Research

Publications and working papers by Mauricio Romero in development, education, and public economics.

Publications (peer-reviewed)

1. Inputs, Incentives, and Complementarities in Primary Education: Experimental Evidence from Tanzania

with Karthik Muralidharan, Isaac Mbiti, Youdi Schipper, Constantine Manda, and Rakesh Rajani · Quarterly Journal of Economics 134(3), 2019

Final Manuscript · AEA RCT registration · NBER WP No. 24876 · Data and code (.zip) · Dataverse · Twaweza policy brief · JPAL’s policy brief · VoxDev coverage · Mundo ITAM coverage (Spanish) · Podcast · Free Policy Brief

We present results from a large-scale randomized experiment across 350 schools in Tanzania that studied the impact of providing schools with (i) unconditional grants, (ii) teacher incentives based on student performance, and (iii) both of the above. After two years, we find (i) no impact on student test scores from providing school grants, (ii) some evidence of positive effects from teacher incentives, and (iii) significant positive effects from providing both programs. Most important, we find strong evidence of complementarities between the programs, with the effect of joint provision being significantly greater than the sum of the individual effects. Our results suggest that combining spending on school inputs (the default policy) with improved teacher incentives could substantially increase the cost-effectiveness of public spending on education.

@article{mbiti2019inputs,
  title={Inputs, incentives, and complementarities in education: Experimental evidence from Tanzania},
  author={Mbiti, Isaac and Muralidharan, Karthik and Romero, Mauricio and Schipper, Youdi and Manda, Constantine and Rajani, Rakesh},
  journal={The Quarterly Journal of Economics},
  volume={134}, number={3}, pages={1627--1673}, year={2019},
  doi={10.1093/qje/qjz010}
}

2. Outsourcing Education: Experimental Evidence from Liberia

with Justin Sandefur and Wayne Sandholtz · American Economic Review 110(2), 2020

Final Manuscript · AEA RCT registration · Executive Summary · Policy Brief · Devex coverage · The Economist · Future Development (blog) · Chris Blattman’s blog · Data and code

In 2016, the Liberian government delegated management of 93 randomly selected public schools to private providers. Providers received US$50 per pupil, on top of US$50 per pupil annual expenditure in control schools. After one academic year, students in outsourced schools scored 0.18 standard deviations higher in English and mathematics. We do not find heterogeneity in learning gains or enrollment by student characteristics, but there is significant heterogeneity across providers. While outsourcing appears to be a cost-effective way to use new resources to improve test scores, some providers engaged in unforeseen and potentially harmful behavior, complicating any assessment of welfare gains.

@article{romero2020outsourcing,
  title={Outsourcing education: Experimental evidence from Liberia},
  author={Romero, Mauricio and Sandefur, Justin and Sandholtz, Wayne Aaron},
  journal={American Economic Review},
  volume={110}, number={2}, pages={364--400}, year={2020},
  doi={10.1257/aer.20181478}
}

3. Communal Property Rights and Deforestation

with Santiago Saavedra · Journal of Development Studies 57(6), 2021

Final Manuscript · Data and code

Almost a third of the world’s forest area is communally managed. In principle, this arrangement could lead to a ‘tragedy of the commons’ and therefore more deforestation. But it may be easier to monitor outsiders’ deforestation of land owned by a community rather than an individual. We present a theoretical framework to examine these trade-offs and empirically study the effect of communal titling on deforestation in Colombia. Our empirical approach uses a differences-in-discontinuities strategy that compares areas just outside and inside a title, before and after titling. We find that deforestation decreased in communal areas after titling, especially in small communities, which is consistent with the model’s predictions. We also find evidence of positive spillovers: titling reduced deforestation in nearby areas outside the title (and thus our estimates are a lower bound of the total effects of communal titling on deforestation).

@article{saavedra2021communal,
  title={Communal property rights and deforestation},
  author={Saavedra, Santiago and Romero, Mauricio},
  journal={The Journal of Development Studies},
  volume={57}, number={6}, pages={1038--1052}, year={2021},
  doi={10.1080/00220388.2020.1817394}
}

4. Local Incentives and National Tax Evasion: The Response of Illegal Mining to a Tax Reform in Colombia

with Santiago Saavedra · European Economic Review 138, 2021

Final Manuscript · Data and code

Achieving a fair distribution of resources is one of the goals of fiscal policy. To this end, governments often transfer tax resources from richer to more marginalized areas. In the context of mining in Colombia, we study whether lower transfers to the locality where the taxed economic activity takes place dampen local authorities’ incentives to curb tax evasion. Using machine learning predictions on satellite imagery to identify mines allows us to overcome the challenge of measuring evasion. Employing difference-in-differences strategies, we find that reducing the share of revenue transferred back to mining municipalities leads to an increase in illegal mining. This result highlights the difficulties inherent in adequately redistributing tax revenues.

