Parallel Trends Assumption Research Articles

Porter in China: A quasi-experimental view of market-based environmental regulation effects on firm performance

This paper exploits a natural Market-based Environmental Regulation (MER) experiment (the 2007 Sulfur dioxide (SO2) Emissions Trading Pilot (ETP) of the 11th five-year plan (FYP)) to assess the empirical relevance of the Porter hypothesis in the context of Chinese firms. MER effects on firms' emissions intensity and innovation performances are quantified by combining a unique sample of 3,198,000 individual firms (including >516,126 SO2 emitting plants) with a Propensity Score Matched Difference-in-Differences (PSM-DID) quasi-experimental framework over the 2001–2014 period. Empirical results show how ETP significantly reduces plant-level SO2 emission intensity while incentivizing polluters to effectively innovate, thus locally reconciling economic and environmental objectives. Findings are robust to replacement variables, restricted samples, parallel trend assumption, and placebo tests, indicating that effective policy and secular smooth time effects are individually captured. Further mechanistic tests suggest that MERs trigger green innovation behaviors of local governments and firms whereas the heterogeneity analysis confirms why large enterprises and regions exposed to strong environmental controls achieve emission reductions more effectively. Implications for global policy are proposed to systematize MER in future environmental planning.

Medicaid Expansion Is Associated With Increases in Medicaid-Funded Total Joint Arthroplasty

BackgroundExisting literature presents competing views concerning the impact of Medicaid expansion on total joint arthroplasty (TJA) utilizations. While some reports demonstrate that expansion does not increase Medicaid acceptance by surgeons, others show increases in Medicaid-funded TJA via limited analyses. We conducted a nationwide, multi-insurance, econometric study to determine if Medicaid-funded and all-funding-source total hip arthroplasty (THA) or total knee arthroplasty (TKA) utilizations increased following expansion. MethodsThis study examined 999,015 THA and 2,099,975 TKA from 2010 to 2017 using a commercially available national payer database. Difference-in-differences analyses, econometric regression methods used to assess the impact of policy change, were used to examine the impact of Medicaid expansion on TJA utilizations, and event analyses were used to confirm the parallel trends assumption, which helps to ensure that the estimated effect is not a result of existing differences in trends between treatment and nontreatment groups. ResultsEvent analyses confirmed parallel trends in the pre-expansion period. Difference-in-differences analyses found a persistent increase in Medicaid-funded THA (40.4%, P = .001, confidence interval [CI]: 12.7, 62.1%), but not THA from all funding sources (4.6%, P = .128, CI: −1.3, 10.8%). Medicaid-funded TKA (35.8%, P < .001, CI: 17.4, 68.0%) increased, but not TKA from all funding sources (3.4%, P = .321, CI: −3.1, 10.1%). ConclusionWhile the number of Medicaid-funded TJAs increased, expansion had no significant effect when examining all funding sources. This suggests that Medicaid expansion primarily affected source of TJA funding, not overall volume. Further research is needed to examine state-specific predictors of response to Medicaid expansion.

The impact of the clean air act on particulate matter in the 1970s

We examine whether counties designated as out of attainment with the National Ambient Air Quality Standards (NAAQS) under the 1970 Clean Air Act (CAA) experienced larger reductions in Total Suspended Particulates (TSP) during the 1970s than attainment counties. We answer this question using the official designation of nonattainment status which, between 1972 and 1978, was by Air Quality Control Region (AQCR). Data from balanced panels of TSP monitors in operation from 1969–78 and from 1971–78 are used to examine the impact of nonattainment status on TSP. We also examine the impact of nonattainment on TSP using the definition in the literature, which designates a county as out of attainment if any of its monitors violated the NAAQS. Using the official (AQCR) nonattainment designation, TSP, on average, fell by 10.2 μg/m3 using the 1969 panel and 9.1 μg/m3 using the 1971 panel. Using the definition of nonattainment in the literature yields smaller reductions: 6.0 μg/m3 using the 1969-78 panel and 7.7 μg/m3 using the 1971-78 panel. Using the correct definition of non-attainment in difference-in-differences (DID) models calls into question whether these results can be interpreted as causal. When counties are characterized using the official nonattainment designation, the parallel trends assumption, crucial to causal inference in the DID context, is violated.

