Effect of Austerity Measures on Infant Mortality: Evidence From Greece Authors: Robert J Kolesar, Rok Spruk Published: 2026 ABSTRACT ABSTRACTGovernments frequently adopt austerity policies when facing economic crises, yet their long‐term consequences for population health remain incompletely understood. This paper examines the impact of large‐scale fiscal austerity on infant mortality by exploiting the Troika‐led economic adjustment program implemented in Greece beginning in 2010 as a quasi‐experimental shock. Using the synthetic control method, we construct a counterfactual for Greece based on OECD and Union for the Mediterranean countries that did not experience austerity of comparable depth or duration. Relative to this counterfactual, Greece experienced a sharp and persistent increase in infant mortality following the onset of austerity. The divergence emerges immediately after 2010, remains statistically significant throughout the post‐intervention period, and shows little evidence of full reversion prior to the COVID‐19 pandemic. The estimated effect corresponds to an average 43 percent increase in the infant mortality rate. Mortality effects are larger for boys than for girls and are concentrated in the neonatal period. Accounting explicitly for the fertility decline, we estimate approximately 854 excess infant deaths cumulatively from 2010 to 2020. Extensive robustness checks support the findings. The results identify the total effect of austerity and highlight the importance of protecting early‐life health during fiscal consolidation.Keywords: austerity, Greece, infant mortality, synthetic control method Introduction Achieving and maintaining equitable service access and vital public health outcomes necessitates large‐scale investments in health systems, capacity, and preventive care (Acemoglu and Johnson 2007). Large fiscal policy shocks to either government expenditure or taxation may have both short‐ and long‐term consequences for public health outcomes that affect equity, quality, and the provision of care (Corsetti 2012; Bailey and Goodman‐Bacon 2015; Fadlon and Nielsen 2021; Romer 2021). In 2009, Greece's infant mortality rate stood at around three deaths per 1000 live births, or roughly one half of the average mortality rate among OECD members. In the same year, the Eurozone's public debt crisis began in Greece (Chodorow‐Reich et al. 2023). To avoid bankruptcy, the country borrowed €256.6 billion from the International Monetary Fund (IMF), the European Union (EU), and the European Central Bank (Kentikelenis et al. 2011; Gourinchas et al. 2017). Greek public spending fell by 36% from 2009 to 2014 (Burki 2018, Commission, Economic, and Affairs 2023). The 2010 Troika Memorandum limited public expenditure on health to 6% of gross domestic product (Papademos et al. 2012), and by 2019 Greece was one of the few OECD member states with public health expenditure below 4% of GDP. The austerity measures introduced under the Troika framework constituted a broad and tightly bundled policy package. They included severe reductions in government health expenditure, public employee salary cuts and layoffs, reductions in minimum wages, pensions, and welfare payments, and increases in taxes and special levies, among other measures. Together, these policies weakened the capacity of the state to address social risks such as unemployment, inequality, and poverty, while exacerbating pre‐existing structural weaknesses in the health‐care system (Kentikelenis and Papanicolas 2012; Adam and Papatheodorou 2016; Economou et al. 2015). Importantly, these changes plausibly affected infant health through multiple, potentially interacting channels, including health‐system capacity and access, household income and employment conditions, and broader macroeconomic stress. In this paper, we exploit the introduction of the Economic Adjustment Program in Greece to examine the overall impact of fiscal austerity on infant mortality by estimating a counterfactual benchmark. Because the Greek adjustment program was implemented as a comprehensive fiscal and institutional package during a period of exceptional economic contraction, isolating any single mechanism, such as health‐care spending reductions, from other contemporaneous effects is not feasible with available aggregate data. Therefore, our estimand represents the total causal effect of the austerity episode as a whole on infant mortality, relative to a plausible counterfactual scenario in which such large‐scale austerity measures were not implemented. We apply a quasi‐experimental synthetic control approach to address three voids in the literature. First, we use this method to quantify the effect of austerity measures on infant mortality in a setting where randomized or conventional difference‐in‐differences designs are not credible. Second, the analysis provides evidence on the longer‐term public health consequences of large fiscal policy shocks, rather than focusing solely on short‐run responses. Third, by disaggregating the effects by sex and extending the analysis to neonatal and post‐neonatal