No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
How Much Should We Trust Differences-In-Differences Estimates?
Abstract
Most papers that employ Differences-in-Differences estimation (DD) use many years of data and focus on serially correlated outcomes but ignore that the resulting standard errors are inconsistent. To illustrate the severity of this issue, we randomly generate placebo laws in state-level data on female wages from the Current Population Survey. For each law, we use OLS to compute the DD estimate of its "effect" as well as the standard error of this estimate. These conventional DD standard errors severely understate the standard deviation of the estimators: we find an "effect" significant at the 5 percent level for up to 45 percent of the placebo interventions. We use Monte Carlo simulations to investigate how well existing methods help solve this problem. Econometric corrections that place a specific parametric form on the time-series process do not perform well. Bootstrap (taking into account the autocorrelation of the data) works well when the number of states is large enough. Two corrections based on asymptotic approximation of the variance-covariance matrix work well for moderate numbers of states and one correction that collapses the time series information into a "pre"-and "post"-period and explicitly takes into account the effective sample size works well even for small numbers of states.
... As the name implies, the DID methodology is based on a double subtraction: one occurs in the difference between the averages of the outcome variable for both groups between the periods preceding and following the event, and the other occurs in the difference of the first difference calculated between groups (Bertrand et al., 2004). This measurement lets us determine which group has the greatest difference during the analysis period. ...
... Equation 2 Bertrand et al. (2004) state that the DID model's estimations are derived from the ordinary least squares (OLS) method and their associated standard errors. To select a group of companies with characteristics similar to those that have changed due to this regulation, Funchal and Monte-Mor (2016) advised using the Propensity Score Matching (PSM) technique. ...
Purpose
Board members are critical in resolving agency conflicts. How- ever, many are unable to perform their function due to their distance, as they are not present at board meetings. As of Instruction no. 561, the Brazilian Securities and Exchange Commission (CVM) regulated remote voting for Boards of Directors, allowing for greater attendance at meet- ings and, as a result, increased involvement. In this context, this research examines the effect of remote voting by Boards of Directors on execu- tive compensation and financial performance of publicly traded firms in Brazil.
Originality/value
This research is innovative in the sense that it exam- ines the effect of the Board of Directors remote voting on the CEO com- pensation and financial performance of the firm, using an innovative methodology.
Design/methodology/approach
We applied a quasi-experimental method (difference-in-differences) to assess the effects of a given group (treat- ment) before and after the event, significantly reducing endogeneity when considering an exogenous shock to the system.
Findings
As a result, the estimation of the main model reveals statisti- cally significant differences between the effects of treatment and control on profitability and executive remuneration, indicating that remote vot- ing mitigated agency problems by generating a substitution effect for explicit incentives (as evidenced by the decrease in executive remunera- tion) and by providing greater accounting performance for companies.
Keywords:
financial performance; executive compensation; differ- ence-in-differences; remote voting; board of directors
... To formally characterize the baseline specification for the causal analysis of the Municipal Advisor Rule in a standard difference-in-differences (Bertrand et al., 2004) equation, consider: y b,i,a,t = α + β 0 * P ost t × N ego. b,i,a + β 1 * P ost t + β 2 * N ego. ...
... Importantly, before the SEC Rule, I find that the treated and control groups tend to follow nearly parallel trends with issuer fixed effects. This lends useful support to the main identification assumption (Bertrand et al., 2004) that the treated group would follow the control group in the absence of the regulation. From July 2014 onward, I find that the offering yields for the treated bonds tend to decrease in comparison to the control bonds. ...
I examine whether the imposition of fiduciary duty on municipal advisors affects bond yields and advising fees. Using a difference-in-differences analysis, I show that bond yields reduce by $\sim$9\% after the imposition of the SEC Municipal Advisor Rule due to lower underwriting spreads. Larger municipalities are more likely to recruit advisors after the rule is effective and experience a greater reduction in yields. However, smaller issuers do not experience a reduction in offering yields after the SEC Rule. Instead, their borrowing cost increases if their primary advisor exits the market. Using novel hand-collected data, I find that the average advising fees paid by issuers does not increase after the regulation. Overall, my results suggest that while fiduciary duty may mitigate the principal-agent problem between some issuers and advisors, there is heterogeneity among issuers.
... Difference-in-Difference (DID) regression analysis is most applied for evaluating the effect of treatment (Kellogg & Wolff, 2008;Kotchen & Grant, 2011), typically instigated by a new policy (Bertrand et al., 2004). Two key data were designed, 'Treated' (region) and 'Time' (before and after policy implementation), that is, whether a region implements DST. ...
(Double) Daylight Saving Time (DST/DDST) synchronizes human daily activities with sunlight hours, aiming to save energy. China implemented DST/DDST in 1986 and annulled it in 1991, which has not since been observed. This study aims to revisit the household lighting energy-saving potentials of DST/DDST implementation in six representative highly urbanized Chinese cities across different latitudes and within different climate zones, based on the electricity consumption behaviour of Chinese people in the 2020s. We constructed country-specific Multiple Linear Regression (MLR) analysis models interrelated with the local characteristic features of the DST/DDST in China, including latitude, age structure, industrial structure, get-up and bedtime schedules. The results showed the following: (1) Both DST and DDST can effectively reduce household lighting consumption in all the studied case cities, while DDST has a superior energy-saving potential; (2) Considering China-specific lifetime schedules, April to October is the most recommended duration for DST/DDST implementation; (3) Northern cities in China exhibit greater potential for energy savings in lighting than southern regions. (4) Cities with larger portions of primary industry have more household lighting-saving potential in China. Based on the obtained results, monthly data and separate statistical data on household lighting could help promote the precision of regression models. ARTICLE HISTORY
... Recent studies on the effects of tourism on housing prices have employed the DID methodology (Sommerville et al., 2020; Barron et al., 2021;Franco & Santos, 2021), yielding promising results. However, econometricians have criticised the methodology, arguing that the fixed effects OLS estimator underestimates standard deviations due to serial correlation (Bertrand et al., 2004), as seen in the previous section. Parks (1967) had already developed a Feasible Generalised Least Squares (FGLS) estimator that allowed for serial correlation in seemingly unrelated regression equation systems (Zellner, 1962). ...
