Modelling Human Social Security During War

This article is a summary of the scientific work published by Olena Bazhenova, Zakharii Varnalii, Oksana Cheberyako, and Oksana Mykytiuk
The original source can be accessed via DOI 10.33111/nfmte.2023.111

‍

This paper explores the role of human social security as a fundamental driver of economic stability and development, particularly in the context of war and post-war recovery. Social security, understood as the protection of vital socio-economic interests and rights, operates at both macro and micro levels—enhancing societal well-being, mitigating the effects of structural change, and supporting individual livelihoods through assistance and income guarantees.

In peacetime, indicators such as life expectancy and GDP per capita reflect the level of social development and national prosperity. However, during periods of military conflict, the focus shifts toward indicators of protection and survival, including military personnel levels, defense spending, and the number of displaced persons. These shifts underscore the need to rethink how social security systems function under extreme conditions.

Given its significant impact on human development and economic performance, the paper argues for the use of economic and mathematical models—specifically cluster analysis and panel VAR modeling—to identify, forecast, and respond to threats to social stability. The goal is to better understand the dynamics of social security during wartime and to provide policy tools for long-term recovery and resilience.

The structure of the paper includes an introduction to the topic, a literature review, theoretical foundations, methodology and results, and concluding insights with suggestions for future research.

Theoretical Background

Life expectancy at birth is one of the most meaningful indicators when assessing the level of human social security. It encapsulates the quality of life support systems, societal well-being, access to education and healthcare, and the population’s overall capacity for self-preservation and development. As such, it serves not only as a proxy for social conditions but also as a reflection of broader economic and institutional development.

While economic growth is generally expected to enhance life expectancy, this relationship is not always straightforward. In cases where development is not underpinned by strong social policies, rapid economic gains may fail to translate into improved health outcomes or longer lives. Conversely, high life expectancy often correlates with resilient institutions, robust public services, and a stable socio-political environment.

Importantly, life expectancy is influenced not only by domestic economic and social policies but also by exposure to armed conflict, hybrid warfare, and geopolitical instability. As of 2021, global life expectancy figures varied drastically, from as low as 52.5 years in Chad to as high as 85.5 years in Hong Kong. These disparities underscore the need for a structured comparative analysis to identify countries with similar life expectancy profiles and to understand the socio-economic dynamics behind them.

To capture these dynamics, countries were grouped based on their life expectancy at birth using the k-means clustering method. This approach enables the segmentation of countries into distinct groups sharing similar social security characteristics. The analysis was conducted using 2021 data for 209 countries, resulting in four clusters with centroids at 54.21, 63.40, 73.73, and 82.31 years respectively. These clusters provide a useful framework for identifying common patterns and development trajectories across different national contexts.

The clustering results reveal not only stark contrasts in the global distribution of life expectancy but also deeper structural divisions in how countries provide for and protect their populations. This segmentation serves as the foundation for further analysis into the relationship between social security, conflict, and long-term economic development.

The clustering results reveal a pronounced disparity in life expectancy across income groups and geopolitical contexts. The first cluster, with an average life expectancy of just 54.21 years, includes seven low-income African nations experiencing protracted armed conflict. These countries suffer from systemic poverty, fragile institutions, and a lack of access to basic healthcare. The second cluster, centered at 63.40 years, includes 58 lower-middle-income countries, mainly in Africa and Asia, with India being the most populous representative.

A third cluster encompasses the majority of upper-middle-income economies—107 countries with an average life expectancy of 73.73 years. These nations are typically more politically stable, with better-developed health and education systems. At the top of the hierarchy is the fourth cluster, composed of 37 high-income countries where life expectancy averages 82.31 years, reflecting strong social protection, advanced healthcare, and high living standards.

These findings confirm the existence of what may be called a "trap of social insecurity"—a self-reinforcing mechanism where low income leads to underinvestment in human capital and inadequate social protection. This, in turn, reduces life expectancy, further undermines productivity, and slows economic growth. Military conflicts significantly accelerate this downward spiral, eroding both human development and institutional resilience.

