EVALUATING FORECASTING MODELS FOR
UNEMPLOYMENT RATES BY GENDER IN
SELECTED EUROPEAN COUNTRIES
Ksenija Dumičić
http://orcid.org/0000-0001-7131-9455,
Berislav Žmuk
http://orcid.org/0000-0003-3487-1376 and
Anita Čeh Časni
http://orcid.org/0000-0002-2742-8619
Zagreb, Croatia
INDECS 15(1), 16-35, 2017 DOI 10.7906/indecs.15.1.2 Full text available here. |
Received: 23 January 2017 |
ABSTRACT
The unemployment can be considered as one of the main economic problems. The aim of this article is to examine the differences in male and female unemployment rates in selected European countries and to predict their future trends by using different statistical forecasting models. Furthermore, the impact of adding a new data point on the selection of the most appropriate statistical forecasting model and on the overall forecasting errors values is also evaluated. Male and female unemployment rates are observed for twelve European countries in the period from 1991 to 2014. Four statistical forecasting models have been selected and applied and the most appropriate model is considered to be the one with the lowest overall forecasting errors values. The analysis has shown that in the period from 1991 to 2014 the decreasing trend of unemployment rates in the short-run is forecasted for more Eastern Balkan than the EU-28 countries. An additional data point for male and female unemployment rates in 2014 led to somewhat smaller forecasting errors in more than half of the observed countries. However, the additional data point does not necessarily improve forecasting performances of the used statistical forecasting models.
KEY WORDS
Balkan countries, forecasting error criteria, unemployment rates by gender, statistical forecasting methods, unemployment
CLASSIFICATION
JEL: C53, E27, J29