Econometrics and Data Science
Econometrics and Data Science
The Econometrics and Data Science Research Group covers a wide number of topics within the areas of modern econometric theory and applications, as well as data science in economics. On the econometrics side, the group’s research interests include: the econometrics of networks, panel data econometrics, identification and semiparametric econometrics, macroeconometrics and financial econometrics. On the data science side, the group is interested in, among other topics, machine learning, artificial intelligence, high-dimensional econometrics and text analysis. Such research is often motivated and applied to problems in other fields, including those in industrial organisation, labour economics, political economy, macroeconomics and finance.
The group organises an Econometric seminar that takes place every two weeks on Mondays at 2pm. The group also participates in the CAGE seminar in applied economics, which runs every two weeks on Tuesdays at 2pm, and engages with other seminars in the Department. Students and faculty of the group present their work in progress in two brown bag seminars which run weekly on Tuesdays and Wednesdays at 1pm. The group also co-organises annual workshops, including the Econometrics Workshop, which is a one-day event coupled with an econometrics masterclass.
Our activities
Econometrics Seminar
Monday afternoons
For faculty and PhD students at Warwick and other top-level academic institutions across the world. For a detailed scheduled of speakers please see our upcoming events.
Organisers: Kenichi Nagasawa and Ao Wang
Work in Progress Seminars
Tuesdays and Wednesdays: 1.00-2.00pm
Students and Faculty of the group present their work in progress in two brown bag seminars. For a detailed scheduled of speakers see our upcoming events.
Organiser: Chris Roth
People
Academics
Academics associated with the Reseach Group Name research group are:
Events
Applied & Development Economics Seminar - Erzo Luttmer (Dartmouth)
Title: Living Large or Long? Preference Estimates from Completed-Life Stories (joint with Joshua Schwartzstein, Tomáš Jagelka, Amitabh Chandra)
The paper will not be ready for sharing prior to the seminar, an abstract is as follows:
———Extended Abstract ———
The value of a year of life plays a crucial role in informing policy decisions related to healthcare, the environment, innovation, and safety regulations. Given the significance of a life year’s value for policymaking, estimates should accurately reflect people’s preferences. Currently, estimates are derived from individuals’ observed choices involving small risks of death and monetary compensation, such as their willingness to take riskier jobs for higher pay. However, these estimates have significant limitations: individuals may not correctly perceive small risks, may deviate from rationality for choices involving small probabilities, and those making such choices, like soldiers or firefighters, represent a non-representative sample. Furthermore, unobserved job characteristics related to risk can bias these estimates. Our paper introduces an alternative method for estimating the value of a life year, attempting to overcome these limitations. We present a broadly representative sample of U.S. respondents with choices between pairs of “completed-life stories,” asking which life they would prefer for themselves. By randomizing lifetime earnings and ages of death in each story, we estimate how the probability of choosing a particular story increases with higher lifetime earnings versus a later age of death. We find that a 7% increase in lifetime earnings increases the choice probability as much as a 1% increase in longevity, implying a Life Year Value (LYV) elasticity of 7. Consequently, a typical individual with median earnings of $80,000 annually values an additional life year at age 80 as equivalent to receiving an additional $5,700 per year for all working years, totaling approximately $225,000 over their lifetime. We validate our estimates by demonstrating high test-retest reliability across survey waves and establishing their robustness through a series of specification checks and additional randomizations. Furthermore, we estimate markedly higher LYV elasticities for individuals whose subjective attitudes indicate that they place an above-average value on an additional life year. A key methodological advantage of our approach is that respondents engage in a natural task—assessing whether a life was desirable—which avoids triggering cultural or social norms about paying for life extension. Moreover, our method does not entail decisions under uncertainty, thereby avoiding many known behavioral biases. An important substantive advantage of our approach is that it allows us to estimate how the Life Year Value varies with respondent characteristics, age of death, and lifetime earnings in a broadly representative population. We find that the LYV increases with lifetime earnings and decreases with later life years. Even when holding constant lifetime earnings and age of death, there is notable heterogeneity in the LYV across individuals. This suggests that tailoring policies to individual preferences can result in welfare improvements.
——————