Journal of Time Series Analysis最新文献

We develop a finite-sample theory for estimating the coefficients and for the prediction of multiple stable autoregressive processes that (i) share an unknown lag order but (ii) can differ in their sample sizes. Our technique is based on penalisation similar to hierarchical, overlapping group-Lasso but requires a new mathematical set-up to accommodate (i) and (ii). The set-up differs from existing work considerably, for example, in that we estimate the common lag order directly from the data rather than using extrinsic criteria. We prove that the estimated autoregressive processes enjoy stability, and we establish rates for both the estimation and prediction error that can outmatch the known rates in our setting. Our insights on lag selection and stability are also of interest in the case of individual autoregressive processes.

We develop a finite-sample theory for estimating the coefficients and for the prediction of multiple stable autoregressive processes that (i) share an unknown lag order but (ii) can differ in their sample sizes. Our technique is based on penalisation similar to hierarchical, overlapping group-Lasso but requires a new mathematical set-up to accommodate (i) and (ii). The set-up differs from existing work considerably, for example, in that we estimate the common lag order directly from the data rather than using extrinsic criteria. We prove that the estimated autoregressive processes enjoy stability, and we establish rates for both the estimation and prediction error that can outmatch the known rates in our setting. Our insights on lag selection and stability are also of interest in the case of individual autoregressive processes.

This paper introduces a panel threshold mixed data sampling model with a covariate-dependent threshold and unobserved individual-specific threshold effects (PTMIDAS-CDT), in which we allow for a covariate-dependent threshold effect in the relationship between dependent and independent variables sampled at different frequencies, and allow for unobserved individual-specific threshold effects. Based on the Chamberlain–Mundlak correlated random effects (CRE) device and Markov chain Monte Carlo (MCMC) technique, we develop the estimator of model parameters and suggest test statistics for threshold effect, threshold constancy, the equal weighting scheme, and unobserved individual-specific threshold effects. We establish the asymptotic properties of the proposed estimator in the small-threshold-effect framework and derive the limiting distributions of the suggested test statistics. Monte Carlo simulations are conducted to examine the performance properties of the estimation and testing procedures. The simulation results point out that the estimation procedure works well in finite samples, and the test statistics have good size and power properties. The model is illustrated with an application to the nexus between climate change and economic growth.

This paper introduces a panel threshold mixed data sampling model with a covariate-dependent threshold and unobserved individual-specific threshold effects (PTMIDAS-CDT), in which we allow for a covariate-dependent threshold effect in the relationship between dependent and independent variables sampled at different frequencies, and allow for unobserved individual-specific threshold effects. Based on the Chamberlain–Mundlak correlated random effects (CRE) device and Markov chain Monte Carlo (MCMC) technique, we develop the estimator of model parameters and suggest test statistics for threshold effect, threshold constancy, the equal weighting scheme, and unobserved individual-specific threshold effects. We establish the asymptotic properties of the proposed estimator in the small-threshold-effect framework and derive the limiting distributions of the suggested test statistics. Monte Carlo simulations are conducted to examine the performance properties of the estimation and testing procedures. The simulation results point out that the estimation procedure works well in finite samples, and the test statistics have good size and power properties. The model is illustrated with an application to the nexus between climate change and economic growth.

Estimating conditional quantiles plays a crucial role in modern risk management and other various applications. However, the quantile regression (QR) estimation of Poisson autoregressive (PAR) models, count-type models, remain an unresolved challenge. In this study, we propose a novel approach that employs a jittering smoothing method and a novel transformation strategy to convert this complex problem into an easily implementable quantile regression problem for continuous-type regression models. The asymptotic theory of the estimator is derived under some regularity conditions and the applications to four popular and classical PAR models are considered. Additionally, a novel $��h�$-step prediction method ($��h�$-QRF) is developed to forecast the $��h�$-step conditional distribution. The finite sample performance of the method is examined, and its advantages over existing methods are illustrated by simulation studies and an empirical application to the daily stock volume dataset of Technofirst.

Estimating conditional quantiles plays a crucial role in modern risk management and other various applications. However, the quantile regression (QR) estimation of Poisson autoregressive (PAR) models, count-type models, remain an unresolved challenge. In this study, we propose a novel approach that employs a jittering smoothing method and a novel transformation strategy to convert this complex problem into an easily implementable quantile regression problem for continuous-type regression models. The asymptotic theory of the estimator is derived under some regularity conditions and the applications to four popular and classical PAR models are considered. Additionally, a novel $��h�$-step prediction method ($��h�$-QRF) is developed to forecast the $��h�$-step conditional distribution. The finite sample performance of the method is examined, and its advantages over existing methods are illustrated by simulation studies and an empirical application to the daily stock volume dataset of Technofirst.

This article proposes a new ordinary least squares (OLS)-based procedure for retrospectively dating the emergence and collapse of bubbles. We first consider a data generating process that entails a switch from a unit root regime to an explosive regime followed by a collapse and subsequent return to unit root behavior. We demonstrate analytically that the standard OLS estimates are inconsistent and date both the origination and implosion points with a delay in large samples. A simple modification that involves omitting the residual corresponding to the implosion date is shown to yield consistent estimates. We also develop an efficient dating algorithm that can accommodate a framework with multiple bubbles. The algorithm exploits the explicit form of the unit root restrictions to directly embed them into the recursive optimization problem which obviates the need to rely on an iterative scheme that requires initial values. Extensive simulation experiments indicate that our proposed procedure typically delivers estimates with lower bias and root mean squared error relative to competing alternatives. An empirical illustration is included.

Crises, such as the Covid-19 epidemic, can affect economic time series by distorting historical trends and seasonal patterns with extreme values. The identification and adjustment of such extremes is important for the production of seasonally adjusted data; an ongoing challenge is the identification of a crisis' end, where the time series returns to more typical pre-crisis dynamics. A secondary challenge is to model and analyze time series data with both missing values and extremes. This paper extends the maximum entropy framework to a generalized class of extreme values, including level shifts, temporary changes, and seasonal outliers. These outliers are described as a particular type of stochastic process that is latent, or unobserved, such that its removal increases the time series entropy. The proposed methods allow one to model and fit time series data using conventional tools in the presence of specified streams of extreme values, as well as missing values. Extreme value adjustment, with quantification of mean squared error, can then be obtained along with the seasonal adjustment; there is also a test statistic to directly compare two specifications of extremes. The techniques are illustrated in weekly employment data.