Threshold Autoregressive Moving-average Models Research Articles

Robust estimation for Threshold Autoregressive Moving-Average models

Threshold autoregressive moving-average (TARMA) models extend the popular TAR model and are among the few parametric time series specifications to include a moving average in a non-linear setting. The state dependent reactions to shocks is particularly appealing in Economics and Finance. However, no theory is currently available when the data present heavy tails or anomalous observations. Here we provide the first theoretical framework for robust M-estimation for TARMA models and study its practical relevance. Under mild conditions, we show that the robust estimator for the threshold parameter is super-consistent, while the estimators for autoregressive and moving-average parameters are strongly consistent and asymptotically normal. The Monte Carlo study shows that the M-estimator is superior, in terms of both bias and variance, to the least squares estimator, which can be heavily affected by outliers. The findings suggest that robust M-estimation should be generally preferred to the least squares method. We apply our methodology to a set of commodity price time series; the robust TARMA fit presents smaller standard errors and superior forecasting accuracy. The results support the hypothesis of a two-regime non-linearity characterised by slow expansions and fast contractions.

Incorporating Russo-Ukrainian war in Brent crude oil price forecasting: A comparative analysis of ARIMA, TARMA and ENNReg models

This article investigates the performance of three models - Autoregressive Integrated Moving Average (ARIMA), Threshold Autoregressive Moving Average (TARMA) and Evidential Neural Network for Regression (ENNReg) - in forecasting the Brent crude oil price, a crucial economic variable with a significant impact on the global economy. With the increasing complexity of the price dynamics due to geopolitical factors such as the Russo-Ukrainian war, we examine the impact of incorporating information on the war on the forecasting accuracy of these models. Our analysis shows that incorporating the impact of the war can significantly improve the forecasting accuracy of the models, and the ENNReg model with the inclusion of the dummy variable outperforms the other models during the war period. Including the war variable has enhanced the forecasting accuracy of the ENNReg model by 0.11%. These results carry significant implications regarding policymakers, investors, and researchers interested in developing accurate forecasting models in the presence of geopolitical events such as the Russo-Ukrainian war. The results can be used by the governments of oil-exporting countries for budget policies.

Revisiting the Canadian Lynx Time Series Analysis Through TARMA Models

The class of threshold autoregressive models has been proven to be a powerful and appropriate tool to describe many dynamical phenomena in different fields. In this work, we deploy the threshold autoregressive moving-average framework to revisit the analysis of the benchmark Canadian lynx time series. This data set has attracted great attention among non-linear time series analysts due to its asymmetric cycle that makes the investigation very challenging. We compare some of the best threshold autoregressive models (TAR) proposed in literature with a selection of thresholdautoregressive moving-average models (TARMA). The models are compared under different prospectives: (i) goodness-of-fit through information criteria, (ii) their ability to reproduce characteristic cycles, (iv) their capability to capture multimodality and (iii) forecasting performance. We found TARMAmodels that perform better than TAR models with respect to all these aspects.

On the Ergodicity of First‐Order Threshold Autoregressive Moving‐Average Processes

We introduce a certain Markovian representation for the threshold autoregressive moving‐average (TARMA) process with which we solve the long‐standing problem regarding the irreducibility condition of a first‐order TARMA model. Under some mild regularity conditions, we obtain a complete classification of the parameter space of an invertible first‐order TARMA model into parametric regions over which the model is either transient or recurrent, and the recurrence region is further subdivided into regions of null recurrence or positive recurrence, or even geometric recurrence. We derive a set of necessary and sufficient conditions for the ergodicity of invertible first‐order TARMA processes.

Testing a linear ARMA model against threshold-ARMA models: A Bayesian approach

ABSTRACTWe introduce a Bayesian approach to test linear autoregressive moving-average (ARMA) models against threshold autoregressive moving-average (TARMA) models. First, the marginal posterior densities of all parameters, including the threshold and delay, of a TARMA model are obtained by using Gibbs sampler with Metropolis–Hastings algorithm. Second, reversible-jump Markov chain Monte Carlo (RJMCMC) method is adopted to calculate the posterior probabilities for ARMA and TARMA models: Posterior evidence in favor of TARMA models indicates threshold nonlinearity. Finally, based on RJMCMC scheme and Akaike information criterion (AIC) or Bayesian information criterion (BIC), the procedure for modeling TARMA models is exploited. Simulation experiments and a real data example show that our method works well for distinguishing an ARMA from a TARMA model and for building TARMA models.

Bayesian Analysis of Two-Regime Threshold Autoregressive Moving Average Model with Exogenous Inputs

We consider Bayesian analysis of threshold autoregressive moving average model with exogenous inputs (TARMAX). In order to obtain the desired marginal posterior distributions of all parameters including the threshold value of the two-regime TARMAX model, we use two different Markov chain Monte Carlo (MCMC) methods to apply Gibbs sampler with Metropolis-Hastings algorithm. The first one is used to obtain iterative least squares estimates of the parameters. The second one includes two MCMC stages for estimate the desired marginal posterior distributions and the parameters. Simulation experiments and a real data example show support to our approaches.

Bayesian subset selection for threshold autoregressive moving-average models

Optimal subset selection among a general family of threshold autoregressive moving-average (TARMA) models is considered. The usual complexity of model/order selection is increased by capturing the uncertainty of unknown threshold levels and an unknown delay lag. The Monte Carlo method of Bayesian model averaging provides a possible way to overcome such model uncertainty. Incorporating with the idea of Bayesian model averaging, a modified stochastic search variable selection method is adapted to consider subset selection in TARMA models, by adding latent indicator variables for all potential model lags as part of the proposed Markov chain Monte Carlo sampling scheme. Metropolis–Hastings methods are employed to deal with the well-known difficulty of including moving-average terms in the model and a novel proposal mechanism is designed for this purpose. Bayesian comparison of two hyper-parameter settings is carried out via a simulation study. The results demonstrate that the modified method has favourable performance under reasonable sample size and appropriate settings of the necessary hyper-parameters. Finally, the application to four real datasets illustrates that the proposed method can provide promising and parsimonious models from more than 16 million possible subsets.

Statistical Properties of Threshold Models

This article focuses the attention on the Self Exciting Threshold Autoregressive Moving Average model (SETARMA) proposed in Tong (1983). The stochastic structure of the model is discussed and different specifications are presented. Starting from one of them, we give sufficient conditions for the weak stationarity of the model that are discussed and critically compared to other results given in literature. In particular, after showing that the SETARMA model belongs to the class of the Random Coefficients Autoregressive models, widely discussed in Nicholls and Quinn (1982), we give some issues on the weak stationarity of its stochastic structure that are more general than those given in the existing literature and appear not affected by the moving average component.