- Research by first-year McCombs faculty has revealed new insights into the worlds of social media, sports, and housing finance
- Several new professors specialize in building statistical models that reveal patterns in large volumes of data, which can be used for everything from e-commerce to gene sequencing
The newest members of the McCombs School of Business faculty are a tough group to label. Their research focuses on a variety of concepts including social media economics, golf scoring statistics, kidney donation programs, mortgage debt, and machine learning, among many other issues. By contributing such a fresh set of analytical skills, these first-year professors add to McCombs’ growing arsenal of researchers who thrive in an increasingly data-driven world.
Assistant Professor Naveed Chehrazi joined the of Information, Risk, and Operations Management department in July 2013 upon receiving his Ph.D. in management science and engineering at Stanford University. Chehrazi's doctoral dissertation and his current research focus on consumer behavior in the unsecured debt market. In the U.S., this consists mainly of credit card debt amounting to about $600 billion.
In addition to their importance to consumers, credit cards play a major role in financing small businesses. Indeed, more than 80 percent of bank financing extended to public firms is in the form of revolving lines of credit. Chehrazi's research combines tools from different disciplines including statistics, economics, and operations research to develop mathematical models that can explain and predict consumers' behavior in terms of their repayment pattern when they are in default.
"A good understanding of consumers' repayment behavior at the collection phase not only allows banks to tailor their collection strategy to match consumers' financial status but also helps them to change their underwriting practice to limit their exposure to consumers' default in the first place," Chehrazi says. This approach can decrease banks' cost of capital, in turn leading to more lending and easier access to credit for both consumers and small businesses. "The challenge, however, is to develop a model that can predict behavior at the individual level" because each newly delinquent cardholder may or may not behave like previously delinquent cardholders, Chehrazi explains.
In addition to his research, Chehrazi teaches an introductory class on operations management. "Teaching at McCombs has been a fun experience so far," he says. "I like my students. They are intellectually curious, motivated, and engaging."
Chehrazi also enjoys cycling, swimming, playing squash, and traveling. In fact, during his graduate studies, he traveled to Quebec, Switzerland, France, Germany, Turkey, and the UAE.
Doug Fearing is excited to be back in Austin, which is “without a doubt” his favorite place he’s lived. After graduating from Carnegie Mellon University in 1999 with an undergraduate degree in computer science, Fearing was employed locally by Trilogy as the manager of a software development team. After five years in Austin, he returned to school at MIT, where he earned a Ph.D. in operations research. In his first faculty position, Fearing taught at Harvard Business School for three years.
Now, he’s an assistant professor in the Department of Information, Risk, and Operations Management, where he teaches working professionals and doctoral students. “Ever since the year that I was on the market for my Ph.D., I’ve been looking for an opportunity to get back” to Austin, he says. “I was very fortunate that last year this position opened up.”
Although his prior work experience isn’t directly connected to his current research, which focuses on operational responses to disruptions, Fearing says his time in industry influences his efforts to address real-world problems. One major area of his research deals with how the airline industry can better deal with disruptions and delays.
He also studies professional sports — specifically, baseball and golf. Among his recognitions, Fearing and his co-author won an MIT sports analytics research paper award for their look at how baseball teams can mitigate the disruption risk from injuries by having players who can fill in at more than one position. Since completing his Ph.D., Fearing has served as a senior advisor to the research and development department of Major League Baseball’s Tampa Bay Rays.
As for golf, his research with co-authors into modeling putting on the PGA Tour as a Markov chain spurred the adoption of the “strokes gained putting” statistic by the PGA. “It’s the only thing I can show my parents that I’ve done,” he jokes.
What’s the value of a social media connection? Rajiv Garg, an assistant professor in the IROM department, is investigating various methods companies can use to gather that information and then use it to develop more effective business strategies.
Garg researches the economic significance of data that can be generated on the Internet — especially from social networks such as LinkedIn, Facebook, and Twitter. In one study, he measured the level of influence LinkedIn connections have in the job search process. In another, he examined sales rankings of smartphone applications in Apple’s iTunes App Store, calculating the extent to which an app’s daily download totals could be inferred from its overall ranking.
“There’s all this data out there, and companies don’t know what to do with that data,” he says. In an effort to remedy that problem, Garg is organizing a workshop on social media and business analytics in March.
Garg’s new post at McCombs is his first faculty position, but he’s no stranger to the world of higher education. He has earned five academic degrees: a bachelor’s from the Indian Institute of Technology; three Master’s (one from Carnegie Mellon University and two from the University of Southern California); and a Ph.D. from Carnegie Mellon.
His professional endeavors are equally diverse. In the last twelve years he has worked as a consultant, an electrical engineer, and a project manager, among other positions in a broad range of industries. Garg says his interests are constantly evolving with the emergence of new technologies, an instinct that has drawn him to projects involving robotics, artificial intelligence, virtual reality, product strategy, and more recently to economics of information.
“I love education, and I love learning,” he says. “There’s so much to do and so much to learn.”
