Results 1 
9 of
9
Evidence on Structural Instability in Macroeconomic Time Series Relations
 Journal of Business and Economic Statistics
, 1996
"... An experiment is performed to assess the prevalence of instability in univariate and bivariate macroeconomic time series relations and to ascertain whether various adaptive forecasting techniques successfully handle any such instability. Formal tests for instability and outofsample forecasts from ..."
Abstract

Cited by 117 (7 self)
 Add to MetaCart
An experiment is performed to assess the prevalence of instability in univariate and bivariate macroeconomic time series relations and to ascertain whether various adaptive forecasting techniques successfully handle any such instability. Formal tests for instability and outofsample forecasts from 16 different models are computed using a sample of 76 representative U.S. monthly postwar macroeconomic time series, constituting 5,700 bivariate forecasting relations. The tests for instability and the forecast comparisonsuggest that there is substantial instability in a significant fraction of the univariate and bivariate autoregressive models.
Nonlinear and NonGaussian StateSpace Modeling with Monte Carlo Techniques: A Survey and Comparative Study
 In Rao, C., & Shanbhag, D. (Eds.), Handbook of Statistics
, 2000
"... Since Kitagawa (1987) and Kramer and Sorenson (1988) proposed the filter and smoother using numerical integration, nonlinear and/or nonGaussian state estimation problems have been developed. Numerical integration becomes extremely computerintensive in the higher dimensional cases of the state vect ..."
Abstract

Cited by 17 (4 self)
 Add to MetaCart
Since Kitagawa (1987) and Kramer and Sorenson (1988) proposed the filter and smoother using numerical integration, nonlinear and/or nonGaussian state estimation problems have been developed. Numerical integration becomes extremely computerintensive in the higher dimensional cases of the state vector. Therefore, to improve the above problem, the sampling techniques such as Monte Carlo integration with importance sampling, resampling, rejection sampling, Markov chain Monte Carlo and so on are utilized, which can be easily applied to multidimensional cases. Thus, in the last decade, several kinds of nonlinear and nonGaussian filters and smoothers have been proposed using various computational techniques. The objective of this paper is to introduce the nonlinear and nonGaussian filters and smoothers which can be applied to any nonlinear and/or nonGaussian cases. Moreover, by Monte Carlo studies, each procedure is compared by the root mean square error criterion.
Testing for Reduction to Random Walk in Autoregressive Conditional Heteroscedasticity Models
, 2000
"... A class of time series models which allow for conditional heteroscedasticity and autoregression reduce to random walk or white noise when some of the conditional volatility parameters take their boundary values of 0, and the autoregressive component takes the form of a unit root or is not in fact pr ..."
Abstract

Cited by 5 (1 self)
 Add to MetaCart
A class of time series models which allow for conditional heteroscedasticity and autoregression reduce to random walk or white noise when some of the conditional volatility parameters take their boundary values of 0, and the autoregressive component takes the form of a unit root or is not in fact present. Under mild assumptions on the residual distribution we calculate the asymptotic distributions of pseudologlikelihood ratio statistics for testing hypotheses like these, and assess their finite sample performances by simulations. The results apply in particular to the commonly used ARARCH and ARGARCH models, and have application in the analysis of asset price series and random coefficient models.
17. ConstraintBased Neural Network Learning for Time Series Predictions
"... In this chapter, we have briefly surveyed previous work in predicting noisefree piecewise chaotic time series and noisy time series with high frequency random noise. For noisefree time series, we have proposed a constrained formulation for neural network learning that incorporates the error of eac ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
In this chapter, we have briefly surveyed previous work in predicting noisefree piecewise chaotic time series and noisy time series with high frequency random noise. For noisefree time series, we have proposed a constrained formulation for neural network learning that incorporates the error of each learning pattern as a constraint, a new crossvalidation scheme that allows multiple validations sets to be considered in learning, a recurrent FIR neural network architecture that combines a recurrent structure and a memorybased FIR structure, and a violationguided back propagation algorithm for searching in the constrained space of the formulation. For noisy time series, we have studied systematically the edge effect due to lowpass filtering of noisy time series and have developed an approach that incorporates constraints on predicting lowpass data in the lag period. The new constraints enable active training in the lag period that greatly improves the prediction accuracy in the lag period. Finally, experimental
Financial Time Series Prediction Using Nonfixed and Asymmetrical Margin Setting with Momentum
 in Support Vector Regression. Neural Information Processing: Research and Development
, 2004
"... Abstract. Recently, Support Vector Regression (SVR) has been applied to financial time series prediction. Typical characteristics of financial time series are nonstationary and noisy in nature. The volatility, usually timevarying, of the time series therefore contains some valuable information abou ..."
Abstract

