This chapter starts by reviewing available surveys to highlight factors influencing displaced persons’ intentions to return. It then presents original data analysis, exploring whether intentions diminish over time, how intentions translate into actual returns, and what is known about current returnees. Particular attention is given to housing, employment, and access to public services including health and education – all constitutionally guaranteed rights in Ukraine and areas where policy action can meaningfully support return and reintegration. The chapter concludes with simulations illustrating how many Ukrainians may return under different hypothetical scenarios, providing insights into the preparedness of the current systems to support return and reintegration and identifying areas where further support and progress are needed.
Ukraine’s Strategic Response to the Displacement Crisis
Abstract
Return intentions are diverse and evolving over time. Surveys show that the share of individuals intending to return has declined since 2022, with notable differences by host region and individual profile.
Security and family unity are the strongest determinants of intention to return. However, housing, employment and access to services are the most influential factors within the reach of policymakers.
Housing and employment conditions strongly shape return intentions. Intact or reparable housing and the prospect of decent work encourage return, while destroyed housing and weak labour market opportunities act as major barriers.
Public services influence return intentions in both positive and negative ways. Healthcare access in Ukraine is often viewed more favourably than abroad, making it both a pull factor and a reason for short-term visits. By contrast, the continuity of children’s education in host countries is frequently cited as a reason for delaying or avoiding return.
From the earliest months of Russia’s aggression, questions about whether and when displaced Ukrainians might return have been at the centre of discussions in both Ukraine and host countries. Alongside efforts to provide immediate support, authorities in Ukraine and abroad sought to anticipate the future course of displacement. Numerous surveys were launched in early 2022 and in subsequent years by international organisations, as well as analytical and research institutions in Ukraine and in host countries, to understand the needs of displaced populations and to capture their views on eventual return.
By 2025, more than 60 surveys containing questions on intentions to return had been conducted among individuals displaced by the aggression, across Ukraine and host OECD countries (see Annex 4.C). While the results are not directly comparable across the majority of surveys due to differences in definitions, sampling methodology, timeframe and location, they nevertheless allow for a general picture of the key factors underpinning return intentions to emerge. They provide valuable insights into how displaced persons weigh different aspects of life when considering return. This information is critical for policymakers, who must plan for the eventuality of return and reintegration even before wider returns begin.
Return intentions change over time and are diverse
The review of surveys undertaken for this report has highlighted not only the diversity of displaced persons’ experiences, but also the fluidity of their intentions. They shift over time as security conditions, the situation in Ukraine, and prospects in host countries evolve.
Overall, the observed trend across intention surveys is that the share of individuals expressing an intention to return is gradually declining over time. For example, in UNHCR’s surveys among Ukrainians displaced abroad, 83% expressed hopes of return in 2022. This share fell to 76% in 2023 and to 61% in 2024 (UNHCR, 2024[1]). Similarly, the survey by the Centre for Economic Strategy, a Ukrainian think tank, recorded a drop from 74% in early 2023 to 63% later that year, and to 43% by the end of 2024 (Centre for Economic Strategy, 2025[2]). As shown in the following sections of this chapter, however, caution is needed in interpreting this trend as a decline in the intentions to return of specific individuals, as different respondents are surveyed each time.
Differences are also apparent between host regions. The SAM-UKR survey conducted in 2023 found that nearly two in three respondents in neighbouring Central and Eastern European countries (62%) were certain they would return, compared with only one in three (33%) in Scandinavian countries (EUAA/OECD, 2024[3]).
Intentions also differ by profile, with higher levels of expressed return among single‑parent female‑headed households, older people, households with dependants, and those maintaining stronger ties with Ukraine (UNHCR, 2024[1]).
Security situation and family reunification remain the key factors shaping return intentions, though other considerations matter as well
Not surprisingly, the security situation, the course of the war and the threat of active hostilities remain the key factors reducing intentions to return. Across surveys, an improved security situation and family reunification consistently emerge as the main reasons for considering return (IMPACT, 2023[4]; EUAA/OECD, 2024[3]; MPI and IOM, 2024[5]). At the same time, these figures illustrate the limitations of relying on headline numbers alone. Expressions of hope, plans, or reluctance to return often fluctuate with the immediate context and should not be interpreted as forecasts of future movements.
While safety and family unity are consistently the most important considerations, they are largely beyond the direct influence of policymakers. By contrast, factors such as the condition and availability of housing, access to employment opportunities, and the provision of public services are areas that remain within the reach of policymakers and where policies and specific measures can have an impact. Numerous surveys show that, indeed, beyond security, these factors are the next major elements that shape return intentions. Understanding how displaced persons perceive these factors, and how they interact with broader considerations, is therefore essential for preparing policies that can support safe, voluntary, and sustainable returns when safety conditions permit.
In this light, the rest of this section explores these three domains in greater depth. To improve comparability, the discussion focusses on seven large‑scale surveys of displaced Ukrainians that contained sufficiently detailed and comparable questions on access to housing, employment, and social services. Where surveys were conducted in several rounds, the most recent results are presented, with earlier rounds referenced only where they provide unique insights not captured later. A full overview of the selected surveys and the methodology for selecting them is provided in Annex 4.C.
Housing as a factor shaping return intentions
Accommodation is one of the most frequently mentioned factors shaping displaced persons’ intentions to return after security concerns. Taken together, the reviewed surveys suggest that the condition of housing in Ukraine is closely linked to intentions to return. Intact or reparable housing in Ukraine increases the likelihood of displaced persons considering return, while destroyed, inaccessible, or unaffordable housing is often described as a decisive barrier. Access to alternative housing, sometimes combined with financial or in-kind support, can encourage return even under less favourable circumstances.
Intact housing and ownership make displaced persons more likely to plan a return
Evidence indicates that displaced persons with undamaged or reparable housing more often report plans or hopes to return than those without. Responses in the sixth round of UNHCR’s survey of intentions and perspectives of refugees show that externally displaced persons who owned a property that was not damaged were more likely to report plans to return within 12 months (UNHCR, 2024[1]). The YOUkraina survey shows a similar pattern among those abroad, with a larger share planning or hoping to return where housing was intact than where it was destroyed (60.6% compared with 52.8%) (Factum Group, 2023[6]; Vox Ukraine, 2024[7]).
Even damaged housing appears to have a more positive effect on the intention to return than uncertainty or lack of ownership. In the same UNHCR survey, externally displaced persons with damaged property more often reported hopes to return one day than those who did not know the status of their property or who did not own one (64% compared with 50%). The pattern was even more pronounced among internally displaced persons. Ownership of undamaged property was associated with a higher share planning to return within 12 months, while those without property were less likely to hope to return one day than others (56% compared with 70%).
Destroyed, inaccessible or unaffordable housing is a frequently cited barrier
Several surveys identify damaged, inaccessible, or unaffordable housing as a factor behind decisions not to return, including changes of mind among those who had previously intended to do so. In the fifth round of UNHCR’s intentions survey (UNHCR, 2024[8]), among externally displaced persons whose prior plans to return permanently did not materialise, 20% cited property that was damaged or inaccessible. Among those who had previously hoped to return one day but no longer did, 23% gave the same reason. For internally displaced persons, the corresponding figures were 7% and 12%. More generally, for IDPs, destroyed or inaccessible housing in the areas of origin was cited as the most important barrier for return. The YOUkraina survey found that among respondents not planning to return, 14% reported having no place to live and 24% cited the high cost of housing. Among those whose housing was destroyed, 40% reported not planning to return (Factum Group, 2023[6]; Vox Ukraine, 2024[7]).
These concerns extend beyond individual housing to occupation and broader issues of infrastructure and community viability. In mid-2023, among externally displaced persons who were not planning to return, 11% stated that this was because their locality was under occupation (Factum Group, 2023[6]; Vox Ukraine, 2024[7]). SAM-UKR responses reference fears that not only homes but entire communities had been destroyed (EUAA/OECD, 2024[3]). Community-level assessments in IOM’s Conditions of Return Assessment further illustrate the scale and variation of residential damage. Most assessed locations reported up to 40% of residences destroyed, with a smaller share reporting higher levels of destruction. These locations encompassed areas where the vast majority of returnees resided (95%) (IOM, 2024[9]).
Access to suitable housing and support can encourage return once safety improves
While the end of hostilities remains the overriding condition, surveys show that access to property or alternative housing is often among the most important factors enabling return. In UNHCR Round 6, access to property or alternative housing was cited as the second most important factor for externally displaced persons after the end of war and occupation (UNHCR, 2024[1]). For internally displaced persons, it was the most important factor, reported more often by those originating from the east, where property damage was more prevalent.
The third wave of the CES survey found that 19% of respondents said free housing in a safer region could encourage them to return to another part of Ukraine, with this share rising to 29% among older respondents. In addition, 17% would be encouraged by a job offer that included accommodation and 11% by rent support (Centre for Economic Strategy, 2024[10]). In the fourth wave, housing and infrastructure items continued to be mentioned as potential enablers, including the loss of temporary shelter abroad and the restoration of infrastructure or one’s own home, although they ranked below safety and livelihood considerations (Centre for Economic Strategy, 2025[2]).
Employment as a factor shaping return intentions
Just as accommodation strongly shapes return intentions, employment opportunities are another central element influencing whether displaced persons consider returning to Ukraine. Findings from the different surveys show that concerns about economic conditions and the lack of decent work are among the most frequently cited barriers to return. At the same time, access to employment is consistently reported as one of the most important conditions that could enable return. Employment and income levels in host countries also influence return intentions, with those less integrated in labour markets abroad showing a greater inclination to return.
Limited economic opportunities are a major barrier for those considering return
Multiple surveys point to the economic situation and the lack of suitable work as important reasons for not contemplating immediate return. In UNHCR’s fifth round, among externally displaced persons whose previous plans to return permanently did not materialise, 18% cited concerns about economic opportunities (UNHCR, 2024[8]). Among those who had previously hoped to return one day but no longer did, 37% cited the worsening of economic opportunities. For internally displaced persons, the corresponding figures were 6% and 17%. Consistent with this, the YOUkraina survey reports that among those not planning to return, 58% cited the weak Ukrainian economy, 37% mentioned low wages in their field, and 30% pointed to the absence of a job or difficulty finding one (Factum Group, 2023[6]; Vox Ukraine, 2024[7]).
More recent evidence suggests that economic considerations are beginning to surpass other barriers. The latest SAM-UKR survey results indicate that the deterioration of the economic situation in Ukraine is viewed as the main barrier to return by 63% of those abroad, ahead even of security concerns at 62% (EUAA, 2025[11]).
Available employment is one of the most important conditions affecting return intentions
When asked what could facilitate return, respondents frequently highlight work opportunities. In UNHCR’s sixth round, access to employment in areas of return was the most important factor for externally displaced persons beyond the end of war and occupation, cited by 43% (UNHCR, 2024[1]). It was also the second most important factor for internally displaced persons at 21%, following access to housing. Externally displaced persons participating in the labour force in host countries mentioned work opportunities as an enabling factor more often than those outside the labour force, at 50% compared with 25%.
The CES survey reinforces this picture. In the third wave, 26% said the opportunity to find a well-paid job quickly would encourage return to a safer region rather than their home region, rising to 32% among those aged 18 to 24 and 31% among those aged 35 to 49 (Centre for Economic Strategy, 2024[10]). In the fourth wave, 35% said a well-paid job in Ukraine would persuade them to return and 40% pointed to a higher standard of living in Ukraine, with both incentives especially salient for younger adults (Centre for Economic Strategy, 2025[2]).
Labour market integration abroad also shapes intentions to return
Individual circumstances abroad are closely linked to differences in stated intentions. UNHCR’s sixth round shows that externally displaced persons in the host-country labour force were less likely to express hopes of returning than those outside the labour force, at 52% compared with 69% (UNHCR, 2024[1]). Those outside the labour force were also more likely to report plans to return within the next 12 months, with 7% of caregivers and retirees doing so, compared to 4% of those in the labour force.