Two-way fixed effects and differences-in-differences estimators with several treatments

We study two-way-fixed-effects regressions (TWFE) with several treatment variables. Under a parallel trends assumption, we show that the coefficient on each treatment identifies a weighted sum of that treatment’s effect, with possibly negative weights, plus a weighted sum of the effects of the other treatments. Thus, those estimators are not robust to heterogeneous effects and may be contaminated by other treatments’ effects. We further show that omitting a treatment from the regression can actually reduce the estimator’s bias, unlike what would happen under constant treatment effects. We propose an alternative difference-in-differences estimator, robust to heterogeneous effects and immune to the contamination problem. In the application we consider, the TWFE regression identifies a highly non-convex combination of effects, with large contamination weights, and one of its coefficients significantly differs from our heterogeneity-robust estimator.

Association of state policies regulating short-term limited-duration insurance plans and cancer stage at diagnosis.

1589 Background: In 2018, federal regulations expanded access to short-term limited duration insurance (STLD) plans. STLD plans generally limit coverage of medical services and have no cost-sharing caps, which can delay a cancer diagnosis. Furthermore, uptake of STLD plans can destabilize individual insurance markets, leading to increased prices and subsequent barriers to care for comprehensive insurance plan enrollees. Several states introduced legislation prohibiting and/or limiting STLD plans, but it is unclear whether STLD plans or policies restricting their proliferation have affected cancer stage at diagnosis. Methods: Individuals ages 18-64 years with staged first primary malignancies diagnosed 2016-2019 were identified from the Surveillance, Epidemiology, and End Results (SEER) database. Difference-in-differences (DID) analyses evaluated changes in early-stage (stages 0 [breast and bladder only] through II) diagnoses from pre- (2016-2017) to post- STLD expansion (2019; 2018 was excluded as washout/phase-in period) between states with policies prohibiting and/or limiting STLD plans vs. states with no policies regulating STLD plans. In some states limiting but not prohibiting STLD plans, insurers stopped selling STLD plans; we conducted additional analyses by state STLD plan availability in 2019. Linear probability models utilized cluster-robust standard errors (clustered at state level) and adjusted for fixed state and year effects, sociodemographic covariates, and cancer site. Sensitivity analyses excluded states that expanded and/or substantially changed Medicaid eligibility during the study period. The plausibility of the parallel trends assumption was tested in adjusted models restricted to the pre-policy period (2016-2017) and was satisfied for all analyses. Results: 398,190 individuals were included. In adjusted DID analyses, early-stage diagnoses increased 0.95 percentage points (95% CI = 0.53 to 1.38, p&lt;.001) in states that prohibited vs. did not regulate STLD plans, but there was no relative change in early-stage diagnoses in states with other policies limiting STLD plans (p=.25). Relative to states where STLD plans remained available, there was a nonsignificant increase in early-stage diagnoses in states where legislation stopped the sale of STLD plans, with greater evidence for an increase after excluding states with Medicaid eligibility changes (1.03, 95% CI = 0.71 to 1.35, p&lt;.001). There were no differential changes in early-stage diagnoses by state group in falsification tests limited to individuals &gt;65 years-old, who were Medicare-age-eligible. Conclusions: State legislation prohibiting and/or leading to the cessation of STLD plans was associated with an increased rate of early-stage cancer diagnoses. Alternatively, the proliferation of STLD plans was associated with a differential decreased rate of early-stage cancers.