mortality, we provide additional evidence on sex‐related and age‐specific disparities in vulnerability to major macro‐fiscal interventions. By comparing Greece with a stable donor pool of OECD and Union for the Mediterranean member states (excluding Syria and Lebanon due to instability and conflict) that did not undergo prolonged austerity programmes of comparable severity, we estimate the counterfactual scenario using the synthetic control estimator (Abadie 2021). A key advantage of the synthetic control method as a causal inference tool is that it flexibly accounts for both observed and unobserved confounders by matching the entire pre‐intervention outcome trajectory, rather than relying on the parallel‐trends assumption. For example, pre‐existing economic trends are implicitly controlled for through matching on the pre‐austerity infant mortality path, which reflects underlying economic and social conditions, including income dynamics (Barr et al. 2022). The method further allows unobserved confounders to vary over time by re‐weighting the counterfactual unit to closely reproduce the treated unit's pre‐intervention characteristics (Kreif et al. 2016; Barr et al. 2022). Against this backdrop, we document an excellent quality of fit between the infant mortality trajectories of Greece and its synthetic control group prior to the onset of austerity. The results reveal that, relative to the synthetic counterfactual, the austerity episode in Greece is associated with a marked and persistent increase in infant mortality of approximately 43% in the post‐austerity period. The increase is around one third larger for boys and remains elevated for nearly a decade. These findings are robust to an extensive battery of sensitivity analyses and placebo tests. While the adjustment program formally ended in 2017 and infant mortality rates declined somewhat thereafter, the trajectory remained substantially above its counterfactual benchmark and did not fully revert to pre‐crisis trends by the end of the sample period. Finally, we note an important interpretive consideration that is especially salient in the Greek context. Austerity coincided with a sharp decline in fertility, which may affect infant mortality rates mechanically through changes in the number of births and potentially through changes in the composition of births. Throughout the paper, our primary estimates are defined in terms of infant mortality rates. We therefore examine fertility dynamics in supplemental analyses and interpret mortality effects accordingly, distinguishing the overall impact of austerity on infant mortality from specific mechanisms operating through changes in risk conditional on birth versus changes in the composition of births. The remainder of the article is organized as follows. Section 2 reviews the literature and policy background; Section 3 presents the identification strategy; Section 4 discusses the data and sample; Section 5 presents and discusses the results, robustness checks, and limitations; and Section 6 offers concluding remarks. Background Austerity Austerity measures refer to policies that combine reductions in public expenditure with increases in taxation. Such measures can affect population health through multiple, potentially interacting pathways. A useful conceptual distinction in the literature separates direct effects on the health‐care system from indirect effects operating through broader socioeconomic conditions. Direct effects include cuts to healthcare services, reductions in coverage, and restrictions in access to care, often referred to as the health‐care effect (Stuckler et al. 2017). For example, increases in out‐of‐pocket payments can create access barriers and undermine financial risk protection (Grigorakis et al. 2018). In the Greek case, the World Health Organization reported that increases in healthcare user fees and co‐payments resulted in nearly one‐third of the poorest households incurring catastrophic medical expenditures by 2014 (Burki 2018). Indirect effects of austerity operate through labor markets, household incomes, and social protection, and are commonly referred to as the social risk effect (Stuckler et al. 2017). Rising unemployment, income loss, and poverty are all associated with deteriorations in physical and mental health. For example, increases in unemployment are associated with higher probabilities of diabetes, infarction, ulcer, cirrhosis, and nervous disorders (Colombo et al. 2018). In Greece, adult unemployment rose from 6.6% in May 2008 to 16.6% in May 2011, while youth unemployment increased from 18.6 to 40.1% over the same period. At the same time, rising morbidity and reduced ability to afford private healthcare led to increased reliance on public health services (Kentikelenis et al. 2011). This surge in demand occurred alongside sharp reductions in public expenditure, placing additional strain on the system (Kentikelenis and Papanicolas 2012). Beyond health outcomes, the three successive Economic Adjustment Programmes implemented in Greece are estimated to have had persistent negative effects