Motivation
The existing no-arbitrage regime-switching literature prices bonds (DSY 2007, ABW 2008, Bikbov-Chernov) or, separately, real estate (Leather-Sagi) — but never both under the same stochastic discount factor with a common regime structure. The CRE topic page explicitly identifies this as an unfilled gap. The Leather- Sagi model already prices Treasury strip yields and CRE cap rates jointly, so the infrastructure for a cross-asset comparison of regime-specific risk premia exists — it just hasn’t been extracted and analyzed.
Hypothesis
Regime-specific risk premia for CRE cap rates and Treasury term premia share a common macro driver under the joint SDF: both are highest in the Active-Flexible monetary policy regime and lowest in Passive-Rigid, and their cross-asset correlation across regimes exceeds 0.7. Published bond-only term premium estimates (ABW 2008, DSY 2007) are consistent with the Treasury-side term premia implied by the joint CRE+Treasury estimation.
Approach sketch
- Extract regime-conditional risk premia from the existing MAP/MLE estimates
(or from
ep06c_polishedon simulated DGP):- Treasury term premia for 2Y, 5Y, 10Y maturities, per compound regime
- CRE risk premia (expected excess return over risk-free) for each property type, per compound regime
- Cross-asset comparison:
- Correlations of CRE risk premia vs. term premia across the 4 regimes
- Regime-specific Sharpe ratios for CRE vs. Treasuries
- Decompose: how much of the cross-asset co-movement comes from the common regime structure vs. regime-specific loadings?
- External validation: compare the implied term premia to published estimates from ang-bekaert-wei-2008-real-rates-expected-inflation and dai-singleton-yang-2007-regime-shifts-term-structure. Consistency is a check on the model; discrepancy identifies where the joint estimation differs from bond-only estimation.
- Restriction test: estimate a version where CRE and Treasury risk premia are forced to have identical regime loadings (same risk-price vector) vs. the unrestricted version. LR test quantifies whether asset-class-specific regime risk pricing is needed.
Expected outcome
A published cross-asset comparison of regime risk premia that is the first of its kind in the no-arbitrage MS-TSM literature. If CRE and bond risk premia co-move strongly across regimes, this supports the structural claim that a common macro regime drives all asset prices. If they diverge, it identifies where the model needs asset-specific channels (e.g., the regulatory capital channel for CRE).
Risks
- Extracting regime-conditional risk premia requires careful accounting of the risk-neutral vs. physical measure distinction and the exponential-quadratic pricing factors. The CRE risk premium is not as cleanly defined as the bond term premium.
- The simulated-DGP risk premia are by construction consistent with the model. Real-data risk premia are the interesting case but require the real-data estimation to be finalized.
- If the model’s bond-side term premia are far from published estimates, this may indicate misspecification rather than a cross-asset finding.
Pilot results
(empty)
Lessons learned
(empty)