Congratulations to Michelle Lewis, who was awarded the 2015 A. R. Bergstrom Prize in Econometrics for her paper “Forecasting with Macro-Finance Models: Applications to United States and New Zealand”. The Bergstrom Prize can be awarded every two years (although a three year gap ensued this time) and aims to reward the achievement of excellence in econometrics, as evidenced by a research paper in any area of econometrics.
Michelle Lewis’s Masters Thesis employs macro-finance models, which incorporate macroeconomic and timely financial market data, to forecast macroeconomic variables and the yield curve for New Zealand and the United States. The macro-finance models use the arbitrage-free Nelson-Siegel approach to represent yield curve data with just several components, and those components are combined with the macroeconomic variables of economic activity, inflation, and policy interest rates in a joint vector autoregression to produce forecasts.
The key contribution to the literature is that Michelle’s forecasting analysis is undertaken in a genuine real-time setting. That is, the model estimation and forecasts use the actual macroeconomic data that was available at each historical point in time, which realistically allows for an unavoidable uncertainty faced by practitioners. Conversely, the comparable literature to-date uses quasi-real-time macroeconomic data, which simply truncates the final available macroeconomic data series to estimate the model and produce forecasts over history. While showing promising forecasting benefits from macro-finance models, quasi-real-time analysis is unrealistic because it implicitly assumes that future revisions to historical macroeconomic data are already known at each historical point in time.
Fortunately, Michelle’s results show that, even in real time, there are still substantial forecasting benefits from using macro-finance models. The forecast improvements are most significant and robust for inflation and the policy rate, and economic activity for longer horizons. Furthermore, theoretically motivated restrictions on the yield curve dynamics improve the forecast performance of macroeconomic variables, and the yield curve itself.
However, for economic activity at short-term horizons, the forecasts from macro-finance models do not outperform forecasts from a standard vector autoregression of the macroeconomic variables. This result is at odds with the analogous quasi-real-time analysis, hence illustrating that quasi-real-time analysis can overstate the forecasting benefits of macro-finance models.
In their assessment, the adjudicators Professors Alfred Haug and Les Oxley noted: “The thesis is a substantial piece of empirical research that involved constructing new data and applying sophisticated econometric techniques that were skilfully mastered. Overall, it is an excellent piece of empirical econometrics. The author needs to be congratulated on her achievements.”