To better understand this phenomenon, the paper focuses on five post-Soviet countries—Armenia, Azerbaijan, Georgia, Moldova, and Ukraine—all of which have experienced either hybrid wars or direct military conflicts over the past three decades. These conflicts, often rooted in ethnic tensions or territorial disputes, have varied in intensity and duration but share long-term socio-economic consequences.

The Nagorno-Karabakh conflict between Armenia and Azerbaijan has persisted since 1988, fluctuating between armed clashes and diplomatic standoffs. Moldova’s unresolved tensions with the self-declared Transnistrian Republic since the early 1990s illustrate a long-term hybrid warfare scenario. Georgia has been engaged in conflict with Russia since the early 1990s, punctuated by direct military confrontation in 2008. In Ukraine, the full-scale war with Russia began in 2022, although hybrid warfare tactics had been employed since 2014.

The impact of these conflicts on life expectancy is reflected in national health trends over the period 1991–2021. According to World Bank data, Ukraine began the period with the highest life expectancy among the five countries (69.3 years in 1991) but ended only slightly higher at 69.65 years in 2021. In contrast, Armenia achieved the highest life expectancy by 2021 (72.04 years), while Moldova recorded the lowest (68.85 years). A significant dip in life expectancy for all countries occurred in 2020, linked to the global Covid-19 pandemic.

‍

These trends suggest that even modest gains in life expectancy in conflict-affected countries can be vulnerable to external shocks and are often shaped by the effectiveness of economic and social policies. The resilience or stagnation of these indicators provides valuable insight into a country’s broader development trajectory and institutional strength.

Results

To examine the impact of public economic and social policies on life expectancy in countries affected by hybrid warfare and military conflict, a panel data model was constructed using key macroeconomic and policy-related indicators. The analysis focuses on Armenia, Azerbaijan, Georgia, Moldova, and Ukraine—countries that have experienced varying degrees of armed conflict or hybrid war since the early 1990s. Understanding the socio-economic mechanisms that influence human security and well-being in these contexts is essential for effective policymaking and strategic post-conflict recovery planning.

The core dependent variable in this analysis is life expectancy at birth, which serves as a comprehensive indicator of human development and social security. To explain its variation, a set of explanatory variables was selected to represent the structure and orientation of governmental economic and social policies. These include GDP per capita (PPP), current health expenditure per capita, military expenditure, public spending on education, the proportion of armed forces personnel within the labor force, and social contributions as a share of government revenue.

In addition, inflation, unemployment, and domestic credit to the private sector were included as broader economic control variables. To capture the effects of conflict, a dummy variable was introduced to indicate years marked by hybrid warfare or direct military aggression within the studied countries. The panel dataset encompasses five countries over a 21-year period (2000–2020), with annual data obtained from the World Bank’s World Development Indicators. This panel structure enables both cross-sectional and longitudinal analysis, making it possible to identify country-specific effects as well as temporal dynamics.

This comprehensive variable set provides the foundation for assessing how national policy instruments and conflict dynamics shape health outcomes and life expectancy in complex geopolitical environments. The next section of the study presents the estimation techniques and empirical findings based on this dataset.

Before constructing a predictive or explanatory model, it is essential to understand the distribution and variability of the key variables included in the panel dataset. Descriptive statistics for all variables used in this study are presented in Table 3, capturing the central tendencies, spread, asymmetry, and kurtosis of each indicator over the 2000–2020 period across the five post-Soviet countries under analysis.

The statistics reveal several notable trends. Life expectancy (our dependent variable) shows relatively low variation across the sample (standard deviation of 2.28 years), suggesting overall convergence within the group despite conflict exposure. Economic and fiscal variables such as GDP per capita and health expenditure show much higher dispersion, indicating substantial cross-country differences in resource availability and allocation. Inflation, in particular, displays extreme skewness and kurtosis, reflecting volatility in some of the observed economies.

