Dynamic Models for Predicting a Changing World
If models serve as the vehicles of economic analysis, data is where the rubber meets the road. And just as roads can change in terrain over a long drive, so too can data—an important reality for econometricians, according to Robert Engle. Dynamic models of things like risk must hold up to data that have yet to be collected if they are expected to reliably predict shifts in the economy over the long term.
Engle, a 2003 Nobel laureate in Economics, learned that lesson many years ago from celebrated econometrician Ta-Chung Liu, his former thesis advisor. Now he is sharing his macroeconomic modeling expertise as the institute’s inaugural Ta-Chung Liu Distinguished Visitor.
The post bringing Engle to the institute in April 2014 honors Liu’s contributions to econometric theory, supporting future generations of economists that advance his data-driven approach to modeling causality in the macroeconomy.
Engle says his mentor taught him to prize empirical models built using the freshest data available. “Liu had a wonderful blend of econometric theory and a pragmatic approach to model building,” recalls Engle.“I continue to use that mixture.”
Among his many contributions to the field of econometrics, Liu pioneered the use of high-frequency macroeconomic data in econometric analysis. He moved modeling from annual data to using quarterly, monthly, and eventually weekly data from different economic sectors. That wealth of information allowed Liu to build the only monthly econometric model of the U.S. in existence at the time of his death in 1975.
The trend Liu forged toward richer, more frequent data-gathering only accelerated. “Now in financial markets, we can go to daily data or even intra-daily data to try to see which variables move first, which ones are having an impact on the other ones,” says Engle.
The Ta-Chung Liu Distinguished Visitor is supported by a generous gift from Ernest and Joan Liu, who sought to honor Ta-Chung Liu’s memory by supporting econometricians who will advance his legacy of data-driven analysis. “The institute takes a strong role in the development of econometricians,” says his son Ernest Liu. “Perhaps Liu’s focus on [joining] the practical world and the theoretical field may inspire young econometricians to do the same.”
High-frequency data analysis is essential in Engle’s work—especially in financial analysis of systemic risk which is observable within the time-varying relationships between financial firms and the global economy. “The econometric questions that I’m dealing with relate to different ways of measuring multivariate models that are continually changing,” says Engle.
The pace of change within those relationships is lightning-quick, and Engle says that means the models have to be re-estimated and re-evaluated to reflect what new data tells us. To reflect that pace, Engle publishes his models and the literature supporting them online for public viewing, logging updates and adjustments over time.
“That’s a much more demanding standard for what a good model is–it not only has to work once, but every time, on data that you haven’t even seen yet,” he says. But the result is not only a rigorously validated model, but also a record of what’s happening in the economy over time that serves as a valuable resource for regulators, policy makers and academics alike.
This new research approach reflects the innovative thinking required to structure valid investigation into the role of risk within the economy. “Systemic risk is interesting as a topic; there was basically no research on it five years ago,” says Engle. “The financial crisis hit, and now many academics are working on it, many practitioners are working on it and many government agencies are tasked with making progress on it. It’s an amazing proliferation of research, which kind of goes all over the place. What is the best way of coordinating it?”
Coordinating with policymakers was one of Liu’s great accomplishments. According to his son, he took great pride in leveraging findings from his academic work to motivate economic growth in his home country of Taiwan. “He considered as his greatest accomplishment to be aiding the Taiwan government in turning from an agrarian economy to one that concentrated on technology industries,” says the younger Liu. A comprehensive tax reform package architected by Liu brought the transparency and equity in enforcement required for industry to thrive in the 1960s.
Engle says that his former professor understood that the hard questions for legislators are ones that economists are uniquely positioned to answer. For Liu, it was in aiding with the transition to a modern economy; for Engle and his econometrician colleagues, the hard questions relate to how, in the light of the 2008 financial crisis, another financial meltdown can be averted. “Academics and regulators are continually trying to figure out what the costs are of restricting the free flow of innovation and operation of banks compared to the benefits of stability,” says Engle.
Policymakers today debate macroprudential tools–policy instruments like capital buffer requirements–as an answer to this question, but economists like Engle have a vital role to play in providing dynamic models that can accurately predict when such tools should be used, and what the impact of their use could possibly be.
Engle says that the unknowns represent another set of questions for he and his colleagues to examine. “There’s a lot of discussion about it among regulators, but as economists, we should have something useful to add.”
That combination of pragmatism and public service that marked Liu’s storied career is reflected in the work that Engle is advancing today. “We do research because we want to help all sorts of people better understand how pieces of the economy fit together.”