Forecasting Inflation in a Data-Rich Environment: The Benefits of Machine Learning Methods

Abstract

Inflation forecasting is an important but difficult task. Here, we explore advances in machine learning (ML) methods and the availability of new datasets to forecast US inflation. Despite the skepticism in the previous literature, we show that ML models with a large number of covariates are systematically more accurate than the benchmarks. The ML method that deserves more attention is the random forest model, which dominates all other models. Its good performance is due not only to its specific method of variable selection but also the potential nonlinearities between past key macroeconomic variables and inflation. ResumenProyectar la inflación es una tarea importante pero difícil. Aquí, exploramos los avances de los métodos de machine learning (ML) y la disponibilidad de nuevas bases de datos para proyectar la inflación de EE.UU. A pesar del escepticismo en la literatura previa, demostramos que los modelos de ML con un amplio número de covariables son sistemáticamente más precisos que los modelos referenciales. El método de ML que merece más atención es el modelo random forest, el cual domina a todos los otros modelos. Su buen desempeño se debe no solo al método específico de selección de variables, sino que también a las potenciales no linealidades entre variables macroeconómicas clave y la inflación. We are very grateful to the coeditor, Christian B. Hansen; the associate editor; and two anonymous referees for very helpful comments. We thank Federico Bandi, Anders B. Kock and Michael Wolf for comments during the workshop "Trends in Econometrics:

FRED-QD: A Quarterly Database for Macroeconomic Research

“…In contrast, we find that when forecasting nominal price series, forecast accuracy is typically better when using factors estimated using the unit root-based codes. This result coincides with evidence provided by Medeiros et al (2019) and Coulombe et al (2019), who find that treating price inflation as I(0) leads to better forecasts of inflation than treating it as I(1) -which is precisely what the benchmark transformation codes recommend.…”

“…FRED-MD has been successful. It has been used as a foil for applying big data methods including random subspace methods (Boot and Nibberin, 2019), sufficient dimension reduction (Barbarino and Bura, 2017), dynamic factor models (Stock and Watson, 2016), large Bayesian VARs (Giannone, Lenza, and Primiceri, 2018), various lasso-type regressions (Smeekes and Wijler, 2018), functional principal components, (Hu and Park, 2017), complete subset regression (Kotchoni, Lerous, and Stevanovich, 2019), and random forests (Medeiros, Vasconcelos, Veiga, and Zilberman, 2019). In addition, these various methods have been used to study a wide variety of economic and financial topics including bond risk premia (Bauer and Hamilton, 2017), the presence of real and financial tail risk (Nicolò and Lucchetta, 2016), liquidity shocks (Ellington, Florackis, and Milas, 2017), recession forecasting (Davig and Hall, 2019), identification of uncertainty shocks (Angelini, Bacchiocchi, Caggiano, and Fanelli, 2019), and identification of monetary policy shocks (Miranda-Agrippino and Ricco, 2017).…”

Googling Unemployment During the Pandemic: Inference and Nowcast Using Search Data

“…As an example, we retrieved 178 keywords in Italy compared with 63 data-points in the unemployment rate monthly time series. Medeiros et al (2019) explore the performance of different machine learning methods in a forecast race targeted at predicting US in ation using a wide set of covariates. The authors show that, in a data rich environment, the Random Forest algorithm outperforms all the considered alternative high-dimensional models (including the linear LASSO and RIDGE regressions), as well as state-of-the-art dynamic factor models widely used in time series modelling.…”

“…This information can then be used for interpretation purposes. As an example, Medeiros et al (2019) use Random Forest variable importance to show that one possible explanation for the better performance of the algorithm is its ability to capture the importance of predictors which are neglected by other linear and non-linear methods.…”

Macroeconomic forecasting using factor models and machine learning: an application to Japan

“…Nakajima and Sueishi (2019) employed a lasso-VAR model for forecasting Japanese macroeconomic series. The usefulness of the ensemble machine learning based on regression trees, such as random forests, was emphasized in Medeiros et al (2019) and Chen et al (2019). Bai and Ng (2009) examined the e¤ectiveness of boosting in forecasting in ‡ation, interest rate, industrial production, employment and the unemployment rate by using a large set of U.S. macroeconomic data.…”