To identify deeper patterns and classify countries according to their socio-economic characteristics, cluster analysis was conducted using the ten remaining explanatory variables. The objective is to construct a generalized profile of life expectancy at birth that reflects variations in public policy, institutional frameworks, and broader economic conditions.

For this purpose, Kohonen self-organizing maps (SOMs) are utilized—a neural network-based technique particularly effective for clustering in high-dimensional data environments. SOMs not only reduce the dimensionality of complex datasets but also generate visually interpretable clusters that reveal underlying structural similarities among the analyzed entities. The resulting visual layout facilitates intuitive comparisons, illustrating how each object is positioned relative to others within the multidimensional indicator space.

In this study, each country-year observation is mapped based on its values across the ten explanatory variables. The location of each object on the Kohonen map reflects the relative development of the underlying features that influence life expectancy. This approach enables a nuanced, data-driven assessment of how closely or distantly aligned these countries are in terms of human social security under conditions of conflict and transformation.

Using the ten previously identified explanatory variables (excluding life expectancy, which served only as a reference), an unsupervised clustering of five post-Soviet countries—Armenia, Azerbaijan, Georgia, Moldova, and Ukraine—was performed for the period 2000 to 2020. The clustering utilized Kohonen Self-Organizing Maps (SOM) configured with a 16×12 hexagonal grid and optimized over 5,000 training epochs. A Gaussian neighborhood function was applied to define the radius of influence during the learning process.

The result of this process was a clear segmentation of the country-year observations into six distinct clusters, as visualized in Figure 2.

Importantly, life expectancy was not included in the clustering process, but was instead used as a benchmark for interpreting the spatial layout of the resulting clusters. In the bottom right SOM labeled "Кластеры," the colors correspond to different clusters, and the colored ruler below the map provides their numeric labels.

A particularly illustrative case is Ukraine, which is divided into two distinct temporal segments. Observations from 2000 to 2013 are grouped in the cyan-colored Cluster 2 located in the lower left of the map. This isolated positioning is primarily due to the absence of active military conflict in Ukraine during that period, which is reflected in the dummy variable WtW_tWt​ holding a value of zero. In contrast, all other countries were already engaged in conflict or hybrid warfare throughout the observation period, hence their WtW_tWt​ values remain consistently at one.

From 2014 to 2020, following the onset of Russian aggression, Ukraine's observations shift to Cluster 1, where they are grouped together with all records for Moldova. This indicates a structural convergence in economic and social policy patterns in both countries during wartime. These clusters are characterized by several shared features:

• Above-average public expenditures on education (EtE_tEt​) and social contributions (CTRtCTR_tCTRt​)
• Elevated inflation rates (INFtINF_tINFt​)
• Low unemployment levels (UtU_tUt​)
• In Moldova's case, particularly low military spending (MEtME_tMEt​) and low armed forces employment (FtF_tFt​)

Despite these differences in inputs, life expectancy (LtL_tLt​) within these two clusters remains at a moderate level across the entire study period. This suggests that while conflict-affected states may preserve some aspects of social investment, these do not necessarily translate into immediate health improvements, especially under persistent military and economic pressure.

The remaining clusters include observations for Armenia, Georgia, and Azerbaijan—each with unique profiles shaped by prolonged conflict exposure and differing public spending strategies. The SOM framework thus not only reveals distinct developmental trajectories but also highlights how geopolitical shocks, institutional responses, and policy prioritization intersect to influence public health outcomes.

To explore how public economic and social policy variables influence life expectancy at birth, a panel vector autoregressive (panel VAR) model was developed. The model incorporated data from Armenia, Azerbaijan, Georgia, Moldova, and Ukraine, spanning the years 2000 to 2020. All non-stationary variables were differenced to ensure model stability, as confirmed through tests on the characteristic polynomial. The optimal lag structure was selected using the Akaike, Schwarz, and Hannan-Quinn information criteria.

