Paper short abstract:

The aims are to identify factors affecting called NEET-youth and predicting risks of becoming NEET in Tajikistan, implementing econometric methods and artificial intelligence algorithms. It based on data 13.000 data on family, individual and institutional factors collected during 2014-2022.

Paper long abstract:

The aim of the paper is to identify factors affecting called NEET-youth (NEET- Not in Education, Employment, or Training) and predicting risks of becoming NEET in Tajikistan. It helps to answer the question why the level of youth unemployment is the highest in Tajikistan in the Post-Soviet countries.

Annually 120-150 thousand of young people leave or drop educational system to enter the labor market of Tajikistan, but only 25-35 percent of them are able to find job in Tajikistan, the rest part goes abroad or ends up in so-called economically inactive group - they form a category NEET, which is growing each year. Level of NEET is 35-40% in Tajikistan. Traditional methods for estimating of unemployment are weak particularly for developing countries with young population, where labor market is not able to create necessary number of productive jobs, though we have suggested new methodology for counting youth unemployment - NEET. NEET includes not only unemployed youth, but economically inactive as well. It is important for the cases when youth frequently switches between unemployment and inactivity states.

Implementing econometrical methods and algorithms of machine learning (artificial intelligence) we have tried to build models to predict risks of becoming NEET. We choose machine learning algorithms for multinomial dependent variable, and built models that based on given information predict whether a person is NEET or not, or show probability of becoming NEET. Also, in this paper we will check factors on importance, using feature importance tests.

And, models were built using big data from several large-scale quantitative data which were collected in Tajikistan during 2014-2022, including retrospective quantitative data on 2000 youth of Tajikistan from 2017, data on 5000 youth collected in 2018. Models will be updated by data of 6000 youth which is currently being collected.

Summarizing our findings, we came to conclusion that family and individual factors, like financial conditions of the family, education level of parents, migration, level of education, accessibility of jobs nearby have significant effect on youth employment. The role of gender is the highest.