Armed conflict and family formation: How the Nagorno-Karabakh conflict affected fertility and early marriage in Azerbaijan

With levels of political violence spiking dramatically in the last decades, addressing the consequences of armed conflict for population dynamics is of paramount importance. In this post, Orsola Torrisi presents findings from research on family formation in Azerbaijan, a country embroiled in a violent, yet mostly forgotten conflict with Armenia since the early 1990s.

In the morning of 27 September 2020, media reporting on Covid-19 statistics was interrupted abruptly by the news that intense fighting had resurfaced between Armenia and Azerbaijan in and around the mountainous region of Nagorno-Karabakh. If nothing else, this severe escalation in violence has revived interest in a region – the South Caucasus – where complex conflict dynamics are rarely discussed in relation to demographic research.

Literature on recent population change in this part of the world is thin (for instance, one of the few detailed and up-to-date accounts of Georgia’s demographic transformation can be found in another post in this blog), and has largely overlooked the consequences of antagonisms that, while kept dormant under communism, degenerated into armed conflicts as the USSR collapsed.

In my PhD project, I seek to fill part of this gap by examining how armed violence has affected fertility and early marriage patterns in Azerbaijan after its regained independence.

Does armed conflict affect childbearing transitions?

An intriguing question about the link between armed violence and family formation is whether, and how, conflict influences fertility. Theoretical expectations are ambiguous. On the one side, people in conflict settings may expedite the transition to childbearing and, hence increase overall fertility, for various direct and indirect reasons, for instance to buffer future economic uncertainty, compensate for child loss or because of macro-level factors like nationalist pro-natalism. Increases in childbearing may also be expected if sexual violence is used as a weapon of war. One the other side, there is reason to think that forced displacement, couple separation or intentional trade-offs between child quantity and quality (i.e. waiting until child-rearing conditions are favourable) may lead to postponement and declining birth rates.

Existing empirical evidence is inconclusive, perhaps because different types of conflict (e.g. interstate vs. intrastate) induce diverse fertility responses. Yet prior research has generally been limited to aggregate trends and has ignored the possibility that relationships between conflict and childbearing may depend on the fertility transition (i.e. the stage of the transition for a given context), and that responses to conflict might therefore vary by parity.

In my own study, recently published in Population and Development Review, I explored this hypothesis in the context of the Nagorno-Karabakh conflict between Azerbaijan and Armenia. First, given the absence of detailed previous studies, I retraced the evolution of Azerbaijan’s fertility history to determine its stage in the fertility transition. Then I employed event history models to explore the association between transitions to first, second, and third birth – and Azerbaijani women’s exposure to the conflict.

Fig 1  Total fertility rate estimated from various sources, Azerbaijan 1991–2005

Source: 2006 Azerbaijan DHS (women aged 15-39), TrasMonEE vital registration data (women age 15-49), Human Fertility Collection (ODE Collection). Shaded area highlights key conflict years.zerbaijan 1991–2005

Azerbaijan’s period TFR (Total Fertility Rate) dropped from above 3 children per woman before full-scale hostilities in Nagorno-Karabakh (which resurged in 1992), to slightly below replacement (around 2) at the start of the 21st century (Fig.1, DHS estimates). The decline was essentially in two phases. The first phase occurred during years of deteriorating economic conditions and conflict violence. As in several other ex-Soviet republics that were hardly hit by the transition to the market economy, this stage was characterised by a “stopping-sooner” behaviour, where declines mainly concerned third-order fertility. The second phase of the decline, which surfaced only a decade later (after the fall of the USSR and onset of the Karabakh conflict), is characterised by delayed first birth (Fig.2).

Fig 2   Parity progression ratios, Azerbaijan 1991–2005

Source: 2006 Azerbaijan DHS (ever married women). 3-year moving average. Shaded area highlights key conflict years.

