Melissa Dell (Department of Economics), "Deep Learning for Political Economy"
Abstract
Deep learning provides a robust method for learning a mapping between unstructured data (e.g., text, images, audio) and computable representations that can power downstream analyses. These methods, which have already transformed a variety of disciplines, allow us to process traditional data sources at an unprecedented scale and to bring completely new types of data into political economy analyses. Yet taking existing methods off-the-shelf often has significant limitations - particularly for historical applications or those in non-Western societies – given the domain shift from the pre-training corpora that power much of deep learning. This talk will provide an overview of work developing novel datasets and methods for using deep learning to examine social science questions. These include a series of user-friendly open-source packages for deep learning-powered document layout analysis, OCR, record linkage, and other data wrangling tasks, designed to be highly extensible to a diversity of societies. I will also introduce massive-scale open-source text datasets that we curated by applying deep learning to historical newspapers. These are useful both for large-scale pre-training and for social science research. Finally, I will discuss deep learning methods designed to examine the influence of historical media.... Read more about Melissa Dell (Alesina Seminar)
Amanda Coston (Microsoft Research), "Addressing confounding in decision-making algorithms"
Abstract
Machine learning algorithms are used for decision-making in societally high-stakes settings from child welfare and criminal justice to healthcare and consumer lending. These algorithms are often intended to predict outcomes under a proposed decision. It is challenging to evaluate how well these algorithms perform because we only observe the relevant outcome under a biased sample of the population. In this talk, we explore how to use...
Kiara Hernandez, "Firm-level Ethnoracial Diversity and Support for Unionization"
Abstract
In the United States today, mass preferences for fiscal and social spending policies appear minimally responsive to rising earnings inequality and rapidly deteriorating job protections. Theories of political behavior and political economy maintain that because individuals’ preferences for redistribution depend on whether they perceive racial outgroups to be policy beneficiaries, racial animus may explain the mismatch between contemporary inequality and...
Alex Fouirnaies (University of Chicago), "Can Interest Groups Influence Elections? Evidence from Unions in Great Britain 1900-2019"
Abstract
Unions sponsor electoral candidates around the world, yet little is known about the consequences of these arrangements. I study how union sponsorship affected the electoral prospect of British parliamentary candidates throughout the 20th century. From archival material, I collect new data on the universe of union-sponsored candidates. Employing a difference-in-differences design, I document that sponsorship caused a six percentage-point increase in candidate vote shares. I outline theoretical mechanisms and examine whether sponsees improved their electoral fortune because of changes in constituencies, opponents, resources, mobilization, or information. The evidence supports the constituency and resource mechanisms: Sponsorship helped candidates get nominated in attractive constituencies, accounting for two-thirds of the effect, and caused an inflow of campaign resources into constituency-party organizations. Overall, sponsorship promoted the representation of union-friendly candidates in parliament, but it only led to moderate shifts in the balance of power between parties.... Read more about Alex Fouirnaies (Alesina Seminar)
Phillip Heiler, "Heterogeneous Treatment Effect Bounds under Sample Selection with an Application to the Effects of Social Media on Political Polarization" (Link to paper)
Abstract
We propose a method for estimation and inference for bounds for heterogeneous causal effect parameters in general sample selection models where the treatment can affect whether an outcome is observed and no exclusion restrictions are available. The method provides conditional effect bounds as functions of policy relevant pre-treatment variables... Read more about Workshop in Applied Statistics (Gov 3009)
Stephen Ansolabehere presents "American Mosaic," a book project on public opinion.
The American Politics Research Workshop (Gov 3004) meets all academic year, Tuesdays, 12:00 - 2:00 PM, in CGIS K354. This workshop presents an opportunity for graduate students and Harvard faculty to present and receive feedback on their current research. The workshop highlights key theoretical and empirical findings from Harvard affiliates on topics related to American politics.
All interested Harvard affiliates are invited to...
Leander Heldring (Northwestern University), "Bureaucracy as a tool for Politicians: Evidence from Germany" (link to PDF)
Abstract
This paper studies the impact of a well-functioning bureaucracy on the effectiveness of repression, in the context of Germany’s Nazi regime. I compare former Prussian to non-Prussian municipalities within unified Germany in a regression discontinuity framework. When the Nazis persecuted the German Jews, Prussian areas implemented deportations of Jews more efficiently. During the Weimar republic, when Jews were legally protected, violence against Jews is lower in former Prussian areas. In both periods, Prussian local governments had greater ‘capacity’: They were more effective at raising taxes and collecting trash. Capacity derived from greater specialization and better information processing rather than from effort. Specialization may have created the moral wiggle room to implement repugnant directives.... Read more about Leander Heldring (Alesina Seminar)