The dynamic effects of policy shocks were assessed through impulse response functions, which describe the time-path of life expectancy following a one-time innovation in each explanatory variable. The graphical representation of these responses is presented in Fig. 3.

The response of life expectancy to changes in GDP per capita proved to be the most pronounced. A positive shock to GDP resulted in a slight but noticeable increase in life expectancy, peaking at 0.011% in the first year following the shock, before gradually returning to baseline by the third year. This suggests that economic growth, while not transformative in the short term, exerts a small yet measurable influence on public health outcomes.

A similar pattern was observed for public expenditure on education, where a one-time increase led to an estimated rise in life expectancy of approximately 0.004% in the initial post-shock period. However, this effect too was temporary, dissipating within three years. Health expenditure per capita also generated a marginal short-run gain, with a minor improvement in life expectancy visible in the second year after the shock.

In contrast, an increase in unemployment produced a modest decline in life expectancy in the short term. This negative impact persisted for up to three years, after which the effect was largely neutralized. Hybrid war and military conflict shocks, captured using a dummy variable, induced a slight decrease in life expectancy during the year immediately following the shock, but the effect did not continue into the longer term.

Other variables, including the share of armed forces personnel and public spending on education beyond the initial impulse, showed minimal or no sustained influence on life expectancy over the ten-year horizon analyzed.

Overall, the model suggests that while multiple policy instruments affect life expectancy to varying degrees, the most substantial impact in the short run stems from changes in GDP per capita. The influence of military conflict appears relatively limited in the observed data, likely because full-scale hostilities, particularly the Russian invasion of Ukraine in 2022, fall outside the period covered. These dynamics highlight the importance of economic development as a core component of human well-being, with targeted social investments playing a complementary but less immediately potent role.

This interpretation is further supported by the forecast error variance decomposition of life expectancy at birth, presented in Table 4, which confirms that GDP shocks explain the largest share of future fluctuations in this outcome.

Such results emphasize that macroeconomic stability and growth not only improve income levels but also contribute meaningfully to population health and longevity, particularly in countries undergoing political transition and periodic conflict.

This interpretation is reinforced by the forecast error variance decomposition in Table 4, which shows that GDP per capita shocks explain the largest share of variation in life expectancy at birth—rising from 8.3% in the first period to 8.7% by the tenth. Public expenditure on education follows, accounting for around 5%, while unemployment contributes just over 2%. Other variables, including military and health spending or conflict indicators, explain less than 1%, suggesting their limited standalone impact. These results confirm that economic growth remains the strongest short-term driver of life expectancy in the region.

Conclusions

In the context of hybrid war and military aggression, the protection of human life and the preservation of basic social rights become critical priorities. These conditions underscore the urgency of establishing a robust system for ensuring human social security—one that includes continuous monitoring, diagnosis, and timely intervention. Life expectancy at birth, as a comprehensive indicator, serves as a reliable proxy for assessing the level of human social security and, more broadly, the developmental maturity of a society.

This study identifies a “trap of social insecurity,” a negative feedback loop in which low national income erodes human capital and life expectancy through insufficient social protection, thereby stalling economic development and reinforcing poverty. This mechanism becomes even more pronounced in times of military conflict, exacerbating existing vulnerabilities and undermining prospects for recovery.

Given the strategic importance of social security for long-term economic resilience, the paper aimed to assess how shifts in public policy influence life expectancy in countries experiencing prolonged hybrid warfare or military aggression—specifically Armenia, Azerbaijan, Georgia, Moldova, and Ukraine. For this purpose, we employed a combination of clustering techniques and econometric modeling.

Clustering was conducted using Kohonen self-organizing maps, which enabled the identification of structural similarities and developmental patterns across the five countries. The analysis covered the period from 2000 to 2020 and offered insights into the interaction of economic, demographic, and social indicators under conflict conditions. While military conflicts were not found to have a significant immediate impact on average life expectancy during this period, the cluster patterns demonstrated that economic growth—particularly as measured by GDP per capita—remains a strong predictor of higher life expectancy.