@article{saavedra2021local,
  title={Local incentives and national tax evasion: The response of illegal mining to a tax reform in Colombia},
  author={Saavedra, Santiago and Romero, Mauricio},
  journal={European Economic Review},
  volume={138}, pages={103843}, year={2021},
  doi={10.1016/j.euroecorev.2021.103843}
}

5. Direct vs Indirect Management Training: Experimental Evidence from Schools in Mexico

with Juan Bedoya, Rafael de Hoyos, Marcela Silveyra, and Monica Yanez-Pagans · Journal of Development Economics 154, 2022

Final Manuscript · Data and code · Harry Patrinos’s blog · Original WP

We use a large-scale randomized experiment (across 1,198 public primary schools in Mexico) to study the impact of providing schools directly with high-quality managerial training by professional trainers vis-à-vis through a cascade-style “train the trainer” model. The training focused on improving principals’ capacities to collect and use data to monitor students’ basic numeracy and literacy skills and to provide feedback to teachers on their instruction and pedagogical practices. After two years, the direct training improved schools’ managerial capacity by 0.13 standard deviations (p-value 0.018) (relative to “train the trainer” schools), but had no meaningful impact on student test scores (we can rule out an effect greater than 0.08 standard deviations at the 95% level).

@article{deHoyos2022direct,
  title={Direct vs indirect management training: Experimental evidence from schools in Mexico},
  author={de Hoyos, Rafael and Bedoya, Juan and Romero, Mauricio and Silveyra, Marcela and Yanez-Pagans, Monica},
  journal={Journal of Development Economics},
  volume={154}, pages={102779}, year={2022},
  doi={10.1016/j.jdeveco.2021.102779}
}

6. Cross-Age Tutoring: Experimental Evidence from Kenya

with Lisa Chen and Noriko Magari · Economic Development and Cultural Change 70(3), 2022

Final Manuscript · Data and code

Interventions that tailor teaching to students’ learning levels are among the most effective for improving learning, but teachers often lack the time to provide personalized instruction. Cross-age tutoring—in which older students tutor younger ones—is a low-cost alternative. We present results from a randomized controlled trial in more than 180 schools in Kenya (roughly 15,000 tutees and 15,000 tutors), in which schools were randomly assigned to implement a daily tutoring program in either English or mathematics. Math tutoring (relative to English tutoring) raised math test scores by 0.063σ (p = 0.068), with the largest effects for students in the middle of the ability distribution (0.13σ for the third quintile); English tutoring had no detectable effect on English scores. We find no evidence of effects—positive or negative—on the tutors themselves.

@article{chen2022crossage,
  title={Cross-age tutoring: Experimental evidence from Kenya},
  author={Chen, Lisa and Magari, Noriko and Romero, Mauricio},
  journal={Economic Development and Cultural Change},
  volume={70}, number={3}, pages={1145--1175}, year={2022},
  doi={10.1086/713940}
}

7. Beyond Short-term Learning Gains: The Impact of Outsourcing Schools in Liberia after Three Years

with Justin Sandefur · Economic Journal 132(644), 2022

Final Manuscript · AEA RCT registration · Devex coverage · The Economist · Quartz coverage · Chris Blattman’s blog · Data and code · Replication package

Outsourcing the management of ninety-three randomly-selected government primary schools in Liberia to eight private operators led to learning gains of 0.18 standard deviations after one year, but these effects plateaued in subsequent years (reaching 0.2 after three years). Beyond learning gains, the programme reduced corporal punishment (by 4.6 percentage points from a base of 51%), but increased dropout (by 3.3 percentage points from a base of 15%) and failed to reduce sexual abuse. Despite facing similar contracts and settings, some providers produced uniformly positive results, while others presented trade-offs between learning gains, access to education, child safety, and financial sustainability.