This paper aims to help improve the estimations produced by researchers who rely on conventional housing market pricing models to determine housing prices. The widespread use of panel data in estimating housing prices is justified by the richness of cross-sectional regional or metropolitan data analysed over several periods. Unfortunately, panel data has slope coefficient heterogeneity and cross-sectional dependence, producing inconsistent and misleading estimates of the coefficients using the Ordinary Least Squares (OLS) estimator. Recent advances in econometrics address these panel data limitations, producing better estimates. We analysed the empirical application of these new estimators on housing market panel data, showing that the Fully Modified Ordinary Least Squares Augmented Mean Group (FMOLS-MG) estimator produces the best estimates of the long-term housing market equilibrium and that the Dynamic Common Correlated Effects Mean Group (DCCE-MG) estimator produces the best estimates of the housing market's short-term dynamics. Adopting a trending methodology like Difference-in-Differences (DID) in housing market research to explain the effects of policy decisions on housing prices also has complications related to using the OLS estimator with fixed effects when the data has serial correlation. We show these problems can be overcome using the Feasible Generalised Least Squares estimator in a Seemingly Unrelated Regression Equations (FGLS-SURE) system. Recent econometric developments produce more accurate housing price determinant estimates than conventional econometric methods. These new methodologies can help researchers better estimate the effects of fundamental economic changes and policy decisions on housing prices, which can, in turn, support policymakers in implementing better housing policies.
... There is no statistical test for the parallel trend assumption. A visual inspection is the best way for verifying this assumption (Bertrand et al., 2004;Hill et al., 2018). Fig. 1 displays the visual inspection of the parallel trend of the dependent variables: 1) total deposits to total assets and 2) interbank deposits to total assets. ...
... Standard errors in a difference-in-difference regression are likely serially correlated across time, while the law change is persistent. Standard errors that do not account for this will overstate the precision so, following Bertrand, Duflo & Mullainathan (2004), we adjust by clustering the standard errors by state. ...
We estimate the effect of state-level policies enacting universal free full-day kindergarten on mothers’ labor supply using a life-cycle analysis. Similar to previous research on childcare and labor supply, we find that free full-day kindergarten increases labor force participation rates for mothers whose youngest child is kindergarten-aged by 4.3 to 7.1 percentage points. We find that for mothers whose youngest child is an infant, labor force participation increases by 7.2 to 9.8 percentage points, and for women whose youngest child is 3 to 4 years old labor force participation increases by 5.9 to 7.9 percentage points. The fact that the policies impact the labor supply for mothers of younger-than-kindergarten-age children by even more than for mothers of kindergarten-aged children is important for understanding the full effect of subsidized childcare. This is consistent with a life-cycle model of labor supply where wages and prices in future periods impact mothers’ labor force attachment.
... Finally, is the error term. We estimate both equations using a linear model clustering the error term at the state level (s) to address the non-independence of observations from the same area over time (Bertrand et al. 2004). In all models, we align policy-effective dates with the relevant month of the interview. ...
We provide new evidence on the economic and health impacts of government- mandated non-pharmaceutical interventions (NPIs) during the COVID-19 pandemic. Apart from labor force participation, unemployment, and hours worked, we provide novel results on work absence due to illness. We also examine the heterogeneity of these results by demographic and employment groups. We use recent innovations in the difference-in-differences methodology to capture the dynamic effects of these orders that were staggered in nature. Our findings show that states’ social distancing measures increased unemployment and lowered labor market participation and hours worked. The adverse labor market effects were more pronounced for single parents and those working non-teleworkable jobs. We find some evidence that workers’ health improved as absence from work due to illness significantly decreased, suggesting that NPIs protected many vulnerable workers.
... Lastly, to investigate the diverse characteristics of our sample and explore the mechanisms potentially influencing our principal findings, we use a triple interaction model (T reat i ×Exp t ×θ i ) where θ i denotes individual motivations and preferences to buy OPs, awareness of climate issues, confidence in scientific evidence, as well as the correct answer to the control question verifying their understanding of the information presented. In all specifications, we cluster standard errors at the individual (treatment) level (Bertrand et al., 2004). ...
... First, bootstrap samples are constructed to keep the covariate distribution in each cluster identical to that of the original sample. Bertrand et al. (2004) show that bootstrap inference can become unreliable otherwise. Second, the bootstrap samples are constructed by imposing the null hypothesis. ...
Home sharing platforms have experienced a rapid growth over the last decade. Following negative publicity, many cities have started regulating the short-term rental market. Regulations often involve a cap on the number of days a property can be rented out on a short-term basis. We draw on rich data for short-term rentals on Airbnb and for the long-term rental market to examine the impact of short-term rental regulations with a day cap on various stakeholders: hosts, guests, the platform provider, and residents. Based on a difference-in-differences design, we document a sizable drop in Airbnb activity. Interestingly, not only targeted hosts (i.e., hosts with reservation days larger than the day cap), but also non-targeted hosts reduce their Airbnb activity. The reservation days of non-targeted hosts decrease between 26.27% and 51.89% depending on the treatment. Targeted hosts experience a similar decline. There is, nevertheless, significant non-compliance: more than one third of hosts do not comply with enacted short-term rental regulations. Additional analyses show that few properties are redirected from short-term rental to long-term rental use and that there is no significant drop in long-term rents. Drawing on a theoretical model, we tie the estimated effects to changes in stakeholders’ welfare: Regulations significantly reduce the welfare of hosts, and the loss ranges between 46.30% and 9.02%. The welfare loss of the platform provider is proportional to the loss of the hosts. Welfare of guests decreases moderately ranging between 4.5% to 4.1%. The welfare of residents increases minimal. These results question the effectiveness and desirability of the studied short-term rental regulations.