Regression analysis based on the intention questions of the CES third wave further indicates that people who were students, employed, running their own business, or actively seeking employment were less likely to want to return than those outside the labour force, by 57%, 31%, 70% and 47% respectively (Centre for Economic Strategy, 2024[10]). Higher income abroad was also linked with lower willingness to return, with those in the highest income group 85% less likely to want to return than those in the lowest group.
Public services as a factor shaping return intentions
Beyond housing and employment, access to public services including healthcare and education, and various forms of social assistance, also plays a significant role in shaping displaced persons’ intentions to return. Healthcare access in Ukraine is often viewed more favourably than abroad, making it both a pull factor and a reason for short-term visits. By contrast, the continuity of children’s education in host countries is frequently cited as a reason for delaying or avoiding return, while the availability of safe and good-quality schooling in Ukraine can encourage return or settlement in particular areas.
Access to basic services is essential, and gaps can drive further movement
Access to basic services is seen by many displaced persons as indispensable both for sustaining displacement and for enabling return. In UNHCR’s intentions survey, 19% of externally displaced persons and 11% of internally displaced persons identified access to services as a key condition for return, with health and education specifically cited by 12% and 6% (UNHCR, 2024[1]). At the same time, 60% of those abroad said that if the war continued, difficulties accessing rights and services in host countries could force them to return, highlighting the degree of reliance on stable support.
Healthcare access can encourage short-term visits from abroad and potentially even return
Healthcare is consistently reported as a significant factor in return decisions from abroad. In UNHCR’s intentions survey, 33% of externally displaced persons cited access to healthcare as a reason for short-term visits, compared with 5% IDPs returning to home regions for this reason (UNHCR, 2024[1]). The YOUkraina survey similarly found that 39% of returnees said better healthcare in Ukraine, including more accessible medication, influenced their decision to return, although only 1% identified it as the main reason (Factum Group, 2023[6]). The same survey shows that dissatisfaction with healthcare in host countries grew between 2022 and 2023, with 46% of respondents reporting discontent compared with 40% the year before. Issues cited included long waiting times, high costs, language barriers and restricted access to medicines.
Education can be both a barrier to and a facilitator of return
Schooling considerations weigh heavily in decisions, but they can pull in different directions. In SAM-UKR, 37% of those not planning to return cited their children’s enrolment in host country schools as a reason for staying abroad (EUAA/OECD, 2024[3]). Meanwhile, the YOUkraina survey indicates that 26% of returnees mentioned the desire for their children to study in Ukraine as one of the reasons for returning (Factum Group, 2023[6]). Education also influences settlement choices after return. In IMPACT’s survey, 21% of home returnees and 20% of those who became internally displaced after returning said they remained in their current location because of schooling (IMPACT, 2024[12]). This suggests that parents prioritise continuity and stability in their children’s education, limiting their desire for further moves.
Individuals’ intentions regarding return are often subject to change. Those initially considering staying may later contemplate returning or even proceed with return. This highlights the inherent uncertainty embedded within migration intentions.
Sample composition of displaced individuals changes over time. A sizeable number of individuals have returned to Ukraine or moved to other countries, especially in earlier periods. As a result, a proportion of displaced individuals reporting an intention to stay in a host country increases over time, while the absolute number of such individuals may actually decline.
Individuals displaced at different points in time may have different intentions to stay from the outset. For example, those displaced closer to the start of an academic year may have an intention to stay for at least the duration of the academic year.
Even when these considerations are accounted for, the longer individuals remain in a host country, the more likely they are to express an intention to continue staying.
From the reviewed surveys it seems that intentions to return to Ukraine tend to decline over time. However, this interpretation warrants caution. Most of the reviewed surveys are conducted at a single point in time, meaning they capture different samples of individuals. Later surveys may include more recent arrivals who may differ in their motivations, expectations of stay and the time they took to prepare their move, as compared to those who arrived earlier. Moreover, later surveys do not include those individuals who have already departed, that is, those who have acted upon the wish to return that they may have expressed earlier. This shift in sample composition complicates the interpretation of trends over time.
A more robust approach involves the use of longitudinal data, which tracks the intentions of the same individuals at multiple time points. This allows for a clearer understanding of how return intentions evolve. Although such data are rare, they do exist. The IMPACT longitudinal survey is one such valuable example (for details, see Annex 4.A and (IMPACT Initiatives, 2025[13])). IMPACT Initiatives has been running a longitudinal survey of Ukrainians displaced abroad from the onset of Russia’s aggression, including those who have remained abroad and those who have returned. Designed to capture the evolving nature of displacement, it provides valuable insights into the migration dynamics, living conditions, return intentions and return decisions of the same individuals over time.
Sample composition of displaced individuals has altered over time, with a sizeable number of individuals returning to Ukraine, especially in earlier periods
Using four rounds of IMPACT data, thereafter “IMPACT longitudinal data”, our computations show how intentions of the same individuals to stay abroad evolve over time. Figure 4.1 focusses on individuals who provided responses to the survey question “What is your intention to stay in a host country within the next 6 months”1 in four periods: August 2022, June 2023, May to June 2024, and September to November 2024.
The starting point of Figure 4.1 is Panel A. In the earliest period with available data, August 2022, 747 Ukrainians displaced abroad reported their intentions, with 69% stating that they intend to stay and 31% stating that they do not intend to stay in their current destination within the next six months or are uncertain.2 By June 2023, 22.9% of this initial sample had returned to Ukraine (including 44% of those who did not plan to stay as of August 2022, or 13% of the total sample; and 15% of those who actually initially planned to stay, or 9.9% of the full initial sample). A small but significant percentage also changed their intention, and either started contemplating staying (10.4% of the original sample) or on the contrary no longer intended to stay (5.9%). Panels B and C are constructed in a similar way, but taking, as a starting point, the remaining displaced individuals (in the same location as in the first period).
Note: Panel A: Total sample in period 1: 747 individuals. Panel B: Total sample in period 2: 569 individuals. This includes those who still intend to stay, as compared to period 1, and those who switched their intention between period 1 and 2, as shown in Panel A. Panel C: Total sample in period 3: 147 individuals. This includes those who still intend to stay, as compared to period 2, and those who switched their intention between period 2 and 3, as shown in Panel B.
Source: Calculations based on IMPACT longitudinal data.
Taken together, Panels A to C of Figure 4.1 demonstrate that the share of individuals who intend to stay, as a proportion of all displaced individuals in the same host country, increased over two years (from 69% in August 2022, to 79.6% in September and October 2024). However, the absolute number of such individuals actually declined within the same original sample (from 529 in August 2022, to 495 in September/October 2024). In other words, part of the relative increase is explained by a mechanical change of the sample composition, due to the fact that some people returned back home, and others left to another country.
Importantly, individuals’ intentions regarding return are often fluid and subject to change. Those initially considering staying may later contemplate returning or even proceed with return. This highlights the inherent uncertainty embedded within migration intentions.
Individuals displaced at different points in time have different intentions to stay
In addition, comparison of intentions to stay across various intention surveys conducted at different points in time suffers from the fact that such surveys contain individuals who might have arrived at different moments. In other words, such individuals may belong to different arrival cohorts. Those who arrived later may have distinct motivations influenced by unobservable factors, such as experience of prior displacement, greater economic and social strain caused by protracted aggression, or having had more time to prepare for their move. Moreover, as the war of aggression continues, other considerations besides immediate security may enter into individuals’ decision to flee, such as fatigue from living under the war, economic hardship caused by the war, or the desire to reunite with family members and friends displaced earlier. As such, individuals arriving at a later date may have different intentions regarding their stay or return as compared to those who arrived earlier.
Calculations based on IMPACT longitudinal data support this hypothesis. Individuals displaced in July 2022 show a slightly higher intention to remain in the host country in the coming six months as compared to individuals displaced between February and June 2022 (Figure 4.2). While the differences are not sizeable, they are statistically significant (Box 4.1).
The months refer to the time of arrival in the current host country, 2022
Note: Average of responses to the question: “are you intending to stay in the current destination in the next 6 months”, across four rounds.
Source: Calculations based on IMPACT longitudinal data.
The longer individuals remain in a host country, the more likely they are to express an intention to stay longer, even when accounting for cohorts of displacement
A regression analysis based on IMPACT longitudinal data further helps disentangle the effects of both the timing of displacement (cohort effects) and the duration of displacement, while also accounting for individual characteristics (Box 4.1). It confirms the importance of accounting for cohort effects by showing that individuals who arrived in July 2022, as compared to February to June 2022, tended to show a slightly stronger inclination to remain at their destination in the coming 6 months. The period covered by the data is too short, however, to claim that all individuals who arrive later in the war of aggression have a different motivation to stay. It is possible that there is a seasonality effect: those arriving in July might be aiming to stay at least over the coming academic year, while not excluding later return or the possibility to change their mind.
Even once the cohort effects are accounted for, the duration of displacement matters as well. The longer individuals remain displaced, the more likely they are to report an intention to stay in their current location.
Going forward, these results suggest that the longer individuals stay, the more they grow attached to their new place of residence, integrate socially or economically, or feel that return is unfeasible or undesirable. At the same time, because different cohorts may have different motivations, cohort-specific strategies may be envisaged, recognising that displacements taking place at different moments in time may be associated with different intentions regarding the duration of stay, at least in the short run.
Research question
Many intention surveys conducted at different periods show that the share of displaced individuals who report that they still wish to return has diminished over time. Given this, it is tempting to think that, for many displaced individuals, the decision to return to Ukraine fades away over time. But can the decline in the shares of individuals who wish to return be taken as evidence that intentions change for the same individual as the time passes, while returns also take place? Or perhaps individuals, who arrived at a later date, had different return intentions from the outset?
Data and methodology
In order to disentangle the effects of “duration of displacement” and “cohort of arrival” on the intention to leave or to stay, longitudinal data are needed. Based on IMPACT longitudinal data (Annex 4.A), econometric panel data analysis is applied to estimate the following equation:
Intentionit = 𝛼𝑖 + 𝛽1Duration_of_displacement𝑖t + 𝛽2Cohort_of_displacement𝑖 + 𝛽3𝑋𝑖t + 𝜀𝑖t
In this equation, the dependent variable Intentionit is a dichotomous variable equal to one if, in wave t, the respondent i reports planning to stay in the current host country over the next six months (“Yes”), and zero if he or she reports not planning to stay or is uncertain. Independent variables of interest are: “duration of displacement” (measured in days) and the vector of dummy variables reflecting the “cohort of displacement” (measured in months, between February and July 2022, with February as the reference category). 𝑋𝑖t is a set of individual socio‑economic characteristics that include age, education, number of household members, and the number of children. These characteristics are measured at the beginning of each wave. Finally, 𝜀𝑖t is the error term. Given the dichotomous nature of the dependent variables, the chosen method of estimation is probit regression (panel data, random effects).
Dependent variable: dichotomous variable equal to 1 if individual reports an intention to stay in current destination over the coming 6 months, and 0 otherwise
Note: Estimation method: probit regression; panel data, random effects. Standard errors are in parentheses. The symbol (***) represents statistical significance at p<0.1; (**) at p<0.05, (*) at p<0.1. Number of observations: 305. R2 = 0.134.
Source: Computations based on IMPACT longitudinal data.
Results
Results reported in Table 4.1 show that a longer duration in displacement is associated in a significant and robust manner with an increased propensity to report an intention to stay in the coming six months. In addition, individuals displaced in July 2022, as compared to earlier arrivals, are more likely to express an intention to stay in their host countries in the coming 6 months.
These results are robust to the stepwise inclusion of additional variables, such as the presence of a family member with a disability (insignificant), employment status in host country (insignificant), employment status in Ukraine on the eve of the invasion (insignificant), income (insignificant), oblast of origin in Ukraine (insignificant), destination country (insignificant), and reason for displacement in addition to security (such as family or economic; insignificant).