on GDP per capita and long‐run growth (Efthimiadis et al. 2013). Sharp declines in disposable income combined with large increases in unemployment translated into a substantial rise in absolute poverty (Mitrakos 2014). Together, these developments underscore that the Greek austerity episode constituted a broad macro‐fiscal shock with likely impacts on health through multiple channels simultaneously. Health System Effects Prior to the 2009 economic crisis, the Greek health system faced significant structural challenges, including fragmented financing and governance, reliance on informal payments, inequities in access and quality of care, and inefficiencies such as oversupply of specialists alongside shortages of nurses (Mossialos et al. 2005, Davaki and Mossialos 2005, Charalambos Economou and Organization 2010, Economou et al. 2015, Commission, Economic, and Affairs 2023). These weaknesses shaped both the system's vulnerability to fiscal contraction and the way austerity measures translated into service delivery outcomes. Health‐related components of the Greek adjustment programmes combined expenditure reductions with structural reforms aimed at improving efficiency and cost‐effectiveness. These included the creation of a national health insurance fund to pool resources, harmonization of contribution rates, centralized procurement of supplies, and cost‐containment measures such as reference pricing, expanded use of generics, electronic prescribing, and annual spending caps (Commission, Economic, and Affairs 2023). While these reforms were intended to address long‐standing inefficiencies, they were implemented during a period of acute fiscal retrenchment. The immediate focus of health‐related austerity in Greece was on expenditure reduction, with uncertain short‐ and long‐term consequences for public health and health‐care delivery (Simou and Koutsogeorgou 2014). Large cuts to public health expenditure affected hospital staffing, operating budgets, and availability of essential supplies, exacerbating pre‐existing constraints (Ifanti et al. 2013). For example, health‐care staff salaries were reduced twice in 2010, maternal and child health services were cut by 73% between 2009 and 2012; and funding for public hospitals declined by roughly 50% between 2009 and 2015 (Burki 2018). These reductions coincided with rising hospital admissions as patients shifted from private providers and the number of uninsured individuals increased (Kentikelenis et al. 2011; Adam and Papatheodorou 2016). Consistent with this strain, mortality due to adverse events during medical treatment has been estimated at 242 excess deaths per month beginning in September 2008 (Laliotis et al. 2016). The crisis further eroded social insurance revenues as unemployment rose, wages declined, and working hours were reduced (Burki 2018). Indicators of system performance reflect these pressures, including increased patient dissatisfaction and rising unmet need for healthcare (Zavras et al. 2016; Keramidou and Triantafyllopoulos 2018; Carr and Wolfe 2019; Filippidis et al. 2017). At the same time, it has been noted that the adjustment programmes also produced some positive institutional changes, such as standardization of benefits, restoration of universal coverage, pharmaceutical cost containment, and reforms to procurement and hospital payment systems (Economou et al. 2017). This mixed record highlights the complexity of evaluating health‐system performance during periods of deep fiscal consolidation. Austerity, Income and Health Outcomes A strong association between income inequality and infant mortality is well documented in the literature (Siddiqi et al. 2015). However, this relationship is not necessarily causal. Avendano (2012) argues that social policies that reduce infant mortality tend to cluster in countries with lower income inequality, even if their effects do not operate directly through income itself. This observation motivates closer examination of specific policy environments rather than reliance on cross‐sectional correlations. At the micro and macro levels, infant mortality has been linked to both relative and absolute income as well as to unemployment (Dallolio et al. 2012). There is also growing evidence that stalled or reversed mortality improvements in several high‐income countries may be partly attributable to austerity policies following the global financial crisis (McCartney et al. 2022). Many European countries adopted fiscal consolidation measures after 2008, though with substantial heterogeneity in timing, severity, and institutional context (Staehr 2013). Figure 1 illustrates the evolution of per capita general government health expenditure for OECD and Union for the Mediterranean member states from 2000 to 2019. Comparative analyses of European responses to the crisis indicate that Greece, Spain, and Portugal implemented particularly strict austerity programmes, experienced deeper and longer economic contractions, and faced greater strain on their health‐care systems (Karanikolos et al. 2013). In contrast, Iceland rejected severe austerity