-
Dynamic Term Structure Models with Nonlinearities using Gaussian Processes
Authors:
Tomasz Dubiel-Teleszynski,
Konstantinos Kalogeropoulos,
Nikolaos Karouzakis
Abstract:
The importance of unspanned macroeconomic variables for Dynamic Term Structure Models has been intensively discussed in the literature. To our best knowledge the earlier studies considered only linear interactions between the economy and the real-world dynamics of interest rates in DTSMs. We propose a generalized modelling setup for Gaussian DTSMs which allows for unspanned nonlinear associations…
▽ More
The importance of unspanned macroeconomic variables for Dynamic Term Structure Models has been intensively discussed in the literature. To our best knowledge the earlier studies considered only linear interactions between the economy and the real-world dynamics of interest rates in DTSMs. We propose a generalized modelling setup for Gaussian DTSMs which allows for unspanned nonlinear associations between the two and we exploit it in forecasting. Specifically, we construct a custom sequential Monte Carlo estimation and forecasting scheme where we introduce Gaussian Process priors to model nonlinearities. Sequential scheme we propose can also be used with dynamic portfolio optimization to assess the potential of generated economic value to investors. The methodology is presented using US Treasury data and selected macroeconomic indices. Namely, we look at core inflation and real economic activity. We contrast the results obtained from the nonlinear model with those stemming from an application of a linear model. Unlike for real economic activity, in case of core inflation we find that, compared to linear models, application of nonlinear models leads to statistically significant gains in economic value across considered maturities.
△ Less
Submitted 18 May, 2023;
originally announced May 2023.
-
On Unspanned Latent Risks in Dynamic Term Structure Models
Authors:
Tomasz Dubiel-Teleszynski,
Konstantinos Kalogeropoulos,
Nikolaos Karouzakis
Abstract:
We explore the importance of information hidden from the yield curve and assess how valuable the unspanned risks are to a real-time Bayesian investor seeking to forecast excess bond returns and maximise her utility. We propose a novel class of arbitrage-free unspanned Dynamic Term Structure Models (DTSM), that embed a stochastic market price of risk specification. We develop a suitable Sequential…
▽ More
We explore the importance of information hidden from the yield curve and assess how valuable the unspanned risks are to a real-time Bayesian investor seeking to forecast excess bond returns and maximise her utility. We propose a novel class of arbitrage-free unspanned Dynamic Term Structure Models (DTSM), that embed a stochastic market price of risk specification. We develop a suitable Sequential Monte Carlo (SMC) inferential and prediction scheme that guarantees joint identification of parameters and latent states and takes into account all relevant uncertainties. We find that latent factors contain significant predictive power above and beyond the yield curve, providing improvement to the out-of-sample predictive performance of models, especially at shorter maturities. Most importantly, they are capable of exploiting information hidden from the yield curve and translate the evident statistical predictability into significant utility gains, out-of-sample. The hidden component associated with slope risk is countercyclical and links with real activity.
△ Less
Submitted 13 October, 2022; v1 submitted 29 April, 2022;
originally announced May 2022.
-
Sequential Learning and Economic Benefits from Dynamic Term Structure Models
Authors:
Tomasz Dubiel-Teleszynski,
Konstantinos Kalogeropoulos,
Nikolaos Karouzakis
Abstract:
We explore the statistical and economic importance of restrictions on the dynamics of risk compensation from the perspective of a real-time Bayesian learner who predicts bond excess returns using dynamic term structure models (DTSMs). The question on whether potential statistical predictability offered by such models can generate economically significant portfolio benefits out-of-sample, is revisi…
▽ More
We explore the statistical and economic importance of restrictions on the dynamics of risk compensation from the perspective of a real-time Bayesian learner who predicts bond excess returns using dynamic term structure models (DTSMs). The question on whether potential statistical predictability offered by such models can generate economically significant portfolio benefits out-of-sample, is revisited while imposing restrictions on their risk premia parameters. To address this question, we propose a methodological framework that successfully handles sequential model search and parameter estimation over the restriction space in real time, allowing investors to revise their beliefs when new information arrives, thus informing their asset allocation and maximising their expected utility. Empirical results reinforce the argument of sparsity in the market price of risk specification since we find strong evidence of out-of-sample predictability only for those models that allow for level risk to be priced and, additionally, only one or two of these risk premia parameters to be different than zero. Most importantly, such statistical evidence is turned into economically significant utility gains, across prediction horizons, different time periods and portfolio specifications. In addition to identifying successful DTSMs, the sequential version of the stochastic search variable selection (SSVS) scheme developed can be applied on its own and also offer useful diagnostics monitoring key quantities over time. Connections with predictive regressions are also provided.
△ Less
Submitted 13 October, 2022; v1 submitted 22 April, 2022;
originally announced April 2022.
