A time deformation model and its time-varying autocorrelation: An application to US unemployment data

A time deformation model and its time-varying autocorrelation: An application to US unemployment data

Chu-Ping C. Vijverberg
Wichita State University, Wichita, Kansas, United States
International Journal of Forecasting 25 (2009) 128–145

Keyword : G-Lambda, ARIMA,Unemployment, ARFIMA, STAR

Review :

Introduction
This paper explores the time-varying behavior of the G-Lambda model. Simulation results indicate that it is possible to distinguish between the G-Lambda model and other better-known models such as the ARIMA, ARFIMA and STAR models. Applying the model to US unemployment data, the performance of the G-Lambda model varies as the start of the forecast periods changes.

Problem and Hypothesis
The study of time deformation models in a different direction, applying it to a variable (the U.S unemployment rate) that does not exhibit dominant cyclical behavior. The unemployment rate is one of several important measures that economists use to gauge economic performance. The purpose of this paper is to explore the time-varying characteristic of the time deformation model, to model the US unemployment rate with a time deformed model, and, in the process, to compare it with other models.

Methodology
Through a quantitative survey of 24 studies about unemployment and secondary data from US Government persistence. This paper therefore compares the forecasting ability of the time deformation model with those of the ARIMA, ARFIMA and STAR models.

Conclusion
1. It reveals another attribute of the time deformation model, which has not been explored before.
2. This paper uses the Laplace method in finding a discrete state transition matrix within a continuous Kalman filtering model.
3. Comparing the forecast capabilities of the continuous GL (CGL), ARIMA, ARFIMA and STAR models, a different insight is revealed about the US unemployment data.

A time-varying ACF model – the case of real roots comparing with The dual relationship between CGL and CAR models This paper shows that, even with a time-varying ACF, data application of the CGL model is feasible because of the duality between the CGL model and a classical stationary CAR model. Simulation results indicate that it is possible to empirically distinguish between CGL and other better-known models (ARIMA, ARFIMA and STAR).

Criticism
In my point of view ,the choice between the CGL model and the other models is open to debate. Some may contend that the long-term forecasts offer valuable information,
while others may prefer to emphasize the short-range performance. However, the results of the sign test and the Diebold Mariano test indicate that the G-Lambda model has significantly letter long-term forecasts than other models.

Reference :
Chu-Ping ,C. Vijverberg .2009. A time deformation model and its time-varying autocorrelation: An application to US unemployment data. International Journal of Forecasting No.25.

.Copeland, L. 2008, Exchange Rates and International Finance. 5th edition.Prentice Hall.USA.

Knight,K.G. 1987.Unemployment : economic analysis. Barnes and noble book.USA


DENI DANIAL KESA
9904608