Semester Undergraduate Program for Economics Research (SUPER)

Harvard Economics is pleased to launch a new round of SUPER. Peruse available projects below, and enter your application here by 8/10. RA matches will be announced on August 21, 2022 and RAships will begin on/around the week of September 5, 2022.

Applications require a transcript, résumé, list of relevant qualifications, and a ranking of your preferred project(s). Successful applicants will be placed in a semester-long RA position on one of their preferred projects (pay is $15 per hour, and hours vary by project - see description). 

Applications from members of under-represented or disadvantaged groups are especially welcome, but SUPER is very much open to everyone.


Why Aren’t American Roads Smoother? (Lindsey Currier, Ed Glaeser, Gabriel Kreindler) - Support work using Uber data on road roughness and survey data on road repaving.  The RA will directly work for Ms. Currier (PhD student). Weekly hours: 10. Prereqs: Ec 1011a valued but not required.  

Policies to Address High Prescription Drug Prices (Leemore Dafne) - Work with professor and graduate student to research a range of topics related to drug pricing, including efforts by employers to contain spending, efforts by pharmaceutical companies to increase spending, and relevant regulations and potential policies. Initially, research will be qualitative, but as our focus narrows we expect to utilize individual or aggregated insurance claims data to evaluate our hypotheses and - where possible - to simulate reforms. Weekly hours: 8. Prereqs: voracious appetite for healthcare economics, skill reading and synthesizing articles, Ec 1123/1126; Ec 1011a; R/Stata not required but definitely preferred.

American Communities Computable Newspaper Database (Melissa Dell) - We have developed a deep learning pipeline to extract structured text from over 50 million page scans drawn from over 10,000 historical U.S. newspapers (1880-1978). We are now using cutting edge NLP methods to understand what content different newspapers printed, the sources they used (i.e. locally generated versus newswire or syndicated content), the topics covered, the biggest stories across time, who they talk about (disambiguated entities), the sentiment of their coverage, and what factors influenced this choice of content. Tasks include building data necessary to train NLP models, embedding content, running scripts to train models, visualization, and validation of results. RA will work in a team of graduate and undergraduate researchers headed by Melissa Dell. Weekly hours: 8. Prereqs: Python fluency.

Historical Memory (Melissa Dell) - This project seeks to quantify who gets remembered - and why - relative to prominence in historical print media sources. We have used deep learning to extract structured text from around 50 million historical newspaper scans, comprising over a billion bounding boxes. The project entails training and deploying named entity recognition at scale, the development of novel methods for cross-document coreference resolution (to quantify which mentions refer to the same person), disambiguating these entity clusters to Wikidata, and disambiguating to a variety of other sources such as textbooks. We will use these data to study the formation of historical memory. RA will work in a team of graduate and undergraduate researchers headed by Melissa Dell. Weekly hours: 8. Prereqs: Python fluency.

New Data on the Rental Housing Market (Winnie van Dijk) - The share of Americans who rent their homes is rising and nearly half of renters pay more than 30% of their income towards rent. Yet data on rental prices within cities is sparse, making it difficult to study impacts of housing policies, gentrification, zoning, migration, and new technologies. The aim of this project is to fill the gap in data on rents using prices and characteristics from an online apartment listing platform. We’re looking for someone with a computer science background and an interest in economic applications. The aim of the RA-ship would be to experiment with scraping data, and if feasible, to implement large-scale, real time data collection. The RA would work with Winnie Van Dijk, assistant professor, and Lindsey Currier, graduate student. Weekly hours: 8. Prereqs: Advanced knowledge of python (or a different web-scraping language) and demonstrable previous experience web-scraping is strongly preferred.

What Happened to Mom and Pop? (Michael Droste – Ph.D. candidate) - This project (which will probably be my job market paper) investigates the longitudinal impacts of business exit on local economies in the US. In particular, I'm looking at the shift in composition away from small single-establishment firms over the last 50 years. I'm largely working on this alone using administrative data right now, but would very much like to work with an eager student who would like some exposure to empirical work using large publicly available data (like the American Community Survey, decennial censuses, and public use tax data). Weekly hours: 5. Prereqs: Ec 1123/1126 (or stats); Ec 1011b; Proficiency at least one of: Stata, R, Python, or Julia for data analysis.

Why Women Won (Claudia Goldin) - There is a long history of legislation and court decisions regarding women's economic, legal, and social rights. Why did these changes occur at particular moments in US history? This project involves getting an accurate sense of the chronology of the laws and decisions (at the federal and state levels) and also using survey data on opinions and election results for the US to understand why change occurred. There is also a possibility of extending the project to other countries. Weekly hours: max 10. Prereqs: Ec 1123/1126; Stata fluency; 1010a or 1011a.


Optimal Hedging for Homeowners and Future Homebuyers (David Laibson) - We are building a model of the risks associated with owning a home and the risks associated with not owning a home. The RA will help us with the numerical solution of this model and the analytic and numerical design of optimal financial hedging instruments for both home owners and (future) home buyers. Note - we are in an exploratory phase of this project. We are also considering a project on private paternalism (which would involve designing, implementing, and analyzing the results of a field experiment regarding self-selected incentives for productivity). Weekly hours: minimum of 6. Prereqs: Ec 1123/1126; Stat 104/Ec 20/Stat 110; Ec 1011a; Ec 1011b; R or Stata needed; Matlab also helpful.

