It’s bizarre to go to a university where it’s practically a given that your classmates will your mind when they tell you about their summer. This could be daunting, as not all of us have the resources for a big internship or trip around the world. However, you don’t need to travel to have a story worth sharing, a fact that seventeen Wesleyan students took advantage of this summer. The QAC’s Summer Apprenticeship is a program in which students partner with a faculty mentor to work on a data-based research project. I spoke to a couple of participants and asked them to tell me about their work.
Frederick Corpuz ’20 worked with Professor Boulware of the economics department studying decoupling in emerging Asian markets. Decoupling is an economics theory looking at potential deviation from association that might occur between relevant factors For instance, a lot of exchange rates tend to be highly correlated — but what decoupling suggests is an absence in such a correlation can be just as important as its presence.
“[We’re trying to see] if the world economies are still being driven by the developed countries or emergent countries,” Frederick explained. “Specifically, we were looking at how much of an impact the emerging Asian markets – so this is Thailand, Malaysia, Philippines, Singapore, etc. – had on the world economy, specifically during the time period of the Asian Financial Crisis.”
The aforementioned crisis refers to the huge regional market destabilization that occurred in 1998, beginning with devaluation of the Thai currency, the baht. A successful decoupling of currencies would suggest that these markets are becoming less dependent on Western economies. Performing time series tests on a period of about 15 years, Frederick aimed to see if the stop indices of the aforementioned countries pointed to the presence of similar economic trends between the nations over time.
The results? There was some evidence of co-integration, supporting this notion — but problem was a little more complicated than initially supposed.
“Late in the game, we realized that we didn’t take into account big shocks like the 2008 crisis…. Since everyone is in a crisis, everything just falls.” According to Frederick, this led to breaks within the data set that resulted in less reliable calculations. More exposure to statistical knowledge was necessary to be able to adapt to this kind of challenge. Rather than viewing this as a setback, Frederick saw an opportunity, and said that one of the biggest takeaways from this project for him was gaining agency in being able to face such daunting tasks. “It was just the comfort level of knowing I can do this,” he said, “and if I can’t do it right now it will take me a week to learn it.”
Kelly Jamrog ‘19, who did research with economics professor Abigail Hornstein, faced a different problem. Kelly chased after the question of fluctuations in mutual funds. “Traditionally the board of directors would be like old white retired men — that’s just the overarching matter. And frequently they’re retired CEOs, or have some kind of personal history with corporations. The point was just to see — were funds more inclined to invest in companies they had prior involvement with?”
Kelly’s job consisted of two tasks: filling in a gap in a data set documenting members of corporate boards between 2007 and 2014, and joining this information with purchased data on board members’ employment histories. In our conversation, Kelly expressed a lot of enthusiasm for the later data manipulation and the thrill of actively working with data. In a document transcribed for many years by a huge variety of students, Kelly pointed to the challenge in a lack of standardization in data transcription. “The example we were talking about was like, John Doe, John J. Doe, J. Doe — do [those transcribing] write their middle names, do abbreviations?” She solved the problem via fuzzy matching, a technique involving the calculating the similarity between two items and providing a confidence interval evaluated by the researcher. Along with a couple of QAC professors, she turned to Fril, an open-source, point-and-click program that made the join a simple task.
The initial data collection for this merge, however, at times felt more like manual labor than data science.
“It was pretty brutal, I’m not going to lie,” she laughed, recounting finding all of the information and transcribing it by hand. But deep in the throes of her work, she stumbled upon something pretty cool.
According to Kelly, the understanding of all the researchers, assistants and professors, was that boards don’t really change unless someone ages out of the company and is forced to retire, or dies. This was the basis of a lot of previous research. She, however, found this to be false.
“In the end I believe that the data set I was working on has the potential to go to like four different papers. A lot of them are like work in progress, contingent on the data set being completed and is still being collected to this day and has been in the process of being collected for years.” Still, there’s a potential that a summer’s work could have huge ramifications.
Information on Asian Financial Crisis found on: http://www.investopedia.com/terms/a/asian-financial-crisis.asp
Information on Decoupling found on: http://www.investopedia.com/terms/d/decoupling.asp