measures by popular vote and experienced few discernible adverse health effects. Consistent with this pattern, Greece's general government health expenditure per capita declined by 39% between 2009 and 2014, from 1953 to 1191 international dollars. With the exception of Luxembourg, Greece is the only country in the donor pool to have implemented such dramatic and sustained reductions. By comparison, Iceland's health expenditure declined by 9.7% between 2009 and 2011 and began to recover thereafter. General government health expenditure per capita for OECD and Union for the Mediterranean member states from 2000 to 2019, in international dollars (at purchasing power parity). Source: (World Bank 2019; excludes Syria and Lebanon). A substantial body of evidence documents deterioration in self‐reported health, access to care, and service availability in Greece during the crisis, accompanied by increases in morbidity and mortality (Vandoros et al. 2013, K.N. Fountoulakis and Theodorakis 2014). Numerous studies have examined mortality outcomes during this period, including infant mortality, suicide, circulatory and infectious diseases (Kentikelenis et al. 2014a, 2014b; Laliotis et al. 2016; Filippidis et al. 2017; Kotzamanis et al. 2022; Zafeiris and Kostaki 2019; Zilidis and Hadjichristodoulou 2020; Siahanidou et al. 2019; Doetsch et al. 2023; Kubrin et al. 2022). A systematic review of 39 studies covering the period 2009–2013 concludes that the crisis was associated with deteriorations in population health, including increased mental health problems, suicides, epidemics, and worsening self‐rated health (Simou and Koutsogeorgou 2014). Infant Mortality Child mortality is heavily concentrated in the first year of life, with nearly half of all child deaths occurring among newborns (Quah 2016; World Health Organization 2022). The infant mortality rate, defined as deaths before the first birthday per 1000 live births, is widely used as a summary indicator of population health and social conditions (Centers for Disease Control 2024). This measure is particularly sensitive to structural factors such as living conditions, health‐system capacity, and socioeconomic development (Reidpath and Allotey 2003; Schell et al. 2007). In contrast, aggregate indicators such as life expectancy can mask important subgroup‐specific changes, including increases in suicide or cardiovascular mortality during periods of economic hardship (Granados and Rodriguez 2015; Fountoulakis et al. 2022). The relationship between economic downturns and infant health is complex and context dependent (Bacigalupe et al. 2016). The literature reports heterogeneous findings, reflecting the coexistence of countervailing mechanisms (Stuckler et al. 2015). For example, evidence from the United States suggests that infants conceived during periods of high unemployment may exhibit lower rates of low birth weight, congenital malformations, and post‐neonatal mortality, potentially reflecting selection and behavioral responses among parents (Dehejia and Lleras‐Muney 2004). In Greece, nationwide evidence shows disparities in infant mortality trends by human development index and rural residence (Siahanidou et al. 2019). At the same time, multiple studies document adverse effects of economic crises and unemployment on infant mortality across countries (Alexander et al. 2011; Cruces et al. 2012; Zilidis and Hadjichristodoulou 2020). Catalano et al. (2011) describe this as a “net effect” framework, in which overall health outcomes reflect the balance of pro‐ and countercyclical forces operating across different population groups. Adverse birth outcomes represent an important proximate risk factor. Pre‐term births and low birth weight are strongly associated with infant mortality, particularly when both conditions are present (Pusdekar et al. 2020). In Greece, the preterm birth rate increased by 16% between 2008 and 2010 (Vlachadis, Loukas, et al. 2024), alongside a 19% increase in low birth weight and a 23% rise in very low birth weight births (Vlachadis et al. 2025). Evidence also points to increases in low‐birth weight deliveries independent of maternal age and origin, as well as higher stillbirth rates among younger women during crisis periods (Zografaki et al. 2018). Cross‐country evidence further indicates that lower rates of low birth weight are associated with broader health‐system capacity, including coverage, hospital availability, and health‐care workforce density (Erasun et al. 2021). At least 11 published studies examine infant mortality in Greece following the economic crisis. Seven document increases in infant mortality after the onset of the crisis (Filippidis et al. 2017; Kotzamanis et al. 2022; Doetsch et al. 2023; Zilidis and Hadjichristodoulou 2020; Siahanidou et al. 2019; Fountoulakis et al. 2022; Zafeiris and Kostaki 2019). Three find increases followed by partial recovery (Michas et al. 2014, K.N. Fountoulakis and Theodorakis 2014, Vlachadis, Loukas, et al. 2024). One study reports a temporary decline in infant mortality between 2008 and 2010 followed by a rebound (Rajmil et al. 2018), though the crisis onset in late 2009 complicates interpretation. Another study finds no evidence of a health crisis when comparing 2003–2007 with 2008–2012, again using periods that do not align closely with the timing of the crisis (Granados and Rodriguez 2015). Supporting Information S1: Appendix 1 summarizes data sources, methods, and findings across these studies. Male Disadvantage According to the United Nations Inter‐Agency Group for Child Mortality Estimation, child mortality is higher among boys than girls in all countries (You et al. 2015). A large body of literature documents excess male vulnerability in the perinatal and infant periods (Naeye et al. 1971; Ulizzi and Zonta 2002; Lawn et al. 2005; Sawyer 2012; Carlsen et al. 2013; Fottrell et al. 2015). Girls exhibit a biological survival advantage in early life, with lower susceptibility to birth complications, infections, and certain congenital conditions. As a result, the ratio of infant mortality among boys to girls exceeds one in settings where access to food and medical care is broadly equal (Sawyer 2012). Environmental conditions before and during pregnancy may further amplify sex differences in infant outcomes. Preconception and prenatal stressors have been shown to disproportionately affect male fetal development (Pongou 2013). Consistent with this mechanism, a large study examining sex‐specific responses to macroeconomic conditions around birth finds that high unemployment rates are associated with a small but statistically significant reduction in birthweight among boys (Alessie et al. 2018). These patterns motivate the examination of sex‐disaggregated infant mortality responses in periods of severe economic stress. Identification Strategy Setup This study examines the effect of the Greek austerity episode on the infant mortality rate by estimating the missing counterfactual trajectory in the absence of the post‐2009 adjustment program. To this end, we apply the synthetic control (SC) estimator (Abadie et al. 2010; Kreif and Diaz Ordaz 2019; Spruk and Kovac 2020; Bonander 2021; Gilchrist et al. 2023). The synthetic control approach constructs a weighted combination of comparison countries whose infant mortality dynamics closely reproduce Greece's pre‐austerity trajectory and uses this synthetic unit to approximate how infant mortality would have evolved had the austerity program not been implemented. Because the Greek Economic Adjustment Program consisted of a bundled set of fiscal and institutional changes implemented during a period of severe macroeconomic contraction, our findings can be interpreted as the total causal effect of the austerity episode as a whole on infant mortality. This analysis does not isolate the contribution of any single policy component, such as reductions in health‐care spending, from other contemporaneous changes affecting households and the health system. Accordingly, the estimand throughout the paper is the overall difference between observed post‐2010 infant mortality in Greece and the counterfactual trajectory implied by the synthetic control. Formally, we estimate the counterfactual infant mortality path of Greece by exploiting the implicit attributes of infant mortality dynamics in other OECD and Union for the Mediterranean member countries that share similar pre‐austerity characteristics but were not subject to austerity programmes of comparable severity and duration. This procedure yields a weighted combination of donor countries that best reproduces Greece's infant mortality trajectory prior to the onset of austerity. Details of the implementation are provided in Supporting Information S2: Appendix 2. The central identifying assumption underlying the synthetic control approach is that the synthetic version of Greece's infant mortality trajectory in the pre‐austerity period provides a credible approximation of how mortality would have evolved in the absence of the austerity episode. Following the established literature, the most plausible way to achieve this is to match on the full pre‐intervention outcome path of infant mortality (Hinrichs 2012; Bilgel and Galle 2015; Kešeljević and Spruk 2024). When the outcome trajectory itself summarizes the relevant observed and unobserved determinants of mortality, auxiliary covariates become unnecessary and may even degrade performance (Kaul et al. 2015). Consistent with this logic, Kaul et al. (2022) show that optimization based on the full span of lagged outcomes more plausibly captures latent factors and can improve the performance of the synthetic control estimator while mitigating dimensionality concerns. Inference A well‐known limitation of the synthetic control method is that conventional large‐sample asymptotic inference is not available. To assess statistical significance, we therefore follow Abadie et al. (2010) and implement permutation‐based inference. Specifically, we conduct both in‐space and in‐time placebo tests to evaluate whether the estimated post‐austerity divergence in infant mortality for Greece is unusually large relative to placebo