Treatment Effect Framework Research Articles

Estimation of the Local Conditional Tail Average Treatment Effect

The conditional tail average treatment effect (CTATE) is defined as a difference between the conditional tail expectations of potential outcomes, which can capture heterogeneity and deliver aggregated local information on treatment effects over different quantile levels and is closely related to the notion of second-order stochastic dominance and the Lorenz curve. These properties render it a valuable tool for policy evaluation. In this article, we study estimation of the CTATE locally for a group of compliers (local CTATE or LCTATE) under the two-sided noncompliance framework. We consider a semiparametric treatment effect framework under endogeneity for the LCTATE estimation using a newly introduced class of consistent loss functions jointly for the CTE and quantile. We establish the asymptotic theory of our proposed LCTATE estimator and provide an efficient algorithm for its implementation. We then apply the method to evaluate the effects of participating in programs under the Job Training Partnership Act in the United States.

Heterogeneity, selection, and the policy effect of educational expansion on college graduate earnings in China, 1981–2015

This article examines the impact of social origin on the returns to college education in the context of China’s higher education expansion since 1999. Utilizing a double-treatment setting, a marginal treatment effect framework is adopted to estimate the causal effect of college education on earnings, while a difference-in-differences methodology identifies the policy effect of educational expansion on the college premium. Analysis of data from a series of nationally representative Chinese surveys reveals that the “true” college earnings premium is rather small for the post-expansion cohort, and that much of the observed earnings gap between college and high school graduates after the expansion is due to returns to unobserved abilities. Further analysis shows that the college earnings premium after the expansion declines more for rural-origin children with schooling probabilities in the top percentiles than for their urban counterparts.

Must watch propaganda: the marginal treatment effect of foreign media among always-takers

Abstract Studies of political persuasion often use an exogenous encouragement as an instrument for persuasive messaging. However, for some people, such encouragement is insufficient, while for others, it is unnecessary. These individuals are excluded from methods that only estimate a treatment effect among compliers. Using the marginal treatment effect framework, we extend research finding that exposure to West German television increases support for communism. We find that, because of self-selection, for those who watch West German TV regardless of signal quality, i.e. always-takers, cutting off West German television would have increased support for communism. Our extrapolation shows that media choices reinforce, rather than mollify, political preferences.

Reducing susceptibility to drought under growing conditions as set by farmers: The impact of new generation drought tolerant maize varieties in Uganda

Given the challenges brought about by the increasing frequency of climatic stressors (droughts) and other biotic challenges (pests and diseases), breeding for tolerance to these traits is now seen as an indispensable adjunct to the enhancement of yield potential. Drought tolerant (DT) maize varieties that do well under moderate drought and outperform (or do not underperform) commercial checks under normal rainfall are becoming available. This study examines the role of these maize varieties in mitigating the effects of drought on maize yields in drought-prone areas of eastern Uganda. We estimate the causal impact of these new generations of maize varieties using a multinomial endogenous switching regression treatment effect framework. The average treatment effects of adopting DT maize show that farmers who actually cultivated DT maize achieve 30% more yield than what they would have obtained with non-DT hybrids. Similarly, average treatment effects on the untreated, revealed that farmers who grew non-DT modern and local maize would have 32 and 54% more yield, respectively, if they instead had adopted DT maize. While being superior to all other maize seeds, the magnitudes of the benefits of DT maize varieties were more pronounced in areas with comparatively less rainfall amount providing strong evidence that the yield potential of these varieties is stable across space and a wide range of rainfall conditions. If the genetic gains of these varieties can be secured over the long term, their impacts in improving the resilience of maize farming systems are likely to be considerably large and favorable.

Can mHealth campaigns improve CCT outcomes? Experimental evidence from sms-nudges in Indonesia

Employing a clustered RCT this paper examines the short-term impact of a 12-month mHealth BCC campaign on health practices & outcomes among CCT beneficiaries in Indonesia. Our analysis reveals that the intervention led to substantial improvements in maternal health behavior (postnatal care, child vaccinations, hygiene practices) & outcomes (anemia rates). Adopting a heterogeneous treatment effect framework, we further show that improvements in maternal hygiene practices and anemia rates are closely linked to health knowledge gained by mothers through the sms campaign. In contrast, we provide suggestive evidence that improvements in other health indicators are more likely to be related to the reminder and nudge components of the intervention.

