themes
Drawing upon methods and theories from economics and land system science, my research seeks to addresses how digital data can help predict, detect, and remedy environmental hazards that affect human welfare. One example includes evaluating programs -- whether sustainability standards or satellite-based insurance contracts -- to help farmers enhance their resilience to weather risk. Other work, in partnership with the Stanford RegLab, concerns how public agencies can leverage machine learning to enhance their effectiveness, focusing on environmental noncompliance in the US.
works-in-progress:
Drop a Line, Submit on Time? Randomized Tailored Reminders Improve Pollution Reporting Timeliness (with Daniel E. Ho, Elizabeth S. Ragnauth, Nathanael Jo). [Working Paper link on SSRN, under review]
From transferring knowledge to transferring risk: bridging science and practice to en(in)sure sustainable development in a changing climate (under review, co-lead author with Colby K. Fisher and Michaela Dolk, with additional contributions from Jane Baldwin, Inbal Becker-Reshef, Rosa I. Cuppari, Tobias Dalhaus, Andrew Hobbs, Gregor Leckebusch, Peter Lacovara, Adam H. Sobel, and Elizabeth Tellman)
Get in the Zone: The Risk-Adjusted Welfare Effects of Data-Driven vs. Administrative Borders to Define Agricultural Index Insurance Zones (with Zhenong Jin, Andrew Hobbs, and Michael R. Carter)
Measures and mediators of household vulnerability to weather-induced poverty in the Sahel (with Armine Poghosyan, Odyssia Ng, and Stephanie Brunelin)
Perennial Problems -- Risk Perception and Reaction in a Rural Economic Portfolio with Relatively Inelastic Supply (with Travis Lybbert and Sambath Jayapregasham)
publications
The Distributive Effects of Risk Prediction in Environmental Compliance: Algorithmic Design, Environmental Justice, and Public Policy (with Reid Whitaker, Vincent La, Hongjin Lin, Brandon R. Anderson, and Daniel E. Ho) Peer-reviewed proceedings of the ACM Conference on Fairness, Accountability, and Transparency (FAccT) 2021
Government agencies are embracing machine learning to support a variety of resource allocation decisions. The U.S. Environmental Protection Agency (EPA), for example, has engaged academic research labs to test the use machine learning in support of a national initiative to reduce Clean Water Act violations. We evaluate prototypical risk prediction models that can support compliance interventions and demonstrate how critical algorithmic design choices can generate or mitigate disparate impact in environmental enforcement. First, we show that the definition of which facilities to focus on through this national compliance initiative hinges on arbitrary differences in state-level permitting schemes, causing a shift in environmental protection away from areas with more minority populations. Second, the policy objective to reduce the noncompliance rate is encoded in a classification model, which does not account for the extent of pollution beyond the permitted limit. We hence compare allocation schemes between regression and classification, and show that the latter directs attention towards facilities in more rural and white areas. Overall, our study illustrates that as machine learning enters government, algorithmic design can both embed and elucidate sources of administrative policy discretion with discernable distributional consequences.
Article with Supplemental Info • Press (Stanford News)
“Can Digital Technologies Reshape Rural Microfinance? Implications for Credit, Insurance, and Saving” (with Michael Carter) Applied Economics Perspectives and Policy (AEPP) 2021.
To serve rural communities, microfinance must contend with the triple challenges of isolation, small scale transactions, and risk. These challenges result in information asymmetries and transaction costs that render markets for financial services imperfect, costly, or missing. Grounded in this theoretical framework, this paper examines how emerging digital technologies (e.g., mobile money, digital credit scoring, and earth observation) can reshape rural markets for savings, credit and insurance services. While our synthesis of the literature suggests reason for hope in all three domains; evaluating, monitoring, and regulating the emerging digital technologies will be critical for ensuring that the resulting rural financial system is more efficient and equitable than its predecessor.
Article • Press (UC Davis) • Press (AAEA)
“Uniting Advances in Remote Sensing, Crop Modeling, & Economics for Understanding and Managing Weather Risk in Agriculture” Nature Reviews Earth & Environment 2021. (with Zhenong Jin, Michael Carter, Benson Kenduiywo, Ani Ghosh, Andrew Hobbs, Robert Hijmans, and David Lobell)
The expanding availability of satellite data at higher spatial, temporal, and spectral resolutions has spurred many applications in agriculture and economic development, including agricultural insurance. Yet, effectively using satellite data in this context requires blending technical knowledge about their capabilities and limitations with an understanding of how they influence the value of different risk reduction programs. This paper reviews how approaches to estimate agricultural losses for index insurance have evolved, starting with costly fi eld-sampling based campaigns towards lower cost techniques from weather and now satellite data. We identify advances in satellite data and in crop modeling for estimating crop yield, but reliably and cheaply assessing yield remains a challenge in complex landscapes. A simple case study and diagnostic diagrams illustrate an economic framework to gauge and enhance the value of insurance based on earth observation data. As yield estimation techniques improve, much of their value for the insured depends on how well they capture low yield situations when people suffer most. Strategically improving the collection, archiving, and accessibility of reliable ground-reference data on crop types and production would facilitate this task. Audits to account for inevitable mis-estimation complement efforts to detect and protect against large losses.
Article • Supplemental Info • Press (EurekaAlert) • Plain Language Summary (Twitter)
“Innovations for Environmental Compliance: Emerging Evidence and Opportunities”. Stanford Institute for Economic Policy Research (SIEPR) Policy Brief. February 2020. [Equal authorship with D. E. Ho, and A. McDonough]
Key take-aways: (1) Environmental practices have improved significantly over the last half century, but noncompliance with environmental law remains stubborn. (2) New data sources and machine learning are powerful tools to assess risks, enhance enforcement and improve compliance. (3) Learning how to best combine machine learning with compliance interventions will require collaborative partnerships to rigorously pilot and evaluate impacts.
