Research Experience

Published & Working Papers

Difference-in-Differences for Continuous Treatments and Instruments with Stayers, July 2025.

Arxiv Working Paper

(with Clement de Chaisemartin, Xavier d'haultfoeuille, Felix Pasquier, and Gonzalo Vasquez-Bare)

Submitted.

When one studies the effects of taxes, tariffs, or prices using panel data, the treatment is often continuously distributed in every period. We propose difference-in-differences (DID) estimators for such cases. We assume that between consecutive periods, the treatment of some units, the switchers, changes, while the treatment of other units, the stayers, remains constant. We show that under a parallel-trends assumption, the slopes of switchers' potential outcomes are nonparametrically identified by difference-in-differences estimands comparing the outcome evolutions of switchers and stayers with the same baseline treatment. Controlling for the baseline treatment ensures that our estimands remain valid if the treatment's effect changes over time. We consider two weighted averages of switchers' slopes, and discuss their respective advantages. For each weighted average, we propose a doubly-robust, nonparametric, and square-root(n)-consistent estimator. We generalize our results to the instrumental-variable case. We use our estimators to estimate the price-elasticity of gasoline consumption.

Using did_multiplegt_stat to Estimate Static Treatment Effects in Complex Designs: Overview, and Examples Based on Real Datasets, August 2025.

Link

(with Clement de Chaisemartin, Xavier d'haultfoeuille, Felix Pasquier, and Gonzalo Vasquez-Bare)

This paper presents the did_multiplegt_stat Stata command, available from the SSC repository (ssc install did_multiplegt_stat). The command computes estimators proposed by de Chaisemartin et al. (2025), de Chaisemartin and D’Haultfœuille (2020), and de Chaisemartin and D’Haultfœuille (2023). It can be used with a binary, discrete, or continuous treatment, and the treatment can be continuous at all time periods including the first one in the data. For the command to be applicable, only one condition needs to hold: the data should contain one pair of consecutive time periods between which the treatment of some units, the stayers, does not change. The baseline estimators computed by the command assume a static model: the outcome does not depend on the treatment’s lags. The command has options that one can use to partly relax this assumption. It can also be used when there are several treatment variables, and one wants to estimate the effect of a treatment controlling for the other treatments. Finally, it can compute IV-DID estimators, under a parallel-trends assumption with respect to an instrument rather than the treatment.

Event-Study Estimators and Variance Estimators Computed by the did_multiplegt_dyn Command, July 2024. Paper

(with Diego Cicca, Clement de Chaisemartin, Xavier D'Haultfoeuille, Felix Knau, Mélitine Malézieux)

The paper presents the did_multiplegt_dyn Stata and R commands, respectively available from the SSC and CRAN repositories, which compute the event-study estimators proposed by De Chaisemartin and D'Haultfoeuille (2024), as well as estimators of the variance of those estimators. The main goal of this note is to increase the transparency of the did_multiplegt_dyn command. The companion note also shows extensive simulation results based on two real datasets, that showcase the properties of the estimators and variance estimators computed by the command.

The Effects of Armed Conflict on Women’s Empowerment in Burkina Faso, 2025. Poster, presented at the Annual Meeting of Population Association of America.

(with Jessica Heckert, Jean-Pierre Tranchant, Abdoulaye Pedehombga, Flor Paz, and Aulo Gelli )

Abstract

Previous studies have examined the consequences of armed conflict on women’s wellbeing, as well as women’s role in the resolution of armed conflict. Less work, however, examines how armed conflict affects women’s empowerment.

We analyze the effects of armed conflict in Burkina Faso on multiple domains of women’s empowerment measured in the project-level Women’s Empowerment in Agriculture Index. We use data from the evaluation of Soutenir l’Exploitation Familiale pour Lancer l’Elevage des Volailles et Valoriser l’Economie Rural (SELEVER) a gender- and nutrition-sensitive poultry production intervention in Burkina Faso, which took place during a period of increased conflict (2017-2020) and combine these data with Armed Conflict Location and Event Data project database.

To evaluate the effect of conflict on empowerment, we estimate a difference-in-difference model, separately for women and men, across multiple empowerment indicators, in which the primary explanatory variable indicates whether conflict moved closer to the village during this time period. Then, to determine if the SELEVER program had a protective effect in the context of increased insecurity, we estimate a triple difference model.