Taken together, these results show that both duration of displacement and the cohort of arrival significantly shape intentions to stay in a host country.
Homesickness and desire to reunite with family are the most often reported reasons for return. At the same time, family reunification in the host country is a strong anchor for non-return.
Economic and housing factors are the key enablers of returns. Having employment in a host country significantly reduces the likelihood of return. At the same time, the fact of having been employed on the eve of the aggression in Ukraine increases the likelihood of return: returnees are often returning to their previous employment. Having good housing conditions in Ukraine is another key enabler of return.
Longer duration of displacement reduces the likelihood of return.
As the insecurity situation continues, the role of some factors (such as pre‑war employment in Ukraine) will diminish, while the role of others, such as family reunification in the host country, may be enhanced. Nevertheless, it is reasonable to expect, based on these findings, that having employment arranged and a decent housing option in Ukraine will remain as strong conditions conducive to return.
While often used interchangeably, the intention to return and the decision to return represent distinct stages in the migration cycle (OECD, 2024[14]). Intention reflects a migrant’s aspiration or hope to go back, shaped by emotional ties, perceived opportunities, or changing conditions in both origin and host countries. In contrast, a decision to return implies a concrete action, a commitment, often accompanied by logistical planning and readiness to undertake this act (see also Box 4.2). Understanding this distinction is crucial for designing policies that effectively support voluntary return and reintegration.
Gauging from the IOM General Population Survey (GPS) round 17, conducted in July and August 2024 (IOM, 2024[15]) in Ukraine, there were approximately 4.3 million returnees already at that time. Among them, 24.4% returned from abroad, 47.9% returned from another oblast in Ukraine, and 27.4% returned from displacement within the same oblast.3
The decision to return is shaped by a complex mix of structural, contextual and individual factors (Figure 4.3). Return is rarely a single event or a purely rational decision. It is often the result of evolving considerations, including conditions in both the host and origin countries, or regions in the case of internally displaced persons, as well as personal aspirations, legal status, family circumstances and access to opportunities. In addition to these factors, policies in both host and origin countries can create incentives or disincentives that influence the timing and feasibility of return.
Source: OECD (2020[16]), Sustainable Reintegration of Returning Migrants: A Better Homecoming, https://doi.org/10.1787/5fee55b3-en.
The decision to return is often explained in terms of push and pull factors, or deterrents and incentives to stay (OECD, 2020[16]; OECD, 2024[14]). Conditions in host countries such as access to legal status, employment, housing and public services can encourage individuals to stay. Conversely, lack of integration, unfulfilled hopes, or extended periods without regular status can increase the appeal of return, even if conditions in the origin country are challenging as well.
Most existing research on voluntary return has focussed on labour or economic migrants. As a result, economic and professional considerations are commonly cited as key influences, although the nature of their impact may vary (OECD, 2020[16]). In some cases, return becomes attractive when individuals believe that the skills and experience acquired abroad will be more valuable in the country of origin (Wahba, 2015[17]; Dustmann and Weiss, 2007[18]). For others, return is linked to a perceived failure to meet economic goals in the host country (Constant and Massey, 2002[19]). At the same time, migrants are more likely to remain abroad when the host country offers better economic prospects than the origin country (Song and Song, 2015[20]). Those who have attained economic stability abroad may come to associate return with financial insecurity or a decline in well-being, further reducing the likelihood of return (Toomistu et al., 2024[21]).
Beyond economic factors, return decisions are increasingly understood as socially embedded processes shaped by family, friends and community networks (OECD, 2024[14]). Family reunification, the desire to care for relatives or start a family in the origin country, and feelings of homesickness can encourage return (Konzett-Smoliner, 2016[22]). At the same time, migrants who have lost contact with their home communities or who feel unable to meet social expectations may be less likely to return (IOM, 2018[23]; OECD, 2024[14]). Personal emotional ties to the home country also play an important role. A large‑scale OECD household survey conducted in 11 countries of origin found that the most commonly reported reason for return was an overall preference for the home country (OECD, 2017[24]). Emotional and social dimensions of belonging often carry as much weight as economic considerations in shaping return intentions.
However, return patterns can vary significantly across different migration categories. Labour migrants and students tend to return earlier, often citing completed goals or social ties. By contrast, forcibly displaced persons face greater structural barriers to return. In these cases, the decision is driven less by economic opportunity and more by concerns related to safety, human rights and long-term prospects in the origin country. Research consistently finds that perceptions of safety and stability are among the most decisive factors influencing return among refugees and internally displaced persons (Black and Koser, 1999[25]; Alrababa’h, Casalis and Masterson, 2020[26]). Return is typically only considered when basic security, the rule of law (where it was broken) and access to essential services have been restored. As such, decisions are shaped more by progress in security and the restoration of public trust (in cases where it was broken) in the origin country rather than by policy incentives and the situation in host countries.
Calculations based on these data, thereafter “IOM GPS data round 17” (Annex 4.A), show that sentimental reasons, including homesickness, as well as the desire to be closer to relatives and friends, are the most important reasons for return for displaced Ukrainians, as they were cited by the largest share of returnees (Figure 4.4). Access to affordable housing and employment in original places of living are the next two most important reasons for return, cited by the second and third largest shares of returnees, amidst persisting insecurity and the fact that many returnees returned to areas that are still unsafe or were heavily damaged by Russia’s aggression (see also (IOM, 2024[15])).
Note: This figure is based on responses to the question: “Can you tell me top 3 reasons why you returned to your place of origin, starting from the most important to less important?” The responses are aggregated and weighted according to the priority of each reason: a weight of 3 is assigned to the first (most important) reason, 2 to the second, and 1 to the third.
Source: Computations based on IOM General Population Survey Round 17.
Econometric analysis based on IMPACT longitudinal data (Box 4.3) further confirms the importance of economic and housing factors, alongside a range of other factors, as well as the duration of stay and of cohort effects, in enabling the return decision.
Research question
What factors lead an individual to take a decision to return, that is, making a transition from “being in displacement” to “being back in Ukraine”?
Data and methodology
This question can be best answered, again, with longitudinal data, that is, data that track displaced individuals from one period to another, including those who continue staying and those who have returned between the two survey periods. IMPACT longitudinal data and a model set up similar to the one in Box 4.1 are used. The dependent variable “decision to return” is constructed as equal to one for individuals who report having returned to Ukraine between two survey rounds, and who also report an intention to stay there; and zero otherwise. Independent variables include time in displacement, cohort of displacement, and a set of individual socio‑economic characteristics. The model does not control for gender, as the majority of the returnee sample is comprised of women. The model is built by a further interchangeable inclusion of lagged (past) intentions to stay, reasons for displacement in addition to security reasons, and a range of other factors in host countries and in Ukraine, such as employment and housing. All models are estimated using probit panel data, random effects.
Results
Table 4.2 shows estimation results. In all regressions, non-reported controls include: age (positive and significant effect on the decision to return), educational attainment (insignificant), household size (insignificant), number of children (insignificant), household income (significant in some regressions but not robust), region of return to Ukraine (small positive and significant effect for some oblasts in central Ukraine), host countries (lower propensity to return from the Slovak Republic and Southern European countries, as compared with other European countries).
Column (1) shows the results of a baseline model, which includes duration of stay in the host country (measured as time in displacement, in days) and the cohort of displacement (measured by months of arrival, with February as the reference category). As expected, longer duration at destination has a negative statistically significant effect on the decision to return. Individuals arriving at a later date also have a lower propensity to return.
Column (2) includes a dichotomous variable “lagged intention to stay”, which is equal to one if, in the preceding survey round, the individual reported an intention to stay in the host country in the coming six months, and zero otherwise. Previous intention does not seem to affect the decision to return in a significant way, which confirms the findings of Figure 4.1, which show that intentions may change over time and that some individuals who did not initially intend to return actually do so.
Column (3) investigates whether, in addition to security reasons (common to all surveyed individuals), other reasons affected the displacement decision in the first place. Such reasons include economic motives, study, or family reunification. Reported here is the result for family reunification in the destination country, against security reasons only (non-reported effects of other reasons are insignificant). In other words, individuals who moved to a host country, as a result of Russia’s aggression, because they could join a family member already present there, are less likely to report returning to Ukraine.
Column (4) inquires into the role of selected factors in the host country that may affect the decision to return. Reported are the results for two factors relevant from a policy perspective: being employed (dichotomous variable) and accommodation (rented; versus own, provided for free by friends and family, or provided for free by the host state), both variables are lagged. Of these two, only the fact of being employed in the preceding round affects negatively the probability of having accomplished the return by the following survey round.
Dependent variable: dichotomous variable equal to 1 if individual reports having returned to Ukraine from one survey round to another, and 0 otherwise
Note: Estimation method: probit regression; panel data, random effects. Standard errors are in parentheses. The symbol (***) represents statistical significance at p<0.1; (**) at p<0.05, (*) at p<0.1.
Source: Computations based on IMPACT longitudinal data.
Column (5) investigates the role of selected factors in the home country that may affect the decision to return. Reported are again the results for factors relevant from a policy perspective: being employed on the eve of the aggression (dichotomous variable) and two variables for accommodation (rented; versus own, or provided for free; as well as housing condition: measured on a scale from 1 to 5, from very poor to very good). The fact of being employed on the eve of the aggression is a strong determinant of the decision to return. Most likely, returning individuals are returning to their previous employment. In addition, the condition of accommodation at home, though not the ownership, matters. Individuals are more likely to return to better living conditions.
The influence of a whole range of other factors was analysed, including employment downshifting in a host country as compared to Ukraine, receipt of assistance, satisfaction with the education system in host countries, satisfaction with online schooling through Ukrainian schools, self-assessed level of security, self-assessed language proficiency, number of prior visits to Ukraine, and family reunification in Ukraine, but none were found to be statistically significant, although numerous missing values and the availability of these questions only in some selected rounds might have affected the inference levels. Likewise, a large number of missing observations, and the fact that not all questions featured in columns 2 to 5 were asked in all rounds, precludes the inclusion of all variables into one model.
Results of the analysis in Box 4.3 show that, among the factors in the host country, having employment significantly reduces the likelihood of return. Accommodation type in the host country (rented, own, provided for free by friends and family, or provided for free by the host state) does not seem to affect the return decision in a significant way, once other factors are accounted for. In contrast, family reunification in the host country emerges as a strong anchor: if individuals have moved to rejoin their family, in addition to fleeing Ukraine because of the security situation, they have a lower likelihood of return.
From the home country perspective, employment and good housing conditions in Ukraine appear to be the key enablers of return. In particular, it is the fact of having been employed on the eve of the aggression that matters. Most likely, individuals who make the decision to return do so because they can return to their previous employment.
In addition to this, duration of displacement and timing of arrival are also significant factors affecting decision to return. Those who have spent more time abroad, or moved abroad in July 2022 rather than between February and June 2022, are less likely to have returned by August 2024 (the timing of the last wave of the analysed data).
Interestingly, prior intentions to stay in the host country do not predict actual return behaviour in a significant way, underscoring the dynamic nature of return and stay decision making. Other factors, including satisfaction with services, security perceptions, and language proficiency, do not seem to affect return decisions either, though data limitations may have affected these results.
As the insecurity situation continues, it is clear that the role of some factors (such as pre‑war employment in Ukraine) will diminish, while the role of others, such as family reunification in the host country, may be enhanced. Nevertheless, it is reasonable to expect, based on these findings, that the notion of having secured employment and a decent housing option in Ukraine will remain as strong conditions conducive to return.
Returns that already took place could be anticipated and hence planned for by policymakers. Most of the returns took place soon after the positive news about the Ukrainian Armed Forces’ advancement and the liberation of some temporarily occupied territories. Subsequent returns feature a certain seasonality.
Profiles of returnees are quite distinct from non-displaced individuals and those who are still in displacement. Population movements have altered demographics in the origin and receiving places, requiring adjustments to public services.
