Skip to content

flat_favar_twostep

Jump to bottom
Jaromír Beneš edited this page Mar 20, 2026 · 1 revision

Two-Step FAVAR with Flat Prior (FlatFAVARTwostep)

Overview

The Two-Step FAVAR with Flat Prior implements a Factor-Augmented Vector Autoregression using a two-step estimation procedure with uninformative (flat) priors. This estimator first extracts factors using principal component analysis, then estimates a VAR on the factors and observed variables using flat priors, providing a simple baseline FAVAR approach with minimal prior restrictions.

Methodology

The Flat FAVAR Two-Step specification features:

  • Two-step estimation: factor extraction followed by VAR estimation
  • Principal component factor extraction in first step
  • Flat (uninformative) priors on VAR coefficients in second step
  • Minimal prior restrictions on parameters
  • Baseline FAVAR approach for comparison with other specifications

This approach provides the simplest FAVAR implementation, serving as a natural starting point for factor-augmented analysis.

Parameters

Minnesota-Style Prior Parameters

  • Autoregression: Prior on first-order autoregression

    • Range: [0, 1]
    • Default: 0.8
    • Controls the belief in random walk behavior for own lags

  • Lambda1: Overall tightness of priors

    • Range: (0, ∞)
    • Default: 0.1
    • Smaller values impose tighter priors (more shrinkage)

  • Lambda2: Variable weighting

    • Range: (0, ∞)
    • Default: 0.5
    • Controls relative weights between own-variable and cross-variable lags

  • Lambda3: Lag decay

    • Range: (0, ∞)
    • Default: 1
    • Controls how quickly prior importance decays with lag length

  • Lambda4: Exogenous variable tightness

    • Range: (0, ∞)
    • Default: 100
    • Controls prior tightness on exogenous variable coefficients

  • Lambda5: Block exogeneity shrinkage

    • Range: (0, ∞)
    • Default: 0.001
    • Controls shrinkage when block exogeneity is imposed

Advanced Options

  • Exogenous: Priors on exogenous variables flag

    • Default: false
    • Enables specialized priors for exogenous variables

  • BlockExogenous: Block exogeneity flag

    • Default: false
    • Imposes block exogeneity restrictions

MCMC Settings

  • Burnin: Number of burn-in draws

    • Range: [0, ∞)
    • Default: 0
    • May require burn-in depending on posterior complexity

  • StabilityThreshold: Threshold for maximum eigenvalue magnitude

    • Range: (0, ∞)
    • Default: null (no stability check)
    • Ensures VAR stability by rejecting draws with eigenvalues above threshold

  • MaxNumUnstableAttempts: Maximum number of unstable sampling attempts

    • Range: [1, ∞)
    • Default: 1000
    • Maximum attempts to find stable draw before stopping

Usage Guidelines

When to Use

  • Baseline FAVAR: As starting point for factor-augmented analysis
  • Large datasets: When dealing with many observable series
  • Minimal prior assumptions: When avoiding strong prior restrictions
  • Comparison purposes: As benchmark against more informative FAVAR specifications

When Not to Use

  • Small datasets: When factor extraction is unreliable
  • Strong prior beliefs: When informative priors are preferred
  • Real-time applications: When computational speed is critical
  • Unstable factors: When factor identification is problematic

Practical Considerations

  • Requires sufficient cross-sectional dimension for reliable factor extraction
  • Monitor factor interpretability and stability
  • Consider number of factors to extract (typically determined by information criteria)
  • Assess whether two-step vs. one-step estimation is appropriate

Model Specification

Two-Step Procedure

Step 1: Factor Extraction

F_t = Λ X_t + e_t

where F_t are factors extracted via principal components from large dataset X_t.

Step 2: FAVAR Estimation

[Y_t]   [Y_{t-1}]
[F_t] = A [F_{t-1}] + ε_t

where Y_t are key observable variables and F_t are extracted factors.

Prior Specification

Flat priors applied to VAR coefficients:

vec(A) ~ N(0, κ I)

where κ → ∞ represents uninformative priors.

Computational Characteristics

Performance

  • Speed: Fast due to two-step procedure and flat priors
  • Memory: Moderate requirements depending on dataset size
  • Convergence: Generally good convergence with flat priors
  • Scalability: Scales well with number of observable series

Implementation Features

  • Principal component factor extraction in first step
  • Standard VAR estimation in second step
  • Minimal computational complexity
  • Straightforward implementation

Factor Extraction Considerations

Factor Determination

  • Use information criteria (IC) to select number of factors
  • Consider eigenvalue plots for factor importance
  • Assess factor stability across different samples
  • Validate factor interpretability

Data Requirements

  • Large cross-sectional dimension (N) relative to time (T)
  • Sufficient common variation to identify factors
  • Balanced panel preferred for reliable extraction
  • Consider data transformations (stationarity, normalization)

Comparison with Related FAVAR Approaches

vs. One-Step FAVAR

  • Estimation: Two-step vs. simultaneous estimation
  • Efficiency: Less efficient but computationally simpler
  • Uncertainty: Ignores factor estimation uncertainty
  • Implementation: Easier to implement and debug

vs. Minnesota FAVAR

  • Priors: Flat vs. Minnesota priors on VAR coefficients
  • Shrinkage: No shrinkage vs. systematic shrinkage
  • Performance: May perform worse in small samples
  • Benchmark: Serves as uninformative baseline

Applications

Suitable Contexts

  • Macroeconomic nowcasting with large datasets
  • Monetary policy analysis with information-rich environment
  • Financial market factor analysis
  • Business cycle research with many indicators

Research Applications

  • Factor structure analysis in large datasets
  • Information content evaluation of economic indicators
  • Comparison benchmark for more sophisticated FAVAR models
  • Structural break analysis in factor-augmented settings

Implementation Strategy

Factor Extraction

  • Standardize data before factor extraction
  • Use standard principal component methods
  • Select factors based on information criteria
  • Validate factor stability and interpretability

VAR Estimation

  • Apply flat priors to factor-augmented VAR
  • Monitor convergence of posterior sampler
  • Check stability of estimated VAR system
  • Compare results with standard VAR on observables only

Model Validation

Factor Assessment

  • Examine factor loadings for economic interpretation
  • Assess factor stability across different periods
  • Compare with alternative factor extraction methods
  • Validate factor forecasting performance

FAVAR Performance

  • Compare forecasting accuracy with standard VAR
  • Assess impulse response function stability
  • Evaluate structural interpretation of factor shocks
  • Cross-validate across different specifications

References

  • Bernanke, B.S., J. Boivin, and P. Eliasz (2005). "Measuring the Effects of Monetary Policy: A Factor-Augmented Vector Autoregressive (FAVAR) Approach"
  • Stock, J.H. and M.W. Watson (2002). "Forecasting Using Principal Components From a Large Number of Predictors"
  • Forni, M., M. Hallin, M. Lippi, and L. Reichlin (2000). "The Generalized Dynamic-Factor Model: Identification and Estimation"
  • Bai, J. and S. Ng (2002). "Determining the Number of Factors in Approximate Factor Models"

BEARX Toolbox Wiki

Clone this wiki locally