@article{romero2022beyond,
  title={Beyond short-term learning gains: The impact of outsourcing schools in Liberia after three years},
  author={Romero, Mauricio and Sandefur, Justin},
  journal={The Economic Journal},
  volume={132}, number={644}, pages={1600--1619}, year={2022},
  doi={10.1093/ej/ueab087}
}

8. Designing Teacher Performance Pay Programs: Experimental Evidence from Tanzania

with Isaac Mbiti and Youdi Schipper · Economic Journal 133(653), 2023

Final Manuscript · AEA RCT registration · NBER WP No. 25903 · Twaweza policy brief · GlobalDev brief · GlobalDev Spanish · Data and code

We use a nationally representative field experiment in Tanzania to compare two teacher performance pay systems in public primary schools: a ‘pay-for-percentile’ system (a rank-order tournament) and a ‘levels’ system that features multiple proficiency thresholds. Pay for percentile can potentially induce socially optimal effort among teachers, while levels systems can encourage teachers to focus on students near passing thresholds. Despite the theoretical advantage of the tournament system, we find that both systems improved student test scores across the distribution of initial learning levels after two years. However, the levels system is easier to implement and is more cost effective.

@article{mbiti2023designing,
  title={Designing teacher performance pay programs: Experimental evidence from Tanzania},
  author={Mbiti, Isaac and Romero, Mauricio and Schipper, Youdi},
  journal={The Economic Journal},
  volume={133}, number={653}, pages={1923--1955}, year={2023},
  doi={10.1093/ej/uead010}
}

9. Learning Losses during the COVID-19 Pandemic: Evidence from Mexico

with Enrique Alasino, María José Ramírez, Norbert Schady, and David Uribe · Economics of Education Review 98, 2024

Final Manuscript · Blog (Nexos)

This paper presents evidence of large learning losses and partial recovery in Guanajuato, Mexico, during and after the school closures related to the COVID-19 pandemic. Learning losses were estimated using administrative data from enrollment records and by comparing the results of a census-based standardized test administered to approximately 20,000 5th and 6th graders in: (a) March 2020 (a few weeks before school closed); (b) November 2021 (2 months after schools reopened); and (c) June of 2023 (21 months after schools re-opened and over three years after the pandemic started). On average, students performed 0.2 to 0.3 standard deviations lower in Spanish and math after schools reopened, equivalent to 0.66 to 0.87 years of schooling in Spanish and 0.87 to 1.05 years of schooling in math. By June of 2023, students were able to make up for 60% of the learning loss that built up during school closures but still scored 0.08–0.11 standard deviations below their pre-pandemic levels (equivalent to 0.23–0.36 years of schooling).

@article{alasino2024learning,
  title={Learning losses during the COVID-19 pandemic: Evidence from Mexico},
  author={Alasino, Enrique and Ram{\'i}rez, Mar{\'i}a Jos{\'e} and Romero, Mauricio and Schady, Norbert and Uribe, David},
  journal={Economics of Education Review},
  volume={98}, pages={102492}, year={2024},
  doi={10.1016/j.econedurev.2023.102492}
}

10. The Effect of School Grants on Test Scores: Experimental Evidence from Mexico

with Juan Bedoya, Rafael de Hoyos, Marcela Silveyra, and Monica Yanez-Pagans · Economica 91(363), 2024

Final Manuscript · Online Appendix

We use a randomized experiment (across 200 public primary schools in Puebla, Mexico) to study the impact of providing schools with cash grants on student test scores. Treated schools received on average 16 USD per student each year for two years, an increase of 20% in public spending per child, after teacher salaries. Overall, the grants had no impact on student test scores. Lack of a treatment effect does not seem to be driven by poor implementation or a substitution away from other inputs (e.g. household expenditure).

@article{bedoya2024effect,
  title={The effect of school grants on test scores: Experimental evidence from Mexico},
  author={Bedoya, Juan and de Hoyos, Rafael and Romero, Mauricio and Silveyra, Marcela and Yanez-Pagans, Monica},
  journal={Economica},
  volume={91}, number={363}, pages={1010--1036}, year={2024},
  doi={10.1111/ecca.12523}
}

11. Factorial Designs, Model Selection, and (Incorrect) Inference in Randomized Experiments

with Karthik Muralidharan and Kaspar Wuthrich · Review of Economics and Statistics 107(3), 2025

Published version · Final Manuscript · Online Appendix · NBER WP No. 26562 · WB Blog #1 · WB Blog #2 · Data and code

Factorial designs are widely used to study multiple treatments in one experiment. While t-tests using a fully-saturated “long” model provide valid inferences, “short” model t-tests (that ignore interactions) yield higher power if interactions are zero, but incorrect inferences otherwise. Of 27 factorial experiments published in top-5 journals (2007–2017), 19 use the short model. After including interactions, over half of their results lose significance. Based on recent econometric advances, we show that power improvements over the long model are possible. We provide practical guidance for the design of new experiments and the analysis of completed experiments.