Dynamics of return – and of non-return – is strongly shaped by individual profiles, oblast of origin, and the area of displacement. The largest returns from internal displacement took place for those displaced within the same oblast (usually non-frontline), as well as for individuals returning to relatively safer areas. Understanding how many individuals will and will not return, as well as the profiles of returnees and non-returnees can help in making policy decisions on appropriate future approaches to integration and reintegration policies.
Timing of returns: Returns can be anticipated
The vast majority of those individuals who returned did so in spring/summer 2022, following the liberation of parts of Ukrainian territories by the Ukrainian Armed Forces. This is well-reflected in IOM GPS data round 17 (Figure 4.5): of those surveyed in July-August 2024 who had returned from abroad, two‑thirds had done so 730 days or more prior to the survey, which roughly corresponds to summer 2022. Two smaller spikes in returns are also observed around 450 days prior to the survey, corresponding to the late spring-early summer 2023, and less than three months prior to the survey, corresponding to the late spring/early summer 2024. Similar patterns are also evident among returnees from another oblast and from the same oblast within Ukraine.
Summing up, returns initially took place upon a significant improvement in the security situation and thereafter exhibit a certain seasonality, likely linked to the end of the academic year for families with children. Fewer returns over the colder months may also be linked to ongoing challenges with heating and power outages induced by Russia’s attacks on critical infrastructure. Going forward, these findings also suggest that returns are typically planned, and hence can, to a certain extent, be anticipated by the Ukrainian Government.
Note: Total sample: 1 166 observations. Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 17.
Profiles of returnees are quite distinct from non-displaced individuals and those who are still in displacement
Returnees have relatively distinct profiles compared to those who still remain displaced, and also compared to the non-displaced population.4
Calculations based on IOM GPS data round 17 show that, among returnees from abroad, 86.9 were women, reflecting the fact that the original displacement abroad was female dominated. The shares of women were also higher than those of men, but to a lesser extent, among returnees from a different oblast in Ukraine (66.3%), and returnees from the same oblast (63.8%). Among those who still remained in internal displacement, 66.7% were women. The share of women was the smallest among the non-displaced population: 5.4%.
Returnees from internal displacement in another oblast were older (44 years), compared to other returnees (aged, on average, 41 years) and those still internally displaced (43 years). The non-displaced population, however, were the oldest (their average age is 47.7 years).5
Returnees had a slightly larger household size (3.1 members for returnees from abroad, 2.9 for returnees from another oblast, 3.4 for returnees from the same oblast), compared with 2.7 among the non-displaced. Household size of IDPs was 2.9 members.
Returnees featured a higher proportion of individuals with higher education (53.5% among returnees from abroad, 54.1% among returnees from another oblast, 44.1% among returnees from the same oblast), as compared with IDPs (41.7%), or the non-displaced (37.8%).
Among all returnees, there was a higher share of returnees to urban areas (around 84%), compared to the location of IDPs (76.61% in urban areas) and the non-displaced population (71.21%).
Dynamics of return – and of non-return – are strongly shaped by individual profiles, oblasts of origin, and the area of displacement
Using the information on individual profiles of those who have already returned and those who are still in displacement, as well as information on the time that both groups spent in displacement, can help in understanding the dynamics of returns. Who has returned first? Who is still displaced and is likely to remain in displacement? Which individual characteristics are associated with a greater propensity to return over time? Answering these questions can help in understanding returns that already took place, as well as non-returns. As such, it can also help in better anticipating future returns and understanding not only how many people can return, but what new demographic composition will emerge as a result.
These questions can be answered using duration analysis. The duration analysis does not assume that everyone behaves the same way or returns at the same time, but instead helps identify which groups are more or less likely to return sooner.
Using the IOM GPS data round 17, conducted in Ukraine two and a half years after the start of the aggression on the territories that are under Ukrainian control, it is possible to shed light on the dynamics of return from internal displacement. A starting point is the identification of characteristics that make individuals more or less likely to return sooner from internal displacement, in a significant manner. Based on the duration analysis applied to these data (Box 4.4), such characteristics include: age, education, area of living, employment status, oblast of origin, and whether a person was displaced within the same oblast or outside the oblast of origin. Being a woman (rather than a man), having a larger household size, or a greater income do not appear to be factors that make individuals more or less likely to return sooner from internal displacement, once all other characteristics are accounted for.
Duration analysis, also known as survival analysis, is a statistical method used to study the time until an event occurs, such as returning to the place of origin after displacement. One of the most widely used models in this field is the Cox proportional hazards model.
The Cox proportional hazards model estimates the effect of various factors on the hazard rate, the probability that the event (e.g. return migration) happens at a particular time, given that it has not happened yet. It is called “proportional hazards” because it assumes that the effect of each factor is multiplicative and constant over time. The model can be applied to censored data (e.g. when it is not known whether or when someone will return), and does not require specifying the baseline hazard function, making it semi-parametric and flexible. In the context of the return of displaced individuals, duration analysis can help identify which characteristics are associated with faster or slower return.
In this report, the Cox proportional hazards model is applied to the IOM GPS data round 17. Table 4.3 presents the results of a Cox proportional hazards regression model estimating the hazard of return over time, given the current number of days in displacement for those who have not returned yet, and the days in displacement for those who have already returned, for a sample of 2022 individuals, including current IDPs and returnees from internal displacement. The model estimates the effect of the following covariates on the hazard (likelihood of return): gender, age, household size, education level, area of living, employment status, income level, place of origin, and whether a person is internally displaced within the same oblast or outside the oblast. Hazard ratios above one, when significant, indicate a higher chance of return for individuals with the corresponding characteristics.
Event equal to 1 is the return
Note: Reference categories are as follows: male (for female); age 40+ (for age below 40); secondary or below (for technical and higher education); rural (for urban); out of the labour force (for employee, self-employed, and unemployed); income above the mean (for income below the mean); other regions of Ukraine (for three dummies denoting the type of place of origin); displaced in another oblast (for displaced within the same oblast). Category “technical” includes technical and non-completed higher education. The symbol (***) represents statistical significance at p<0.1; (**) at p<0.05, (*) at p<0.1. Sample restricted to population aged 18‑65.
From this model, there is no significant difference in return hazard between men and women (from internal displacement), nor between individuals with a greater household or a different level of income.
Individuals aged less than 40 years have a 23.9% lower hazard of returning compared to the reference group (40 or above).
Higher education is associated with a 34.3% higher hazard of returning compared to the reference group (secondary education or below).
Individuals currently living in urban areas have a 40.0% higher hazard of returning compared to those in rural areas.
Being self-employed, as compared to any other employment statuses, is associated with a 31.2% higher hazard of returning.
Place of origin variables were grouped into four categories, based on the Ministry of Reintegration’s instruction “On Approval of the List of Territories in which Military Operations are (were) Conducted or Temporarily Occupied by the Russian Federation”, first published on 22.12.2022, with the updated version as of 17 August 2024 taken into account. The first category includes oblasts that were partly occupied (Donetsk, Kherson, Luhansk and Zaporizhzhia oblasts). The second category includes oblasts that were classified as having past or present active hostilities, as well as possible active hostilities (Chernihiv, Dnipropetrovsk, Kharkiv, Mykolaiv, Odesa, Sumy oblasts as well as the oblast of Kyiv). The third category includes all other oblasts. The fourth category includes the city of Kyiv. The results show that individuals who lived in partly temporarily occupied oblasts before the start of the 2022 full-scale aggression had a nearly 80% lower hazard of returning; individuals who lived in oblasts with active hostilities had a nearly 45% lower hazard of returning; and individuals living in the city of Kyiv had a 39.0% higher hazard of returning compared to displaced individuals from other oblasts.
Finally, displacement within the same oblast, as compared to those displaced to a different oblast, is the strongest predictor of return: displacement within the same oblast increases the hazard of return by nearly 5.4 times.1
These estimates are used to construct survival curves representing the estimated probability of remaining displaced over time for groups of individuals, varying characteristics of interest while keeping all other characteristics at their mean or reference values (Figure 4.6 and Figure 4.7).
1. Unfortunately, information on housing could not be included, as it is not contained in this round of data.
Once these characteristics have been identified, survival curves – graphs that show the probability of remaining displaced over time – can be constructed for individuals with and without these characteristics.
Figure 4.6 shows survival curves for individuals that differ only in one of the characteristics at a time, holding all other characteristics constant. Each curve represents the estimated probability of remaining displaced (i.e. not having returned) at each point in time since displacement. A steeper decline in the curve indicates a higher rate of return (shorter displacement duration), while a flatter curve suggests longer displacement durations.
For example, from panel A, the curve for individuals who are younger than 40 years of age lies above the curve for individuals who are aged 40 or older, suggesting that younger individuals remain displaced longer. The age effect, however, disappears over time.
From panel B, the effect of education has a divergent effect on returns over time. Individuals with higher education tend to return faster as compared with others. As a result, two years after displacement, around 65% of displaced individuals with secondary and technical education level remain displaced, in contrast to around 55% of displaced individuals with higher education.
Note: Survival curves based on Cox model predictions (Box 4.4) holding all other covariates at their mean or reference values.
Source: Computations based on IOM General Population Survey Round 17.
From Panel E, around 20% of individuals internally displaced from one of the oblasts partly occupied in August 2024 (including Donetsk, Kherson, Luhansk and Zaporizhzhia oblasts) have returned over 2022 to 2024. Around 30% have returned to oblasts with past, present, or possible active hostilities (Chernihiv, Dnipropetrovsk, Kharkiv, Mykolaiv, Odesa, Sumy oblasts as well as the oblast of Kyiv). The highest returns, at almost 50%, were to the city of Kyiv.6
The most striking result is in Panel F: almost 90% of individuals displaced within the same oblast have returned within two years, compared with only about 30% of individuals displaced to other oblasts. Moreover, the majority of returns happened within the first six months of displacement. It is worthwhile noting that this dynamic may be quite specific to the early months of the war, that will not necessarily be also observed with future returns. The reason for this is that the initial displacement during the early months involved significantly higher shares of individuals displaced within their non-frontline oblasts of origin (differently from subsequent displacements). As a result, many of the early intra‑oblast return movements likely occurred within non-frontline areas, rather than representing large intra‑oblast movements within frontline areas. Similar results are also found in other studies, such as (IOM, 2025[27]), which show that among those who returned a long time ago, returns within non-frontline areas dominated, while returns within frontline areas became prevalent later on.
It is also useful to understand the dynamics of return for individuals who accumulate several traits that amplify the likelihood of return. For this, artificial profiles can be constructed. For example, one such profile can be for individuals who possess all individual characteristics that increase the hazard of return (aged 40 or older, having a higher education level, living in an urban area, being self-employed, returning to the city of Kyiv), while the second profile can be for individuals who possess all characteristics that minimise the hazard of return (younger, having secondary education or lower, living in a rural area, not being self-employed, returning to oblasts with active hostilities). Figure 4.7 shows the probabilities of remaining displaced for these two profiles. Given the strong effect of the oblast of origin and the oblast of displacement (whether within the same oblast or not), it focusses only on internal displacement outside the initial oblast of origin and only on returns to oblasts that are not temporarily partially occupied.
The results of Figure 4.7 suggest that, two years after the start of internal displacement (outside their oblast that was not partially occupied as of August 2024), about 65% displaced individuals with individual traits that increase the likelihood of return have effectively returned. Among individuals with characteristics that do not increase the likelihood of return, over 80% still remain displaced two years after their initial displacement. It is these individuals who are most likely to continue staying in displacement and are likely to be in need of specific targeted integration policies.
Note: Survival curves based on Cox model predictions (Box 4.4) holding all other covariates at their mean or reference values.
Source: Computations based on IOM General Population Survey Round 17.
