Publications

Electric heating and the effects of temperature on household electricity consumption in South Africa
The Energy Journal. 2020. Vol 41, No. 4: 209-230. Forthcoming. 
[Available at IAEE.org] [Ungated copy] [Executive summary]
How does temperature affect household energy demand in low-income countries? I use 132,375,282 hourly electricity consumption observations from 5,975 households in South Africa to estimate the causal effects of short-term temperature changes on household electricity consumption. The estimates flexibly identify a constant log-linear temperature response - for every 1°C increase in temperature, electricity consumption decreases by 4.1% among temperatures below the heating threshold but increases by 8.1% among temperatures above the cooling threshold. This relationship is driven more strongly by seasonal than hourly temperature changes. Holding all else constant, a 3.25°C increase in temperatures would reduce electricity consumption by 1,093.4 kWh (6.2%) per year per household. Widespread use of electric heating due to limited residential gas heating infrastructure likely drives this. These results point to important regional heterogeneity in how temperature increases may affect household energy demand in the coming decades.

​Hardware, apps, and surveys at scale: Insights from measuring grid reliability in Accra, Ghana  N. Klugman, J. Adkins, S. Berkouwer, K. Abrokwah, I. Bobashev, P. Pannuto, M. Podolsky, A. Suseno, R. Thatte, C. Wolfram, J. Taneja, and P. Dutta). ​July 2019. The second annual ACM SIGCAS Conference on Computing and Sustainable Societies (COMPASS '19 ).  [Available here]

​The vision of sensor systems that collect critical and previously ungathered information about the world is often only realized when sensors, students, and subjects move outside the academic laboratory. However, deployments at even the smallest scales introduce complexities and risks that can be difficult for a research team to anticipate. Over the past year, our interdisciplinary team of engineers and economists has been designing, deploying, and operating a large sensor network in Accra, Ghana that measures power outages and quality at households and firms. This network consists of 457 custom sensors, over 3,000 mobile app instances, thousands of participant surveys, and custom user incentive and deployment management systems. In part, this deployment supports an evaluation of the impacts of investments in the grid on reliability and the subsequent effects of improvements in reliability on socioeconomic well-being. We report our experiences as we move from performing small pilot deployments to our current scale, attempting to identify the pain points at each stage of the deployment. Finally, we extract high-level observations and lessons learned from our deployment activities, which we wish we had originally known when forecasting budgets, human resources, and project timelines. These insights will be critical as we look toward scaling our deployment to the entire city of Accra and beyond, and we hope that they will encourage and support other researchers looking to measure highly granular information about our world’s critical systems.

Working Papers

Credit and attention in the adoption of profitable energy efficient technologies in Kenya 
(with J.T. Dean).  [Available here]
What roles do credit constraints and inattention play in the under-adoption of high-return technologies? We study this question in the case of energy efficient cookstoves in Nairobi. Using a randomized field experiment with 1,000 households, we estimate a 300% average annual rate of return to investing in this technology, or $120 per year in fuel savings - around one month of income. Despite this, adoption rates are low: eliciting preferences using an incentive-compatible Becker-DeGroot-Marschak mechanism, we find that average willingness-to-pay (WTP) is only $12. To investigate what drives this puzzling pattern, we cross-randomize access to credit with two interventions designed to increase attention to the costs and benefits of adoption. Our first main finding is that credit doubles WTP and closes the energy efficiency gap over the period of the loan. Second, credit works in part through psychological mechanisms: around one-third of the total impact of credit is caused by inattention to loan payments. We find no evidence of inattention to energy savings. These findings have implications for second-best regulation of pollution externalities using subsidies versus taxes. In low-income contexts, where credit constraints are common, Pigovian taxation alone may not be the optimal policy. Private benefits and avoided environmental damages generate average welfare gains of $600 for each stove adopted and used for two years. A subsidy would have a marginal value of public funds of $19 per $1 spent. 

          Coverage:
​          [VoxDev] [World Bank Development Impact Blog] [Energy Institute at Haas Blog] 
          [US Association for Energy Economics Best Student Paper] 
​          Cited as evidence for ID Insight's Giving Green Recommendation for Jikokoa stoves.

The political economy and governance of rural electrification (with Eric Hsu, Oliver W. Kim, Edward Miguel, Felipe Vial, and Catherine D. Wolfram).
How does political favoritism shape the provision of public goods? We study the implementation of Kenya’s Last Mile Connectivity Project, a large-scale program to connect all households in Kenya to electricity. Using administrative construction data and voting data from the 2013 and 2017 presidential elections, we find evidence that sites in wards that voted for the current president in 2013 are more likely to be included in the program, and that construction at these sites proceeds on average more quickly than those that voted for the opposition---a pattern consistent with clientelistic behavior around the election cycle. To account for unobservables that may threaten identification, we repeat our analysis using only adjacent electoral wards, where unobservable characteristics are less likely to differ, and find similar results.

          Coverage:  [PBS NewsHour]

Research in Progress

Accountability in infrastructure construction: donor conditionality and political incentives (with Eric Hsu, Edward Miguel, and Catherine Wolfram).
Governments in low-income countries frequently outsource public good provision to the private sector, but accountability and governance in this sphere remain poorly understood. We study outsourced public good provision by a low-capacity state in the context of Kenya's national Last Mile Connectivity Project, which aims to provide universal household electricity access by 2020 using World Bank and African Development Bank funding. We focus on three factors in particular as potential drivers of construction quality: donor conditionality, political incentives, and independent monitoring in a repeated-game setting. We collect novel field engineering assessments and household surveys in 300 sites to study the channels affecting leakage and technical construction. Collection of electricity infrastructure construction quality measurements and household survey data is ongoing and has been completed across more than 100 sites. (This study is registered in the AEA RCT Registry, ID 2389) ​​

          Coverage: [DFID Economic Development and Institutions Video]

The economics of long-term grid reliability (with Catherine Wolfram and Steve Puller)
In urban areas of Ghana, a primary barrier to economic growth is the reliability of the grid. Since 2012, persistent power failures have given rise to the term 'Dumsor,' meaning 'lights off-on' in the Akan language. According to a 2013 World Bank report, 61% of firms see electricity reliability as a major constraint, with firms reporting an average of over 700 hours of outages annually, compared to 1.5 hours for firms in the U.S. We first study quasi-randomly assigned infrastructure upgrades to estimate the cost of marginal improvements in reliability. We then exploit spatial network discontinuities in power quality across Accra to understand the socio-economic impacts of long-term reliability improvements. We collect temporally and spatially high-resolution reliability data using the GridWatch technology, and combine these with administrative infrastructure network data and 5,000 on-the-ground firm and household surveys to accurately estimate these outcomes.

          Coverage: ​[Energy Institute at Haas Blog]