to enhance the likelihood of performing a written calculation. The authors could not find any evidence that school attendance affected child labour.Originality/valueThis research is the first causality analysis in sub-Saharan Africa that uses a nationally representative dataset to study the impact of school attendance on child labour and learning outcomes using a regression discontinuity estimator to deal with endogeneity issues.

Local Polynomial Order in Regression Discontinuity Designs

Treatment effect estimates in regression discontinuity (RD) designs are often sensitive to the choice of bandwidth and polynomial order, the two important ingredients of widely used local regression methods. While Imbens and Kalyanaraman and Calonico, Cattaneo, and Titiunik provided guidance on bandwidth, the sensitivity to polynomial order still poses a conundrum to RD practitioners. It is understood in the econometric literature that applying the argument of bias reduction does not help resolve this conundrum, since it would always lead to preferring higher orders. We therefore extend the frameworks of Imbens and Kalyanaraman and Calonico, Cattaneo, and Titiunik and use the asymptotic mean squared error of the local regression RD estimator as the criterion to guide polynomial order selection. We show in Monte Carlo simulations that the proposed order selection procedure performs well, particularly in large sample sizes typically found in empirical RD applications. This procedure extends easily to fuzzy regression discontinuity and regression kink designs.

Scrutinizing the Monotonicity Assumption in IV and fuzzy RD designs*

AbstractWhenever treatment effects are heterogeneous, and there is sorting into treatment based on the gain, monotonicity is a condition that both instrumental variable (IV) and fuzzy regression discontinuity (RD) designs must satisfy for their estimate to be interpretable as a local average treatment effect. However, applied economic work often omits a discussion of this important assumption. A possible explanation for this missing step is the lack of a clear framework to think about monotonicity in practice. In this paper, we use an extended Roy model to provide insights into the interpretation of IV and fuzzy RD estimates under various degrees of treatment effect heterogeneity, sorting on gain and violation of monotonicity. We then extend our analysis to two applied settings to illustrate how monotonicity can be investigated using a mix of economic insights, data patterns and formal tests. For both settings, we use a Roy model to interpret the estimate even in the absence of monotonicity. We conclude with a set of recommendations for the applied researcher.

The unequal impact of natural light on crime

This paper studies the relationship between ambient light and criminal activity. I develop a Becker-style crime model that shows that a sudden increase in ambient light produces a larger reduction in crime in areas with less public lighting. Daylight savings time (DST), the natural experiment used, induces a sharp increase in natural light during crime-intense hours. Using geolocated data on crime and public lighting for the city of Montevideo in Uruguay, regression discontinuity estimates identify a strong and statistically significant decrease in robbery of 17%. The decrease is larger in poorly lit areas. Computing the level of public lighting at which DST has no effect on crime reduction, I identify the minimum level of public lighting that an area should target.

Fighting Failure: The Persistent Real Effects of Resolving Distressed Banks

We study the real effects of resolving distressed banks using quasi-experimental variation in resolutions introduced by a threshold-based rule of the FDIC Improvement Act. Our fuzzy regression discontinuity estimates indicate that resolutions lead to reductions in employment and establishments growth of up to six percentage points. These effects are concentrated in small, less urban counties, and translate to large declines in SME lending and increases in corporate bankruptcies. These results imply that large acquiring banks restrict lending to the small business borrowers of distressed target banks. Overall, current bank resolution policy may have costly externalities for local economic activity.

A regression discontinuity design for ordinal running variables: Evaluating central bank purchases of corporate bonds

Regression discontinuity (RD) is a widely used quasi-experimental design for causal inference. In the standard RD the assignment to treatment is determined by a continuous pretreatment variable (i.e., running variable) falling above or below a prefixed threshold. Recent applications increasingly feature ordered categorical or ordinal running variables which pose challenges to RD estimation due to the lack of a meaningful measure of distance. This paper proposes an RD approach for ordinal running variables under the local randomization framework. The proposal first estimates an ordered probit model for the ordinal running variable. The estimated probability of being assigned to treatment is then adopted as a latent continuous running variable and used to identify a covariate-balanced subsample around the threshold. Assuming local unconfoundedness of the treatment in the subsample, an estimate of the effect of the program is obtained by employing a weighted estimator of the average treatment effect. Two weighting estimators—overlap weights and ATT weights—as well as their augmented versions are considered. We apply the method to evaluate the causal effects of the corporate sector purchase programme (CSPP) of the European Central Bank which involves large-scale purchases of securities issued by corporations in the euro area. We find a statistically significant and negative effect of the CSPP on corporate bond spreads at issuance.