This (pre-conflict) reduction in the propensity to have a third child, accompanied by almost universal motherhood in the early re-independence period, led me to hypothesise that – if the conflict with Armenia had any influence on different parities then – the conflict was likely to have its strongest effect on the transition to second birth. Effectively, this was what I observed in my next analyses. In particular, the risk of a second birth was around 40 percent higher for women who were exposed to the conflict (both those who never migrated from the contested territories and those who were forcibly displaced). Conflict-affected women also had a greater risk of having a second birth in highly violent years (1993-1994), while for the Azerbaijani population as a whole the hazard had already started to decline soon after regaining independence.

Plausible explanations

What might explain these results? One explanation is household risk-insurance strategies. Children may have been viewed as a long-run investment (and future resource for the household, e.g. via paid work). At the same time, the considerable economic assistance provided by the Azerbaijani government to all individuals in conflict-affected households may have exerted short-term influence over the propensity of displaced families to transition to ‘average parity’ (i.e. two children) sooner rather than later. Risk-insurance mechanisms may have also operated at the macro-level through dominant nationalist discourses on the need to maintain a demographic balance with the opposing faction. Alternatively, these results may be explained by replacement-effects. The fact that I observed a robust positive relationship between child loss during conflict years and the transition to childbearing, irrespective of the sex of the prior birth, provides some evidence in support of this mechanism.

Teen marriage?

Changes in the timing of childbearing are often preceded by shifts in marriage patterns in many low- and middle-income countries. A particularly pressing issue in conflict settings, and Azerbaijan in particular, relates to early marriages. The harmful consequences of adolescent unions are well-documented, and most countries with the highest levels of early unions are also involved in armed conflicts. This is notable for Azerbaijan because, despite data fragmentation and underreporting, official figures still indicate a considerable increase in marriages involving teenagers in the last decades (although reliable quantitative or causal evidence on the relationship is again scarce).

By exploiting spatial and cohort variation in exposure to the Nagorno-Karabakh conflict, and adopting a difference-in-difference logic, I then examined whether war affected women’s likelihood of marrying as teenagers. Women exposed to intense and frequent fighting during adolescence were significantly less likely to form unions in their teenage years, with the largest effect for women who spent most of their teens under active conflict conditions (Fig.3).

Fig 3   Predicted probabilities of teen marriage by conflict-related migration status and granular cohorts

Source: 2006 Azerbaijan DHS. As per discrete-time models with ‘frailty’ terms estimating the probability of marrying in teen ages. Notes: Sample consists of women born 1957–1984 (ages 22–49 in 2006), excluding women born 1972–1973. Subjects enter analysis at age 12. Women born 1974–1984 are considered in the “conflict cohort”. Estimates are weighted using provided survey weights.
* p<0.05, ** p<0.01, *** p<0.001.


Although we should be cautious in generalising these findings, they provide some evidence on households’ family formation decision-making in times of violence. Further, they stress the importance of taking into account the ages at exposure to violence and the duration of the experience. While conflict-induced declines in early unions appear as surprising as they are desirable for girls in violent contexts, this does not preclude adverse marriage outcomes occurring, perhaps after adolescence (e.g., Azerbaijan DHS data indicate that more than twice as many conflict-affected women who were teenagers during the conflict period married a man aged 10+ years older). Similarly, while armed violence may not seem to affect aggregate fertility trends, my research demonstrates that it can accelerate the transition to specific parities. Analysing the demographic consequences of violent conflict is of clear academic interest. Perhaps more importantly, in neglected geographical settings, demographic research is essential to guide policy and to understanding the ways that historical events shape society.

Armed Conflict and the Timing of Childbearing in Azerbaijan by Orsola Torrisi was published in Population and Development Review in September 2020.

Orsola Torrisi is a doctoral student at the London School of Economics Social Policy Department. Besides studying the effects of conflict violence on family formation outcomes, her PhD research also looks into the long-term relationship between early-age exposure to armed conflict and intimate partner violence in later life in post-Soviet Eurasia. More information on Orsola’s work can be found here.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: