跳到主要内容

Macro by Mark 的 Cookie.我们使用必要 Cookie 处理登录、语言和安全。经您同意,我们还会将可选 Cookie 用于保存偏好、分析,以及通过 Meta(Facebook)Pixel 衡量广告效果。请阅读我们的Cookie 政策。

Macro by MarkMacro by Mark
  • 首页
加载中…
加载中…
← 模型
Macro概念政策模型

MIDAS Model

Mixed-data sampling regressions that handle indicators arriving at different frequencies inside a single forecasting equation.

概述图表数学参考资料

How do you use high-frequency data to predict a low-frequency target without first aggregating to the lower frequency?

Background

Bridge equations aggregate monthly indicators to the quarterly level before regression, discarding within-quarter timing information. If month-1 industrial production is a stronger predictor of quarterly GDP than month-3, the bridge equation cannot detect this because it averages all three months. Ghysels, Santa-Clara, and Valkanov (2004, 2006) introduced Mixed-Data Sampling (MIDAS) regression to solve exactly this problem. MIDAS regresses the low-frequency target (quarterly GDP) directly on high-frequency lags (monthly indicators), keeping the original monthly observations intact. A parsimonious weighting function constrains the monthly lag coefficients so the model does not explode in parameters.

The core idea: instead of kkk free coefficients on kkk monthly lags, MIDAS parameterizes the lag profile with a smooth weighting function controlled by 2-3 hyperparameters. The exponential Almon lag polynomial is the most popular: w(k;θ)=exp⁡(θ1k+θ2k2)/∑j=0Kexp⁡(θ1j+θ2j2)w(k; \theta) = \exp(\theta_1 k + \theta_2 k^2) / \sum_{j=0}^{K} \exp(\theta_1 j + \theta_2 j^2)w(k;θ)=exp(θ1​k+θ2​k2)/∑j=0K​exp(θ1​j+θ2​j2). This function can represent declining weights (recent months matter more), hump-shaped weights (middle months most informative), or flat weights (equal weighting, which collapses to bridge equation aggregation). The weighting shape is estimated from the data, not imposed a priori.

Foroni, Marcellino, and Schumacher (2015) introduced Unrestricted MIDAS (U-MIDAS), which drops the weighting function and estimates each monthly lag coefficient freely by OLS. U-MIDAS is valid when the number of high-frequency lags is small relative to the sample size (e.g., 3 monthly lags for quarterly prediction). When the frequency mismatch is large (daily to monthly, for instance), the restricted MIDAS with a parameterized weighting function remains necessary.

MIDAS has been extended in multiple directions. MIDAS-AR adds autoregressive lags of the low-frequency target. Markov-switching MIDAS (Guerin and Marcellino 2013) allows regime-dependent weights. Factor MIDAS combines factor extraction from a large monthly panel with MIDAS weighting for the quarterly target. Bayesian MIDAS (Rodriguez and Puggioni 2010) places priors on the weighting function parameters. ECB and Bundesbank staff nowcasting exercises have used MIDAS-type regressions for euro area and German GDP.

Connections to explore

Math ↗The structure and its conditions, step by step.References ↗Academic sources and guidance for using the model.

How the Parts Fit Together

ytLy_t^LytL​

Low-frequency target

The variable to forecast, observed at the lower frequency (quarterly). Indexed by t at the quarterly level.

Inputs come at two frequencies. The low-frequency target ytLy_t^LytL​ (quarterly GDP growth, indexed at the quarterly level) and the high-frequency predictor xτHx_{\tau}^HxτH​ (monthly industrial production, indexed at the monthly level). The frequency ratio mmm links the two: for monthly-to-quarterly, m=3m = 3m=3. Each quarterly observation ytLy_t^LytL​ corresponds to m=3m = 3m=3 monthly observations xtmH,xtm−1H,xtm−2Hx_{tm}^H, x_{tm-1}^H, x_{tm-2}^HxtmH​,xtm−1H​,xtm−2H​ within the quarter, plus additional monthly lags from prior quarters.

The model regresses the quarterly target on a weighted sum of monthly lags: ytL=α+β∑k=0Kw(k;θ)xtm−kH+εty_t^L = \alpha + \beta \sum_{k=0}^{K} w(k; \theta) x_{tm-k}^H + \varepsilon_tytL​=α+β∑k=0K​w(k;θ)xtm−kH​+εt​. The weighting function w(k;θ)w(k; \theta)w(k;θ) assigns different importance to different monthly lags. KKK is the total number of high-frequency lags included (e.g., K=11K = 11K=11 means 4 quarters of monthly lags). The key parameter is θ=(θ1,θ2)\theta = (\theta_1, \theta_2)θ=(θ1​,θ2​), which controls the shape of the weight function. The coefficient β\betaβ scales the overall effect.

Estimation is nonlinear least squares (NLS) when the weighting function is nonlinear in θ\thetaθ (exponential Almon, beta polynomial). The objective function is the sum of squared residuals with respect to (α,β,θ)(\alpha, \beta, \theta)(α,β,θ). NLS requires starting values; a grid search over θ\thetaθ followed by numerical optimization is standard. For U-MIDAS (no weighting function), estimation is OLS. Model selection involves choosing the number of high-frequency lags KKK, the weighting function family, and whether to include autoregressive terms.

Assumptions

Testable

Correct weighting function family

The exponential Almon, beta polynomial, or chosen functional form can approximate the true lag profile between the high-frequency predictor and the low-frequency target.

If violated:If the true lag profile is non-smooth (e.g., only month-2 matters, month-1 and month-3 do not), a smooth weighting function forces weight onto irrelevant lags. U-MIDAS avoids this but requires more data.