... This paper conducts an empirical investigation into the ramifications of e-commerce advancement through the utilization of the double-difference method. Given the batch-wise establishment of pilot cities, this study derives inspiration from literature (Bertrand et al., 2004;Li & Wang, 2022) and constructs the ensuing multi-temporal DID model: ...
Amid escalating global concerns over environmental sustainability and carbon neutrality, the role of technological innovation, particularly e-commerce transformation, in fostering energy conservation and emission reduction within urban landscapes has garnered substantial attention. This study leverages a quasi-natural experiment involving the establishment of national e-commerce demonstration cities to explore the implications of urban e-commerce transformation for energy saving and emission curtailment. Utilizing a multi-temporal double-difference (DID) method, our findings elucidate a significant negative correlation between e-commerce transformation and carbon emission intensity, with emission reductions manifesting from the second year of policy implementation onward. The research highlights the differential impacts of e-commerce transformation across various city types, with pronounced benefits observed in provincial capitals, non-resource-dependent cities, and those with higher levels of digital economic development and innovation capacity. Mechanistic analysis reveals that the primary channel through which e-commerce transformation facilitates emission reduction is through a decrease in energy intensity, serving as a critical mediator in the process. This study contributes to the discourse on the knowledge economy by underscoring the potential of digital innovation and e-commerce as pivotal drivers of environmental sustainability and economic transformation. It calls for policies that further harness the power of digital transformation to achieve broader societal goals of energy conservation and emission reduction, thereby advancing the knowledge economy towards sustainable development.
... The coefficient β on the post-tax dummy captures the treatment effect of the environmental pollution tax on ecological efficiency. It is noteworthy that to overcome the possible temporal correlation of random disturbance terms and reduce the risk of underestimating the standard error, all regressions in this paper adopt a robust standard error clustered at the city level [20]. ...
... In this case, agricultural insurance can be re-characterized objectively. DID (Difference in Difference) method can be employed to evaluate the causal effects of agricultural insurance on green pesticides (Bertrand et al., 2004). In Table 4, significant positive effects of agricultural insurance on the adoption of green pesticides can also be captured. ...
Purpose-Numerical literature shows that agricultural insurance can affect pesticide investments, but few of them are devoted to explain how agricultural insurance affects farmers' selection on green or traditional pesticides. This paper aims to develop a theoretical model about how agricultural insurance influences on green pesticides selections and tests our conclusions by using the data from China land economic survey (CLES) from 2020 to 2021. Design/methodology/approach-We employ probit model to capture the effects of agricultural insurance on green pesticides adoption. Findings-We indicate that green pesticides have a stronger effect on stabilizing yield and increasing income than traditional pesticides, but there are still risks disturbing farmers' decisions on green pesticides usage. By providing premium subsidies after the farmers are affected by natural risk, agricultural insurance improves the farmers' expected income and encourages farmers to use green pesticides. Further, we further confirm these conclusions by considering different scenarios such as climate risks, farmers' entrepreneurship and credit constraints. We find that the effects are more salient if croplands are under higher natural risks and, farmers are equipped with entrepreneurship and formal credit. This paper implies that the agricultural insurance decoupled with green technologies also have salient positive effects on agricultural pollution control. Originality/value-The potential contributions of this paper can be outlined in three aspects in detail. Firstly, this paper aims to revel the effects of agricultural insurance on pesticide selection by structuring a general theoretical model. By using the CLES data from 2020 to 2021, we confirm that agricultural insurance increases the probability for adopting green pesticides. Secondly, this paper discusses the effects of farmers' characteristics on the results and finds that if farmers have entrepreneurship, the effects of agricultural insurance on green pesticide usage will be more salient. Thirdly, it uncovers some practices in China, which will supply experiences for other developing countries. For example, this paper further demonstrates that "insurance þ credit" plan the present Chinese government carried out will be an important measure for strengthening effects of agricultural insurance on green pesticides usage. Moreover, it shows that decouple agricultural policies will also guide farmers to use green technologies eventually if the technologies are reliable and farmers can afford.
... DID estimation is a commonly used experimental design technique in the Econometrics literature to assess causal relationships [3,18]. The key idea behind the DID experimental design is to measure the change in a response variable before and after an intervention for both the affected and unaffected groups [8]. The difference of these changes is the treatment effect. ...
Product assortment selection is a critical challenge facing physical retailers. Effectively aligning inventory with the preferences of shoppers can increase sales and decrease out-of-stocks. However, in real-world settings the problem is challenging due to the combinatorial explosion of product assortment possibilities. Consumer preferences are typically heterogeneous across space and time, making inventory-preference alignment challenging. Additionally, existing strategies rely on syndicated data, which tends to be aggregated, low resolution, and suffer from high latency. To solve these challenges we introduce a real-time recommendation system, which we call \ours. Our system utilizes recent advances in 3D computer vision for perception and automatic, fine grained sales estimation. These perceptual components run on the edge of the network and facilitate real-time reward signals. Additionally, we develop a Bayesian payoff model to account for noisy estimates from 3D LIDAR data. We rely on spatial clustering to allow the system to adapt to heterogeneous consumer preferences, and a graph-based candidate generation algorithm to address the combinatorial search problem. We test our system in real-world stores across two, 6-8 week A/B tests with beverage products and demonstrate a 35% and 27\% increase in sales respectively. Finally, we monitor the deployed system for a period of 28 weeks with an observational study and show a 9.4\% increase in sales.