There are notable disparities in home ownership and housing conditions, including damage, among different groups of the displaced population. Returnees and non-displaced individuals are more likely to be homeowners than other population groups. In contrast, even two years after displacement, IDPs are more likely to rent their current dwelling or be hosted for free. Returnees from internal displacement are more likely than others to live in damaged housing.
IDPs and some returnees often lack legally recognised documents to prove their housing arrangements. This may hinder integration and reintegration efforts.
State programmes to address housing issues are well known. Still, more progress can be made to ease the application process, especially for some groups such as women returning from abroad.
Notable disparities are also observed in employment. IDPs are the least likely to be employed. When they are employed, they are most likely to work as employees, rather than self-employed. Particular attention should also be paid to employment quality: the share of informal and casual employment is sizeable across all groups of displaced persons, as opposed to the non-displaced.
There is a higher share of poorer households among IDPs and the non-displaced as compared to returnees. Yet, income difference across households are actually driven by factors other than displacement status, notably education level and employment. This underscores the importance of designing long-term income support mechanisms that are needs-based rather than strictly status-based, as status alone may not accurately reflect a household’s vulnerability; as well as further promoting education, skill development, and employment-enhancing policies.
Further understanding how returnees are faring on the ground is essential for shaping effective reintegration policies. What are their outcomes in terms of housing, employment and income, which are the key considerations beyond security concerns shaping return intentions and actual returns? The analysis that follows provides answers to these questions. Unless stated otherwise, it is based on the IOM GPS rounds 16 and 17. The analysis distinguishes returnees from abroad and from internal displacement, in order to shed light on potential barriers to reintegration that may exist among these groups and to show whether there is space for public policies to improve their outcomes in a targeted manner. It also compares the outcomes of returnees with the outcomes of non-displaced individuals and of IDPs. Large differences may pose a challenge for social cohesion.
Housing situation among returnees, non-displaced population and IDPs
The full-scale Russian war against Ukraine has created tremendous challenges in the housing sector. As of late 2023, over 10% of the housing stock was destroyed (Miloserdov, 2024[28]; World Bank et al., 2024[29]), and even more was damaged. Areas that received large inflows of IDPs saw increased pressure on their housing sector, pressure that resulted in rising prices in both rental and mortgage markets, increasing unaffordability of housing and creating an imbalance between the quality and the cost of housing (Bobrova et al., 2025[30]).
Moreover, a major change occurred in the housing tenure structure. If, before 2022, only about 4% of housing was rented, this figure rose to 14%, with significant pressures on large cities where renters may account for about a quarter of the population (Bobrova et al., 2025[30]). The absence of a strong regulatory framework (which was of lesser relevance when the rental market was small) exacerbated the extent of the shadow-renting in the years since 2022, and led to an imbalance of power in favour of landlords, leaving tenants, especially vulnerable groups that include IDPs and some returnees, unprotected (Bobrova et al., 2025[30]).
Returnees and non-displaced are more likely to be homeowners, in contrast to IDPs, who are more likely to rent or be hosted
Based on IOM GPS round 16, in the spring of 2024, IDPs exhibited the highest reliance on rental housing, with nearly 59% renting their accommodation. Although lower, rental rates were also notable among returnees, particularly those returning from other oblasts (16.2%), followed by those from the same region (13.3%) and from abroad (12.8%). Homeownership was most prevalent among non-displaced individuals and returnees from abroad, with similar rates of 72.2% and 71.1%, respectively. Additionally, a significant proportion of IDPs – 31.5% – were accommodated without charge, a figure that surpassed all other groups (Figure 4.8).7 These results support the notion that the availability of own housing can be an important factor in return decisions from abroad. The fact that returnees from internal displacement in another oblast had a relatively high rate of renting corroborates the fact that, to a greater extent, returns were driven by necessity and inability to stay in other locations, and that return was happening even to heavily damaged areas.
Note: Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 16.
Returnees from internal displacement are more likely than others to live in damaged housing
Looking at dwelling conditions (Figure 4.9), on average, 15.02% of returnees live in damaged but habitable housing. The highest proportion is among returnees from within the oblast (18.59%) or another oblast in Ukraine (16.87%). The findings again underscore that a significant share of returns occurred to areas heavily affected by damage, highlighting the continued vulnerability of returnees upon their return (IOM, 2024[31]). Among returnees from abroad, 9.2% reported living in damaged housing, a proportion comparable to that of IDPs (10.6%). The lowest, though non-negligeable, share of individuals living in damaged housing was recorded among the non-displaced (7.6%).
Note: Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 16.
IDPs have the greatest difficulty proving their current housing arrangement with legally recognised documents, but so do some returnees
An important consideration for policies to improve housing, as well as for the return and reintegration process, is whether individuals possess legally recognised documentation for their current housing arrangements, such as proof of property ownership or valid rental contracts. The absence of such documentation can hinder access to housing programmes, including compensation for damaged housing, legal protection and public services, and complicate long-term settlement or reconstruction efforts.
The highest share of individuals who do not have legally recognised documents for their current housing arrangement (Figure 4.10) is among IDPs (37.5%). This is followed by returnees from within the oblast (16.1%), another oblast (15.7%), and returnees from another country (14.3%). While a non-negligeable share of the non-displaced population (11.9%) also reports this issue, this share is nevertheless the smallest as compared to other groups.
Note: Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 16.
Across all population groups, individuals renting, rather than owning, their current dwellings face the most critical challenges in terms of securing legally recognised documentation. Among IDPs renting their housing, 39.8% do not have legal documents. For IDPs in particular, this situation may exacerbate their vulnerability. Indeed, 16.8% of IDPs report that the absence of legal documentation limits their ability to stay in their current place as long as they wish (in contrast to around 5.5% among other groups), in addition to 40.6% of IDPs who may not be able to stay due to inability to pay rent and/or utilities (Figure 4.11). Also among the IDPs, there is the greatest proportion of homeowners who lack proper documentation (13.7%, against around 4.0% for other groups).
Note: Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 16.
There is decent awareness of state programmes to address housing issues, though application proves difficult, especially for women returning from abroad
In response to the severe and widespread destruction of residential infrastructure due to the Russia’s full-scale aggression, the Ukrainian Government has implemented targeted compensation measures to support affected populations (described in Chapter 3). One of such measures is eRecovery (eVidnovlennya), a compensation programme for damaged or destroyed housing located in non-occupied territories outside active combat zones. If housing is damaged as a result of military action, the owner can report the damage through Diia app. Following an assessment by a commission, monetary compensation may be granted.
Among IDPs, there is both a higher percentage of individuals who applied for the programme (18.7%) and of those who could apply but did not manage to do so (27.5%), as compared to other groups (Figure 4.12). If general awareness about the programme seems to be quite high, IDPs and returnees from within the oblast appear to have the lowest awareness rate.
Note: Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 16.
Among those who submitted applications for the eRecovery programme, a significant portion reported that the process was very difficult. This sentiment was most prevalent among IDPs, with 56.2% expressing difficulty, followed by returnees from abroad (40.4%) and from other oblasts (39.7%). While other groups also faced challenges, their reported difficulty rates were lower, though still notable at around 30%. Programme design can inadvertently disadvantage IDPs and, more broadly, individuals who fled currently occupied territories, due to the absence of clear mechanisms for verifying property ownership or assessing damage, compared to individuals who did not relocate but whose housing was damaged. They underscore the importance of developing alternative methods for establishing ownership and documenting damage in contexts where standard procedures are unfeasible.
Application to eRecovery (eVidnovlennya) features a strong gender dimension. Among those who could have applied but did not manage, 68.7% are women. Among those who are not aware of the programme, 76.45% are women. This gender disadvantage is particularly striking among women returnees from abroad: among those who could have applied but did not manage 94.1% are women, and among those who are not aware of the programme, 89.9% are women.
Employment situation among returnees, the non-displaced population and IDPs
The war of aggression has led to massive disruption to the labour market, due to physical destruction and damage, relocation of enterprises and workers, and disruption of economic activity (OECD, 2025[32]).
Just as in housing, employment outcomes among displaced and returnee populations in Ukraine also reveal certain disparities. Evidence on post-return experiences points to significant gaps between expectations and local labour market realities. In UNHCR’s sixth round, 53% of returnees reported that access to work opportunities after return was worse than expected, a concern particularly common among those in the south and east (UNHCR, 2024[1]). IOM’s Conditions of Return Assessment underscores the constraints on the demand side: in 53% of assessed locations, key informants reported that few or no residents seeking work could find suitable jobs (IOM, 2024[9]).
IDPs are the most disadvantaged in terms of employment, although some returnees from internal displacement have difficulties too
Employment of individuals may depend on many parameters. These can include the fact of displacement and return, but also a wide range of other factors, such as the level of education or area of living. Multivariate analysis can help in understanding whether, once individual socio‑economic characteristics are accounted for, there is still a difference in employment across different groups of returnees, non-displaced individuals, and IDPs.
Multivariate analysis applied to the IOM GPS round 17 survey data (Box 4.5) shows that, once several socio‑economic characteristics are taken into account, IDPs and returnees from internal displacement in another oblast are the two groups that are more likely to be unemployed than employed, as compared to non-displaced persons. IDPs are significantly less likely to be in employment than non-displaced individuals, with an estimated reduction in probability of about 8.2 percentage points (p.p.). Returnees from displacement in another oblast are also significantly less likely to be in employment than non-displaced individuals, with an estimated reduction in probability of about 4.6 p.p. Probability of being employed is not statistically different for returnees from the same region and from abroad as compared to the non-displaced (Figure 4.13).
Marginal effects with 95% confidence intervals
Note: Marginal Effects with 95% confidence intervals are reported based on the regression results in (Box 4.5). Sample restricted to population aged 18‑65.
Source: Computations based on IOM General Population Survey Round 17.
Research questions
Do employment outcomes differ across returnees, non-displaced individuals, and IDPs, once their individual socio‑economic characteristics are accounted for? Which factors most strongly influence employment outcomes within each group?
Data and methodology
IOM GPS round 17 is used to answer these questions. The following equation is estimated:
Employedi = 𝛼𝑖 + 𝛽1Returnee_from_abroad𝑖 + 𝛽2Returnee_from_another_oblast𝑖 +
+ 𝛽3Returnee_within_ oblast𝑖 + 𝛽4IDP𝑖 + 𝛽3𝑋i + 𝜀𝑖
In this equation, the dependent variable Employedi is a dichotomous variable equal to one if the respondent i is employed, and zero otherwise. Four independent dichotomous variables of interest are: being a returnee from abroad; being a returnee from another oblast; being a returnee from within the oblast; or being an IDP. The reference category is that of a non-displaced individual. 𝑋𝑖t is a set of individual socio‑economic characteristics that include age, age squared, being female (male is the reference category), area of living (urban versus rural), household size and education level. Finally, 𝜀𝑖t is the error term.
This equation is first estimated on the pooled sample (Table 4.4, Column 1) to gauge the difference in employment rates between returnees, IDPs and non-displaced, while accounting for individual characteristics. It is then estimated on subsamples of returnees, IDPs and non-displaced individuals, excluding the displacement status variables (Table 4.4, Columns 2‑4), in order to compare the factors that most strongly affect employment outcomes within each group.
Results
Table 4.4 presents marginal effects from the probit regressions.
From Column (1), coefficients on two variables – being a returnee from abroad and being a returnee from another oblast – are negative and statistically significant. They suggest that, once all socio‑economic characteristics are taken into account, returnees from another oblast and IDPs are disadvantaged in terms of employment compared to the non-displaced. Being a returnee from another oblast reduces the probability of being employed by 4.6 p.p. as compared to the non-displaced. Being an IDP reduces the probability of employment by 8.2 p.p. compared to non-displaced individuals.
From Columns (2) to (4), comparing the effect of various socio‑economic characteristics of returnees, non-displaced individuals, and IDPs on employment, a few differences and similarities are observed. For individuals within each of these groups, employment probability increases with age, but at a decreasing rate. Being female substantially decreases the probability of employment for IDPs and for returnees: the difference between IDP women and men in the probability of being employed is 11.60 p.p.; the difference between returnee women and men in probability of being employed is 5.40 p.p. There is no gender effect on employment for the non-displaced.