Promoting student achievement in high school using school funding: evidence from quantile regression discontinuity design

Even though there are many quasi-experimental research in recent literature, there is still no consensus on whether an increase in school funding improves student achievement. Leveraging a natural experiment in South Korea, this study exploits the discontinuity in school funding rules to identify the impact of increased funding on the test scores of high-school students in a national assessment exam. The setting provides a useful context to study the effect of school funding because students typically attend largely similar schools that follow a standardized curriculum, thus eliminating the possibility of the results being contaminated by idiosyncratic variation in school-level characteristics. This study reports mean regression discontinuity estimates as well as quantile regression discontinuity estimates using a procedure suggested by Frandsen et al. (J Econom 168:382–395, 2012). The findings reveal that an increase in school funding, which is equal to approximately 300,000 won per student, results in improved exam performance, particularly in mathematics. Contrary to the stated purpose of the program, however, the evidence suggests that students in the middle and top of the ability distribution gained the most from the intervention, rather than students who are at the highest risk of failing.

On the interpretation of diploma wage effects estimated by regression discontinuity designs

AbstractSeveral recent papers employ regression discontinuity (RD) designs to estimate the causal effect of a diploma or similar credentials on wages. I build a model of knowledge acquisition, test‐taking and labour market careers that mimics the settings studied in these papers. I show that a positive RD estimate is evidence for information frictions in the labour market, but caution that the relative importance of acquired knowledge and innate talent in the production function cannot be separately identified. While a positive RD estimate does not reveal whether students study too much or too little compared to the social optimum, the rate at which RD estimates decline with labour market experience indicates the speed of employer learning, a parameter that is critical for the extent of inefficiency in study choices.

Do Fiscal Restraints Harm Test Scores? Evidence from Italy

Most countries discipline their public budget through a set of fiscal rules aiming at limiting public debt accumulation. Yet, apart from the direct effect on public finance outcomes, there is limited evidence on whether these policies affect broader socio-economic outcomes. This paper provides regression discontinuity estimates of fiscal rules-induced school spending drops on test scores of Italian students. We show that school spending per-pupil is around 102 euros lower in municipalities subject to fiscal restraints. Using longitudinal data on pupils’ attainment in national test at the beginning and the end of primary school, we find that spending differences lead to a gap in standardized test score gains of nearly 12 percent of a standard deviation. The impact is particularly strong for lower socio-economic groups. We find that both the lack of several basic instructional tools and limited investments in school facilities explain most of the observed achievement gap. Our results reveal how fiscal restraints can create “unintended” consequences for younger generations and exacerbate cross-generation inequalities when governments need to reduce public spending.

Minimum Legal Drinking Age and the Social Gradient in Binge Drinking

Low minimum legal drinking ages (MLDAs), as prevalent in many European countries, are severely understudied. We use rich survey and administrative data to estimate the impact of the Austrian MLDA of 16 on teenage drinking behavior and morbidity. Regression discontinuity estimates show that legal access to alcohol increases the frequency and intensity of drinking, which results in more hospital admissions due to alcohol intoxication. The effects are stronger for boys and teenagers with low socioeconomic background. The policy’s impact is not driven by access. Data from an annual large-scale field study shows that about 25 percent of all retailers sell even hard liquor to underage customers. In line with this, perceived access to alcohol is very high and hardly changes at the MLDA. However, teenagers consider binge drinking at weekends to be less harmful after gaining legal access.

Abandoned Retirement Savings

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 in 2017, 2.7% of 72.5 year-old account-holders in total abandoned $790 million; the median abandoned account held $5,400. Nearly all of these funds remain with plans and are not sent to state unclaimed property. Regression discontinuity estimates show that abandonment is 10 times higher in automatic rollover IRAs, a type of default account. We nest our findings in a model of retirement savings featuring forgetting to derive implications for passive and active savers.

Shirking with Good Reputation? Evidence from Hotel Industry

This paper empirically examines the impact of online reputation on investment in hotel industry. Recent theory suggests that reputation could have ambiguous effects on investments. Using detailed firm-level data on investment expenditures, and online consumer ratings from Taiwanese hotels, I adopt a regression discontinuity design based on TripAdvisor's rating display system and identify treatment effects. The regression discontinuity estimates show that that higher ratings negatively impact investment expenditures while lower ratings tend to encourage investment. The findings are consistent with Board and Meyer-ter-Vehn (2013), in which their good news model predicts that firms shirk when they have good reputation.