Increased exposure to conflict negatively affected intrahousehold decisions in terms of women’s input into livelihood decisions and women’s control over use of income. Interpreting these findings along with trends suggesting increased acceptability of intimate partner violence suggests an alarming shift in intrahousehold dynamics. Somewhat surprisingly, we also found that increased proximity to conflict led to an increase in access to and decisions on credit, as well as an increase in women’s self-efficacy.

These findings may be related to how humanitarian support was being delivered. In examining whether SELEVER had any protective effect, we only find evidence of a protective effect in the area of work balance on both women and men and not on other domains of empowerment.

and more

Involving Influential Men’s Groups to Promote Maternal and Child Nutrition Practices Increased Diet Diversity and Egg Consumption Despite Rising Insecurity in The Sahel, July 2024. Publisher (Current Developments in Nutrition )website:

(with Loty Diop, Aulo Gelli, Malick Dione, Abdoulaye Pedehombga, Henri Some, Rasmané Ganaba, Jean Pierre Tranchant, Jessica Heckert, Elodie Becquey)

Abstract

Objectives: A two-arm cluster randomized trial was designed to compare 2 approaches of behavior change communication for improved nutrition and women’s empowerment practices. Both intervention arms received monthly training sessions using village saving and loans associations (VSLA) as main delivery platform, while in one arm, influential men’s groups (called EBENE) were also trained to promote improved nutrition and women’s empowerment practices.

Methods: Data were collected after one year of implementation in June and August 2023 through phone surveys in a composite sample of women pre-identified before the intervention including members of VSLA, women interested in VSLA and women of the general population. Mixed-effects regression models using fixed effects for treatment exposure and random effects at cluster level, were used to assess differential effects of the interventions on program participation, and knowledge and practices related to diets and women’s empowerment. Adjusted models estimated the effects of insecurity and of the EBENE intervention in the context of insecurity, adjusting for baseline characteristics associated to insecurity prior to its onset, to account for its non-random nature.

Results: Unadjusted analyses found participation rates of 44% and 38% for training on nutrition and gender respectively, with no significant differences between intervention groups. The EBENE intervention was found to increase the likelihood that women felt encouraged by men to improve dietary practices and that women and children consumed poultry and eggs. Adjusted analyses suggested that though insecurity had a negative effect on program participation and diet diversity, the EBENE intervention had a protective effect on diet diversity in areas with higher insecurity. The EBENE intervention was found to have protective effects on women’s mobility, group membership and participation in decisions related to poultry production.

Conclusions: In the context of increasing insecurity, nutrition and women’s empowerment behavior change promoted through VSLA platforms can be boosted by engaging influential men’s to promote improved practices. Further research is needed to better understand the costs and mechanisms involved.

and more

Treatment effect beyond the mean using quantile regression: Experimental evidence from an impact evaluation of an integrated poultry value chain and nutrition intervention in Burkina Faso. (2022) Access to the document.

[As part of my M1 Internship at the International Food Research Policy Institute, IFPRI, Washington DC, United States]

Abstract

Objectives: This paper aims to evaluate a program, subsequently called SELEVER, going beyond the standard framework of average effects by using quantile treatment effect method. SELEVER is a five-year impact evaluation in Burkina Faso designed to address key knowledge gaps on the impact of poultry value chain interventions on the diets, health and nutritional status of women and children. The project is funded by the Bill and Melinda Gates fundation (grant no. OPP1149709) and the CGIAR Research Program on Agriculture for Nutrition and Health (sub-grant no. A4NH- 202004.040.500) via IFPRI.

Methods and Results: This study is as an empirical evidence of the several advantages provided by the Quantile regression modeling when evaluating a program. The evaluation also provides an evidence about how a program such as SELEVER in the value-chain of poultry can help to significantly reduce poultry losses by helping producers to be more attentive about their business and by providing advice on the importance of vaccination etc. Although there is no evidence on cash-flow outcomes, the increased of the use of poultry inputs due to the participation to the intervention, may contribute to long-lasting effects. term on outcomes such as profit.

Page updated

Google Sites

Report abuse