The influence of personal factors, unmet need and service obstacles on the relationship between health service use and outcome after brain injury

BackgroundThis exploratory study aimed to: (i) examine the relationship between health service use and quality of life, psychological wellbeing, global function and participation after discharge from brain injury inpatient rehabilitation, and (ii) determine the influence of personal factors, unmet need for services and service obstacles on the relationship between service use and these outcomes.MethodsUsing a prospective cohort design, 41 adults with acquired brain injury (median age = 46 years; 71% male; 61% severe traumatic injury) were followed for 6-months after discharge from specialist brain injury inpatient rehabilitation. Service use was continuously recorded and obtained through data linkage methods, focusing on the use of: outpatient medical services, outpatient nursing, outpatient allied health; medical acute services; incidents of re-hospitalization; and transitional rehabilitation service use. Outcome questionnaire measures were completed via telephone, at 6-months after discharge, and included: the EuroQol-5D; Depression Anxiety and Stress Scale, Mayo-Portland Adaptability Inventory and Sydney Psychosocial Reintegration Scale. Data were analyzed in a heterogeneous treatment effects framework, using Bayesian Additive Regression Trees.ResultsThere was weak evidence that transitional rehabilitation service use was associated with better psychological wellbeing scores. The posterior probability of lower depression, anxiety and stress scores was .87, .81 and .86, respectively (average treatment effect). There was also weak evidence that re-hospitalization was associated with worse independent living skills scores. The posterior probability of worse scores was .87. However, most re-hospitalizations were due to unavoidable medical complications. We did not find that place of residence at discharge, marital status, unmet need, or service obstacles affected the relationship between service use and the studied outcomes.ConclusionsThis study may highlight the importance of participation in transitional rehabilitation, in the 6-months after discharge from brain injury rehabilitation. Replication in a larger sample size is required to confirm these findings.

Investment and access to external finance in Europe: Does analyst coverage matter?

We aim to determine whether analyst coverage improves European firms’ access to capital markets and investment. Based on a data set that includes firms from several European countries between 2000 and 2015, we implement a treatment effect framework and an instrumental variables (IV) approach, in which the intensity of industry-level waves in coverage is used as an instrument for firm-level coverage. We show that analyst coverage is favorable to firms’ debt and share issuance and their investment expenses. Our paper emphasizes the key role of financial analysts in improving European firms’ financial conditions.

Individual differences in the effects of the ACTION-PAC intervention: an application of personalized medicine in the prevention and treatment of obesity.

There is an increased interest in the use of personalized medicine approaches in the prevention or treatment of obesity, however, few studies have used these approaches to identify individual differences in treatment effects. The current study demonstrates the use of the predicted individual treatment effects framework to test for individual differences in the effects of the ACTION-PAC intervention, which targeted the treatment and prevention of obesity in a high school setting. We show how methods for personalized medicine can be used to test for significant individual differences in responses to an intervention and we discuss the potential and limitations of these methods. In our example, 25% of students in the preventive intervention, were predicted to have their BMI z-score reduced by 0.39 or greater, while at other end of the spectrum, 25% were predicted to have their BMI z-score increased by 0.09 or more. In this paper, we demonstrate and discuss the process of using methods for personalized medicine with interventions targeting adiposity and discuss the lessons learned from this application. Ultimately, these methods have the potential to be useful for clinicians and clients in choosing between treatment options, however they are limited in their ability to help researchers understand the mechanisms underlying these predictions.

Causality in Econometrics: Choice vs Chance

This essay describes the evolution and recent convergence of two methodological approaches to causal inference. The first one, in statistics, started with the analysis and design of randomized experiments. The second, in econometrics, focused on settings with economic agents making optimal choices. I argue that the local average treatment effects framework facilitated the recent convergence by making key assumptions transparent and intelligible to scholars in many fields. Looking ahead, I discuss recent developments in causal inference that combine the same transparency and relevance.

A permutation test for assessing the presence of individual differences in treatment effects.

An important goal of personalized medicine is to identify heterogeneity in treatment effects and then use that heterogeneity to target the intervention to those most likely to benefit. Heterogeneity is assessed using the predicted individual treatment effects framework, and a permutation test is proposed to establish if significant heterogeneity is present given the covariates and predictive model or algorithm used for predicted individual treatment effects. We first show evidence for heterogeneity in the effects of treatment across an illustrative example data set. We then use simulations with two different predictive methods (linear regression model and Random Forests) to show that the permutation test has adequate type-I error control. Next, we use an example dataset as the basis for simulations to demonstrate the ability of the permutation test to find heterogeneity in treatment effects for a predicted individual treatment effects estimate as a function of both effect size and sample size. We find that the proposed test has good power for detecting heterogeneity in treatment effects when the heterogeneity was due primarily to a single predictor, or when it was spread across the predictors. Power was found to be greater for predictions from a linear model than from random forests. This non-parametric permutation test can be used to test for significant differences across individuals in predicted individual treatment effects obtained with a given set of covariates using any predictive method with no additional assumptions.

