Below are my projects that are either in review or near the submission stage. They are in the order that I expect to wrap them up. Please reach out with any questions or comments; I’d appreciate hearing from you.
The random filter identification strategy
The random filter identification strategy estimates a treatment effect by removing variation in treatment that is driven by unobservable confounding factors. This filtering is feasible when the treatment effect mechanism is exogenous and a mediator variable reflecting this mechanism is observable. The approach permits identification of an ATE or ATT in settings where the assumptions underlying other strategies are not met or yield a LATE. In this paper, I derive the identification assumptions for the random filter and demonstrate its applicability to economics research. My empirical example uses the random filter to measure the impact of the launch of a free-to-use mobile banking platform in El Salvador and explores the mechanisms that caused a push for financial inclusion to be unsuccessful.
Keywords: random filter, front door criterion, identification strategies, financial inclusion, digital banking, mobile money, Chivo Wallet
In Review. [Working Paper] [Code available after publication] [Lecture Slides] [Simulation Code]
The value of designer flows in river conservation
With Lucas Bair, Matthew Reimer, Michael Springborn, and Charles Yackulic
Forgoing traditional economic benefits of river management to control invasive species through designer flows—controlled releases of water from dams—can provide an effective, but costly addition to meet conservation goals. Simulated results from an application in the Grand Canyon reveal that without investment in extensive ecosystem monitoring and designer flow experimentation, the opportunity cost of designer flows remains too high to justify implementation.
Keywords: population viability, ecosystem restoration, designer flows, hydropower
[Manuscript and code available by request]
Shifting attention in adaptive management
With Lucas Bair.
A curse of dimensionality in the representation of biological systems for dynamic optimization problems forces researchers to eschew realistic biological dynamics in favor of something more stylized and computationally-feasible. This leaves questions regarding the value of system monitoring and uncertainty reduction unanswered. Recent advances in dynamic programming algorithms permit more faithful biological dynamics with both latent and observed states. This paper values and directs targeted learning efforts concerning observational, parametric, and forecasting uncertainties in the Colorado River ecosystem.
Keywords: adaptive management, targeted learning, attention, approximate dynamic programming
[Manuscript and code available by request]
Budgeting future global warming potential
For 40 years, economics has consistently diluted the severity of the projected damages due to climate change, despite near-universal objection from the scientific community. In this paper, I identify an optimal emissions pathway (and carbon price) driven by the scientific emphasis on avoiding unknown tipping points, without requiring any the heavily-criticized assumptions that burden cost-benefit IAMs. My results do not perpetuate the claim that climate change is something we can afford to address at a snail’s pace.
Keywords: the role of economics in climate change, optimal abatement, risk and irreversibility, tipping points, integrated assessment modeling
[Manuscript and code available by request]
Cost-effective species viability
The modeling choices that go into designing an optimization problem are just as important as the optimization step itself. For example, models concerned with “cost-effective species viability” could take on all sorts of forms, thus the resulting “optimal” policies will naturally vary in expected present cost or long-run viability. Using two representative case studies, I show that without key model components, an “optimal” solution to the viability problem can end up being excessively costly or excessively risky.
Keywords: model design, bioeconomics, endangered species conservation
Encouraging the defragmentation of habitat across privately-owned lands
Land ownership is often fragmented in areas of conservation interest. Fragmented ownership of a public ecological good will naturally inhibit the preservation of a large, contiguous piece of land—the ecological ideal—without an incentive scheme that encourages collective action. I use methods from statistical mechanics to develop a model of landscape value and collective landowner behavior, then use this to inform efficient land-use regulation that rewards smallholders for preserving (1) the social benefits of conserving any one parcel, and (2) the positive network effects that the connection of conserved parcels creates.
Keywords: land use, habitat fragmentation, conservation incentives, policy design