@article{muralidharan2025factorial,
  title={Factorial designs, model selection, and (incorrect) inference in randomized experiments},
  author={Muralidharan, Karthik and Romero, Mauricio and Wuthrich, Kaspar},
  journal={The Review of Economics and Statistics},
  volume={107}, number={3}, pages={597--613}, year={2025},
  doi={10.1162/rest_a_01317}
}

12. The Productivity of Public and Private Preschools (and Schools): Evidence from India

with Petter Berg and Abhijeet Singh · Economic Journal 136(676), 2026

Final Manuscript · Data and code

We study the relative productivity of private and public institutions at the preschool and primary school levels using panel data from 215 villages in Tamil Nadu. Private preschools show higher test score value-added in math and language (0.59–0.74 standard deviations) and outperform government providers in nearly all villages. This productivity difference explains 60% of the socioeconomic test score gap before school entry. These results contrast starkly with primary schooling, where we find no evidence of a private-sector premium in math and negative effects in local language. Test score value-added is positively correlated between private and government options in a village, both at the preschool and primary school levels. Quality is also correlated across levels; villages with more productive primary schools also tend to have more productive preschools. Our findings inform debates on achieving universal foundational skills and highlight the need to improve the quality of preschools available to lower-income families.

@article{berg2026productivity,
  title={The productivity of public and private preschools (and schools): Evidence from India},
  author={Berg, Petter and Romero, Mauricio and Singh, Abhijeet},
  journal={The Economic Journal},
  volume={136}, number={676}, pages={1254--1279}, year={2026},
  doi={10.1093/ej/ueaf089}
}

13. COVID-19 Learning Loss and Recovery: Panel Data Evidence from India

with Abhijeet Singh and Karthik Muralidharan · Forthcoming, Journal of Human Resources

Final Manuscript · RISE Working Paper · NBER WP No. 30552 · Data and code

We use a panel survey of 19,000 primary-school-aged children in rural Tamil Nadu to study ‘learning loss’ after COVID-19-induced school closures, and the pace of recovery after schools reopened. Students tested in December 2021 (18 months after school closures) displayed learning deficits of 0.73 standard deviations in math and 0.34 in language compared to identically-aged students in the same villages in 2019. Two-thirds of this deficit was made up within 6 months after schools reopened. Further, while learning loss was regressive, recovery was progressive. A government-run after-school remediation program contributed 24% of the cohort-level recovery, likely aiding the progressive recovery.

@article{singh2024covid,
  title={COVID-19 learning loss and recovery: Panel data evidence from India},
  author={Singh, Abhijeet and Romero, Mauricio and Muralidharan, Karthik},
  journal={Journal of Human Resources}, note={Forthcoming}, year={2024}
}

14. Preventing School Dropout: Experimental Evidence from Guatemala

with Melissa Adelman, Francisco Haimovich, and Emmanuel Vazquez · Forthcoming, Journal of Labor Economics

Final Manuscript · Earlier WP version · Data and code

We evaluate a randomized dropout prevention program across 4,000 schools in Guatemala, where 30% of children leave school during the primary-to-secondary transition. Schools were assigned to receive a guidance manual and training; the manual, the training and a list of high-risk students; the manual, the training, the list, and behavioral nudges; or control. All treatments reduced dropout in the transition from primary to secondary by 1.2 percentage points from a 34% base, but effects faded after two years (i.e., there is no difference in the likelihood of being enrolled in secondary school after the first year). Still, the program increased years of schooling by 0.012, and its low cost (USD 2–3 per student) and successful large-scale implementation make it a promising, cost-effective approach to increasing schooling in resource-constrained contexts.

