European Union (EU) and governments are monitoring the labor market of young people since many years. We employ a Hidden Markov model to detect the dynamics of those who are Not in Education, Employment or Training (NEET) as well as the Youth Unemployment (YU) rates for the 28 EU countries over a period ranging from 2004 to 2018. The model relies on an unobservable Markov chain to account for the dependence between the observed rates at different time-occasions. The influence of key macroeconomic factors is accounted by considering a suitable parameterization to estimate their effects on the probability of a country to be in a certain state at the first time occasion and to move between states later on time. Maximum likelihood estimation of the model parameters is carried out through the Expectation-Maximization algorithm. Significant effects towards positive improvements in the examined context are estimated for the percentage of part-time contracts, the participation rate in education and training and the GDP growth. Through the decoded states we show that Italy is the worst performing country over time since it is always allocated in the cluster having the highest NEET rates along with Bulgaria. Italy is also the worst country in terms of YU rates along with Greece and Spain.
Pennoni, F., Bal-Domańska, B. (2021). Hidden Markov model to analyze NEET and youth unemployment rates comparing EU countries over time. In Book of Abstracts ASMDA 2021 and Demographics 2021 Workshop.
Hidden Markov model to analyze NEET and youth unemployment rates comparing EU countries over time
Pennoni, F;
2021
Abstract
European Union (EU) and governments are monitoring the labor market of young people since many years. We employ a Hidden Markov model to detect the dynamics of those who are Not in Education, Employment or Training (NEET) as well as the Youth Unemployment (YU) rates for the 28 EU countries over a period ranging from 2004 to 2018. The model relies on an unobservable Markov chain to account for the dependence between the observed rates at different time-occasions. The influence of key macroeconomic factors is accounted by considering a suitable parameterization to estimate their effects on the probability of a country to be in a certain state at the first time occasion and to move between states later on time. Maximum likelihood estimation of the model parameters is carried out through the Expectation-Maximization algorithm. Significant effects towards positive improvements in the examined context are estimated for the percentage of part-time contracts, the participation rate in education and training and the GDP growth. Through the decoded states we show that Italy is the worst performing country over time since it is always allocated in the cluster having the highest NEET rates along with Bulgaria. Italy is also the worst country in terms of YU rates along with Greece and Spain.File | Dimensione | Formato | |
---|---|---|---|
ASMDA_2021_Pennoni1.pdf
accesso aperto
Descrizione: Slides of the presentation at the conference
Tipologia di allegato:
Other attachments
Dimensione
947.04 kB
Formato
Adobe PDF
|
947.04 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.