effects. In the in‐space placebo analysis, the austerity episode is iteratively reassigned to donor‐pool countries that were not subject to deep, Troika‐imposed adjustment programmes, and synthetic controls are constructed for each placebo country. This generates a distribution of placebo treatment effects against which the magnitude of the Greek effect can be compared. In the in‐time placebo analysis, the austerity intervention is deliberately assigned to incorrect pre‐treatment years to test whether similarly large mortality divergences emerge in periods when no policy change occurred. The absence of comparable placebo effects strengthens the interpretation that the observed post‐intervention divergence reflects the impact of the austerity episode rather than spurious pre‐existing trends or coincidental shocks. Data and Samples Outcomes The primary outcome of interest is the infant mortality rate. Using the estimates from United Nations Demographic Yearbook, we approximate the mortality rate as the total number of deaths in a given year of female or male children less than 1 year of age, divided by the total number of female or male live births in the same year, multiplied by 1000. The series is calculated from the estimates and projections of the number of child survivors at the age of one disaggregated by sex. Specifically, we compare the infant mortality trajectories between Greece and its donor pools both for boys and girls to better understand sex‐specific effect disparity of fiscal austerity. 1 Figure 2 depicts the overall infant mortality trajectories in Greece and OECD countries for the period 1991–2020. A broad comparison of trajectories shows that prior to the onset of the fiscal austerity measures, the Greek infant mortality trajectory appears to be substantially below the OECD average. Two detailed patterns emerge after the implementation of austerity measures. First, the austerity measures appear to have pushed the mortality trajectory of Greece closer to the OECD average and have decimated the country's noteworthy and substantial advantage compared to its OECD peers. Second, the infant mortality rate increased in absolute terms following the introduction of the austerity measures. Apart from Syria, Greece was the only nation in the Mediterranean basin and the only OECD member experiencing an absolute increase in infant mortality rate from 2009 until 2017. By the beginning of 2016, the Greek infant mortality trajectory evolved in tandem with the OECD average prior to the onset of the COVID‐19 pandemic. Post‐2017 infant mortality recovery may be attributed to the enactment of the National Health System law establishing two levels of primary healthcare: peripheral clinics and health centers. The later provide specialized services including maternal care and public health (Kampouraki et al. 2023). Consistent with the aggregate, cross‐country nature of the data, the analysis does not observe individual‐level maternal characteristics nor specific causal pathways, but rather focuses on estimating the total effect of the austerity episode on infant mortality. Infant mortality trajectories in OECD and Greece, 1991–2020. Table 1 reports pre‐austerity infant mortality imbalance between Greece, its synthetic control group and the overall donor pool based on the full‐outcome path optimization of Greece and its convex hull of mortality attributes of the other donor pool members. The comparison of outcome balance show that in the absence of convex optimization and synthetic matching between Greece and the donor pool, a simple comparison of pre‐austerity mortality trajectories may not be able to uncover the overall mortality effect associated with the austerity measures. The overall bias between Greece and the donor pool in the absence of the convexity‐based analysis is between 26% and 56% higher than the observed mortality rate values for Greece. By contrast, the estimated bias between Greece and its synthetic control group in the pre‐austerity period is between 0.13% and 1.75%, respectively. Using full outcome path optimization to seek exact mortality attributes of donor pool countries that fall within the convex hull of Greek mortality dynamics nearly erases estimated bias from the comparison. It should be noted that low bias does not appear to be driven by disproportionately highly‐leverage pre‐austerity mortality rate in the benchmark years since the relative weight of each benchmark mortality rate is relatively evenly distributed across the pre‐austerity period, and, as such, does not pose an unresolved caveat for valid inference on the treatment effect of the austerity measures. Figure 3 graphically depicts standardized bias in the mortality rate comparison across the full span of benchmark years in the pre‐austerity period between Greece, its synthetic control group and the donor pool. The comparison indicates that the synthetic control group from the donor pool provides an excellent quality of the fit with the Greek infant mortality trajectory in the pre‐treatment period. TABLE 1Pre‐austerity outcome path imbalance. Vˆ Xt