Meet a RC researcher- Chrystie Burr

            Chrystie Burr is an assistant professor of economics at the University of Colorado, Boulder, focusing on environmental economics. She explains, “My goal is to answer some urgent questions we have in environmental economics, such as policies in terms of addressing climate change issues.” After earning her bachelor’s degree in physics and her master’s degree in industrial engineering, Burr went on to pursue a Ph.D. in economics. Her background in math and science made her comfortable with computer modeling, which she uses for her current research.

            One of Burr’s current projects is working to determine the welfare cost of subsidies for solar power. Tackling climate change requires a slow, expensive transition from coal and natural gas to sustainable energy sources. Although adoption of solar energy has increased exponentially in recent years, it still generates less than 1% of the electricity used in the United States. Burr explains, “When you switch to solar, you’re basically paying your 25-year electricity bill all up front, and the cost is huge.” Doubling the number of solar panels in operation will halve the cost; subsidies are used to encourage early adoption so that the technology can evolve, the cost can go down, and adoption can become more palatable to consumers.

            Economic theory is the basis for Burr’s model. She explains, “Agents have a utility function that they want to maximize, given variables such as the cost of the system, the subsidy amount and the revenue generated from cost-savings.” Theory dictates that consumers are rational and will only switch to solar energy if the long-term savings outweigh the cost. After adding parameters from millions of real-world observations into the model, Burr runs simulations to determine the outcomes of implementing a variety of subsidy programs in different locations. Burr is able to perform counterfactual analysis based on varying locations and subsidy levels.

            To calculate the standard error for the parameters she adds to the model, Burr uses the Janus supercomputer. She has to perform her estimation over 100 times, using randomly selected data. Running this algorithm on her desktop would take years, but the ability to simultaneously run 100 jobs on Janus reduces the time to a couple of months. She says, “This is where the supercomputer is really helpful, in terms of letting me finish my research in a feasible amount of time.” Burr was the first in her department to begin using Janus, and believes that high-performance computing has a place in social science. She explains, “Because we live in the big data era and huge amounts of data are available, computing resources are becoming very important to social scientists.” One of Burr’s data sets contains 40 million observations; opening the data set, or merging it, requires enormous amounts of computational power. Having access to Janus was one of the factors that brought Burr to CU.

 

            Although Burr writes her own code, she often turns to the Research Computing staff for help when she encounters problems with the supercomputer. “They are extremely helpful,” she says. “I keep going back and bugging them about issues I have, and they have helped me out a lot. They really know what they’re doing, and they make sure my research doesn’t get stuck due to a technical bug or anything along those lines.” After over three years of work on the solar project, Burr hopes to submit it this spring. “The end is near,” she says. Going forward, Burr hopes to use similar computational methods to come up with more policy analysis that can address urgent issues relating to climate change.