The three-pass regression filter

A new approach to forecasting using many predictors

Bryan Kelly, Seth Pruitt

Research output: Contribution to journalArticle

38 Citations (Scopus)

Abstract

We forecast a single time series using many predictor variables with a new estimator called the three-pass regression filter (3PRF). It is calculated in closed form and conveniently represented as a set of ordinary least squares regressions. 3PRF forecasts are consistent for the infeasible best forecast when both the time dimension and cross section dimension become large. This requires specifying only the number of relevant factors driving the forecast target, regardless of the total number of common factors driving the cross section of predictors. The 3PRF is a constrained least squares estimator and reduces to partial least squares as a special case. Simulation evidence confirms the 3PRF's forecasting performance relative to alternatives. We explore two empirical applications: Forecasting macroeconomic aggregates with a large panel of economic indices, and forecasting stock market returns with price-dividend ratios of stock portfolios.

Original languageEnglish (US)
Pages (from-to)294-316
Number of pages23
JournalJournal of Econometrics
Volume186
Issue number2
DOIs
StatePublished - Jun 1 2015

Fingerprint

Forecast
Forecasting
Predictors
Regression
Filter
Cross section
Constrained Least Squares
Common factor
Least Squares Regression
Ordinary Least Squares
Dividend
Time series
Partial Least Squares
Least Squares Estimator
Macroeconomics
Stock Market
Economics
Closed-form
Estimator
Target

Keywords

  • Constrained least squares
  • Factor model
  • Forecast
  • Partial least squares
  • Principal components

ASJC Scopus subject areas

  • Economics and Econometrics
  • Applied Mathematics
  • History and Philosophy of Science

Cite this

The three-pass regression filter : A new approach to forecasting using many predictors. / Kelly, Bryan; Pruitt, Seth.

In: Journal of Econometrics, Vol. 186, No. 2, 01.06.2015, p. 294-316.

Research output: Contribution to journalArticle

@article{943582dbb8fa4a6e8b1bdd8126d3d9c9,
title = "The three-pass regression filter: A new approach to forecasting using many predictors",
abstract = "We forecast a single time series using many predictor variables with a new estimator called the three-pass regression filter (3PRF). It is calculated in closed form and conveniently represented as a set of ordinary least squares regressions. 3PRF forecasts are consistent for the infeasible best forecast when both the time dimension and cross section dimension become large. This requires specifying only the number of relevant factors driving the forecast target, regardless of the total number of common factors driving the cross section of predictors. The 3PRF is a constrained least squares estimator and reduces to partial least squares as a special case. Simulation evidence confirms the 3PRF's forecasting performance relative to alternatives. We explore two empirical applications: Forecasting macroeconomic aggregates with a large panel of economic indices, and forecasting stock market returns with price-dividend ratios of stock portfolios.",
keywords = "Constrained least squares, Factor model, Forecast, Partial least squares, Principal components",
author = "Bryan Kelly and Seth Pruitt",
year = "2015",
month = "6",
day = "1",
doi = "10.1016/j.jeconom.2015.02.011",
language = "English (US)",
volume = "186",
pages = "294--316",
journal = "Journal of Econometrics",
issn = "0304-4076",
publisher = "Elsevier BV",
number = "2",

}

TY - JOUR

T1 - The three-pass regression filter

T2 - A new approach to forecasting using many predictors

AU - Kelly, Bryan

AU - Pruitt, Seth

PY - 2015/6/1

Y1 - 2015/6/1

N2 - We forecast a single time series using many predictor variables with a new estimator called the three-pass regression filter (3PRF). It is calculated in closed form and conveniently represented as a set of ordinary least squares regressions. 3PRF forecasts are consistent for the infeasible best forecast when both the time dimension and cross section dimension become large. This requires specifying only the number of relevant factors driving the forecast target, regardless of the total number of common factors driving the cross section of predictors. The 3PRF is a constrained least squares estimator and reduces to partial least squares as a special case. Simulation evidence confirms the 3PRF's forecasting performance relative to alternatives. We explore two empirical applications: Forecasting macroeconomic aggregates with a large panel of economic indices, and forecasting stock market returns with price-dividend ratios of stock portfolios.

AB - We forecast a single time series using many predictor variables with a new estimator called the three-pass regression filter (3PRF). It is calculated in closed form and conveniently represented as a set of ordinary least squares regressions. 3PRF forecasts are consistent for the infeasible best forecast when both the time dimension and cross section dimension become large. This requires specifying only the number of relevant factors driving the forecast target, regardless of the total number of common factors driving the cross section of predictors. The 3PRF is a constrained least squares estimator and reduces to partial least squares as a special case. Simulation evidence confirms the 3PRF's forecasting performance relative to alternatives. We explore two empirical applications: Forecasting macroeconomic aggregates with a large panel of economic indices, and forecasting stock market returns with price-dividend ratios of stock portfolios.

KW - Constrained least squares

KW - Factor model

KW - Forecast

KW - Partial least squares

KW - Principal components

UR - http://www.scopus.com/inward/record.url?scp=84929263289&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84929263289&partnerID=8YFLogxK

U2 - 10.1016/j.jeconom.2015.02.011

DO - 10.1016/j.jeconom.2015.02.011

M3 - Article

VL - 186

SP - 294

EP - 316

JO - Journal of Econometrics

JF - Journal of Econometrics

SN - 0304-4076

IS - 2

ER -