Cotton yield and Climate Change Adaptation in Pakistan: Application of Multinomial Endogenous Switching Regression Model

Purpose: Cotton is the backbone of Pakistan economy, as country is the 4th largest producer of cotton in the world. Despite this importance there is steep decline in cotton production over time due to climate change. The need to evaluate the potential of adaptation in improving cotton yield has necessitated this study.
 Design/Methodology/Approach: This study is based on the farm household survey of four cotton producing districts, two from each Punjab and Sindh that were purposively selected from heat stress regions of Pakistan. Data were analyzed through multinomial endogenous switching regression model and treatment effect framework.
 Findings: Farm management practices were evaluated for their significance in reducing adverse impacts of climatic extremes on cotton yield. Adaptation in the combination of first three strategies observed to be the most successful strategies in increasing yield.
 Implications/Originality/Value: For effective adaptation access to credit and extension, education, farming experience, and sources of information revealed to be important predictors

Cotton Yield and Climate Change Adaptation in Pakistan: Application of Multinomial Endogenous Switching Regression Model

Purpose:Cotton is the backbone of Pakistan economy, as country is the 4th largest producer of cotton in the world. Despite this importance there is steep decline in cotton production over time due to climate change. The need to evaluate the potential of adaptation in improving cotton yield has necessitated this study.
 Design/Methodology/Approach:This study is based on the farm household survey of four cotton producing districts, two from each Punjab and Sindh that were purposively selected from heat stress regions of Pakistan. Data were analyzed through multinomial endogenous switching regression model and treatment effect framework.
 Findings:Farm management practices were evaluated for their significance in reducing adverse impacts of climatic extremes on cotton yield. Adaptation in the combination of first three strategies observed to be the most successful strategies in increasing yield.
 Implications/Originality/Value:For effective adaptation access to credit and extension, education, farming experience, and sources of information revealed to be important predictors

Adoption of dual-purpose sweetpotato varieties under partial population exposure in Rwanda: Insights from an African plant breeding programme

As part of the Green Revolution, plant breeding programmes such as the African Centre for Crop Improvement have trained scientists to breed crop varieties in Africa to address the challenges of low productivity. However, exposure to and adoption of these varieties is somewhat modest, and there is also a scarcity of empirical studies on the drivers of exposure and adoption. Using data from Rwanda, the counterfactual treatment effect framework was used to estimate observed and potential adoption rates, as well as the drivers of exposure and adoption of dual-purpose sweetpotato varieties bred for both food and feed, under partial exposure to information. The results show that if the entire target farming population had been aware of the varieties, the adoption rate could have been up to 70% instead of the observed rate of 42%, indicating a 28% adoption gap due to partial diffusion. The findings indicate that once these varieties have been exposed, there is scope for further expansion of their cultivation. Key drivers of variety exposure and adoption include membership of farmer groups, participation in demonstration fields, and access to extension services. Thus, conscious efforts to minimize information constraints are a prerequisite for unlocking this adoption puzzle.

Exploring heterogeneous returns to collaborative R&D: A marginal treatment effects perspective

I examine returns to collaborative R&D using the marginal treatment effects framework. This framework allows me to examine whether the impacts of participation in collaborative R&D on the benefits of product innovation are homogeneous, or if instead firms derive heterogeneous returns based on unobserved characteristics and expectations. Assuming that returns are indeed heterogeneous, I develop two alternative hypotheses representing different underlying mechanisms driving the connection between collaboration and expected returns: If firms evaluate the pros and cons of collaboration based on idiosyncratic traits and expectations, then it is logical to expect that those most likely to collaborate are also those most likely to derive significant benefits from collaboration. This represents the notion of positive selection. On the other hand, it might be possible that those firms least likely to collaborate are in fact those that would have benefited the most had they chosen to collaborate. This reflects the notion of negative selection. Using anonymized data from the 2006 Community Innovation Survey, I confirm that there exists significant heterogeneity in the returns to collaborative R&D due to both unobservable and observable firm characteristics; moreover, the findings clearly support the hypothesis of negative selection. It appears that collaborative R&D plays an equalizing role on the benefits of product innovation for resource-constrained firms.

Understanding the consequence of higher educational expansion in China: a double-treatment perspective

This article adopts a double-treatment perspective to understand the consequences of higher educational expansion in China. Since the expansion has changed the school population and enabled only a certain group of individuals to enter college, simply doing a before-and-after comparison of the labor market returns to college would be problematic. To overcome such methodological challenges, this article uses a ranking-and-sorting method to determine individuals truly affected by the expansion policy, or the “compliers” in the local average treatment effect framework, and proposes a difference-in-differences approach to identify the policy effect on the labor market returns to college education. Drawing data from the 1996 Life History and Social Change survey and the 2005–2015 waves of the Chinese General Social Survey, the article describes the characteristics of the “compliers” along the process of China’s higher educational expansion. It shows that the 1999 expansion policy has mainly favored females, urban children, singletons, and those with advantaged class backgrounds during the college transition process.

