Institutional Repository of Vilnius University
Lietuvos įmonių darbuotojų kaitą lemiančių veiksnių modeliavimas /
Abstract
dc:descriptionEmployee turnover is a persistent challenge in the Lithuanian labour market, with significant implications for companies' operational costs and long-term productivity. This thesis investigates the company- and sector-level determinants of employee turnover in Lithuanian manufacturing, trade, and construction firms over the period 2023--2025. The empirical analysis is based on "Sodra" panel data comprising 251,649 firm-month observations from 7,816 unique firms. Three methods are applied: linear regression, panel fixed effects regression as a variance decomposition diagnostic, and random forest regression. Unlike existing studies using individual-level human recources data, this study applies random forest to company-level data from a national social insurance registry. The hiring rate is found to be the dominant predictor of turnover across all models. The relative wage ratio is statistically significantly associated with lower turnover, and firm age also has a significant negative effect. Panel fixed effects decomposition confirms that these factors reflect structural between-firm differences rather than within-firm temporal dynamics. The sectoral institutional confidence index is statistically insignificant in linear regression but reveals a modest nonlinear contribution in the random forest model. Comparing the three sectors, the structure of factor importance differs: manufacturing is dominated by hiring dynamics, trade shows a more balanced distribution of importance, and construction exhibits the highest turnover volatility. The random forest models complement linear regression and fixed effects by capturing nonlinear structure that linear methods do not detect.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Kupčikas, Vilius,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD308333185&prefLang=en_US