AUTHOR=Pavlenko Oksana 

TITLE=The smooth transition autoregressive models for the unemployment rate of Latvia

JOURNAL=Frontiers in Applied Mathematics and Statistics

VOLUME=Volume 11 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/applied-mathematics-and-statistics/articles/10.3389/fams.2025.1673247

DOI=10.3389/fams.2025.1673247

ISSN=2297-4687

ABSTRACT=To model potential structural shifts in the data that depend on their historical values, different smooth transition autoregressive models are constructed and compared for the changes in the unemployment rate among 15–75-year-old residents of Latvia, including the popular LSTAR, ESTAR, and LSTAR2 models, as well as the recently introduced ASTAR model with an asymmetric transition function. For their estimation, special modifications of the only available function in the tsDyn package of R software for the classical logistic smooth transition autoregressive model (LSTAR) are used. The constructed models are also compared with a linear autoregressive model (AR), an autoregressive model with Generalized Autoregressive Conditional Heteroscedastic (GARCH) errors, and a self-exciting threshold model. The first lag of the dependent variable and the inflation rate are used as threshold variables. LSTAR2 with the first lag as the threshold variable provides the best fit compared to the other constructed models for these data. However, other STAR models may provide a significantly better out-of-sample forecast. Compared to RMSE, the ASTAR out-of-sample forecast performs better on different horizons. Using the inflation rate as an external threshold variable does not improve the model. The study indicates that the new R functions may be useful for economic data analysis.