Heterogeneous Treatment Eﬀects of Behavioral and Environmental Risk Factors on Infants' Health at Birth - A Causal Machine Learning Approach - Inauguraldissertation

Aug 4, 2026

Upload by Finn

The dissertation analyzes how behavioral and environmental risk factors causally affect infants’ health at birth, emphasizing heterogeneity across groups and conditions. It studies maternal smoking and related policy interventions in Germany, and evaluates temperature and weather shocks using evidence from the US. Causal machine learning methods, including causal forests and effect decomposition, are applied to identify variable treatment effects. Robustness checks cover heavy smokers, low birth weight, pre-pregnancy smoking, and propensity trimming, supplemented by placebo testing and fixed-effects regression.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages176

Reading Time7 h 24 min

AccessFree to download

Introduction

Uncovering Sources of Heterogeneity in the Eﬀects of Maternal Smoking on Infants' Health at Birth

Background

Data

Methods

Empirical Results

Robustness Checks

Discussion and Implications

Conclusion

Eﬀect of Smoking Bans on Smoking during Pregnancy: Evidence from Germany

Introduction

Smoking Bans in Germany

Data and Method

Results

Discussion

Eﬀect of Temperature and Weather Shocks on Health at Birth: Evidence from the US

Introduction

Data

Empirical Strategy

Results

Discussion

Appendix

Heterogeneous Treatment Eﬀects of Behavioral and Environmental Risk Factors on Infants' Health at Birth: A Causal Machine Learning Approach Inauguraldissertation zur Erlangung des Doktorgrades der Wirtschafts-und Sozialwissenschaftlichen Fakultät der Unversität zu Köln 2023 vorgelegt von Johanna Maria Zenzes aus Neuss Referent: Prof. Dr. Tom Zimmermann Korreferent: Prof. Dr. Daniel Wiesen Tag der Promotion: 12.07.2023 Für meine Eltern. Acknowledgments I'm extremely grateful to my main advisor Tom Zimmermann, for his invaluable mentorship and support over the last few years. I want to thank him for all the fruitful discussions, and for generously providing expertise and feedback on this thesis. His encouragement to pursue my research ideas and his advice have been invaluable tomy growth as a researcher. I would also like to express my gratitude to my second advisor Daniel Wiesen, who provided invaluable feedback and guidance on this thesis. The last years would not have been the same without my colleagues at the Data Innovation Lab at AXA and at the Institute of Econometrics and Statistics at the University of Cologne. I am thankful for the great working atmosphere and moral support, which made these last years very special and enjoyable. Above all, I am deeply indebted to my parents who always supported me unconditionally and encouraged me along the way. And special thanks to Kevin, for always having my back and believing in me. Contents 1. Introduction 1 2. Uncovering Sources of Heterogeneity in the Eﬀects of Maternal Smoking on Infants' Health at Birth 3 2.1. Introduction ..................................... 3 2.2. Background ..................................... 7 2.2.1. Smoking and Health at Birth ....................... 7 2.2.2. Heterogeneity in the Smoking Eﬀect ................... 8 2.3. Data ......................................... 10 2.4. Methods ....................................... 15 2.4.1. Setup .................................... 15 2.4.2. Causal Forest ................................ 17 2.4.3. Eﬀect Decomposition ............................ 19 2.5. Empirical Results .................................. 23 2.5.1. Standardized Birth Weight ........................ 25 2.5.2. Apgar Score ................................. 28 2.6. Robustness Checks ................................. 35 2.6.1. Heavy Smokers ............................... 35 2.6.2. Low Birth Weight ............................. 37 2.6.3. Prepregnancy Smoking ........................... 38 2.6.4. Propensity Trimming ............................ 39 2.7. Discussion and Implications ............................ 39 2.8. Conclusion ...................................... 42 3. Eﬀect of Smoking Bans on Smoking during Pregnancy: Evidence from Germany 44 3.1. Introduction ..................................... 44 3.2. Smoking Bans in Germany ............................. 47 3.3. Data and Method .................................. 49 3.3.1. Data Basis ................................. 49 3.3.2. Method ................................... 54 3.4. Results ........................................ 56 3.4.1. Smoking Ban and Smoking Behavior ................... 56 3.4.2. Robustness Checks ............................. 57 Contents 3.5. Discussion ...................................... 62 4. Eﬀect of Temperature and Weather Shocks on Health at Birth: Evidence from the US 65 4.1. Introduction ..................................... 65 4.2. Data ......................................... 70 4.2.1. Birth Data ................................. 70 4.2.2. Weather Data ................................ 72 4.3. Empirical Strategy ................................. 76 4.3.1. Setup and Notation ............................ 76 4.3.2. Identiﬁcation Challenges .......................... 78 4.3.3. Treatment Eﬀect Estimation using Causal Forests ............ 79 4.3.4. Asses

FAQs

What causal machine learning approach does the dissertation use to estimate heterogeneous treatment effects?

It uses causal forests for treatment effect estimation and applies effect decomposition to analyze how effects vary across outcomes and subgroups.

How is maternal smoking studied, and what types of infants’ health outcomes are considered?

Maternal smoking effects are examined using measures such as standardized birth weight and Apgar score, with additional discussion and implications based on the heterogeneity findings.

What robustness checks are performed for the smoking-related analysis?

The dissertation includes robustness checks such as restricting to heavy smokers, focusing on low birth weight, considering pre-pregnancy smoking, and using propensity trimming to address identification concerns.

Aug 4, 2026

Upload by Finn