... DID is an econometric method used to evaluate the effectiveness of policies or interventions. This method estimates the causal effect of a policy by comparing changes before and after the policy is implemented (differences in time) and changes between a group that implements the policy and a control group that does not implement the policy (differences between groups) (Bertrand et al., 2004;Ma et al., 2023b). The principle of DID is to calculate the changes in the experimental group before and after the policy is implemented (the first difference), then calculate the changes in the control group during the same period (the second difference), and finally calculate the difference between the two differences (i.e., the experimental group's change minus the change in the control group). ...
Introduction
There is a strong link between consumer behavior and healthy food consumption. However, how to narrow the gap between consumption intentions and actual healthy food consumption is still under discussion.
Methods
This study takes Chinese families as the research object, based on the family participation in long-term care insurance (LTCI) policy, and constructs an analytical framework including healthy eating behavior, food consumption, and insurance system to discuss how to narrow the gap between consumption intention and actual healthy food consumption. In addition, the intermediary role played by the risk prevention mechanism is also analyzed. Based on data from the China Health and Nutrition Survey (CHNS), this study uses a difference-in-differences analysis framework to empirically examine the impact of changes in consumption behavior on healthy food consumption after households participate in the LTCI pilot.
Results and discussion
The research results show that implementing LTCI can increase the frequency of healthy meal preparation methods by 0.045 units and the frequency of not including processed foods in the meals of households participating in the policy by 0.033 units compared with households that do not participate. The daily meal quantity is increased by 0.198 (converted to 1.219 grams), and 0.198 units increase the healthy food consumption structure. This conclusion holds under a series of robustness tests. Mechanism test shows that LTCI affects healthy food consumption through risk prevention mechanisms. The impact of the LTCI policy will also not be affected by similar competitive policies. The heterogeneity test further proves that LTCI policies are more likely to increase healthy food consumption among urban households, larger households, and households employed in private enterprises. Based on these findings, it is recommended that families participate in LTCI to reduce the financial stress faced by families due to illness and care needs while increasing the demand for and consumption of healthy foods. The findings also provide a valuable reference for current policy formulation on improving family dietary quality in China.
... DID is an econometric method used to evaluate the effectiveness of policies or interventions. This method estimates the causal effect of a policy by comparing changes before and after the policy is implemented (differences in time) and changes between a group that implements the policy and a control group that does not implement the policy (differences between groups) (Bertrand et al., 2004;Ma et al., 2023b). The principle of DID is to calculate the changes in the experimental group before and after the policy is implemented (the first difference), then calculate the changes in the control group during the same period (the second difference), and finally calculate the difference between the two differences (i.e., the experimental group's change minus the change in the control group). ...
... To avoid such a problem, we cluster the standard errors by industry. The resulting estimator of the variance-covariance matrix will be consistent in the presence of any correlation pattern within industries over time (Bertrand et al., 2004). ...
The liberalization initiative commenced in India from 1991 onwards, replacing the four-decade long import substitution policy. The primary objective was to enhance the role of foreign and private investment, in line with the newly embraced outward-oriented growth model. The government had undertaken several policy initiatives since then, especially to strengthen the manufacturing sector which plays an important role in the economic development of any country. The current study evaluates the effects of the liberalization policy in India on industrial outcomes. Recent studies have found that when firm heterogeneity is present in trade models, reforms will lead to a decrease in the number of firms and a rise in their average size (Melitz, 2003). A dataset of 24 manufacturing industries had been used in the current study. We test empirically whether liberalization had led to a rise in the average size of establishments as stated in the literature. We also attempt to analyze the magnitude of trade costs in terms of the impact of reforms on wages and prices. The empirical analysis based on the difference-indifference (DID) estimation method shows that on average, trade reforms do not lead to an increase in the real wages and average size of establishments. In addition, prices appear to increase in the long run due to liberalization , with potential ramifications.
... We employed the placebo method to estimate the standard errors of the treatment effect estimates, accounting for potential heteroskedasticity in the regression model. The placebo method was chosen over alternatives such as bootstrapping or jackknifing due to its ability to provide robust inference in the presence of serial correlation within units, which is a common challenge in panel data settings [31]. This approach helps ensure that our inference is valid despite the potential for within-unit correlations over time [24,25]. ...
Background
Inadequate financing constrains primary healthcare (PHC) capacity in many low- and middle-income countries, particularly in rural areas. This study evaluates an innovative PHC financing reform in rural China that aimed to improve access to healthcare services through supply-side integration and the establishment of a designated PHC fund.
Methods
We employed a quasi-experimental synthetic difference-in-differences (SDID) approach to analyze county-level panel data from Chongqing Province, China, spanning from 2009 to 2018. The study compared the impact of the reform on PHC access and per capita health expenditures in Pengshui County with 37 other control counties (districts). We assessed the reform’s impact on two key outcomes: the share of outpatient visits at PHC facilities and per capita total PHC expenditure.
Results
The reform led to a significant increase in the share of outpatient visits at PHC facilities (14.92% points; 95% CI: 6.59–23.24) and an increase in per capita total PHC expenditure (87.30 CNY; 95% CI: 3.71-170.88) in Pengshui County compared to the synthetic control. These effects were robust across alternative model specifications and increased in magnitude over time, highlighting the effectiveness of the integrated financing model in enhancing PHC capacity and access in rural China.
Conclusions
This research presents compelling evidence demonstrating that horizontal integration in PHC financing significantly improved utilization and resource allocation in rural primary care settings in China. This reform serves as a pivotal model for resource-limited environments, demonstrating how supply-side financing integration can bolster PHC and facilitate progress toward universal health coverage. The findings underscore the importance of sustainable financing mechanisms and the need for policy commitment to achieve equitable healthcare access.
... To examine whether the above observed DIDs between the treatment and the two control groups is statistically significant, drawing on prior research (Bertrand, Duflo, and Mullainathan 2004), we trained an Ordinary Least-Squares (OLS) Regression Model on the MHE scores of the pre-OH and post-OH messages associated with each real or placebo OH message in the three groups. Each MHE score was associated with three variables: g (whether it corresponded to the treatment or one of the control groups), t (whether it corresponded to the pre-OH or post-OH period), and categorical variables based on the OH message ID that were used as dummy variables. ...