Associate Professor of Finance John Hatfield comes to McCombs from Stanford University, where he served as an assistant professor of political economy after earning his Ph.D. there in 2005. Much of his work focuses on economics principles that also play a role in business and politics — many of which have an influence beyond the financial realm.
His research on market design, for instance, examines a variety of systems ranging from kidney donor exchange programs to processes for assigning students to public schools. “Not all markets have money in them,” he explains.
Hatfield also specializes in political economy, an area he describes as “the interactions of business and non-market forces — things like governments, legal bodies, and even activist groups or the media.” A political economist might analyze how the governments of two neighboring cities compete with each other to attract new residents and businesses. “How does that competition influence economic outcomes, citizen welfare, and so on?” Hatfield says.
In addition to doing research, he is currently teaching a class for undergraduates and MBA students called Economic Principles of Managerial Decision Making.
“The goal of the course is to learn how to think like an economist, and use that knowledge to be a better manager,” he says. The class also explores concepts from economics that relate to legal issues commonly encountered in business, such as antitrust regulations, intellectual property law, and trade law.
Another new member of the Finance department is Assistant Professor Tim Landvoigt, who shares a concurrent appointment with the department of Economics in the College of Liberal Arts. Landvoigt holds a Ph.D. in economics from Stanford University, but back in his native Germany, he initially studied computer science and worked as Java developer for SAP before deciding to return to academia.
Landvoigt’s research interests include asset pricing, macroeconomics, and housing finance — particularly related to mortgage securitization.
“My research is about the securitization of mortgage debt, which is when banks take loans and package and sell them so they don’t need to hold them on their balance sheets,” he says. “I also research how mortgage securitization affects the functioning of the banking system and whether or not it makes things more volatile.”
Landvoigt has developed simulations, or macro-models, that show if you give banks the option to package and sell loans to other investors, it is ultimately a good financial choice — as long as those loans are correctly priced and the buyers are fully aware of any risk.
“There is a research agenda of trying to integrate financial frictions and financial intermediaries in the banking sector into macroeconomic models, which is something we have to work on, because older macro-models didn’t incorporate any of that,” Landvoigt says. “They weren’t very useful for explaining the behavior of the economy, say, during the financial crisis. So this is one of the critical issues of macroeconomics right now.”
Landvoigt teaches Money and Capital Markets, an elective for upper division finance majors.
Assistant Professor Sinead Williamson joined the McCombs faculty last semester and holds dual appointments in the IROM department as well as the Division of Statistics and Scientific Computation in the College of Natural Sciences. She has a Master’s and Ph.D. in engineering from Oxford and Cambridge, respectively, and, before coming to UT, was a postdoctoral researcher at Carnegie Mellon University.
“I do probabilistic machine learning, which means trying to assign probabilities to different events,” she says. Williamson’s research specifically focuses on Bayesian nonparametric distributions, which means she develops algorithms responsible for making predictive computational models more adaptive and capable of growing in complexity when exposed to new data. These models can then be applied to anything from genomics and astrophysics to social media networks and e-commerce. For example, probabilistic machine learning enables Amazon’s systems to better predict what products you might like based on what items you hover over, what sits in your cart, or what you’ve purchased in the past.
When Williamson isn’t busy with research or teaching Statistics and Modeling to McCombs undergrads, she’s most likely to be found playing roller derby under the pseudonym Angela Momentum (a play on angular momentum, for the other engineers out there).
“I started skating during my Ph.D. in Cambridge, UK, and joined the Steel City Roller Derby in Pittsburgh while at Carnegie Mellon,” Williamson says. “This year I’m playing with the Texas Rollergirls Rec League and Team Free Radicals.”
And how do roller derbies and statistics mesh? “Actually, I am currently looking to combine statistics and roller derby,” she says. “I am working with a collaborator at Wharton on statistical models for predicting roller derby scores.”
As a new member of the IROM faculty, Assistant Professor Mingyuan Zhou approaches data analysis from the dual perspectives of an engineer and a statistician.
Zhou — who holds a Ph.D. from Duke University, a Master’s from the Chinese Academy of Sciences, and a bachelor’s degree from Nanjing University — says his research focuses on “the intersection between statistics and machine learning,” two fields that complement each other in several ways. “If you talk about machine learning, you also have to talk about statistics,” he says.
Zhou builds nonparametric Bayesian statistical models to reveal patterns in large volumes of data by “self-adjusting” as the sample size increases. His recent focus is to model “count matrices,” which can be used to help retailers understand consumer behavior by identifying patterns in customers’ shopping habits. They can also provide a framework to analyze gene sequencing data, allowing biotechnology researchers to pinpoint genes that may be causing a rare disease, even with a small sample size. That information can then be used to develop a cure that targets that specific gene.
Zhou also teaches the undergraduate course Statistics and Modeling, which he says has allowed him to observe the differences in perspective between business students and those with science backgrounds. To find a common ground, he puts statistics in terms that business students can easily relate to — such as the salary distribution of McCombs graduates.
-- Profiles by Adrienne Dawson, Rob Heidrick, and Jeremy Simon