John Guerard, Jr., Ph.D. recently retired as Director of Quantitative Research at McKinley Capital Management, LLC, a systematic global growth asset management firm based in Anchorage, AK. He now serves as Chairman of the firm’s Scientific Advisory Board which contains 10 industry thought leaders who educate and advise McKinley’s investment team in areas including portfolio optimization and construction, computational finance, AI & machine learning, transaction cost analysis, risk modeling and attribution, sell side and non-sell side analysis, big data and data mining, and healthcare transformation. The Board was founded in 2012 and originally chaired by Harry Markowitz, Ph.D., CFA, Nobel Laureate, and author of Modern Portfolio Theory. John worked Markowitz at Daiwa Securities, in Jersey City, New Jersey, in the early-1990s. John earned his undergraduate degree in Economics from Duke University, master’s degrees from The University of Virginia (Economics) and Georgia Tech (Industrial Management), and Ph. D. in Finance from The University of Texas at Austin. He taught in the Computational Finance Program at the University of Washington and serves an Industrial Affiliate at Georgia Institute of Technology.
How did you become a forecaster?
I studied Econometrics with Rick Ashley at Texas, and we modeled Tax Revenues for the State of Texas using a transfer function. Rick had just published his 1980 Econometrica piece with Clive Granger on transfer functions and causality testing. I thought that the Ashley, Granger, and Schmalensee piece presented a great methodology that is still most useful. While I was finishing my Ph.D. at Texas, I received an offer to teach at The McIntire Scholl of Commerce at The University of Virginia. I taught three courses a term while finishing my dissertation in one year. I only had a one-year contract, and I was required to defend my dissertation in August, which I did. I moved to Lehigh University and taught finance for six years. While I was at Lehigh, I received a grant from The Institute for Quantitative Research in Finance (“The Q-Group”). That grant was the start of a 30-year love affair with forecasting corporate earnings per share (EPS) and testing for market inefficiencies. I left Lehigh and moved to Wall Street in Chicago (for Drexel Burnham Lambert), Jersey City (Daiwa Securities), and Anchorage (McKinley Capital).
What areas of forecasting interest you?
I initially estimated ARIMA models of EPS to compare and combine with analysts’ forecasts. Given that I was forecasting EPS for 600 US firms, I was very interested in automatic time series forecasting, using AutoBox in the mid-1980s and early 1990s. I later used to SCA and OxMetrics to estimate models with outliers, structural breaks, particularly with transfer functions. While at Drexel Burnham, I completed two forecasting projects with Robert (Bob) Clemen, of Oregon and later Duke. Bob and I published a comparison of an ARIMA forecast with the major econometric forecasting services (Wharton, Chase, DRI, and BEA) for US GNP, which appeared in the IJF in 1989. We reported that the ARIMA model forecast, and four econometric forecasts equaled about 1.5 independent forecasts. I was invited to join the board in 1991 by Robert Fildes. I believe that I was one of two practitioners as AEs (Victor Zarnowitz) was the other until his death. In industry, I was forecasting company EPS revisions and breadth (directions on EPS revisions), for Markowitz and later for McKinley Capital, to do produce better, more statistically significant portfolio selection. The IJF Special Issue in 2015 produced by Kajal Lahiri and me reported the use of our EPS forecast work in alternative risk models of Portfolio Selection. We referred to this applied investment research as the McKinley Capital “Horse Race”. I always enjoyed composite forecasting research for company EPS and GDP forecasting. I still believe that practitioner research into EPS forecasting and GDP forecasting, using the Leading Economic Indicators (LEI) is still of great value. Regression and outlier modeling yielded better results for stock selection modeling for portfolio selection than equal weighted models.
What has the International Journal of Forecasting meant to you?
First, I enjoyed and appreciated the composite forecasting literature from the time of Bates and Granger to the present time. New data and forecasting techniques are advanced and yet equal weighting still wins in many, if not most, cases. I have continued to learn from Spyros Makridakis, Robert Fildes, David Hendry, and many IJF AEs and Editors on modeling and testing. I sincerely miss several AEs, including Larry Brown, Victor Zarnowitz, and Herman Stekler. I have dined with Larry in Atlanta when I was at Georgia Tech recruiting quantitative analysts to work at McKinley Capital during the past 15 years. Larry was a great AE at the IJF, and Victor and Herman were truly legends in our field. What am I most proud of with the IJF and ISFs? Bringing Harry Markowitz via “early-Zoom” to the ISF at Nice in 2008 and the 1998 (with Larry) and 2015 (Horse Race) IJF special issues.
COVID and forecasting
The past 27 months of COVID were a very interesting time in financial markets as well as the real economy. During the past 27 months, starting with my initial retirement as Director of Quantitative Research and transition to Chairman of the Scientific Advisory Board at McKinley Capital in June 2020, I was selling a house in Alaska and buying a home in Bluffton, South Carolina, near Hilton Head Island and the Savannah Airport. I completed a Palgrave book on the history and forecasting results of the LEI and accompanying podcast for The Conference Board. My book particularly honored Victor Zarnowitz for his business cycle and LEI research. The outlier and structural break modeling in SAS, SCA, and OxMetrics were particularly useful in the Palgrave book, particularly during COVID. What huge outliers! I finished the third edition of my Springer book on quantitative corporate finance that also had chapters on regression and time series modeling and forecasting. I wrote five hours every morning, from 6-11, worked on house-selling clean-up and later house-hunting and buying details in the afternoon, and returned to the books in the evenings after dinner. It made for very long days, but my academic reviewers were pleased with the books. I expect to continue to use them for classes and workshops at Washington, Georgia Tech, and Rutgers in the coming years. In industry, the consultants and McKinley Capital clients loved to see the IJF Special Issues and the Springer books to demonstrate expertise and modeling statistical significance. All in all, I enjoyed the vast majority of my 30 years as an IJF AE. Upon my retirement, I fully expect the IJF review time to dramatically decrease.