To enhance the analytical depth, a one-lag panel vector autoregressive (panel VAR) model was constructed to assess the impact of economic and social policy variables on life expectancy. The model included indicators such as GDP per capita, expenditures on health and education, military spending, inflation, unemployment, social contributions, and credit availability, along with a dummy variable denoting years of military conflict.

Variance decomposition of the panel VAR model indicated that the majority of life expectancy volatility is attributable to its own historical trajectory. Nevertheless, GDP per capita emerged as the most influential policy-related determinant. Short-term fluctuations in GDP were found to produce measurable improvements in life expectancy, reinforcing the positive relationship between economic growth and population health.

Looking ahead, further research should focus on deepening the understanding of the relationships among income levels, economic development, and social security. Identifying mechanisms to break out of the social insecurity trap—particularly in countries facing political instability or post-conflict reconstruction—will be key to fostering sustainable, long-term economic growth and societal resilience.

‍

References

1. Antohi, V.M., Ionescu, R.V., Dinca, M.S., Zlati, M.L., & Fortea, C. (2023). Dynamics of the Social Security Index in the Context of the Economic Crisis in Romania. Sustainability, 15(4), Article 3616. https://doi.org/10.3390/su15043616
2. Bazhenova, O., Varnalii, Z., & Cheberyako, O. (2023). Sotsialna bezpeka liudyny v umovakh viiskovoi ahresii: indykatory ta otsiniuvannia (na prykladi Ukrainy) [Human Social Security Under Military Aggression Conditions: Indicators and Evaluation (Case of Ukraine)]. Problemy suchasnykh transformatsii. Seriia: ekonomika ta upravlinnia (Problems of Modern Transformations. Series: Economics and Management), 8, Article 07-02. https://doi.org/10.54929/2786-5738-2023-8-07-02 [In Ukrainian]
3. Boháček, R., & Belush, V. (2009). Social Security Reform in a Dynastic Life-Cycle Model with Endogenous Fertility (Working Paper 381). CERGE-EI. https://www.cerge-ei.cz/pdf/wp/Wp381.pdf
4. Fehr, H. (2016). CGE modeling social security reforms. Journal of Policy Modeling, 38(3), 475-494. https://doi.org/10.1016/j.jpolmod.2016.02.007
5. Forni, L. (2005). Social security as Markov equilibrium in OLG models. Review of Economic Dynamics, 8(1), 178-194. https://doi.org/10.1016/j.red.2004.10.003
6. Gonzalez-Eiras, M. (2011). Social security as Markov equilibrium in OLG models: A note. Review of Economic Dynamics, 14(3), 549-552. https://doi.org/10.1016/j.red.2010.07.002
7. International Labour Organization. (2023, August 2). Facts on Social Security. https://www.ilo.org/wcmsp5/groups/public/---dgreports/---dcomm/documents/publication/wcms_067588.pdf
8. Kharazishvili, Yu., Kwilinski, A., Grishnova, O., & Dzwigol, H. (2020). Social Safety of Society for Developing Countries to Meet Sustainable Development Standards: Indicators, Level, Strategic Benchmarks (with Calculations Based on the Case Study of Ukraine). Sustainability, 12(21), Article 8953. https://doi.org/10.3390/su12218953
9. Kharazishvili, Yu.M, & Grishnova, O.A. (2018). Yakist zhyttia v systemi sotsialnoi bezpeky Ukrainy: indykatory, riven’, zahrozy [Quality of life in the system of social security of Ukraine: indicators, level, threats]. Ekonomika Ukrainy (Economy of Ukraine), 61(11-12), 157-171. https://doi.org/10.15407/economyukr.2018.11.157 [In Ukrainian]
