I am currently a Research Assistant at the Vancouver School of Economics, UBC, working with Professor Miguel Ortiz on a multi-wave randomized controlled trial studying the effects of Chinese and American business-training programs on worker productivity among firms in Addis Ababa, Ethiopia, in collaboration with Professor David Wu at Georgia State University.
I recently completed a B.A. in Economics and Statistics, with a minor in Data Science, at the University of British Columbia. Before returning to research, I interned as an investment analyst in London at Arbion and Union Bancaire Privée.
My research interests center on applied microeconomics and development economics: causal inference in field experiments, information efficiency in markets, and behavioral bias in decision-making. My undergraduate thesis examined recency bias in NBA betting markets using data from over 35,000 games.
Looking ahead, I am applying to predoctoral research positions and, eventually, graduate programs in economics.
Analyze fieldwork data from a ~240-firm, ~1,200-worker RCT on Chinese vs. American business-training interventions in Addis Ababa. Work spans enumerator quality scoring, attrition and treatment-implementation analysis, and LASSO-based heterogeneity analysis across waves, in Stata.
Built an M&A prediction pipeline embedding patent portfolios of 7,485 firms via PatentSBERTa and Bayesian Gaussian Mixture Models, then measured technological synergy between firm pairs using the Bhattacharyya Coefficient and a directional complementarity metric, Comp(A → B), that identifies whether a target fills an acquirer's market-relative R&D gaps. Validated the framework against real M&A case studies (AMD/Xilinx, Analog Devices/Maxim, Merck/Acceleron, HP/Poly), correctly distinguishing capability acquisitions from strategic mimicry, and confirmed external validity with a 0.99 correlation to traditional patent citation overlap across 1,000 firm pairs. Extended the framework into a predictive logistic regression on 46 historical M&A events, achieving a Hit@5 of 0.72 against a 23.8% random baseline.
View report (PDF)Tested whether NBA betting markets efficiently price recent team performance using over 35,000 games from 2007 to 2023, comparing market-implied win probabilities (from de-vigged moneyline odds) against a benchmark logit model of true win probability, with team fixed effects and clustered standard errors. Found that markets respond to a one-point increase in recent win rate about two to three times more strongly than is actually justified by game outcomes, a gap that is statistically significant and stable across the full sample period. The bias is asymmetric: it is concentrated in underdogs and driven more by losing streaks than winning streaks, consistent with loss-averse belief updating rather than uniform mispricing.
View paper (PDF)Cleaned and spatially merged Kenya Demographic and Health Survey data from 2008 and 2014 with subcounty shapefiles to examine public health funding discrepancies before and after Kenya's 2010 devolution of health governance to counties, with a focus on Wajir province. Mapped county- and subcounty-level changes in health insurance coverage, finding gains concentrated in rural counties north of Nairobi and the Rift Valley alongside counterintuitive declines in the urban centers of Nairobi and Mombasa. Fit logistic regressions on the underlying survey microdata to identify individual-level determinants of coverage, isolating age, wealth, and education as the strongest predictors.
View notebookEquity and portfolio analysis roles at a $2.6B AUM and $164B AUM firm respectively.
I'm a Research Assistant at UBC and will be applying to predoctoral programs this cycle. If you're a current student, faculty member, or thinking about similar research questions, I'd love to connect — and I'm happy to hear from anyone else, too. Email is the fastest way to reach me.
Email: contact@duncanharrop.com