Estimating the effect of U.S. concealed carry laws on homicide: A replication and sensitivity analysis

In this article, we perform sensible robustness checks and estimation techniques that are broadly applicable to researchers studying the effects of concealed carry laws and apply them to recent work by Moody and Lott (2022). While Moody and Lott claim to have found that shall-issue and permitless-carry laws reduce homicide using data ending in 2014, our event-study analysis demonstrates that their model violates the conditional parallel-trends assumption. Additionally, applying methodology from Broderick et al. (2021), we show Moody and Lott’s results are highly sensitive to the removal of a small number of observations. We examine and reject Moody and Lott’s hypothesis that early- and late-adopting “shall-issue” states experienced different outcomes through sensitivity testing of their early–late threshold and applying the Goodman-Bacon (2021) decomposition. Following De Chaisemartin and d’Haultfoeuille (2020), we show Moody and Lott’s results are biased by heterogeneous treatment effects. Overall, our results highlight the importance of conducting principled validity and sensitivity checks before introducing outlier estimates into the empirical literature.

Centralisation of post-resuscitation care after out-of-hospital cardiac arrest at heart centres - a nationwide follow-up study

Abstract Funding Acknowledgements Type of funding sources: None. Background Recent guidelines recommend centralisation of post-resuscitation care in out-of-hospital cardiac arrest (OHCA) patients in dedicated Cardiac Arrest Centres (CACs) aiming to improve outcome. Purpose To describe the process of implementation of the centralisation policy and impact on 30-day mortality rate. Methods This nationwide study included all adult OHCA patients (&amp;gt;18 years) based on the Danish Cardiac Arrest Registry. Centralisation was introduced in 2009, 2010, and 2012 in four regions (A, B, C, D), and one region has not yet implemented the policy (control). Parallel trends assumption was made and a difference-in-difference model with repeated cross-sections was used to estimate the average treatment effect on the treated patients. The 30-day mortality was compared between patients living in regions that introduced centralisation versus control region without centralisation. Results A total of 20,566 patients with OHCA were identified between 2007 and 2020. The majority were between 50 to 75 years (53.7%) and 67.8% male. Bystander cardiopulmonary resuscitation was performed in 55.3%. Shockable rhythm was present in 34.6% of the patients. The policy of centralisation of OHCA in CACs has been gradually and not yet completely implemented (Figure 1A). Patients with shockable rhythms (72%), and younger patients &amp;lt; 50 years (64%) were more frequently transported to CACs. The 30-day mortality was lower among patients transported to CACs (Figure 1B). Centralisation significantly improved survival by 6% (95% CI: 2%-10%) percentage points compared to the control region after controlling for all known covariates (sex, age, Charlson Comobidity Index, shockable rhythm, and bystander cardiopulmonary resuscitation). Conclusions In this nationwide study, centralisation of post-resuscitation care of OHCA in dedicated CACs has been gradually implemented over time. Although, OHCA patients with favorable prognostic factors were more frequently transported to CACs, centralisation increased survival by 6 absolute percentage points.

Identifying and estimating effects of sustained interventions under parallel trends assumptions.

Many research questions in public health and medicine concern sustained interventions in populations defined by substantive priorities. Existing methods to answer such questions typically require a measured covariate set sufficient to control confounding, which can be questionable in observational studies. Differences-in-differences rely instead on the parallel trends assumption, allowing for some types of time-invariant unmeasured confounding. However, most existing difference-in-differences implementations are limited to point treatments in restricted subpopulations. We derive identification results for population effects of sustained treatments under parallel trends assumptions. In particular, in settings where all individuals begin follow-up with exposure status consistent with the treatment plan of interest but may deviate at later times, a version of Robins' g-formula identifies the intervention-specific mean under stable unit treatment value assumption, positivity, and parallel trends. We develop consistent asymptotically normal estimators based on inverse-probability weighting, outcome regression, and a double robust estimator based on targeted maximum likelihood. Simulation studies confirm theoretical results and support the use of the proposed estimators at realistic sample sizes. As an example, the methods are used to estimate the effect of a hypothetical federal stay-at-home order on all-cause mortality during the COVID-19 pandemic in spring 2020 in the UnitedStates.

Tax authority enforcement and stock price crash risk: Evidence from China

Using the merger of State Tax Bureau and Local Tax Bureaus in China as an exogenous shock, our difference-in-differences estimate shows that an increase in tax authority enforcement significantly decreases stock price crash risk. This relation is robust to a variety of robustness checks, including the parallel-trend assumption, other contemporaneous shocks and industry-level and province-level policies. Furthermore, the active role of tax authority enforcement in lowering stock price crash risk is attenuated when firms have better governance mechanisms and greater tax compliance. Our mechanism tests justify that the reform increases information disclosure quality and reduces the hoarding of bad news.