10. Kharazishvili, Yu.M., & Dron’, Ye.V. (2014). Prohnozuvannia indykatoriv, porohovykh znachen ta rivnia ekonomichnoi bezpeky Ukrainy u serednostrokovii perspektyvi (Forecasting of indicators, threshold values and level of economic security of Ukraine in the medium term perspective). NISD. https://niss.gov.ua/sites/default/files/2014-12/prognoz-5238c.pdf [In Ukrainian]
11. Kohonen, T. (2001). Self-Organizing Maps (3rd ed.). Springer.
12. Kolot, A., Herasymenko, O., Shevchenko, A., & Ryabokon, I. (2022). Employment in the coordinates of digital economy: current trends and foresight trajectories. Neuro-Fuzzy Modeling Techniques in Economics, 11, 78-123. http://doi.org/10.33111/nfmte.2022.078
13. Korzeniowska, A.M. (2021). Heterogeneity of Government Social Spending in European Union Countries. Future Business Journal, 7, Article 38. https://doi.org/10.1186/s43093-021-00084-7
14. Lopez-Velasco, A.R. (2022). Social Security as Markov Equilibrium in OLG Models: Clarifications and Some New Insights. Economic Letters, 217, Article 110707. https://doi.org/10.1016/j.econlet.2022.110707
15. Lopez-Velasco, A.R. (2016). Solving dynamic inefficiency with politically sustainable guest worker programs. Economic Letters, 148, 1-4. https://doi.org/10.1016/j.econlet.2016.09.002
16. Lukianenko, I., Oliskevych, M., & Bazhenova, O. (2020). Regime switching modelling of unemployment rate in Eastern Europe. Ekonomický chasopis, 68(4), 380-408. https://www.sav.sk/journals/uploads/0426175204%2020%20Oliskevych%20a%20kol.%20%20+%20SR.pdf
17. Oliskevych, M., & Lukianenko, I. (2020). European unemployment nonlinear dynamics over the business cycles: Markov switching approach. Global Business and Economics Review, 22(4), 375-401. https://doi.org/10.1504/GBER.2020.107455
18. Paraschiv, D.-M., Manea, D.-I., Țițan, E., & Mihai, M. (2021). Development of an aggregated social inclusion indicator. Disparities in the European Union on inclusion/exclusion social determined with social inclusion index. Technological and Economic Development of Economy, 27(6), 1301-1324. https://doi.org/10.3846/tede.2021.15103
19. Song, Z., & Tong, P. (2022). The Impact of Social Security Expenditure on Human Common Development: Evidence from China’s Provincial Panel Data. Sustainability, 14(17), Article 10946. https://doi.org/10.3390/su141710946
20. Sydorchuk, O. (2020). Intehralne otsiniuvannia rivnia sotsialnoi bezpeky Ukrainy [Integral assessment of the level of social security in Ukraine]. Naukovyi visnyk KhDU. Seriia Ekonomichni nauky (Scientific Bulletin of Kherson State University. Series Economic Sciences), 38, 62-68. https://doi.org/10.32999/ksu2307-8030/2020-38-11 [In Ukrainian]
21. World Bank. (2023). World Development Indicators [Data set]. Retrieved August 19, 2023, from https://databank.worldbank.org/source/world-development-indicators
22. World Bank. (2021, July 01). New World Bank country classifications by income level: 2021-2022. https://blogs.worldbank.org/opendata/new-world-bank-country-classifications-income-level-2021-2022
23. Yew, S. L., & Zhang, J. (2009). Optimal social security in a dynastic model with human capital externalities, fertility and endogenous growth. Journal of Public Economics, 93(3–4), 605-619. https://doi.org/10.1016/j.jpubeco.2008.10.005‍
24. Zhang, J., & Zhang, J. (2007). Optimal social security in a dynastic model with investment externalities and endogenous fertility. Journal of Economic Dynamics and Control, 31(11), 3545-3567. https://doi.org/10.1016/j.jedc.2006.12.005