M.A. candidate, Economics
Erik Drysdale - looking very smart!
Forecasting in the Financial World
by Sharday Mosurinjohn, July 2014
Erik Drysdale measures time in issues of the Economist, forecasts the future of Canada quarter by quarter, and ponders what he’d do if he knew the world were to end in the year 2100. Drysdale is an Economics Masters student in Queen’s intensive one-year coursework program. In his work as in different areas of his life, he operates on the principle of maximizing the value of his time.
Originally from Vancouver, Drysdale chose Queen’s for graduate work after earning an undergraduate degree in economics from Simon Fraser University because not only is Queen’s “one of the best four schools in Canada for economics,” but also because it made the most competitive offer and on account of its location in Kingston. “The weather is a little rough compared to the West Coast,” says Drysdale, but the community environment is a supportive one.
Drysdale’s major research paper has to do with something that can be about as tricky to forecast as the weather; forecasting variables important to the Canadian economy. Taking a combination of macro models used in central banks and international financial agencies—known as Dynamic Stochastic General Equilibrium (DSGE) and Vector autoregression (VAR) models—Drysdale is asking the question: how well do these models predict variables such as output or inflation one or two quarters ahead every quarter since 1998? The results yielded by such predictions are important because they feed into the decision making process about monetary policy.
Drysdale’s project compares four forecasting models referred to as a “Bayesian” DSGE (structured) model, a Bayesian VAR (unstructured) model, a non–Bayesian DSGE(structured), and a non–Bayesian VAR (unstructured). “Structure” refers to models that specify the nature of the relationship between two variables, “for instance, saying that the relationship between consumption and output is that output is the sum of consumption and investment.” Unstructured models name the variables at play, but don’t define how they’re related. “Bayesian” refers to a way of doing statistics that is only starting now to find use in economics, and this is where Drysdale’s research is cutting edge. (If you’re wondering about where the name comes from, Thomas Bayes was an 18th century English clergyman who made a huge contribution to the mathematics of probability.)
You may be most familiar with Bayesian statistics from the domain of machine learning—“the way Google figures out how to anticipate your interests based on each new search you do” is an example of Byesian stats at work. Using a Bayesian model allows economists to specify a value for one of the variables under investigation based on what economic theory predicts or based on previous research. This value is called a “prior” and it “prevents far off predictions by weighting future forecasts.” Like Google, economic models using Bayesian statistics are always updating their predictions based on new information—such as the economic data collected from quarter to quarter as time passes—so that they can come closer and closer to making accurate forecasts in real time.
Drysdale credits a number of pioneers in his field for making this cross-disciplinary work possible, especially for an MA student like him. “Usually this kind of statistics is only taught in PhD programs. I was fortunate that researchers had already done a comparison between structured and unstructured models and their forecasting abilities. And what’s more, I contacted the economists at the Bank of Canada whose paper I’m building on and they were kind enough to recommend a new software program that would help me do the calculations I need to.”
Already Drysdale has a job secured in Ottawa with the Bank of Canada. “Queen’s is good for attracting the Bank of Canada and other institutions. They come around early in the year and students send their resumes around. Then there is a first round of interviews, usually over Skype, usually in November followed by in–person interviews in January. By March or April you know if you have a job in some of these places.” What makes this possible is an extensive alumni network that supports soon-to-be grads in preparing for the interview process. “I talked to people hired in the previous round and I definitely want to give back in the same way next year.”
Drysdale will transition to his Bank of Canada job after the summer, but has ambitions to go back for his PhD eventually—that’s what research positions at the Bank and other financial institutions require. Nearing the end of his final semester, Drysdale is very busy but makes enough time to bond with his cohort over tennis and squash. His advice for other students in intensive programs comes from the perspective of someone who looks back in history for patterns—with an eye on the big picture, Drysdale urges: “be thankful for the extraordinary circumstances” that enabled you to be where you are and, indeed, to be at all.