Cited by 1 (1 self)
 Add to MetaCart
Abstract. Recently, Support Vector Regression (SVR) has been applied to financial time series prediction. Typical characteristics of financial time series are nonstationary and noisy in nature. The volatility, usually timevarying, of the time series therefore contains some valuable information about the series. Previously, we had proposed to use the volatility in the data to adaptively changing the width of the margin in SVR. We have noticed that upside margin and downside margin would not necessary be the same, and we have observed that their choice would affect the upside risk, downside risk and as well as the overall prediction performance. In this work, we introduce a novel approach to adapt the asymmetrical margins using momentum. We applied and compared this method to predict the Hang Seng Index
Learning, Forecasting and Structural Breaks 1
, 2004
"... The literature on structural breaks focuses on ex post identification of break points that may have occurred in the past. While this question is important, a more challenging problem facing econometricians is to provide forecasts when the data generating process is unstable. The purpose of this pape ..."
Abstract
 Add to MetaCart
The literature on structural breaks focuses on ex post identification of break points that may have occurred in the past. While this question is important, a more challenging problem facing econometricians is to provide forecasts when the data generating process is unstable. The purpose of this paper is to provide a general methodology for forecasting in the presence of model instability. We make no assumptions on the number of break points or the law of motion governing parameter changes. Our approach makes use of Bayesian methods of model comparison and learning in order to provide an optimal predictive density from which forecasts can be derived. Estimates for the posterior probability that a break occurred at a particular point in the sample are generated as a byproduct of our procedure. We discuss the importance of using priors that accurately reflect the econometricianâ€™s opinions as to what constitutes a plausible forecast. Several applications to macroeconomic timeseries data demonstrate the usefulness of our procedure.
The TimeVarying Parameter Model Revisited
"... The Kalman filter formula, given by the linear recursive algorithm, is usually used for estimation of the timevarying parameter model. The filtering formula, introduced by Kalman (1960) and Kalman and Bucy (1961), requires the initial state variable. The obtained state estimates are influenced by t ..."
Abstract
 Add to MetaCart
The Kalman filter formula, given by the linear recursive algorithm, is usually used for estimation of the timevarying parameter model. The filtering formula, introduced by Kalman (1960) and Kalman and Bucy (1961), requires the initial state variable. The obtained state estimates are influenced by the initial value when the initial variance is not too large. To avoid the choice of the initial state variable, in this paper we utilize the di#use prior for the initial density. Moreover, using the Gibbs sampler, random draws of the state variables given all the data are generated, which implies that random draws are generated from the fixedinterval smoothing densities. Using the EM algorithm, the unknown parameters included in the system are estimated. As an example, we estimate a traditional consumption function for both the U.S. and Japan.
Testing for Random Walk Coefficients in a Simple State Space Model
, 1994
"... Three different tests for random walk coefficients in linear regression models have become popular: The LaMotte & McWhorter [1978] F tests, the LBI test by Nyblom & Makelainen [1983], and the POI test by King [1980], especially in the version of Shively [1988]. This paper extends these trad ..."
Abstract
 Add to MetaCart
Three different tests for random walk coefficients in linear regression models have become popular: The LaMotte & McWhorter [1978] F tests, the LBI test by Nyblom & Makelainen [1983], and the POI test by King [1980], especially in the version of Shively [1988]. This paper extends these traditional tests for random walk coefficients in linear regression models to more general models. More precisely, the assumption that under the null hypothesis all regression coefficients are constant over time is given up. These adapted tests are designed for situations where we already know that some coefficients follow a random walk. Hereby one will be enabled to test for random walk coefficients successively. As in other similar contexts, the problem arises of how to handle nuisance parameters. Three different methods are proposed and it is shown in a simulation study that these methods hardly affect the good power properties of the tests proposed. These tests will be compared to the Marginal Likel...