@article{adelman2024preventing,
  title={Preventing school dropout: Experimental evidence from Guatemala},
  author={Adelman, Melissa and Haimovich, Francisco and Romero, Mauricio and Vazquez, Emmanuel},
  journal={Journal of Labor Economics}, note={Forthcoming}, year={2024}
}

Comments, conference proceedings, and others

How to Make Replication the Norm

with Paul Gertler and Sebastian Galiani · Nature 554, 2018

Final Manuscript · NBER WP No. 23576 · Data and code · Mirror

@article{gertler2018replication,
  title={How to make replication the norm},
  author={Gertler, Paul and Galiani, Sebastian and Romero, Mauricio},
  journal={Nature}, volume={554}, number={7693}, pages={417--419}, year={2018},
  doi={10.1038/d41586-018-02108-9}
}

Risk Adjustment Revisited using Machine Learning Techniques

with Álvaro Riascos and Natalia Serna · Proceedings of the Brazilian Society of Computational and Applied Mathematics 6(2), 2018

Final Manuscript · Published Version

Risk adjustment is vital in health policy design. Risk adjustment defines the annual capitation payments to health insurers and is a key determinant of insolvency risk for health insurers. In this study we compare the current risk adjustment formula used by Colombia’s Ministry of Health and Social Protection against alternative specifications that adjust for additional factors. We show that the current risk adjustment formula, which conditions on demographic factors and their interactions, can only predict 30% of total health expenditures in the upper quintile of the expenditure distribution. We also show the government’s formula can improve significantly by conditioning ex ante on measures indicators of 29 long-term diseases. We contribute to the risk adjustment literature by estimating machine learning based models and showing non-parametric methodologies (e.g., boosted trees models) outperform linear regressions even when fitted in a smaller set of regressors.

@article{riascos2018risk,
  title={Risk adjustment revisited using machine learning techniques},
  author={Riascos, {\'A}lvaro and Romero, Mauricio and Serna, Natalia},
  journal={Proceeding Series of the Brazilian Society of Computational and Applied Mathematics},
  volume={6}, number={2}, year={2018}
}

Non-economics peer-reviewed articles

On the Optimality of Answer-Copying Indices: Theory and Practice

with Álvaro Riascos and Diego Jara · Journal of Educational and Behavioral Statistics 40(5), 2015

Final Manuscript · Code

Multiple-choice exams are frequently used as an efficient and objective method to assess learning, but they are more vulnerable to answer copying than tests based on open questions. Several statistical tests (known as indices in the literature) have been proposed to detect cheating; however, to the best of our knowledge, they all lack mathematical support that guarantees optimality in any sense. We partially fill this void by deriving the uniformly most powerful (UMP) test under the assumption that the response distribution is known. In practice, however, we must estimate a behavioral model that yields a response distribution for each question. As an application, we calculate the empirical type I and type II error rates for several indices that assume different behavioral models using simulations based on real data from 12 nationwide multiple-choice exams taken by fifth and ninth graders in Colombia. We find that the most powerful index among those studied, subject to the restriction of preserving the type I error, is one based on the work of Wollack and is superior to the index developed by Wesolowsky.

@article{romero2015optimality,
  title={On the optimality of answer-copying indices: Theory and practice},
  author={Romero, Mauricio and Riascos, {\'A}lvaro and Jara, Diego},
  journal={Journal of Educational and Behavioral Statistics},
  volume={40}, number={5}, pages={435--453}, year={2015},
  doi={10.3102/1076998615595628}
}

Misinformation About HIV and Negative Attitudes Toward Homosexuality and Same-Sex Couples’ Rights: The Case of Colombia

with Federico Andrade-Rivas · International Journal of Public Opinion Research 29(3), 2017

Final Manuscript

The rights of homosexuals and same-sex couples are currently the subject of debate on the public stage and in courtrooms around the world. This debate is often colored by prejudice and misconceptions regarding homosexuality, particularly in low- and middle-income countries. This study investigates the opinions of Colombian women about same-sex couples’ rights and women’s attitudes toward a homosexual child (maternal acceptance). Using a nationally representative data set, we find that younger, wealthier, and more educated Colombian women are more likely to support their homosexual child and approve of same-sex couples’ rights, while women with misconceptions regarding the human immunodeficiency virus and homosexuality are less likely to be supportive of a homosexual child and less likely to approve of same-sex couples’ rights.