Testable

Stationarity of both series

Both the low-frequency target and the high-frequency predictor are covariance stationary after appropriate transformations.

If violated:Nonstationary variables produce spurious MIDAS regressions. Cointegrating MIDAS exists (Miller 2014) but is not standard.

Testable

Temporal alignment

The mapping between high-frequency periods and low-frequency periods is correct. Each quarterly observation ytLy_t^LytL​ corresponds to the correct set of monthly observations xtmHx_{tm}^HxtmH​ through xtm−KHx_{tm-K}^Hxtm−KH​.

If violated:Misalignment (off-by-one-month errors) shifts the estimated lag profile and produces biased weight estimates. Common with end-of-quarter versus within-quarter timing conventions.

Testable

Exogeneity of the high-frequency predictor

The monthly predictor is predetermined relative to the quarterly target. No feedback from ytLy_t^LytL​ to contemporaneous xτHx_{\tau}^HxτH​.

If violated:Reverse causality (GDP growth affects monthly indicators within the same quarter) biases the MIDAS coefficient. Restricting the regression to predetermined (lagged) high-frequency observations or using instrumental-variable estimation addresses this.

Testable

Stable relationship over time

The weighting function parameters θ\thetaθ and the slope β\betaβ are constant over the sample period.

If violated:Structural breaks in the relationship between the high-frequency predictor and the quarterly target cause the estimated weights to be a compromise between two regimes. Regime-switching MIDAS handles this.

Testable

Limited serial correlation in errors

The low-frequency regression errors εt\varepsilon_tεt​ are serially uncorrelated or at most weakly dependent.

If violated:Serial correlation in εt\varepsilon_tεt​ indicates omitted low-frequency dynamics. Adding autoregressive terms (MIDAS-AR) is the standard fix.

Applications

MIDAS regressions are a standard tool in institutional nowcasting suites: ECB and Bundesbank staff have used MIDAS-type regressions to nowcast euro area and German GDP from monthly indicators such as industrial production, orders, and survey data, benchmarking them against bridge equations and factor models. The exponential Almon weighting function determines whether early-month or late-month data is more informative. The nowcast updates each time a monthly data release occurs, leveraging the within-quarter timing that bridge equations discard.

Financial applications use daily-to-monthly MIDAS to forecast monthly volatility from daily returns. Ghysels, Santa-Clara, and Valkanov (2006) showed that MIDAS volatility forecasts using daily realized variance outperform GARCH models at monthly horizons. The frequency ratio is m=22m = 22m=22 (trading days per month), and the weighting function determines how quickly past daily information decays. This is the original application that motivated the MIDAS framework.

The OECD's composite leading indicators use a MIDAS-inspired approach to combine monthly financial and survey indicators with quarterly real activity. Marcellino and Schumacher (2010) applied factor MIDAS to German GDP nowcasting, extracting monthly factors from a panel of 100+ indicators and passing them through a MIDAS regression targeting quarterly GDP. The factor-MIDAS combination handles both the large-panel dimension reduction and the mixed-frequency aggregation.

MIDAS regressions fail when the weighting function is misspecified and the sample is too short to detect the misspecification. With K=12K = 12K=12 monthly lags and a 2-parameter weighting function, the NLS optimization may converge to local minima. Starting-value sensitivity is a known issue. The restricted MIDAS also struggles when the true lag profile is discontinuous or multi-modal - situations where U-MIDAS with free coefficients is preferable if the data permits.

Extensions and open questions

Variants

  • Exponential Almon MIDAS: the standard specification. Weights are parameterized as w(k)=exp⁡(θ1k+θ2k2)/∑j=0Kexp⁡(θ1j+θ2j2)w(k) = \exp(\theta_1 k + \theta_2 k^2) / \sum_{j=0}^{K} \exp(\theta_1 j + \theta_2 j^2)w(k)=exp(θ1​k+θ2​k2)/∑j=0K​exp(θ1​j+θ2​j2). Two parameters control the lag profile shape.
  • Beta polynomial MIDAS: weights based on the Beta distribution density. Can produce U-shaped or J-shaped lag profiles. Two parameters (a,b)(a, b)(a,b) with flexible shapes.
  • U-MIDAS (unrestricted): no weighting function. Each monthly lag gets its own OLS coefficient. Works when K is small relative to T.
  • MIDAS-AR: augments the MIDAS regression with autoregressive lags of the low-frequency target. Handles serial correlation in the forecast error.
  • Factor MIDAS: extract factors from a large monthly panel, then use factors as high-frequency predictors in a MIDAS regression targeting the quarterly variable.
  • Markov-switching MIDAS (Guerin and Marcellino 2013): the weighting function parameters switch between regimes, allowing recession vs. expansion dynamics.

Open questions

  • Optimal choice of weighting function: exponential Almon and beta polynomial are the most common, but there is no theoretical basis for preferring one over the other. Data-driven model selection (AIC/BIC on the NLS objective) is the practical approach.
  • Real-time versus pseudo-real-time evaluation: most MIDAS evaluations use final-release data with timing mimicking real-time availability. True real-time vintage data may tell a different story, especially for heavily revised indicators.
  • MIDAS with many high-frequency predictors: extending MIDAS to more than 3 or 4 monthly predictors creates a high-dimensional NLS problem. Penalized MIDAS (LASSO-MIDAS) and factor MIDAS address this differently. No consensus on which dominates.

Keep going

Read next

Continue with a related concept, model, or example.

Theory model

IS-LM Model

a well-connected theoretical model two steps from here

New angle: theoretical model →
Macro by Mark
帮助
隐私政策使用条款伦理与合规

© 2026 Mark Jayson Nation LLC