Online harassment negatively impacts mental health, with victims expressing increased concerns such as depression, anxiety, and even increased risk of suicide, especially among youth and young adults. Yet, research has mainly focused on building automated systems to detect harassment incidents based on publicly available social media trace data, overlooking the impact of these negative events on the victims, especially in private channels of communication. Looking to close this gap, we examine a large dataset of private message conversations from Instagram shared and annotated by youth aged 13-21. We apply trained classifiers from online mental health to analyze the impact of online harassment on indicators pertinent to mental health expressions. Through a robust causal inference design involving a difference-in-differences analysis, we show that harassment results in greater expression of mental health concerns in victims up to 14 days following the incidents, while controlling for time, seasonality, and topic of conversation. Our study provides new benchmarks to quantify how victims perceive online harassment in the immediate aftermath of when it occurs. We make social justice-centered design recommendations to support harassment victims in private networked spaces. We caution that some of the paper's content could be triggering to readers.
... Afterwards, the NDRC, in conjunction with various government departments, announced the second, third, and fourth batches of NEDC in 2011, 2014, and 2017, respectively. Considering the conducive environment for quasi-natural experiments provided by the progressive reforms in China, we adopt the design proposed by Bertrand et al. to formulate the following model [61]: ...
The impact of urban e-commerce transformation on economic resilience can help a country improve its ability to resist risks and seize the initiative in economic development. This study examines the impact of the construction of the National E-commerce Demonstration City (NEDC) on economic resilience using the staggered different-in-differences approach using a sample of 282 Chinese cities from 2006 to 2020. The results show NEDC construction significantly strengthens urban economic resilience. This result remains robust after undergoing placebo test, exclusion of other policies interference, and examining endogeneity. Furthermore, noteworthy heterogeneity exists in the effect of NEDC construction on urban economic resilience, particularly in eastern, developed regions, and cities with high Internet penetration. The mechanisms analysis indicates that NEDC construction enhances urban economic resilience by expanding the scale of urban employment and enhancing market dynamism. Overall, this study refines the causal relationship between e-commerce development and urban economic resilience, providing empirical evidence and policy insights for China and other countries to enhance urban economic resilience and stabilize macroeconomic fluctuations.
... Therefore, the DID method was employed to control the unobserved individual heterogeneity of each sample in the time dimension and eliminate the influence of unobserved and time-varying factors in the cross-section dimension. This approach effectively addresses endogeneity issues (Bertrand et al. 2004). The DID method was used to explore the impact of TPP on rural household income in China. ...
This paper explores how migrant density and diversity influence international franchising through a knowledge‐based perspective. Using data from the World Franchise Network and U.N. immigration statistics from 1993 to 2013, the study finds that higher migrant density positively impacts international franchising activities, suggesting that a substantial presence of migrants facilitates knowledge transfer and network benefits. Conversely, migrant diversity tends to have a negative effect, indicating potential challenges in harmonizing diverse knowledge bases. However, this diversity can also enhance the positive impact of migrant density on franchising, as it enriches the cultural and knowledge pool. These findings suggest that firms can leverage migrant communities for international franchising success by tapping into their networks and expertise, and that immigration policies could be tailored to support this dynamic. Managers should consider engaging with migrants to navigate cultural nuances and market entry barriers in foreign territories.
This article investigates if, and to what extent, banking structural reforms may affect the top income shares over time. Canada and Italy are used as case studies, as both countries undertook a major deregulation and liberalization process within their banking sector in the early 1990s. These banking policies aimed at privatizing the banking sector and reintroducing the quasi universal banking model. The evaluation of these policy packages is undertaken by implementing the Synthetic Control Method. Findings point out, overall, a robust and substantial increase of some of the top income shares in both countries, over the post-deregulation period. This work contributes by also identifying the main potential mechanisms—both direct and indirect—via which banking deregulation might have operated.
Background
Hypertension and diabetes are interconnected highly prevalent chronic conditions in adults particularly among older adults. They contribute to the very high burden of noncommunicable diseases (NCDs) in low- and middle-income countries (LMICs). The awareness, diagnosis, treatment, and control of these conditions are poor and access to quality care for hypertension and diabetes remains limited, particularly in rural areas. Strengthening primary health care systems for NCD care delivery is critical to addressing this rising burden. Digital health interventions for NCD care have shown promising results in pilot studies, but reliable evidence of their benefits remains elusive. Little is known about how digital technology can be utilized to support decentralized primary care to improve accessibility and bridge the gaps in the care continuum in LMICs. In this study, we aim to generate data on the effectiveness and the cost-effectiveness of multicomponent decentralized primary care on hypertension and diabetes care continuum compared with usual care and to digital health intervention alone in rural Bangladesh, and to evaluate factors influence the implementation of the interventions.
Methods
We will implement a type 2 effectiveness-implementation hybrid with a dual focus on testing of effectiveness of a digital technology supported decentralized primary care model and implementation strategies. A three-arm mixed-methods quasi-experimental design will be used to evaluate implementation fidelity, processes, and effectiveness outcomes. The study will be implemented in three subdistricts of Dinajpur district, Rangpur division in northern Bangladesh. Multicomponent, decentralized primary care model will include components of healthcare provider training, digital health, decentralization with task shifting, and community-based care. The key interventions in the multicomponent model comprise expanding the scope of screening, routine monitoring, and dispensing of medication refills from a doctor-managed subdistrict level NCD clinic to nonphysician health worker managed village level community clinics, supported by a digital platform (Simple app) for electronic health records, point-of-care support, referrals when indicated, and routine patient follow-up. The digital health only model includes training and support in subdistrict NCD clinic for incorporating the Simple app. The primary endpoint of the study is changes in the treatment success rates for hypertension and diabetes.