For all three groups, the chance of being employed is higher if individuals live in urban rather than rural areas; this effect is amplified for IDPs. The number of household members does not seem to affect employment probability. Among non-displaced and IDPs, individuals with technical education have greater employment chances compared to individuals with lower education levels. For returnees, having technical education does not seem to affect employment probability in a significant way. Across all three groups, individuals with higher education have the greatest employment probability, and especially IPDs: the difference between an IDP with and without higher education in the probability of being employed is 21.70 p.p.
Dependent variable: dichotomous variable equal to 1 if individual is employed, and 0 otherwise
Note: Estimation method: probit regression. The reference categories are as follows: male (for female); rural (for urban); secondary education or below (for technical and higher education); non-displaced (for IDPs and returnees, in pooled regressions). The category “technical” includes technical and non-completed higher education. Samples are restricted to individuals aged 18‑65. Standard errors are in parentheses. The symbol (***) represents statistical significance at p<0.1; (**) at p<0.05, (*) at p<0.1. Regressions control for oblast of origin.
Source: Computations based on IOM General Population Survey Round 17.
The findings of the multivariate analysis also highlight some compounded vulnerabilities (Box 4.5): IDP women and IDPs in rural areas have particularly lower employment chances compared to all other groups. They require continued targeted policy attention. In contrast, higher education is the best employment guarantee for all groups, and technical education also offers strong protection for non-displaced individuals and for IDPs, highlighting changing labour market needs in favour of technical specialisation.
In addition to these findings, a closer look at status in employment8 shows that these groups have a different composition of status in employment. Among employed individuals, returnees from another country and from another oblast have the highest proportions of self-employed rather than employees. The share of self-employed is the smallest among employed IDPs (Figure 4.14).
Note: Tabulations are weighted to account for survey design and ensure representativeness. Sample restricted to employed population aged 18‑65.
Source: Computations based on IOM General Population Survey Round 17.
Returnees and IDPs are at a higher risk of precarious employment compared to the non-displaced
In addition to status in employment, the type of employment that the different population groups have matters too. For employees, the data allow determining whether employment is permanent in nature, or whether it is temporary or casual (including informal).9 Permanent employment is often considered superior in work quality to temporary and casual employment, as it not only offers more stability but is also typically associated with better wages and protections (OECD, 2019[33]).
Descriptive statistics in Figure 4.15 show that, at 28.08%, temporary and casual employment is most prevalent among returnees from another city or village within the same oblast. In other words, this population group faces the greatest precarity of employment. The outcomes of this group are most similar to IDPs, 26.32% of whom are employed on a temporary or casual basis. Returnees from abroad and from another oblast also have relatively high, albeit lower, rates of temporary or casual employment. The non-displaced have the greatest rate of employment in permanent jobs.
Note: Total sample: 1 806 observations. Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 17.
These results are corroborated by the multivariate analysis (Box 4.6), which confirms that returnees from any type of displacement and IDPs are more prone to temporary and casual informal employment, rather than permanent employment, as compared to the non-displaced population. IDPs are the most disadvantaged: being an IDP reduces the probability of being employed in a permanent rather than temporary job by 11.9 p.p. compared to non-displaced individuals. Taken together, these findings point to a need for employment support measures that go beyond job placement and encourage also better quality, stable, and formal employment.
Multivariate analysis with a similar set-up to the one in Box 4.5 is applied to the survey data of returnees, non-displaced individuals and IDPs, in order to gauge differences in employment types between these population groups. The only difference is in the dependent variable: it is equal to 1 if an individual is employed in a permanent job, and 0 if they are employed in a temporary or casual job.
From Table 4.5, Column (1), the marginal effects on the dummy variables for returnees from all three types of destinations are negative and statistically significant, with relatively similar magnitudes of these effects. The marginal effect on the dummy variable for IDPs is the largest, and suggests that, ceteris paribus, the difference between an IDP and a non-displaced individual in the probability of being employed in a permanent job is minus 11.9 p.p.
From Columns (2) to (4), individual characteristics that correlate with employment type are gender, household size, and education. IDP and returnee women have a higher probability of being employed in permanent rather than non-permanent jobs compared to men. A greater household size is associated with a lower probability of having a permanent job, hinting at the necessity of employment in non-permanent jobs for workers with larger families. Across all groups, individuals with higher education have a greater probability of being employed in permanent jobs, with the greatest effect of education observed for returnees.
Dependent variable: dichotomous variable equal to 1 if individual is employed in a permanent job, and 0 if employed in a temporary or a causal job
Note: Estimation method: probit regression. The reference categories are as follows: male (for female); rural (for urban); secondary education or below (for technical and higher education); non-displaced (for IDPs and returnees, in pooled regressions). The category “technical” includes technical and non-completed higher education. Samples are restricted to individuals aged 18‑65, employees only. Standard errors are in parentheses. The symbol (***) represents statistical significance at p<0.1; (**) at p<0.05, (*) at p<0.1. Regressions control for oblast of origin.
Source: Computations based on IOM General Population Survey Round 17.
Income situation among families of returnees, the non-displaced population, and IDPs
Understanding income differences between returnees, non-displaced individuals and IDPs can provide valuable insight into the economic disparities and vulnerabilities faced by each group.
There is a higher share of poorer individuals among non-displaced populations and IDPs
Figure 4.16 shows that the distributions of household monthly income for population groups with different displacement history. Reported household income includes income from all sources, including government assistance such as IDP allowances.
From Figure 4.16, the distributions of household monthly income of non-displaced and IDPs are quite similar. Both groups have a higher share of individuals with relatively low incomes compared to other groups. For returnees, the distribution is flatter, implying less significant concentration of very high or very low incomes. Still, there is a small spike of relatively well-off individuals. The medians of these income distributions are as follows: UAH 10 000 for the non-displaced population; UAH 12 000 for IDPs; UAH 15 000 for returnees from abroad, from another oblast in Ukraine, and from the same region.
Note: UAH: Ukrainian hryvna. Densities and medians are weighted to account for survey design and ensure representativeness. The total sample is restricted to the 1st to 99th percentile of the income distribution to mitigate the influence of outliers.
Source: Computations based on IOM General Population Survey Round 17.
Yet, household incomes are influenced by factors beyond displacement status
The results of the regression analysis (Box 4.7) suggest that, once a range of other individual characteristics is accounted for, household incomes do not statistically differ across individuals with different displacement statuses. This result needs to be interpreted in light of how household monthly income is measured: even if total household income appears comparable across groups, there remains a strongly displacement-related component to income vulnerability because IDPs are significantly more reliant on external assistance than others. In other words, total household incomes are comparable in part because IDP incomes were toped up by government assistance.
Beyond this, the most significant differences between respondents are driven by factors other than displacement status, notably their education level, employment status, and area of living (urban rather than rural).
These findings underscore the importance of provision income support to IDPs but also the value of designing longer-term income support mechanisms that are needs-based rather than strictly status-based, as status alone may not accurately reflect a household’s and individuals’ vulnerability. A needs-based approach can allow for a more comprehensive assessment of factors that go beyond displacement status, ensuring that assistance reaches those facing the greatest hardship. It can also help to forge a stronger sense of justice and social cohesion.
Multivariate analysis with a set-up similar to the one in (Box 4.5) is applied to the survey data of returnees, non-displaced individuals and IDPs, in order to gauge differences in incomes between these population groups. The dependent variable reflects household income on a logarithmic scale. Estimations are conducted using ordinary least squares regressions (OLS).
Note: Estimation method: OLS regression. The reference categories are as follows: male (for female); rural (for urban); secondary education or below (for technical and higher education); out of the labour force (for employee, self-employed, and unemployed), non-displaced (for IDPs and returnees, in pooled regressions). The category “technical” includes technical and non-completed higher education. Standard errors are in parentheses. The symbol (***) represents statistical significance at p<0.1; (**) at p<0.05, (*) at p<0.1. Regressions control for oblast of origin.
Source: Computations based on IOM General Population Survey Round 17.
From Table 4.6, Column (1), no statistically significant effect is found on the dummy variables for returnees, nor for IDPs. From Columns (2) to (4), characteristics that affect household income are the respondent’s age, gender, area of living, household size (also in logs), education, and employment status. Women across all three groups report lower household income, with IDP women reporting the lowest household income. Individuals in urban areas, as compared to rural areas, report higher incomes; this effect is the greatest for returnees. Household income naturally increases with household size. Individuals with technical and higher education report higher household incomes, the latter group being particularly better off across all groups. The self-employed also report the highest household incomes across all groups, as compared to individuals out of the labour force. Being a wage employee, too, significantly improves household incomes.
Sustainability of returns
For returnees, return is not just a question of movement, but of long-term safety, dignity, and opportunity. Ensuring the sustainability of returns is essential to prevent repeated displacement and migration to other countries, and to support lasting reintegration in countries of origin.
Computations based on IOM GPS round 17 survey data show that, among returnees, only a very small share considers leaving their current location in the nearest future (Figure 4.17). Only 3.17% of returnees from another country contemplate leaving, and fewer than 2% of individuals from other groups. There is, however, a certain degree of uncertainty: 8.01% of returnees from another oblast are not sure whether they might consider leaving, and the proportions fluctuate around 5% for other groups. Security concerns are the most commonly reported primary reason for considering new movement (IOM, 2025[27]).
Note: Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 17.
There are some minor differences in the intentions to stay for individuals depending on occupancy status (Annex Figure 4.B.1), degree of housing damage (Annex Figure 4.B.2), or availability of legal documents for housing (Annex Figure 4.B.3). Generally, there is a higher share of individuals not intending to stay among those who are hosted for free or are renting their housing, as well as among those who do not have the legal documents for their housing, especially among returnees as contrasted to the non-displaced. The small number of overall positive responses, however, precludes a meaningful multivariate analysis based on IOM data to determine statistical significance of these factors.
Some further insights can be drawn from IMPACT Round 28 survey, which highlights that accommodation is often the main reason for staying in the current location among different groups of returnees and internally displaced persons. Needs for housing support and repair vary by context, including among those returning from or to frontline areas. In these locations, accommodation plays an especially prominent role, with 74% of returnees to frontline areas reporting that available housing was the reason they planned to remain (IMPACT, 2024[34]).
Evidence from IMPACT surveys also suggests that employment is an additional “anchoring” factor of stay. Employed returnees tend to stay even when conditions are not safe. Employment is mentioned as a key reason for staying in the current location after return by 48% of returnees, though assistance with job search is also commonly needed (IMPACT, 2024[34]).
Beyond the sustainability of returns, an important longer-term question is also the sustainability of livelihoods. Evidence shows that unmet needs after return can create vulnerability. Over 2023-2024, more than half of returnees reported urgent unmet needs, particularly cash and food, with higher levels among those returning to their place of origin though not to their homes (IMPACT, 2023[4]; IMPACT, 2024[12]). There is also a high reliance on social protection benefits, with 72% of internally displaced persons and 62% of returnees receiving benefits (UNHCR, 2024[1]) – though for the latter group this may also be due to the age composition. IOM’s CoRA (Conditions of Return Assessment) further highlighted tensions around the distribution of humanitarian aid in affected communities hosting the majority of returnees (IOM, 2024[9]).
Access to services is not uniform, either. IOM’s findings suggest that 23% of returnees from abroad reported difficulties accessing health services and 25% reported problems accessing medicines, figures broadly similar to or slightly lower than for internally displaced and non-displaced persons. However, access to mental health services appeared more challenging: 23% of returnees from abroad reported difficulties, a higher share than in other groups (IOM, 2024[9]).
As with healthcare, supply-side constraints in education are evident in Ukraine. IOM’s Conditions of Return Assessment records that 21% of assessed locations hosting 59% of returnees had damaged schools, although key informants reported that in 98% of locations most school-aged children were nonetheless able to attend lessons (IOM, 2024[9]).
