| Abstract [eng] |
To ensure additional financial security in retirement, Lithuania operates a second-pillar pension accumulation system, whose fund performance is closely linked to the dynamics of global financial markets. Since financial markets are characterized by a complex structure of interdependencies, fluctuations across different markets, risk transmission effects and structural changes can directly affect pension fund returns and their risk levels. Traditional econometric methods do not always adequately capture such dependency structures; therefore, this study applies network theory methods. In this research, the interaction between Lithuanian second-pillar pension funds and external financial markets is analyzed using correlation networks, the graphical LASSO method, and the Diebold-Yilmaz shock transmission framework. Correlation networks are used for an initial assessment of the co-movement of financial returns, while the graphical LASSO method allows the identification of direct conditional dependencies between assets. The Diebold–Yilmaz method is applied to analyze directional financial shock transmission relationships. The correlation analysis shows that returns of pension funds managed by the same management company are typically strongly interconnected; therefore, representative funds from different managers were selected for further analysis. Partial correlation networks constructed using the graphical LASSO approach made it possible to identify direct conditional dependencies between pension funds and external financial instruments, eliminating the influence of indirect relationships. Meanwhile, the Diebold–Yilmaz framework enabled the assessment of directional channels of financial shock transmission and the identification of which assets act as major risk transmitters and receivers within the financial network. The results show that Lithuanian second-pillar pension fund returns are closely related to global financial market dynamics; however, the structure of these dependencies is not stable and changes over time. Dynamic network analysis reveals clear structural breaks coinciding with significant macroeconomic and geopolitical shocks that have affected the sensitivity of pension funds to external markets. The findings also indicate that market signals extracted from financial networks improve the accuracy of short-term return forecasting, especially when information on directional shock transmission relationships is incorporated into the models. This suggests that network theory methods are useful not only for analyzing financial dependency structures, but also for practical risk monitoring and short-term forecasting tasks. |