Entrepreneurial Diffusion: The Transmission of New Ventures (Josh Lerner) - High-growth entrepreneurship has spread rapidly around the globe.  Many of these new businesses have emulated successful new ventures in the United States and China. While there has been a substantial literature examining the international diffusion of patents and technology, the rate and direction of entrepreneurial diffusion remains much more poorly understood. Using data from Pitchbook and Refintiv on entrepreneurial business, as well as interconnections of a given nation with the U.S. and China, Jacob Moscona, David Yang, and I will explore how business models diffuse and where they are successful. Ideal for a student interested in economic research, entrepreneurship, and/or venture capital. Weekly hours: TBD – likely 10-12. Prereqs: Stata and R fluency.

Analyzing New Models of Psychology and Economic Theory (Matthew Rabin) - I have been working out the implications of some recent formal models attempting to embed more realistic psychology into economic analysis.  Such models attempt to capture various improvements to the psychological realism of economics with the same rigor as in modern economic theory. Topics of these models include understanding limits to rationality and finding better ways to capture people’s true preferences and goals (see my 2013 papers in the Journal of Economic Literature and the AER for outlines of long-term agenda motivating this research). Potential research assistants would help solve some of these new models across scenarios and write up the solutions under my own guidance and that of other project members, and to help create usable online calculators to allow others to solve the models. Weekly hours: 6-8. Prereqs: A strong background in microeconomic theory and math (although none of the exercises would involve special advanced training); a background in psychology or behavioral economics would likely also be useful. Stat 104/Ec 20/Stat 110;Ec 1011ab: R and Python fluency useful. Knowing (or willing to learn) latex (via overleaf or scientific word) to write up solutions useful. Ec 1123/1126; Stat 104/Ec 20/Stat 110; Ec 1011a.

Policing as Entertainment: The Distortionary Effect of Reality TV & Copaganda (Emma Rackstraw, Ph.D. candidate) - Many police departments across the United States have welcomed reality TV camera crews in recent decades. Extra scrutiny from the presence of cameras may reduce proactive policing. On the other hand, the cameras may create an incentive for officers who want to be featured on the show to seek opportunities to create entertaining footage. In particular, reality TV filming may raise the relative marginal benefits of arrests for some low social cost crimes, such as drug possession, by making their entertainment value salient and relevant. I use a difference-in-differences strategy to consider the potential distortionary effects of these cameras on a wide variety of outcomes, including arrest behavior, civilian attitudes towards police, electoral outcomes, police recruitment efforts, and more. This project involves a massive data acquisition and cleaning effort, particularly in acquiring novel data directly from police departments through the Freedom of Information Act. The RA will report to Emma Rackstraw and will oversee the process of submitting and following up on FOIA requests, cleaning and standardizing novel data, and aiding with analysis. Weekly hours: 3-5. Prereqs: Experience in Python and/or Arc GIS are a plus due to the potential need to scrape data and utilize geographic data. A great candidate will be interested in using applied microeconomic techniques to better understanding policing in the United States. Ec 1123/1126; Stat 104/Ec 20/Stat 110; Ec 1010a; Ideally, R and Stata fluency but more important is are attention to detail and experience with big data management and cleaning using any software.

Statistical Methods for Applied Microeconomics (Jesse Shapiro) - I envision the undergraduate research assistant participating in a collaborative project to build, test, and extend statistical software, implementing the recommendations in some of my methodological research. Open to discussing other opportunities depending on availability and interest. To learn more about my current and ongoing research projects, please see: More about my research software packages here: Weekly hours: flexible – greater investment will yield greater learning. Prereqs: Comfort in at least one of {python, R, Stata} would be very helpful. Comfort in more than one would be a plus.

Superstar Similarity and Long-Run Firm Valuations (Andi Wang, Ph.D. candidate) - Investors often think of new enterprises in relation to businesses that they already know well. "The Uber of xxx" "the Apple in xxx" is a familiar pitch of startups, and also appears in professional stock analysts' reports that provide recommendations for whether a newly listed stock is worth buying. The idea of reasoning through similarity is both intuitive and features heavily in neuroscience research on memory formation. From an asset pricing point of view, similarity to superstar firms may capture important information of investors' long-run expectation on firms, that is not captured by current numeric measures of long-run growth. In this project, the RA will work directly with me (4th year Business Economics student) to extract information from textual data on superstar similarity and relate them to asset price dynamics. In increasing order of skill level and commitment, the RA will 1) read sell-side equity reports and assess potential for computational analysis 2) gather financial data in support of the analysis 3) write code to process textual data 4) analyze the data and get involved in the design of the project. Weekly hours: 10. Prereqs: Python fluency and experience with NLP methods is a plus.

Describing and Understanding China’s Engagement with Africa (David Yang) - Over the last two decades, China has emerged as a major player in trade, foreign direct investment, and aid throughout the African continent. As China rapidly displaces Africa’s traditional partners such as the US and Europe, the Chinese engagement with Africa poses a set of timely and fascinating political economy questions. In this project, we ask: does Chinese involvement in Africa (re)shape locals’ political ideology? Do Chinese projects and China as a role model challenge the perceived conventional wisdom that democracy fosters growth and prosperity? The primary tasks of the RA will involve: (a) collect systematic information on Chinese projects in Africa; (b) use GIS tools to conduct spatial analyses of the projects’ locations; and (c) synthesize surveys and related data on Africans’ political opinions. Weekly hours: 10. Prereqs: Ec 1123/1126; R, Stata, and Arc GIS fluency.