Joint Bayesian Analysis of Parameters and States in Nonlinear non‐Gaussian State Space Models

Abstract

Standard-Nutzungsbedingungen:Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden.Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen.Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents inSiem Jan Koopman, (a,c), ‡ André Lucas A particular feature of our approach is that smoothed estimates of the states and the marginal likelihood are obtained directly as an output of the algorithm. Our method provides a computationally efficient alternative to several recently proposed algorithms. We present extensive simulation evidence for stochastic volatility and stochastic intensity models. For our empirical study, we analyse the performance of our method for stock returns and corporate default panel data.JEL classification: C11, C15, C22, C32, C58.

The reaction to news in live betting

“…However, while discrete states may be easy to interpret, the use of a continuous-valued state process provides more flexibility as a corresponding model can capture gradual changes in the underlying market activity level, which makes more sense conceptually as there is no clear justification for a finite number of market activity levels. This reasoning is analogous to that of corresponding modelling approaches in finance, where share returns are typically modelled via stochastic volatility models, with a continuous-valued state process (see, e.g., Aït-Sahalia et al, 2007;Langrock et al, 2012;Barra et al, 2017). We thus consider a discrete-time SSM with continuous state space to model the time series of stakes placed.…”

“…Financial time series data are typically driven by not necessarily directly observable states such as a market's activity level, the nervousness of the financial market, the state of the economy, or the regulatory environment. In many of the corresponding analyses, it thus makes conceptual sense to consider state-space (SSMs) or related models for relating the observed quantity of interest to underlying (latent) states (see, e.g., Jacquier et al, 1994, Al-Anaswah and Wilfling, 2011, Warne et al, 2017, and Barra et al, 2017.…”

A Bayesian Non-linear State Space Copula Model to Predict Air Pollution in Beijing

“…An alternative Bayesian approach for jointly estimating parameters and states in non-linear non-Gaussian state space models is presented by Barra et al (2017), who designed flexible proposal densities for the independent Metropolis-Hasting and the importance sampling algorithms.…”

Data cloning estimation for asymmetric stochastic volatility models

“…Amidst all these estimation methods, one of the best alternatives is to use an MCMC algorithm, although MCMC is very computationally demanding and requires hard code, which makes its implementation difficult for practitioners and researchers. A very recent paper by Barra et al (2017), based on importance sampling weighted expectation maximization, argues that this approach is computationally efficient and can be regarded as an effective alternative to MCMC methods. However, it requires, again, heavy programming and there is not free available software that implements the method.…”