On the national level, returnees who returned more recently report poorer access to services, while those who returned earlier are more likely to report access levels comparable to the non-displaced (IOM, 2025[35]).
Together, these findings suggest that when housing, employment, and access to basic services and protections are absent or limited, displaced persons, including returnees, resort to coping mechanisms. In the longer term, this may pose risks of moving again in search of support or, for the most vulnerable, on the contrary, compromise the possibility to move in the face of immediate danger.
New outflows from Ukraine
More recent survey data from Ukraine, collected in 2025, suggest that the majority of Ukrainian women (79%) were not planning to leave the country for an extended period (Centre for Economic Strategy, 2025[36]). Family ties (91%) and patriotism (88%) dominated as reasons for staying. Despite a strong sense of attachment to Ukraine, around one in five women would like to relocate abroad but lacked the financial means to do so. This proportion reached 27% among IDP women. In the fourth year of the war, 15% of women were actively considering migration, primarily motivated by the prospect of a better standard of living (75%) and better security (73%). The highest inclination to leave was observed among younger individuals, those with limited income and among Russian-speaking residents in frontline and western regions. Furthermore, individuals with prior experience living abroad were more likely to leave again (ibid).
Another survey, among women and men, showed that 96.3% would remain in the country upon the end of the hostilities either under any circumstances or if economic conditions were to improve, highlighting the central role of socio‑economic stability and prospects in migration decisions (CASE Ukraine, 2025[37]). Young people under 30, especially men, showed the greatest sensitivity to economic opportunities.
Return dynamics are highly scenario dependent. Modelled outcomes range from negligible returns under a prolonged status quo to up to two-thirds of those displaced in Europe returning under more favourable security scenarios, most within the first year after hostilities end.
Household circumstances shape return decisions more than individual demographics. Families with children, particularly those reunited abroad, are less likely to return, while older-person households and single adults show higher return propensities.
Housing conditions are a decisive factor for both the likelihood and timing of return. Access to intact or reparable housing strongly accelerates return, while destroyed housing, lack of property, or uncertainty over property conditions significantly delays or discourages return across all scenarios. These constraints ease under a more favourable security scenario.
Employment status in host countries is likely to play a more limited role than often assumed. Differences in return behaviour between employed and unemployed Ukrainians displaced abroad are modest, reflecting widespread underemployment and the greater importance of future job quality and prospects in Ukraine than current labour market attachment.
Host country context influences return propensities. Those displaced in neighbouring and Central-Eastern European countries are more likely to return than those in Western and Northern Europe, reflecting household composition, oblast of origin, and self-selection effects.
Returns are likely to reshape Ukraine’s internal geography. Many returnees are projected to settle outside their oblast of origin, particularly in urban centres with better housing, services, and jobs. This underscores the need for place-based reintegration planning.
The eventual and voluntary return and reintegration of Ukrainian nationals abroad,10 when people determine it is safe to do so, is a cornerstone of the government’s ambition to rebuild its human capital, mitigate worsening demographic imbalances, and support Ukraine’s recovery.11 Understanding past intentions and return decisions is crucial for anticipating future return patterns. Such understanding can inform policy responses, particularly in areas where reintegration capacity or service delivery could become stretched. As noted earlier, return decisions are shaped by a complex interplay of factors, including individual characteristics, developments in Ukraine related primarily to security as well as to economic and institutional conditions, and circumstances in host countries such as legal status, access to work and prospects for family reunification. Importantly, return decisions result from a complex combination of push and pull factors that can, in some cases, lead the most vulnerable refugees, including those with limited access to housing, income or family support abroad, to be among the first to return.
To better understand who is likely to return, under what conditions and to which parts of Ukraine, the UNHCR and Brunel University of London have developed a prototype agent-based model (ABM) (Box 4.8), specifically tailored to the Ukrainian forced displacement context (UNHCR and Brunel University of London, 2025[38]). While the model simulates the return patterns of approximately 4.4 million refugees from Ukraine hosted across the EU‑27 and Moldova, its results are extrapolated to reflect the broader displaced Ukrainian population across Europe, which exceeds 5 million.
The model’s core value lies in its ability to capture the diversity of experiences, motivations, and needs that shape return decisions. It reflects the different weight attached to factors such as safety, housing, work and access to services across different population groups (see Box 4.8). This approach provides a more nuanced understanding than aggregate forecasting by identifying where returns are more likely to occur and which categories of displaced persons are more inclined to return under different post-war conditions.
Crucially, the model operates in a dynamic manner, simulating how preferences and behaviours evolve over time in response to changing realities such as shifts in security, legal frameworks, work opportunities or access to public services. While it does not aim to predict exact outcomes, it supports evidence‑based decision making by identifying the relative influence of key drivers, potential threshold effects and interactions across different scenarios. This level of analytical granularity is essential for anticipating varied reintegration needs, prioritising interventions and informing long-term recovery planning.
The results presented below are based on simulations provided by the UNHCR and the Brunel University of London specifically for this report. Three hypothetical scenarios are modelled:
Ukraine’s Victory: All temporarily occupied territories are reclaimed by the end of 2026, triggering sustained investments in Ukraine’s recovery, reconstruction and the reintegration of returnees. From a legal perspective, Temporary Protection in the EU ends in March 2027, with a gradual transition to alternative legal statuses for refugees who choose to remain in host countries (for example, work or study permits or other forms of residency).
Prolonged War (Status Quo): Frontline hostilities and aerial attacks continue through late 2029, resulting in very limited or no investments, with the response remaining largely humanitarian in nature. From a legal perspective, the model assumes the extension of the current legal status whilst refugees transition into alternative forms of legal status.
Fragile Peace with Concessions: Hostilities end by late 2026, with currently occupied areas remaining under the control of the Russian Federation, but no further territorial losses occurring. This situation allows for moderate levels of investment in areas under Ukrainian control, supporting recovery, reconstruction, and the reintegration of returnees.
In the model, the transition of refugees toward alternative legal statuses, while possible throughout all simulations, accelerates in the months preceding the materialisation of each scenario and voluntary returns are simulated through to December 2029. For refugees from oblasts currently under temporary occupation, the model assumes they would not be compelled to return under adverse conditions, and that alternative legal pathways or protection statuses would be made available.
Agent-based modelling (ABM) offers a powerful and flexible computational approach for simulating how forcibly displaced individuals may behave under evolving conditions. ABM is increasingly applied in humanitarian contexts to anticipate refugee and returnee movements, as evidenced in recent applications (Ghorbani et al., 2024[39]; Suleimenova, Bell and Groen, 2017[40]). Unlike traditional forecasting methods based on aggregate trends, ABM takes a bottom-up approach by simulating individual agents, each representing a displaced person with distinct demographic, legal and socio‑economic characteristics. These agents interact in dynamic virtual environments that reflect real-world conditions, allowing macro-level return patterns to emerge from diverse micro-level decisions.
The UNHCR – Brunel ABM was designed to support scenario-based policy and programme development in the context of Ukraine’s post-conflict recovery. It simulates the location choices of over 5 million forcibly displaced individuals across Europe, each modelled as a computational agent, who periodically assess whether to remain in their current host country or return to Ukraine. These decisions rely on an individualised utility scoring function for each location, reflecting agent-specific preferences and external conditions. Agents vary in how they assess factors like safety, work, housing, healthcare, education, and social ties, based on their socio-demographic profile. For example, caregivers may prioritise school access, while older adults may value healthcare or familiar environments.
These decision rules are applied across hypothetical post-war scenarios, each reflecting alternative paths for conflict resolution, recovery, or legal arrangements, enabling the model to track how return intentions shift under different plausible futures.
Unlike traditional statistical models that rely on historical aggregates, one of ABM’s methodological advantages is its capacity to integrate both structured data and expert qualitative insights without undermining analytical rigour.
This flexibility makes ABM particularly well-suited to humanitarian contexts, where key behavioural drivers are not always captured by official statistics. The model draws on multiple data sources: UNHCR’s Refugee Intentions Surveys across European host countries,1 the Socio-Economic Insights Survey (SEIS) (UN Interagency Partners, 2025[41]) on refugee socio‑economic conditions in hosting countries, UNHCR’s Protection Monitoring inside Ukraine, its Skills Survey and broader recovery data, including the World Bank – EU – UN Rapid Damage and Needs Assessment (RDNA 4) (World Bank et al., 2024[42]). The model development, including assumptions about parameters, was further guided by an informal advisory board comprising of technical experts from international financial institutions and development actors, relevant EU Directorates, the OECD and Ukrainian institutions such as the Ministry of Economy, and academic partners including the Kyiv School of Economics.
The model is implemented using high-performance computing infrastructure and simulates the period from January 2026 to late 2029. It covers all oblasts of Ukraine, EU host countries and Moldova. Each hypothetical scenario is run five times to mitigate stochastic variation, with averaged results used for analysis. Each agent in the model represents ten real individuals. Simulations are conducted using the SEAVEA toolkit and the ARCHER2 supercomputing platform.
By simulating the decisions of a diverse, data‑informed population rather than relying on aggregate behavioural assumptions, the UNHCR – Brunel ABM provides a robust platform for stress-testing return scenarios and identifying policy-sensitive dynamics in recovery planning.
A forthcoming Methodological Note will equip decision makers with a clear and accessible understanding of the model’s functioning, without requiring engagement with the full technical specifications.
1. UNHCR, in partnership with Ipsos SA, “Lives on hold. Intentions and Perspectives of Refugees from Ukraine” survey series, rounds 1‑6.
The findings illustrate that return trends may differ not only in scale but also in location and socio-demographic profile, depending on how the situation evolves. This deeper understanding can help guide recovery strategies that are both realistic and inclusive.
Results are expected to evolve over time. As refugees’ socio-demographic profiles change, through factors such as household composition (notably following family reunification), access to skilled employment in host countries, and age‑related factors such as education, retirement, or the risk/potential for conscription, projected return figures and profiles will also shift, even under the same scenario assumptions. The forecasting model will therefore continue to be updated as new data become available, and assumptions are revised, enhancing the accuracy and policy relevance of projections.
Return dynamics are highly scenario-dependent, with the model projecting vastly different outcomes by 2029 depending on the security trajectory in Ukraine (Figure 4.18). Under the Ukraine Victory scenario, an estimated 3.5 million refugees, representing 68% of the total refugee population in Europe, are projected to return by the end of 2029.12 In contrast, under the Fragile Peace with Concessions scenario, projected returns reach 2.3 million (44%). Finally, under a prolonged Status Quo, net returns remain negligible at just 40 000 (1%), which aligns with the downward trend in returns observed since the beginning of the war,13 as well as the escalation of aerial and drone attacks recorded since mid-2024.
These differences have significant implications for policymaking, not only in terms of the scale and absorption capacity required, but also regarding the pace and concentration of returns. Higher-return scenarios suggest concentrated return movements within a short timeframe, particularly during the first year after the war. Conversely, lower-return scenarios highlight the importance of considering that a substantial share of refugees are likely to remain in host countries, even after Temporary Protection ends.
Source: UNHCR – Brunel ABM.
Due to the very low number of projected returns under the Status Quo scenario, this case will not be analysed in detail in the model outputs relating to returnee profiles or return locations. The small sample size would limit the robustness and reliability of these disaggregated results.
Returns will not happen all at once but will increase over time. This increase will be shaped by multiple interrelated factors specific to each scenario, including perceived and actual security, access to housing and services, social ties both in Ukraine and in host countries, and comparative living standards. Under these scenarios, around three‑quarters of the total estimated returns are expected to occur in the initial year after hostilities end under these scenarios, with the majority taking place between March 2027 and January 2028. While not indicative of a recurring seasonal trend, a noticeable peak in returns is observed in June 2027 (Figure 4.19). This is likely influenced by the end of the school year, highlighting how return decisions may be shaped by families’ efforts to avoid disrupting their children’s education.