A More Credible Approach to Parallel Trends

Abstract This paper proposes tools for robust inference in difference-in-differences and event-study designs where the parallel trends assumption may be violated. Instead of requiring that parallel trends holds exactly, we impose restrictions on how different the post-treatment violations of parallel trends can be from the pre-treatment differences in trends (“pre-trends”). The causal parameter of interest is partially identified under these restrictions. We introduce two approaches that guarantee uniformly valid inference under the imposed restrictions, and we derive novel results showing that they have desirable power properties in our context. We illustrate how economic knowledge can inform the restrictions on the possible violations of parallel trends in two economic applications. We also highlight how our approach can be used to conduct sensitivity analyses showing what causal conclusions can be drawn under various restrictions on the possible violations of the parallel trends assumption.

Fighting against hunger: A country-wide intervention and its impact on birth outcomes

In 2013 the Mexican government launched Sin Hambre (SH), a large-scale food assistance policy, with which to fight hunger, and which the government expressly intended to improve nutrition and access to food. This study applies a difference-in-differences approach using timing and regional variations in exposure to evaluate the impact of SH, particularly on average birthweight. Because municipalities were not selected randomly, the approach employs a method proposed by Imai et al. (2021) that improves the confidence to satisfy the parallel trend assumption under staggered adoption. Findings reveal that exposure to SH leads to, at most, moderate impacts on birthweight across various specifications.

Managers’ rank & file employee coordination costs and real activities manipulation

We investigate the effect of managers' rank & file employee coordination costs on real activities manipulation (RAM). We identify exogenous variation in managers' rank & file employee coordination costs using the adoption of 99 wrongful dismissal laws across 47 U.S. states between 1970 and 1999. We first find that RAM declines when managers' rank & file employee coordination costs increase. We further document that managers’ rank & file employee coordination costs reduce decentralized RAM but affect neither centralized RAM nor centralized accrual-based earnings management, both of which are less likely to require rank & file employee coordination. Event-time tests corroborate and document the validity of the parallel trends assumption in our setting. Cross-sectional tests document predictable variation in our main result across firms and over time. Consistent with rank & file employee coordination costs constraining RAM, managers miss earnings thresholds more often after rank & file employee coordination costs increase. Our nuanced inferences inform the literature investigating factors that constrain RAM and the extent to which rank & file employees determine firm outcomes.

When Is Parallel Trends Sensitive to Functional Form?

This paper assesses when the validity of difference‐in‐differences depends on functional form. We provide a novel characterization: the parallel trends assumption holds under all strictly monotonic transformations of the outcome if and only if a stronger “parallel trends”‐type condition holds for the cumulative distribution function of untreated potential outcomes. This condition for parallel trends to be insensitive to functional form is satisfied if and essentially only if the population can be partitioned into a subgroup for which treatment is effectively randomly assigned and a remaining subgroup for which the distribution of untreated potential outcomes is stable over time. These conditions have testable implications, and we introduce falsification tests for the null that parallel trends is insensitive to functional form.

The Nonimpact of Opportunity Zones on Home and Business Lending

Opportunity Zones (OZs) promised to stimulate investment in over 8,500 “distressed” neighborhoods. Have OZs increased neighborhood investment and, if so, what types of neighborhoods have benefitted? This study uses a difference-in-differences (DID) design to compare small business and residential lending outcomes in designated OZs with areas that were eligible but not designated. Census tracts are stratified by pretreatment social and economic indicators of distress to search for heterogeneity in effects by neighborhood type. An event study framework is used to check the parallel trends assumption and census tracts are then matched to improve the counterfactual. Finally, an adjusted interrupted time series (AITS) analysis is introduced to further evaluate differences in outcome indicator levels and trends between target and control neighborhoods pre and post OZ. DID and AITS estimates suggest that OZs have had no statistically significant effects on business or residential loan growth. Heterogeneity modeling confirms a noneffect across neighborhood distress type. In conclusion, study limitations and future extensions for both policy and research are discussed.