“Machine learning for environmental monitoring." Nature Sustainability. 2018. [with M Hino* N Brooks. *Joint first author with MH]
Public agencies aiming to enforce environmental regulation have limited resources to achieve their objectives. We demonstrate how machine-learning methods can inform the efficient use of these limited resources while accounting for real-world concerns, such as gaming the system and institutional constraints. Here, we predict the likelihood of a facility failing a water pollution inspection and propose alternative inspection allocations that would target high-risk facilities. Implementing such a data-driven inspection allocation could detect over seven times the expected number of violations than current practices. When we impose constraints, such as maintaining a minimum probability of inspection for all facilities and accounting for state-level differences in inspection budgets, our reallocation regimes double the number of violations detected through inspections. Leveraging increasing amounts of electronic data can help public agencies to enhance their regulatory effectiveness and remedy environmental harms. Although employing algorithm-based resource allocation rules requires care to avoid manipulation and unintentional error propagation, the principled use of predictive analytics can extend the beneficial reach of limited resources.
Article • Supplemental Info • Press (Stanford) • Review (Nature)
"Oil palm land conversion in Pará, Brazil, from 2006–2014: evaluating the 2010 Brazilian Sustainable Palm Oil Production Program." Environmental Research Letters. 2018. [with LM Curran, M Cochrane, A Venturieri, R Franco, J Kneipp, A Swartos.]
Global models of biophysical suitability for oil palm consistently rank Brazil as having the greatest potential for expansion, with estimates as high as 238 Mha of suitable lands. In 2010, Brazil launched the Sustainable Palm Oil Production Program (SPOPP) to incentivize oil palm development without deforestation on as much as 30Mha. Here we examine oil palm expansion before and after the SPOPP’s launch. In Para, the major oil palm producing state in Brazil, we analyze the extent and change in oil palm cultivation from 2006−2014 using satellite imagery, ground-truthed verification, site-based interviews, and rural environmental (land) registration data. Between 2006−2014, oil palm area (≥9 ha) expanded >200% to ∼219 000 ha. Of the ∼148 000 ha of oil palm developed, ∼91% converted pasturelands while ∼8% replaced natural vegetation, including intact and secondary forests. Although >80% of all oil palm parcels rest <0.5 km from intact forests, direct conversion of intact forests declined from ∼4% pre-SPOPP (2006−2010) to <1% post-SPOPP (2010−2014). Despite low and declining deforestation rates associated with oil palm expansion in Para, our results also show a low area of oil palm development overall compared with reported land suitability. To explore potential contributing factors, we conducted semi-structured interviews with researchers, company representatives, and government officials involved in the sector to characterize the perceived factors influencing oil palm development and the role of agro-ecological suitability mapping among them. Interviews indicated that: (1) individual effects of suitability mapping efforts to encourage oil palm expansion on cleared areas, i.e. without deforestation, cannot be disentangled from pre-existing public and private deforestation reduction initiatives; and, (2) socio-economic constraints, e.g. high relative production costs and limited familiarity with this crop, appear to partially explain the major discrepancy between estimated potential suitable areas with realized oil palm development.
“Energy-based economic Development." Renewable and Sustainable Energy Reviews. 2011 [with S Carley, S Lawrence, A Brown, A Nourafshan]
The fields of economic development and energy policy and planning have converged in recent years to form an emerging discipline, which we term “energy-based economic development” (EBED). Despite the significant amount of stimulus funds, as well as state and local funding, that are being allocated to EBED initiatives in the United States, the emerging discipline has received scant attention in the energy, policy, and development literature. The link between energy and economic development in the literature is still theoretical, mostly focused on the need for and the potential benefits of EBED, and rarely applied. Furthermore, funding for EBED has outpaced understanding of the discipline, development of rigorous technical approaches, and meaningful ways to measure impact. Such information would not only help practitioners and policymakers more thoroughly understand the confines of the discipline and shape goals and approaches accordingly, but also help researchers identify, track, and evaluate a variety of activities in the field. With national and international attention focused on the convergence of these fields, researchers and practitioners have a rare opportunity to develop and implement the tools necessary to evaluate and communicate the potentially broader impacts that EBED may hold for society. If ways to leverage and sustain the injection of funds in this discipline are not identified, the opportunity may end before we can achieve either energy policy or economic development goals. In an attempt to respond to this need, this analysis explores the connection between energy and economic development, beginning with a review of the trends in each field and the goals that each seeks to achieve. On the basis of this information, we define the discipline of EBED, review the existing literature on it, and offer insights and perspectives on its emergence.
selected manuscripts in progress
working papers:
The Producer Response to Eco-Certification: Evidence from the Quality, Quantity, and Consistency of Brazilian Coffee Production (with Robert Heilmayr)
Voluntary sustainability standards (VSS) aim to support more socially and environmentally responsible practices without the need for command-and-control regulations. However, due to their voluntary nature, VSS must yield clear benefits for producers to encourage their continued participation. These benefits could include direct incentives in the form of preferential pricing or more indirect benefits that improve a farm’s production quality and efficiency. Here we use a detailed panel of coffee sales between over 500 farms and a major Brazilian coffee cooperative to estimate responses to Rainforest Alliance certification. We separate the revenue impacts of price premia from the financial benefits that result from induced changes in production quality, quantity, and consistency. While previous literature has discussed changes in quality and income stability as a potential benefit, we examine how VSS measurably alter the primary economic activity behind the certification: agricultural production.