Source: UNHCR – Brunel ABM.
Return patterns reflect household circumstances more than demographic characteristics
The demographic profile of returnees remains broadly similar to that of the overall refugee population, yet key differences emerge across age and household composition (Figure 4.20). The share of older persons among returnees (7% under Fragile Peace with Concessions and 8% under Ukraine’s Victory) notably exceeds their proportion among stayers (3% and 5%, respectively), confirming a higher likelihood of return among older refugees, likely linked to property ownership and family ties in Ukraine, and a lower degree of integration in host countries.
At the household level, more pronounced differences emerge (Figure 4.21). Households with at least one child are consistently less represented among returnees than among those likely to remain abroad: 48% versus 53% under Fragile Peace with Concessions, and 45% versus 63% under Ukraine’s Victory. This likely reflects families’ preference for stability and educational continuity abroad. This is particularly true of families with two or more adults and dependants, who are estimated to remain abroad in higher proportions (28% versus 16% of returnees under Fragile Peace with Concessions, and 34% versus 17% under Ukraine’s Victory). These are likely to be nuclear families who have already reunited in host countries, suggesting a higher degree of stability and forward planning.
Among those estimated to be more likely to return are households composed only of older persons (17‑18% among returnees versus 10‑14% among those remaining abroad). Additionally, single‑adult households (18‑59), including both those with dependants and those without, show a higher tendency to return. For single adults without dependants, this may stem from the higher cost of living and limited social support networks in host countries, which make remaining abroad less sustainable. For single adults with dependants, challenges in accessing the labour market due to limited childcare options may act as a driver for return.
Source: UNHCR – Brunel ABM.
Source: UNHCR – Brunel ABM.
Employment status shows limited influence on return intentions
Employment in host countries appears to play a limited role in influencing return decisions. At the household level, there is a modest tendency to remain in the host country when at least one household member is employed. This trend is more pronounced under the Ukraine’s Victory scenario, with 70% of households opting to stay being those with at least one member employed, compared to 61% among returnees. In contrast, under the Fragile Peace with Concessions scenario, the share of households with at least one member employed would be 64% among those that stay abroad and 62% among those who would return (see Figure 4.22).
Among households with at least one unemployed member, the inclination to return is slightly stronger, with a share of 14% among returnees under both scenarios. Meanwhile, the proportion declines to 12% among those that would stay under Fragile Peace with Concessions and 11% under Ukraine’s Victory. For households with no member in the labour force, a small tendency to return is observed under both scenarios, with a more notable difference under Ukraine’s Victory (26% of returnees versus 19% of those who stay abroad) than under Fragile Peace with Concessions (25% versus 24%).
Source: UNHCR – Brunel ABM.
Source: UNHCR – Brunel ABM.
At the individual level, among working-age refugees, differences are even less pronounced (Figure 4.23). For those employed locally, the likelihood of return is nearly identical across scenarios, 51% under Fragile Peace with Concessions and 50% under Ukraine’s Victory. For those employed remotely, the figures are 6% versus 5%, respectively. Unemployment shows no impact under Fragile Peace with Concessions (15% return in both cases), while under Ukraine’s Victory, a slight preference for return is observed (16% versus 14%).
Refugees who are not participating in the labour force tend to prefer staying in host countries under both scenarios: 28% versus 30% under Fragile Peace with Concessions, and 28% versus 31% under Ukraine’s Victory.
These findings may seem counterintuitive, as employment does not appear to be a major determinant of return decisions. However, this likely reflects the persistence of underemployment in host countries, where many refugees work below their skill levels and earn, on average, around two‑thirds of local wages.14 Consequently, job satisfaction and the prospect of working in one’s field of expertise upon returning to Ukraine may influence future decisions. Increased foreign investment and subsequent job creation could further encourage return, particularly among refugees with specialised skills.
These dynamics are also expected to evolve over time. Language proficiency plays a key role in employment outcomes, and as refugees’ command of host-country languages improves, return patterns may shift.15
The limited impact of educational attainment on current return intentions is consistent with the prevalence of low-skilled employment among refugees. There are minimal differences between those returning and those staying. The only notable variation appears among specialists, as under Fragile Peace with Concessions, 28% of returnees would have a specialisation degree compared to 30% among those who would stay. Meanwhile, under Ukraine’s Victory, the figures converge at 29% each, mirroring the overall refugee population.
A shifting geography of return
The model results suggest that a significant share of returnees are likely to resettle in a different oblast than their place of origin. Under the Fragile Peace with Concessions scenario, almost one in five returnees is projected to do so, rising to around one in three under the Ukraine’s Victory scenario. This difference reflects not instability, but the interaction of housing availability, levels of destruction, and the relative attractiveness of urban areas – factors explicitly built into the model and reflective of plausible post-conflict dynamics.
Under the Fragile Peace with Concessions scenario, returns concentrate in oblasts under government control less affected by the war, which together host around 1.3 million returnees. The largest shares are expected in Dnipropetrovsk oblast (19%), the city of Kyiv(18%), Odesa oblast (12%), and Lviv oblast (7%). These regions combine functioning infrastructure, accessible services, and economic opportunities, and also absorb many people from eastern and southern oblasts who cannot yet return to their original homes.
Under the Ukraine’s Victory scenario, the geography of return broadens eastward, with major urban centres and reopened eastern oblasts accounting for around 2.7 million returnees in total. The city of Kyiv remains the primary destination (34%), followed by the oblasts of Kharkiv (12%), Dnipropetrovsk (8%), and Odesa (8%). Newly accessible areas, such as the oblasts of Donetsk (7%), Kherson (4%), Zaporizhzhia (3%), and Luhansk (1%), also attract returnees. Lviv’s share declines to around 3%, reflecting a smaller role as a western refuge once broader security is restored.
These outcomes reflect the model’s assumptions and parameters, based on available data and expert judgment, particularly concerning the scale of war-related housing destruction and the uneven pace of reconstruction across oblasts, which are likely key drivers of these patterns. In many eastern and southern areas, extensive damage, combined with mine contamination and other remnants of war, is expected to delay recovery and reconstruction, including private housing. At the same time, urban centres are likely to offer comparatively larger housing availability, infrastructure, and job opportunities, further reinforcing their pull as destinations for returnees.
Most returns will likely take place early, though timing varies by property status, region of origin, and family type
Under the Fragile Peace with Concessions scenario, returns are projected to occur relatively quickly but not uniformly. By the end of 2027, around 70% of all returns will have taken place, indicating a strong initial movement back, mostly to areas under Ukrainian administration, supported by gradual improvements in security, access to services, and moderate levels of investment in reconstruction. The model nonetheless shows moderate differences in timing across macro-regions,16 property situations, and household types, pointing to distinct though overlapping rhythms of return.
When considering regions of return, the model identifies three broad trajectories that emerge under the Fragile Peace with Concessions scenario. By the end of 2027, the fastest pace of return is recorded in Kyiv (79%) and the Western region (75%), followed by the South (72%) and the Centre (72%), while the North (63%) and the East (58%) register slower progress. These variations likely reflect differences in proximity to former conflict zones, infrastructure recovery, and local security perceptions.
Under the Ukraine’s Victory scenario, the geography of return shifts in line with Ukraine’s full restoration of territorial control. Returns to the East and Centre accelerate significantly, with around 63% and 76% of returns taking place in 2027. In contrast, the South shows a slower pattern, with the 2027 share of returns falling from 72% to 61%, while Kyiv (77%) and the West (76%) remain broadly stable. This pattern likely reflects a redistribution of return flows: under Fragile Peace with Concessions, part of the early movement toward the South may have involved people originating from still-occupied eastern and southern oblasts seeking to establish themselves elsewhere within Ukraine. Under Ukraine’s Victory, these individuals can instead return directly to their home oblasts, reducing movements toward southern regions and smoothing return flows more evenly across 2027‑2028 (see Figure 4.24 and Figure 4.25).
Source: UNHCR – Brunel ABM.
Source: UNHCR – Brunel ABM.
In the Fragile Peace with Concessions scenario, the condition of housing emerges as the strongest factor influencing the timing of return (see Figure 4.26 and Figure 4.27). On average, individuals with intact or partially damaged dwellings are expected to return earliest, with around 70% of their total returns projected in 2027. Those whose homes are fully damaged, who do not own property, or who do not know the condition of their house, likely more frequently located along the frontline or in areas remaining under Russian control, return more gradually, averaging around 62% in 2027 and extending further into 2028. This pattern reflects both the physical constraints of damaged housing and the importance of housing availability as a precondition for deciding to return.
In the Ukraine’s Victory scenario, returns accelerate and become more uniform across property conditions. With Ukraine regaining control over all occupied territories, refugees who previously could not verify the state of their homes are likely to be able to access and assess them. As a result, the timing gap between those with intact dwellings and those with damaged or unknown housing conditions narrows considerably, with roughly 67‑70% of all groups returning by 2027. The importance of housing availability is therefore accompanied by broader security expectations, as improved conditions and governance enable returns even in areas where housing is likely to be more frequently damaged or destroyed.
Overall, in both scenarios the model suggests that housing conditions remain a key determinant of return timing. Yet, under more favourable circumstances, such as full territorial recovery, constraints linked to property damage lose some of their delaying effects, as refugees gain both access and confidence to return. However, effective access to housing in Ukraine continues to be crucial across scenarios. Refugees without property or whose property has been partially or fully destroyed tend to remain in host countries, while those with access to intact property are more likely to return.
Source: UNHCR – Brunel ABM.
Source: UNHCR – Brunel ABM.
Although smaller, some differences by household composition are still discernible (see Figure 4.28 and Figure 4.29). On average, the model results show that single adults (18‑59) and older-person households (aged 60+) return slightly earlier, with around 77% of their total returns taking place in 2027. By comparison, households with two or more adults without dependants average 68%, while those with dependants, whether single or with two or more adults, average 66%. A marked peak among single‑caregiver households (32%) in early summer 2027 suggests that, while these often women-led families are generally more socio‑economically vulnerable and therefore expected to return earlier, they are also motivated by family reunification and tend to wait until the end of the school year before returning. This underlines the importance of educational continuity for their children. Under the Ukraine’s Victory scenario, these patterns become even more pronounced, with an acceleration in returns among older persons (80%) and single adults (82%).
Source: UNHCR – Brunel ABM.
Source: UNHCR – Brunel ABM.
Host country variations matter
Estimated returns differ markedly across scenarios and host-country contexts (Figure 4.30). Under the Fragile Peace with Concessions scenario, refugees in neighbouring countries display an average projected return rate of 50%, compared to 41% in non-neighbouring countries. The pattern remains consistent under the Ukraine’s Victory scenario, rising to 73% and 66%, respectively, consistent with the assumption that Ukraine regains all occupied territory.
Percentage of returnees
Source: UNHCR – Brunel ABM.
Differences are also observed at the sub-regional level. Central and Eastern European countries, such as the Republic of Moldova, Bulgaria and Hungary, record the highest return propensities under both scenarios (ranging between 55‑65% and 75‑85%, respectively), while Western and Northern European hosts, including Germany, Estonia and Belgium, show lower rates (35‑45% and 60‑65%). Southern European countries, including Greece, Italy, and Spain, record moderate return rates (45‑55% and 70‑75%), positioned between the higher levels observed in Central and Eastern Europe and the lower ones in Western and Northern Europe.
These differences reflect a combination of geographic, demographic, and socio‑economic factors. Household composition plays a significant role: nuclear families with children, who are less likely to return, are more prevalent in non-neighbouring countries, whereas single caregivers, more frequently living in neighbouring countries, display a higher return likelihood. The area and oblast of origin also shape intentions, particularly under the Fragile Peace with Concessions scenario, as countries such as Germany, Poland and Estonia host a higher share of refugees from heavily affected or occupied eastern and southern oblasts. