Statement
Observed boom-bust cycles in U.S. commercial real estate capitalization rates are driven primarily by shifts in required risk premia (the discount-rate channel) rather than by changes in expected NOI growth (the cash-flow channel). The risk-premium shifts are in turn associated with (a) movements in general capital-market risk premia (Baa-Treasury spread), (b) regulatory capital requirements on CMBS/CRE lending, and (c) real risk-free interest rates. Expected rent growth contributes very little to the cyclical variation in cap rates.
Evidence summary
Two complementary papers establish this from different data and methods:
-
Duca-Ling (2015) use Johansen VECM on RERC survey data (1996-2014) for four property types and find that the decomposition of equilibrium cap rates attributes the mid-2000s compression mainly to declining risk premia (associated with Basel II capital requirement relaxation), the 2008-09 reversal to spiking general risk premia, and the post-2010 recovery to low real Treasury yields. Expected rent growth contributes almost nothing.
-
Plazzi-Torous-Valkanov (2010) use a structural present-value model on NCREIF transaction data (1985-2003) across 53 MSAs and show that cap rate variation for apartments, retail, and industrial is dominated by expected return variation. Office is the exception: its rent growth covaries strongly with expected returns, so the cap rate fails to capture return predictability.
Conditions and scope
- Strongest for apartments, industrial, and retail. Office is a partial exception.
- The regulatory capital channel (Basel II / CMBS) is institutionally specific to the U.S. and to the 2002-2010 period.
- The finding is in-sample; out-of-sample validation covering the 2020-2025 period (COVID office shock, Fed tightening) has not been published.
- Both papers use U.S. data only; international applicability is unknown.
Counter-evidence
- For office properties, Plazzi-Torous-Valkanov (2010) find that expected rent growth IS an important driver of cap rate variation, contradicting the “risk premia only” narrative for that property type.
- The Duca-Ling regulatory capital measure (RegCap) is hand-constructed and may conflate regulatory changes with correlated market conditions.
- Survey-based measures of required returns and expected rent growth (RERC/PWC) may not accurately reflect marginal investors’ true expectations.
Linked ideas
- out-sample-cre-cap-rate-forecasting — out-of-sample forecasting validates whether regime-switching risk premia predict cap rate movements better than cash-flow models
Open questions
- Does the dominance of the discount-rate channel persist through the 2020-2025 period, or did the remote-work shock to offices create a cash-flow-driven cap rate event?
- Can the regulatory capital arbitrage channel be embedded in a structural no-arbitrage model (like the Leather-Sagi MS-RE framework) as a regime-switching parameter?
- What is the structural economic mechanism that makes office rent growth more correlated with discount rates than other property types?