Skip to content

Economic Model

Overview

The model is a life-cycle problem where a household chooses consumption and portfolio allocation each period, facing risky labor income, time-varying equity returns, and asymmetric consumption adjustment costs.

State Variables (Scaled Mode)

Variable Description
age Current age
\(x_t\) Equity premium state (AR(1))
\(m_t\) Cash-on-hand / persistent earnings
\(cm_t\) Lagged consumption ratio

Controls

  • \(c_t\): consumption (scaled)
  • \(\theta_t\): stock share of liquid wealth, \(\theta \in [0, 1]\)

Key Equations

Equity premium transition: $\(x_{t+1} = \bar{x} + \phi_x (x_t - \bar{x}) + \xi_{t+1}\)$

Realized stock return: $\(R_{stock} = r_f + x_t + N_{CF} - N_{DR}\)$

Consumption adjustment cost (asymmetric, applies only when \(c_t < c_{t-1}\)): $\(\Phi_C = \frac{\phi_c}{2} \frac{(\max(0, cm_t - c_t))^2}{cm_t}\)$

Epstein-Zin utility recursion: $\(V_t = [(1-\beta) c_t^\rho + \beta \text{CE}_t^\rho]^{1/\rho}\)$

where \(\rho = 1 - 1/\psi\) and \(\text{CE}_t = (\mathbb{E}[V_{t+1}^{1-\gamma}])^{1/(1-\gamma)}\).

Paper vs. Package

Aspect Paper Package
Income process params Estimated from confidential tax data Public approximation from qualitative targets
Crash mixture Calibrated to data Parametric approximation
Grid resolution Not specified Configurable via GridSpec
Quadrature "Gaussian quadrature over 4 shocks" Gauss-Hermite with configurable nodes

Approximations

All income process slope parameters (lambda coefficients linking persistent earnings shocks to return news) are derived from the paper's qualitative targets rather than confidential estimation. These are marked with source="public_approximation" in the calibration metadata.