Discussion
Our study is among the first to evaluate the effectiveness and implementation strategy of a decentralized primary care model for integrated hypertension and diabetes management in a LMIC. Using repeated cross-sectional community-based surveys combined with facility-based longitudinal data, our study will provide rich data on clinical and behavioral outcomes, various measures across the care continuum, and implementation processes, including costs. Implementation fidelity and process evaluation will be guided by the UK Medical Research Council guideline on process evaluation of complex intervention, and the WHO’s Noncommunicable Disease Facility-Based Monitoring Guidance, and the RE-AIM framework. We will document the factors that may explain how the interventions influence hypertension and diabetes management and explore barriers and facilitators to delivering and sustaining interventions. The results will have important implications for policy making and programmatic efforts for hypertension and diabetes prevention and management.
Trial registration
ClinicalTrials.gov, NCT06258473. Registered on 06 February 2024.
Low emission zones (LEZ) represent one of the strategies to reduce, or limit, emissions of air pollutants from road traffic within specific spatial areas in cities. This study aims to assess the effectiveness of Madrid LEZ by quantifying the reduction in ambient and background pollution, the latter representative of the pollution to which the population is chronically exposed. This latter fraction was estimated after clustering the analysed time series with Hidden Markov Models. For that purpose, ten monitoring stations were selected, and the behaviour of five key pollutant concentrations was studied through phases of the LEZ implementation in Madrid. The average reduction in the NO2, PM10 and PM2.5 ambient pollution concentration was 12.1 μg·m−3, 5.2 μg·m−3 and 3.4 μg·m−3, respectively. Regarding background pollution, these reductions were 9.6 μg·m−3 for NO2, 0.5 μg·m−3 for PM10, and 1.2 μg·m−3 for PM2.5. Diff-in-Diff analysis compares the changes in outcomes over time between a population enrolled in a program (sites within LEZ zone - the treatment group-) and a population that is not (sites outside the LEZ zone - the control group). This analysis showed that the reduction of NO2 background pollution is less affected by LEZ implementation and that meteorological factors could have slightly influenced the decrease in NO2 concentration. Road traffic contributions were estimated to be reduced by 12.8 μg·m−3 for NO2 and 5.3 μg·m−3 for PM10. The number of times NO2 and PM10 concentrations exceeded the limit values established by the Directive 2008/50/EC was also reduced, but not sufficiently for achieving legal compliance. Even though our results show that LEZ implementation in Madrid has been effective and represented an improvement in the air quality of the city, it still can be considered insufficient. Further, restrictive measures remain to be targeted for its impact on urban health to be more pronounced.
Political and economic tensions, which often jeopardise trade, are rising among the world's major powers, and countries like China are more frequently using food‐related trade actions to deal with deteriorating political relations. Using an event study approach, this paper investigates how importers respond to lasting political tensions by examining China's seafood importers' responses to the 6‐year Norway–China political tensions after Norway awarded Liu Xiaobo, a Chinese political dissident, a Nobel Peace Prize in 2010. Our results reveal firm‐level responses at both the intensive and extensive margins. At the intensive margin, firms that imported Norwegian fresh salmon before the sanction saw a 20% persistent decline in their fresh salmon import value and an 80% decrease in the import share of Norwegian fresh salmon products over our study period. At the extensive margin, we find a trade diversion effect that firms imported fresh salmon from Norway to other countries and regions, but also a consistent ‘political hedging’ effect 3 years after sanction with a 20% decline in the maximum import share from any particular country or region, even if not Norway.
Many IT professionals seek to improve their job prospects by engaging as leaders of online communities, for example, by serving as a moderator or admin. We investigate whether such community leadership leads to (causal) improvements in individuals’ careers. We assemble a data set, including job histories of IT professionals who have sought election as moderators (mods) in Stack Exchange question-and-answer communities, between 2010 and 2020. We estimate the career benefits of moderatorship under two complementary identification strategies: difference-in-differences (DID) and regression discontinuity (RD). We observe qualitatively consistent results under each design, finding that election to a moderator role has a significant, causal, positive effect on job mobility. We estimate that moderatorship increases the probability of a job change by between 4.7 and 12.3 percentage points over the two years following the election. We also report a series of secondary analyses that speak to associated salary increases and show evidence consistent with the notion that social capital and signaling play a role. Our findings help us understand the benefits of online community leadership, and they extend our understanding of the motivations for online community engagement.
This paper was accepted by Chris Forman, information systems.
Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2019.03252 .
With the increasing digitization and networking of medical data and personal health information, information security has become a critical factor in vendor selection. However, limited understanding exists regarding how information security influences vendor selection. Drawing from the attention‐based view (ABV), this study examines the potential impact of data breaches on hospitals' selection of electronic medical record system (EMRS) vendors. To test our hypotheses, we compile a unique dataset spanning 12 years of observations from US hospitals. Utilizing a coarsened exact matching (CEM) technique combined with a difference‐in‐differences (DiD) approach, our study shows that hospitals tend to replace their EMRS vendors after experiencing data breaches. Moreover, breached hospitals tend to prioritize information security in such a vendor replacement process by switching to star vendors and migrating towards a single‐sourcing configuration. Further post‐hoc analyses reveal that these impacts of data breaches are mitigated as the relationship between breached hospitals and vendors matures or when hospitals belong to large healthcare systems. Additionally, we find that the effects of data breaches are contingent on the scale of the breach and are short‐term in nature. This research underscores the significance of information security as a crucial consideration in vendor selection for both academia and practitioners.
This paper uses OECD data to examine whether winning the men's FIFA World Cup boosts GDP growth, as claimed by analysts and media outlets concomitantly with every edition of this football competition. By implementing both an event‐study design and a synthetic difference‐in‐difference strategy, the analysis shows that winning the World Cup increases year‐over‐year GDP growth by at least 0.48 percentage points in the two subsequent quarters. This result seems primarily driven by enhanced export growth, which is consistent with a greater appeal enjoyed by national products and services on the global market after victory in a major sporting event.