Why do firms purchase directors and officers liability insurance? – a perspective from short selling threats

This study examines the role of increased short selling threats in firms’ directors and officers liability insurance (D&O insurance) purchase decision against the backdrop of the Chinese deregulation of short sale. Using a difference-in-differences (DiD) research design, we demonstrate a positive effect of short selling threats on firms’ likelihood of purchasing D&O insurance policies after controlling for the known determinants of D&O insurance. We then perform tests to validate the DiD analysis result, including a test of the parallel trend assumption and placebo tests. To shed light on the mechanism through which the effect of short selling takes place, we perform a path analysis. The results reveal that firms’ litigation risk explains the effect of short selling pressure on D&O insurance purchase decision. Further cross-sectional analyses show that the positive effect of short selling threats on D&O insurance is ameliorated when firms have strong internal control or have great analyst coverage.

Is the Centralization of Complex Surgical Procedures an Unintended Spillover Effect of Global Capitation? - Insights from the Maryland Global Budget Revenue Program.

To determine if global budget revenue (GBR) models incent the centralization of complex surgical care. In 2014, Maryland initiated a statewide GBR model. While prior research has shown improvements in cost and outcomes for surgical care post-GBR implementation, the mechanism remains unclear. Utilizing state inpatient databases, we compared the proportion of adults undergoing elective complex surgeries (gastrectomy, pneumonectomy/lobectomy, proctectomies, and hip/knee revision) at high-concentration hospitals (HCHs) in Maryland and control states. Annual concentration, per procedure, was defined as hospital volume divided by state volume. HCHs were defined as hospitals with a concentration at least at the 75 th percentile in 2010. We estimated the difference-in-differences (DiD) of the probability of patients undergoing surgery at HCHs before and after GBR implementation. Our sample included 122,882 surgeries. Following GBR implementation, all procedures were increasingly performed at HCHs in Maryland. States satisfied the parallel trends assumption for the centralization of gastrectomy and pneumonectomy/lobectomy. Post-GBR, patients were more likely to undergo gastrectomy (DiD: 5.5 p.p., 95% CI [2.2, 8.8]) and pneumonectomy/lobectomy (DiD: 12.4 p.p., 95% CI [10.0, 14.8]) at an HCH in Maryland compared with control states. For our hip/knee revision analyses, we assumed persistent counterfactuals and noted a positive DiD post-GBR implementation (DiD: 4.8 p.p., 95% CI [1.3, 8.2]). No conclusion could be drawn for proctectomy due to different pre-GBR trends. GBR implementation is associated with increased centralization for certain complex surgeries. Future research is needed to explore the impact of centralization on patient experience and access.