@article{andradeRivas2017misinformation,
  title={Misinformation about HIV and negative attitudes toward homosexuality and same-sex couples' rights: The case of Colombia},
  author={Andrade-Rivas, Federico and Romero, Mauricio},
  journal={International Journal of Public Opinion Research},
  volume={29}, number={3}, pages={426--448}, year={2017},
  doi={10.1093/ijpor/edw009}
}

Working papers

The Incidence of Affirmative Action: Evidence from Quotas in Private Schools in India

with Abhijeet Singh · Revise and resubmit, Review of Economic Studies

Working Paper · RISE Working Paper · Online Appendix

The incidence of redistributive policies is central to whether they meet their stated goals. We study this in the context of one of the world’s largest affirmative action programs in schooling: a 25% quota in all Indian private schools for students from disadvantaged groups. We use lottery-based estimates to show that, although students admitted under the quota attend more expensive and preferred schools on average, the distribution of program benefits is very regressive. Program applicants are concentrated among more-educated and better-off households. Consequently, 7.4% of the program spending accrues to the bottom socioeconomic quintile, compared to 24.3% to the top quintile. In total, two-thirds of the per-child cost of a quota seat is inframarginal for school choice. We use rich survey data to show that low application rates for poorer children are not driven by preferences and beliefs. Instead, information constraints and application frictions appear to be key. Finally, we use a randomized intervention to confirm the importance of these frictions and further demonstrate that alleviating a single constraint (e.g., information) may not reduce regressive selection, even if it boosts application rates substantially. Our results demonstrate how constraints facing potential applicants can make redistributive policies regressive in practice. Appropriate policy interventions must consider the joint incidence of these constraints to reduce regressivity.

@unpublished{singh_romero_affirmative,
  title={The incidence of affirmative action: Evidence from quotas in private schools in India},
  author={Singh, Abhijeet and Romero, Mauricio},
  note={Working paper (revise and resubmit, Review of Economic Studies)}
}

Toxic Recycling: The Cost of Used Lead-Acid Battery Processing in Mexico

with Erin Litzow, Bianca Cecato, and Tatiana Zarate-Barrera

Working Paper

@unpublished{litzow_toxic,
  title={Toxic recycling: The cost of used lead-acid battery processing in Mexico},
  author={Litzow, Erin and Cecato, Bianca and Romero, Mauricio and Zarate-Barrera, Tatiana},
  note={Working paper}
}

Work in progress

Resting papers

Benefit Plans, Insurer Competition, and Pharmaceutical Prices: Evidence from Colombia

Public health benefit plans must choose what services are covered with public funds. This coverage choice may affect the prices of covered services through multiple channels. First, coverage reduces out-of-pocket expenditures, making consumers less sensitive to the cost of treatment. In an environment where suppliers have market power (as is often the case with pharmaceutical drugs) this could result in higher prices. The second channel is an increase in competition among drugs listed in the benefit plan with the same therapeutic properties, which could result in lower prices. Thus, the net effect on prices is unclear and depends on consumer sensitivity to prices and the level of competition among drugs. Using a difference-in-difference strategy, I study the effect of including a pharmaceutical drug in the national benefit plan of Colombia, a country with a competitive health insurance market in which all insurance companies offer the same plan (the national benefit plan) and charge the same premium. Drug prices decrease by 14% on average after they are listed in the benefit plan and sales increase by 123%. However, if a drug faces no competition and is listed in the benefit plan its price increases. Coverage also affects the prices of unlisted services: Within a therapeutic class, the prices of drugs which are not listed in the benefit plan decrease as the market share of competing drugs listed in the benefit plan increases. I conclude with a discussion of the role of financial incentives in health care markets.

@unpublished{romero_pharma,
  title={Benefit plans, insurer competition, and pharmaceutical prices: Evidence from Colombia},
  author={Romero, Mauricio},
  note={Working paper}
}

Using Instrumental Variables under Partial Observability of Endogenous Variables for Assessing Effects of Air Pollution on Health

with Tarik Benmarhnia and Prashant Bharadwaj

This paper highlights a common issue with instrumental variables (IV) techniques in evaluating the impact of air pollution on health. Since pollutants are almost always produced alongside other pollutants, instrumenting a single endogenous pollutant requires additional assumptions for correct inference and interpretation. We clarify these assumptions and propose that researchers place more structure on the relationship between the instrument and all the relevant pollutants.

@unpublished{romero_partialiv,
  title={Using instrumental variables under partial observability of endogenous variables for assessing effects of air pollution on health},
  author={Benmarhnia, Tarik and Bharadwaj, Prashant and Romero, Mauricio},
  note={Working paper}
}

The Effect of Gold Mining on the Health of Newborns

with Santiago Saavedra