The dissertation consists of three essays on responsible and sustainable finance. I show that local communities should be seen as stakeholders to decisions made by corporations. In the first essay, I examine whether the imposition of fiduciary duty on municipal advisors affects bond yields and advising fees. Using a difference-in-differences analysis, I show that bond yields reduce by 9\% after the imposition of the SEC Municipal Advisor Rule. Larger municipalities are more likely to recruit advisors after the rule is effective and experience a greater reduction in yields. However, smaller issuers do not seem to significantly benefit from the SEC Rule in terms of offering yield. Using novel hand-collected data, I find that the average advising fees paid by issuers does not increase after the regulation. In the second essay, we analyze the impact of \$40 billion of corporate subsidies given by U.S. local governments on their borrowing costs. We find that winning counties experience a 15.2 bps increase in bond yield spread as compared to the losing counties. The increase in yields is higher (18 -- 26 bps) when the subsidy deal is associated with a lower jobs multiplier or when the winning county has a lower debt capacity. However, a high jobs multiplier does not seem to alleviate the debt capacity constraints of local governments. Our results highlight the potential costs of corporate subsidies for local governments. In the third essay, we provide new evidence that the bankruptcy filing of a locally-headquartered and publicly-listed manufacturing firm imposes externalities on the local governments. Compared to matched counties with similar economic trends, municipal bond yields for affected counties increase by 10 bps within a year of the firm’s bankruptcy filing. Counties that are more economically dependent on the industry of the bankrupt firm are more affected and do not immediately recover from the negative impact of the corporate bankruptcy. Our results highlight that local communities are major stakeholders in public firms and how they are adversely affected by corporate financial distress. The final essay examines whether managers walk the talk on the environmental and social discussion. We train a deep-learning model on various corporate sustainability frameworks to construct a comprehensive Environmental and Social (E\&S) dictionary. Using this dictionary, we find that the discussion of environmental topics in the earnings conference calls of U.S. public firms is associated with higher pollution abatement and more future green patents. Similarly, the discussion of social topics is positively associated with improved employee ratings. The association with E\&S performance is weaker for firms that give more non-answers and when the topic is immaterial to the industry. Overall, our results provide some evidence that firms do walk their talk on E\&S issues.
This study focuses on the impact of floods on agricultural yields and farmers’ mental wellbeing in flood-prone areas of Sri Lanka. Food security is facing heightened risks with the increasing frequency and severity of natural hazards globally, including storms, floods, droughts, extreme temperatures, heat waves, and sea-level rise. To assess these risks, the study conducted a data collection exercise in Sri Lanka, gathering detailed information on crop production, farmers’ demographics, and agricultural inputs. The collected data will be used for future economic analyses to evaluate potential losses and damages caused by floods and to understand the likely effects on the wellbeing of affected farmers. The study aims to provide insights into climate-inclusive agricultural risk assessment and resilience mechanisms for farmers facing flood-related challenges. By understanding the potential impact of floods on agricultural production and farmers’ psychological wellbeing, strategies can be developed to mitigate risks and enhance resilience in flood-prone areas.
The major contributions of twentieth century econometrics to knowledge were the definition of causal parameters within well-defined
economic models in which agents are constrained by resources and markets and causes are interrelated, the analysis of what
is required to recover causal parameters from data (the identification problem), and clarification of the role of causal parameters
in policy evaluation and in forecasting the effects of policies never previously experienced. This paper summarizes the development
ofthese ideas by the Cowles Commission, the response to their work by structural econometricians and VAR econometricians,
and the response to structural and VAR econometrics by calibrators, advocates of natural and social experiments, and by nonparametric
econometricians and statisticians.
Over the past three decades, much research has attempted to identify the determinants of the natural rate of unemployment. The authors reach two main conclusions about this body of work. First, there has been considerable theoretical progress over the past thirty years. A framework emerged that can be used to think, for example, about the relation between technological progress and unemployment. Second, empirical knowledge lags behind. Economists don't have a good quantitative understanding of the determinants of the natural rate, either across time or countries. The authors look at two issues, the relation of wages to unemployment and the rise of European unemployment. Copyright 1997 by American Economic Association.
Using research designs patterned after randomized experiments, many recent economic studies examine outcome measures for treatment groups and comparison groups that are not randomly assigned. By using variation in explanatory variables generated by changes in state laws, government draft mechanisms, or other means, these studies obtain variation that is readily examined and is plausibly exogenous. This article describes the advantages of these studies and suggests how they can be improved. It also provides aids in judging the validity of inferences that they draw. Design complications such as multiple treatment and comparison groups and multiple preintervention or postintervention observations are advocated.
Models for time series of cross-sections with fixed effects and with intertemporal correlation are considered. The regression coefficients, and their standard errors, can be estimated by generalized least squares applied to a transformed model. The procedure is given a conditional likelihood interpretation.
This chapter focuses on observational studies and nonrandomized experiments. In an observational study, treatments are not randomly assigned. If there is no hidden bias, the analysis of an observational study is not fundamentally different from the analysis of a stratified or matched randomized experiment. If there is no hidden bias, care is needed to appropriately adjust for observed covariates, but careful adjustments will remove overt biases. Addressing hidden bias is therefore the central concern in an observational study. Two tasks are central to the analysis of observational or nonrandomized comparisons of treated and control groups. First, if the groups are seen to differ prior to treatment—that is, if there is overt bias—then adjustments are made to correct for these initial differences. Adjustments may use matching, stratification, model based procedures such as covariance adjustment, or a combination of two or more techniques. Second, groups may differ prior to treatment in ways that were not observed—that is, there may be a hidden bias.