Heterogeneous Effects of Place-Based Policies and a Practical Treatment Assignment Rule

We empirically assess the treatment effects of a place-based policy in Japan and propose a socially desirable and practically feasible treatment assignment rule for the program. To account for heterogeneity of treatment effects, we estimate the conditional average treatment effect for each region using a marginal treatment effects framework and construct treatment rules based on regional characteristics such as demography, degree of suburbanization, and local production network. Our results indicate that targeting cities where the number of household cars and the share of the elderly population are moderate is the most appropriate for improving per-capita income and per-worker retail sales.

Policy Evaluation with Multiple Instrumental Variables

Marginal treatment effect methods are widely used for causal inference and policy evaluation with instrumental variables. However, they fundamentally rely on the well-known monotonicity (threshold-crossing) condition on treatment choice behavior. Recent research has shown that this condition cannot hold with multiple instruments unless treatment choice is effectively homogeneous. Based on these findings, we develop a new marginal treatment effect framework under a weaker, partial monotonicity condition. The partial monotonicity condition is implied by standard choice theory and allows for rich heterogeneity even in the presence of multiple instruments. The new framework can be viewed as having multiple different choice models for the same observed treatment variable, all of which must be consistent with the data and with each other. Using this framework, we develop a methodology for partial identification of clearly stated, policy-relevant target parameters while allowing for a wide variety of nonparametric shape restrictions and parametric functional form assumptions. We show how the methodology can be used to combine multiple instruments together to yield more informative empirical conclusions than one would obtain by using each instrument separately. The methodology provides a blueprint for extracting and aggregating information about treatment effects from multiple controlled or natural experiments while still allowing for rich heterogeneity in both treatment effects and choice behavior. Institutional subscribers to the NBER working paper series, and residents of developing countries may download this paper without additional charge at www.nber.org.

WEATHER AT DIFFERENT GROWTH STAGES, MULTIPLE PRACTICES AND RISK EXPOSURES: PANEL DATA EVIDENCE FROM ETHIOPIA

This study investigates the effects of combinations of climate smart agricultural practices on risk exposure and cost of risk. We do this by examining the different risk components — mean, variance, skewness, and kurtosis — in a multinomial treatment effects framework by controlling weather variables for key stages of crop growth. We found that adoption of combinations of practices is widely viewed as a risk-reducing insurance strategy that can increase farmers’ resilience to production risk. The hypothesis of equality of weather parameters across crop development stages is also rejected. The heterogeneous effects of weather across crop growth stages have important implications for climate change adaptation to maximize quasi-option value. For a country that has the vision to build a climate-resilient economy, this knowledge is valuable to identify a combination of climate smart practices that minimizes production risk under variable weather conditions.

Estimations of the Conditional Tail Average Treatment Effect

We study estimation of the conditional tail average treatment effect (CTATE), defined as a difference between conditional tail expectations of potential outcomes. The CTATE can capture heterogeneity and deliver aggregated local information of treatment effects over different quantile levels, and is closely related to the notion of second order stochastic dominance and the Lorenz curve. These properties render it a valuable tool for policy evaluations. We consider a semiparametric treatment effect framework under endogeneity for the CTATE estimation using a new class of consistent loss functions recently developed by Fissler and Ziegel (2016). We establish asymptotic theory of our proposed CTATE estimator and provide an efficient algorithm for its implementation. We then apply the method to the evaluation of effects from participating in programs of the Job Training Partnership Act in the US.

Uptake of Climate-Smart Agricultural Technologies and Practices: Actual and Potential Adoption Rates in the Climate-Smart Village Site of Mali

Understanding the level of adoption of Climate-Smart Agriculture (CSA) technologies and practices and its drivers is needed to spur large-scale uptake of CSA in West Africa. This paper used the Average Treatment Effect framework to derive consistent parametric estimators of the potential adoption rates of eight CSA technologies and practices in the Climate-Smart Village (CSV) site of Mali. A total of 300 household heads were randomly selected within the CSV site for data collection. Results showed significant differences in the observed and potential adoption rates of the CSA technologies and practices (drought tolerant crop varieties, micro-dosing, organic manure, intercropping, contour farming, farmer managed natural regeneration, agroforestry and climate information service). The most adopted technology was the organic manure (89%) while the least adopted was the intercropping (21%). The observed adoption rate varied from 39% to 77% according to the CSA options while the potential adoption rates of the technologies and practices ranged from 55% to 81%. This implies an adoption gap of 2% to 16% due to the incomplete diffusion (lack of awareness) of CSA technologies and practices which must be addressed by carrying out more actions to disseminate these technologies in the CSV. Results showed that education, number of workers in the household, access to subsidies, and training have a positive effect on the adoption of most of the CSA technologies and practices. The adoption of drought tolerant varieties and micro-dosing are positively correlated with access to subsidies and training. The study suggests that efforts should be focused concomitantly on the diffusion of CSA options as well as the lifting of their adoption barriers.