Skip to contentScholarGate
LibraryBookshelfDeskReview StudioAssistant
Sign in
On this page
IntuitionHow it worksWhen to use itStrengths & limitationsCommon pitfallsApplicationsFrequently asked🔒 Read the full methodSourcesRelated methods
Cite this pageSpotted an issue on this page? Report or suggest a fix →
Home›Econometrics›MIDAS Regression: Forecasting Across Mixed Data Frequencies
Regression modelForecasting

MIDAS Regression: Forecasting Across Mixed Data Frequencies

Mixed Data Sampling (MIDAS) Regression · Also known as: Mixed Frequency Regression, Mixed Data Sampling Model, High-Frequency Forecasting Regression, MIDAS Regresyonu

MIDAS (Mixed Data Sampling) Regression is an econometric framework that directly incorporates high-frequency predictors into models for lower-frequency outcome variables without requiring temporal aggregation of the regressors. Introduced by Eric Ghysels, Arthur Sinko, and Rossen Valkanov in 2007, MIDAS uses parsimoniously parameterized lag polynomials — such as the Beta or Exponential Almon weighting schemes — to summarize the information content of many high-frequency lags while avoiding parameter proliferation.

ScholarGate
  1. Regression model
  2. v1
  3. 1 Sources
  4. PUBLISHED
Cite this page →
Tools & resources
Download slides
Learn & explore

Read the full method

Members only

Sign in with a free account to read this section.

Sign in

Method map

The neighbourhood of related methods — select a node to explore.

MIDAS Regression
ARIMADynamic Factor ModelVAR Model

When to use it

MIDAS regression is appropriate when a low-frequency outcome (quarterly GDP, annual mortality rates) must be forecast using one or more high-frequency indicators (daily financial prices, weekly claims). It is most valuable in nowcasting and short-run forecasting contexts where timeliness of the high-frequency signal matters. The method assumes a stable linear relationship between the aggregated high-frequency signal and the outcome, stationarity (or cointegration), and a correctly specified lag length. When the number of high-frequency predictors is large, factor-augmented MIDAS variants should be considered. Alternatives include temporal aggregation followed by standard OLS, the unrestricted distributed-lag model (U-MIDAS) for small m, and state-space mixed-frequency models.

Strengths & limitations

Strengths
  • Uses high-frequency data directly, preserving intra-period timing information that aggregation would destroy.
  • Parsimonious weighting schemes (Beta, Exponential Almon) keep the parameter count small even with hundreds of lags.
  • Produces real-time nowcasts that can be updated as each new high-frequency observation arrives.
  • Extends naturally to multiple predictors, nonlinear weighting, and two-step factor-augmented specifications.
Limitations
  • Nonlinear estimation of the weighting parameters requires careful initialization and is sensitive to local minima.
  • Model selection (choice of weighting scheme, lag length m, and K) can substantially affect forecasting performance.
  • Assumes a linear conditional mean; does not accommodate regime changes or time-varying parameters without extensions.
  • With very large m (e.g., daily within annual), the unrestricted U-MIDAS benchmark becomes infeasible, making correct specification of the weight function critical.

Frequently asked

How does MIDAS differ from simply aggregating high-frequency data to the low frequency before running OLS?

Temporal aggregation (e.g., averaging daily data to monthly) imposes equal weights on every high-frequency observation within a period, discarding information about timing. MIDAS estimates the weights from the data, allowing, for instance, more recent observations within a period to receive higher weight if they are more informative about the outcome.

What is the difference between MIDAS and U-MIDAS?

U-MIDAS (Unrestricted MIDAS) places no constraint on the lag polynomial weights, estimating each freely by OLS. This is feasible only when the frequency ratio m is small (e.g., 3 for monthly-to-quarterly). When m is large, the unrestricted approach leads to parameter proliferation and poor forecasts, making the parsimonious MIDAS weighting schemes preferable.

Can MIDAS handle multiple high-frequency predictors?

Yes. The model extends straightforwardly to multiple regressors, each with its own weighting polynomial and lag length. When the predictor count is large relative to the sample size, dimension reduction via factor-augmented MIDAS — extracting a small number of common factors from the high-frequency panel — is recommended to avoid overfitting.

Sources

  1. Ghysels, E., Sinko, A., & Valkanov, R. (2007). MIDAS regressions: Further results and new directions. Econometric Reviews, 26(1), 53–90. DOI: 10.1080/07474930600972467 ↗

How to cite this page

ScholarGate. (2026, June 2). Mixed Data Sampling (MIDAS) Regression. ScholarGate. https://scholargate.app/en/econometrics/midas-regression

Related methods

ARIMADynamic Factor ModelVAR Model

Which method?

Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.

  • ARIMAEconometrics↔ compare
  • Dynamic Factor ModelEconometrics↔ compare
  • VAR ModelEconometrics↔ compare
Compare side by side →

Referenced by

Dynamic Factor Model

Similar methods

U-MIDASGARCH-MIDASDCC-MIDASTime-varying parameter WLSTime-varying parameter GLSDynamic Factor ModelTime-varying parameter OLSTime-varying Parameter Panel Data Analysis

Related reference concepts

EconometricsFinancial EconometricsMultilevel and Partial Pooling ModelsTime-Series Models • Dynamic Quantile Regressions • Dynamic Treatment Effect Models • Diffusion ProcessesPartial Least Squares RegressionEconometric Modeling

Spotted an issue on this page? Report or suggest a fix →

ScholarGate — MIDAS Regression (Mixed Data Sampling (MIDAS) Regression). Retrieved 2026-07-21 from https://scholargate.app/en/econometrics/midas-regression · Dataset: https://doi.org/10.5281/zenodo.20539026
Quick facts
Originator
Eric Ghysels, Arthur Sinko & Rossen Valkanov
Year
2007
Type
Parametric mixed-frequency forecasting model
Subfamily
Forecasting
Weighting Schemes
Almon, Beta, Exponential Almon, Unrestricted
Data Requirement
At least two time series sampled at different frequencies
Related methods
ARIMADynamic Factor ModelVAR Model
ScholarGate

A content-first reference library for research methods — what each one is, how it works, and where it comes from.

Open data (CC-BY)

Explore

  • Library
  • Search the library…
  • Browse by field
  • Fields
  • Journey
  • Compare
  • Which method?

Reference

  • Subjects
  • Atlas
  • Glossary
  • Methodology
  • Philosophy

Your tools

  • Bookshelf
  • Desk
  • Chat

Company

  • About
  • Pricing
  • Contact
  • Suggest a method

Entries are compiled from published sources for reference. Verifying the accuracy and suitability of any information for your own use remains your responsibility.

© 2026 ScholarGate · A research-method reference library
  • Privacy
  • Cookies
  • Terms
  • Delete account