Statistics is a subject of many uses and surprisingly few effective practitioners. The traditional road to statistical knowledge is blocked, for most, by a formidable wall of mathematics. The approach in An Introduction to the Bootstrap avoids that wall. It arms scientists and engineers, as well as statisticians, with the computational techniques they need to analyze and understand complicated data sets.
We examine inference in panel data when the number of groups is small, as is typically the case for difference-in-differences estimation and when some variables are fixed within groups. In this case, standard asymptotics based on the number of groups going to infinity provide a poor approximation to the finite sample distribution. We show that in some cases the t-statistic is distributed as t and propose simple two-step estimators for these cases. We apply our analysis to two well-known papers. We confirm our theoretical analysis with Monte Carlo simulations. Copyright by the President and Fellows of Harvard College and the Massachusetts Institute of Technology.
Many economic researchers have attempted to measure the effect of aggregate market or public policy variables on micro units by merging aggregate data with micro observations by industry, occupation, or geographical location, then using multiple regression or similar statistical models to measure the effect of the aggregate variable on the micro units. The methods are usually based upon the assumption of independent disturbances, which is typically not appropriate for data from populations with grouped structure. Incorrectly using ordinary least squares can lead to standard errors that are seriously biased downward. This note illustrates the danger of spurious regression from this kind of misspecification, using as an example a wage regression estimated on data for individual workers that includes in the specification aggregate regressors for characteristics of geographical states. Copyright 1990 by MIT Press.
This paper describes a simple method of calculating a heteroskedasticity and autocorrelation consistent covariance matrix that is positive semi-definite by construction. It also establishes consistency of the estimated covariance matrix under fairly general conditions.
[eng] Transportation costs and monopoly location in presence of regional disparities. . This article aims at analysing the impact of the level of transportation costs on the location choice of a monopolist. We consider two asymmetric regions. The heterogeneity of space lies in both regional incomes and population sizes: the first region is endowed with wide income spreads allocated among few consumers whereas the second one is highly populated however not as wealthy. Among the results, we show that a low transportation costs induces the firm to exploit size effects through locating in the most populated region. Moreover, a small transport cost decrease may induce a net welfare loss, thus allowing for regional development policies which do not rely on inter-regional transportation infrastructures. cost decrease may induce a net welfare loss, thus allowing for regional development policies which do not rely on inter-regional transportation infrastructures. [fre] Cet article d�veloppe une statique comparative de l'impact de diff�rents sc�narios d'investissement (projet d'infrastructure conduisant � une baisse mod�r�e ou � une forte baisse du co�t de transport inter-r�gional) sur le choix de localisation d'une entreprise en situation de monopole, au sein d'un espace int�gr� compos� de deux r�gions aux populations et revenus h�t�rog�nes. La premi�re r�gion, faiblement peupl�e, pr�sente de fortes disparit�s de revenus, tandis que la seconde, plus homog�ne en termes de revenu, repr�sente un march� potentiel plus �tendu. On montre que l'h�t�rog�n�it� des revenus constitue la force dominante du mod�le lorsque le sc�nario d'investissement privil�gi� par les politiques publiques conduit � des gains substantiels du point de vue du co�t de transport entre les deux r�gions. L'effet de richesse, lorsqu'il est associ� � une forte disparit� des revenus, n'incite pas l'entreprise � exploiter son pouvoir de march� au d�triment de la r�gion l
There are numerous empirical studies that exploit variation in policies over space and time in the U.S. federal system. If state policy making is purposeful action responsive to economic and political conditions within the state then it is necessary to identify and control for the forces that lead to these policy changes. This paper investigates the implications of policy endogeneity for a specific policy context--workers' compensation benefits. We contrast different methods of estimation and their pros and cons in this context.
This paper estimates the risk preferences of cotton farmers in Southern Peru, using the results from a multiple-price-list lottery game. Assuming that preferences conform to two of the leading models of decision under risk--Expected Utility Theory (EUT) and Cumulative Prospect Theory (CPT)--we find strong evidence of moderate risk aversion. Once we include individual characteristics in the estimation of risk parameters, we observe that farmers use subjective nonlinear probability weighting, a behavior consistent with CPT. Interestingly, when we allow for preference heterogeneity via the estimation of mixture models--where the proportion of subjects who behave according to EUT or to CPT is endogenously determined--we find that the majority of farmers' choices are best explained by CPT. We further hypothesize that the multiple switching behavior observed in our sample can be explained by nonlinear probability weighting made in a context of large random calculation mistakes; the evidence found on this regard is mixed. Finally, we find that attaining higher education is the single most important individual characteristic correlated with risk preferences, a result that suggests a connection between cognitive abilities and behavior towards risk.
The purpose of this note is to explain how to use standard packages to calculate heteroskedasticity and serial c orrelation consistent standard errors for within-groups estimators of a linear regression model from panel data. The within-groups estimat or is calculated as the least squares estimator in a transformed mult ivariate regression with cross-equation linear restrictions. The Whit e standard errors obtained in this way are the desired ones. Copyright 1987 by Blackwell Publishing Ltd
Inference methods that recognize the clustering of individual observations have been available for more than 25 years. Brent Moulton (1990) caught the attention of economists when he demonstrated the serious biases that can result in estimating the effects of aggregate explanatory variables on individual-specific response variables. The source of the downward bias in the usual OLS standard errors is the presence of an unobserved, state-level effect in the error term. More recently, John Pepper (2002) showed how accounting for multi-level clustering can have dramatic effects on t statistics. While adjusting for clustering is much more common than it was 10 years ago, inference methods robust to cluster correlation are not used routinely across all relevant settings. In this paper, I provide an overview of applications of cluster-sample methods, both to cluster samples and to panel data sets. Potential problems with inference in the presence of group effects when the number of groups is small have been highlighted in a recent paper by Stephen Donald and Kevin Lang (2001). I review different ways of handling the small number of groups case in Section III.