TL;DR
This paper introduces a continuous-time affine model for the term structure of interest rates that explicitly accounts for regime shifts. By incorporating regime-switching risk, the model offers a more nuanced understanding of yield curve dynamics and can better explain observed market phenomena like time-varying risk premiums.
What Problem Does This Paper Address?
The term structure of interest rates, which describes the relationship between interest rates and time to maturity, is a fundamental concept in finance. Traditional models often assume a single, stable economic regime. However, economies are known to experience recurrent shifts between different phases (e.g., expansion and recession), which can significantly impact interest rate behavior. Existing models that incorporate these regime shifts often struggle with tractability or offer limited insights into the pricing of regime-switching risk. This paper aims to bridge this gap by developing a tractable continuous-time model that can capture the effects of regime shifts on the yield curve and provide a clear interpretation of regime-switching risk premiums.
Methods
The authors develop a continuous-time affine model for the term structure of interest rates. A key innovation is the use of a marked point process to represent Markov regime shifts, which allows for an analytically simple representation. This framework is flexible enough to accommodate:
- Priced regime-switching risk: The model explicitly accounts for the risk associated with shifts between economic regimes.
- Time-varying transition probabilities: The likelihood of switching between regimes can change over time.
- Regime-dependent mean reversion coefficients: The tendency for interest rates to revert to a long-term average can differ across regimes.
- Stochastic volatilities within each regime: Interest rate volatility can vary even within a given economic regime.
The model is implemented using the Efficient Method of Moments (EMM), a statistical technique for estimating parameters of complex models by matching simulated moments to empirical moments. A two-factor version of the model is estimated.
Datasets
The study uses monthly U.S. Treasury zero-coupon bond yields from June 1964 to December 2001, obtained from the Center for Research in Security Prices (CRSP). This dataset includes eight interest rates with maturities ranging from 1 month to 5 years.
Key Insights and Ideas
- Regime shifts add volatility: The model demonstrates that regime shifts introduce an additional source of volatility that can affect both short and long ends of the yield curve, helping to explain why long-term interest rates don't always attenuate in volatility as predicted by simpler models.
- Two components of bond risk premiums: Under regime shifts, bond risk premiums are shown to have two components: a regime-dependent diffusion risk premium and a regime-switching risk premium. The latter arises from the covariations between bond prices and the stochastic discount factor across regimes, capturing the systematic risk of recurrent shifts.
- Regimes align with business cycles: The estimated regimes from the term structure model closely correspond to the NBER-identified business cycle expansions and recessions, suggesting that the latent regimes in the model have a clear economic interpretation.
- Regime-switching risk is economically important: The empirical results suggest that regime-switching risk is priced and plays a significant role in explaining bond returns, particularly in accounting for the failure of the expectation hypothesis.
Key Results
- The model is able to account for many salient features of the U.S. yield curve.
- The estimated regimes align well with historical business cycle data, with Regime 1 generally corresponding to expansions and Regime 2 to recessions.
- Interest rate volatility is found to be higher in Regime 2 (recessions) than in Regime 1 (expansions).
- The model's flexible specification of bond risk premiums, including the regime-switching component, helps replicate the negative regression coefficients typically found when testing the expectation hypothesis, which are often attributed to time-varying risk premiums.
Limitations and Future Work
- Tractability constraints: To obtain a closed-form solution for the term structure in an affine model, the authors had to restrict the risk-neutral regime-switching probabilities to be constant. This implies that the model might force regime-switching risk to be priced if physical transition probabilities are time-varying.
- Structural interpretation of regimes: The regimes identified by the model lack clear structural interpretations.
- Observable regimes: The current model assumes regimes are observable to bond investors. A natural extension would be to assume regimes are unobservable, requiring investors to learn them through other observable state variables, such as monetary policy stances.
- Incorporating macroeconomic fundamentals: Another extension is to embed the term structure model within a well-specified macroeconomic model with regime shifts to identify and interpret regimes in terms of macroeconomic fundamentals.
- Assessment of regime-switching risk premiums: The authors note that while their results suggest regime-switching risk is priced, there are still uncertainties regarding its magnitude, warranting further research with more general dynamic models.
Summary updated