Cancer Mortality and the 2014 Medicaid Expansions under the Affordable Care Act

<h3>Purpose/Objective(s)</h3> Prior data have linked Medicaid expansions under the Affordable Care Act with improved insurance coverage among patients with cancer, earlier stages at diagnosis, and prolonged survival. Decreased cancer mortality rates were observed in connection with the limited early Medicaid expansions in 2010-2011, but changes associated with the more comprehensive and more recent Medicaid expansions starting in 2014 are not clear. We sought to test our hypothesis that Medicaid expansion was associated with decreased cancer mortality rates. <h3>Materials/Methods</h3> Data for cancer deaths among 20–64-year-old individuals from 2004-2018 were obtained from the National Center for Health Statistics. We obtained annual state-level cancer mortality rates covering all 50 U.S. states for deaths for all cancers combined and for deaths due to breast, cervical, colorectal, liver, lung, pancreatic, and prostate cancers, stratified by age group. State-level demographic data summarizing, age, employment, education, poverty, race/ethnicity, insurance, smoking, and obesity were also obtained and linked with the mortality data. We implemented a difference-in-differences analysis utilizing generalized estimating equations with state-level clustering to compare changes in mortality rates after 2014 between Medicaid expansion and non-expansion states. Models were adjusted for age, year, early (2010-2011) Medicaid expansion effects, and the aforementioned state-level demographic covariates. The common or parallel trends assumption was evaluated by examining for differential temporal trends in mortality rates between state groups and was satisfied for all analyses. Sensitivity analyses with interrupted time series analyses comparing trends in state groups were performed. <h3>Results</h3> Mortality rates for all cancers combined decreased from 92.2 to 85.5 per 100,000 in Medicaid expansion states compared to 98.4 to 93.5 per 100,000 in non-expansion states after 2014. In adjusted difference-in-differences analyses for all cancers combined, cancer mortality rates decreased after Medicaid expansion in expansion relative to non-expansion states (OR: 0.982, 95% CI = 0.965-0.999, p=.038), translating to 5174 cancer deaths that were averted in Medicaid expansion states in the post-expansion period. Expansion-associated reductions in mortality were also observed for breast (OR: 0.970, 95% CI = 0.947-0.993, p = 0.012), liver (OR: 0.939, 95% CI = 0.894-0.986, p = 0.011), and pancreatic cancers (OR: 0.968, 95% CI = 0.942-0.995, p = 0.019). Sensitivity analyses yielded similar results. <h3>Conclusion</h3> These data are the first to show statistically significant associations between state Medicaid expansions after 2014 and decreased overall cancer mortality rates. We observed expansion-associated reductions for all cancers combined and for breast, liver, and pancreatic cancers. Further advocacy efforts focusing on the 12 states that have not yet adopted Medicaid expansion may further reduce cancer mortality.

The European Union's withdrawal of trade preferences for Cambodia

AbstractFollowing Cambodia's violation of human and labour rights, the European Union (EU) decided in February 2020 to withdraw trade preferences for the country temporarily from August 2020. This paper estimates the short‐run impact of preference withdrawal on EU imports from Cambodia. As the EU withdrew preferential tariffs only for certain products and maintained duty‐free quota‐free access for others, I adopt a difference‐in‐differences regression framework and provide supporting evidence for the parallel trend assumption. The results show that preference withdrawal has a significantly positive impact on EU import of affected goods from Cambodia in July 2020, consistent with a last‐minute shipment effect to avoid tariff increases. Subsequently, tariff increases have a significantly negative impact on the EU import of affected goods. These effects are heterogeneous across knitted garments, woven garments, and footwear products.

National poverty alleviation strategy and policy regulation exemption: A quasi-natural experiment based on the green credit guidelines

This paper evaluates the real effects of the environmental policy at the firm level. Using the promulgation of the “Green Credit Guidelines" and national poverty alleviation strategy as a quasi-natural experiment, we find that:1) the Guidelines significantly inhibit the scale and maturity of debt financing for polluting companies; 2) the participation of enterprises in poverty alleviation efforts can effectively offset the restraining influence of the Guidelines. Our results are robust for parallel trend assumption and propensity score matching estimation. Further, we find that state-owned enterprises have decreased more significantly. These results indicate that the national poverty alleviation strategy could play a role in debt financing for polluting companies. Thus providing timely implications for regulators concerned with environmental protection.

Aggressive policing and undermined legitimacy: assessing the impact of police stops at gunpoint on perceptions of police in São Paulo, Brazil

ObjectivesTest the effects of a recent police stop and a recent police stop at gunpoint on changes in attitudes towards the police among residents of Brazil’s biggest city.MethodsA three-wave longitudinal survey of São Paulo residents (2015–2019) measured people’s beliefs about police legitimacy, expectations of police procedural fairness, effectiveness, and overpolicing, whether they were recently stopped by the police, and whether officers had pointed a gun at them during that stop. A novel causal estimand focused on the effect of change in treatment status is estimated using matching methods for panel data combined with difference-in-differences.ResultsWhile estimates are too imprecise to suggest an effect of a recent police stop on attitudinal change, recent police stops at gunpoint decrease public expectations of procedural fairness, increase expectations of overpolicing, and harm public beliefs of police legitimacy.ConclusionsUnder a credible conditional parallel trends assumption, this study provides causal evidence on the relationship between aggressive policing practices and legal attitudes, with implications to public recognition of legal authority in a major Global South city.