normal_wishart_panel

Normal-Wishart Panel VAR (NormalWishartPanel)

Overview

The Normal-Wishart Panel VAR applies the conjugate Normal-Wishart prior framework to panel VAR models, providing computational efficiency while allowing for information sharing across units. This estimator combines the analytical tractability of Normal-Wishart priors with the benefits of panel data, making it suitable for multi-unit VAR analysis where fast, efficient estimation is important.

Methodology

The Normal-Wishart Panel VAR specification features:

• Unit-specific VAR coefficients with Normal-Wishart priors
• Conjugate prior structure enabling analytical posterior computation
• Information pooling across panel units
• Efficient estimation through closed-form solutions
• Minnesota-type hyperprior on mean coefficients

This approach provides the computational advantages of conjugate priors while leveraging information across multiple units in the panel.

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 minimal burn-in due to conjugate structure

• StabilityThreshold: Threshold for maximum eigenvalue magnitude

  • Range: (0, ∞)
  • Default: null (no stability check)
  • Ensures VAR stability for each unit

• MaxNumUnstableAttempts: Maximum number of unstable sampling attempts

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

Usage Guidelines

When to Use

• Computational efficiency: When fast panel VAR estimation is required
• Moderate panel sizes: Works well with typical panel dimensions
• Conjugate framework: When analytical tractability is valued
• Standard panel applications: Good baseline choice for panel VAR

When Not to Use

• Complex panel structures: When sophisticated cross-unit modeling is needed
• Non-standard priors: When non-conjugate priors are required
• Large heterogeneity: When units are very different
• Time-varying parameters: When parameters evolve over time

Practical Considerations

• Benefits from balanced panel structure
• Conjugate structure provides computational advantages
• Consider whether Normal-Wishart assumptions are appropriate
• Monitor convergence even with conjugate priors

Model Specification

Panel VAR Structure

For unit i:

y_{i,t} = X_{i,t} β_i + ε_{i,t}

Conjugate Prior

β_i | Σ_i ~ N(B̄, Σ_i ⊗ Ω)
Σ_i ~ IW(S, ν)

Hierarchical Structure

B̄ ~ N(M, V)

where M and V are based on Minnesota-type restrictions.

Computational Characteristics

Performance

• Speed: Very fast due to conjugate structure
• Memory: Moderate requirements for panel storage
• Convergence: Excellent convergence properties
• Scalability: Scales well with panel dimensions

Analytical Benefits

• Closed-form posterior distributions
• No MCMC burn-in issues
• Predictable computational time
• Numerical stability

Comparison with Related Panel Approaches

vs. Mean OLS Panel

• Framework: Conjugate vs. hierarchical structure
• Efficiency: More computationally efficient
• Flexibility: Less flexible in prior specification
• Covariance: Explicit covariance matrix modeling

vs. Hierarchical Panel

• Computation: Much faster than full hierarchical models
• Prior structure: More restricted prior framework
• Information sharing: Less sophisticated information pooling
• Analytical properties: Closed-form vs. simulation-based

Implementation Advantages

Conjugate Benefits

• No convergence monitoring required
• Predictable computational resources
• Analytical posterior moments
• Easy model comparison through marginal likelihoods

Panel Efficiency

• Leverages cross-unit information
• Reduces individual unit estimation uncertainty
• Provides regularization for short time series
• Natural framework for panel forecasting

Diagnostic Considerations

Model Assessment

• Compare unit-specific coefficient estimates
• Assess reasonableness of covariance matrix estimates
• Evaluate improvement over individual unit estimation
• Check posterior concentration

Economic Validation

• Verify economic plausibility of pooled estimates
• Compare with alternative panel specifications
• Assess cross-unit parameter similarity
• Validate forecasting performance

References

• Kadiyala, K.R. and S. Karlsson (1997). "Numerical Methods for Estimation and Inference in Bayesian VAR-models"
• Canova, F. and M. Ciccarelli (2009). "Estimating Multi-Country VAR Models"
• Koop, G. and D. Korobilis (2010). "Bayesian Multivariate Time Series Methods for Empirical Macroeconomics"
• Banbura, M., D. Giannone, and L. Reichlin (2010). "Large Bayesian Vector Auto Regressions"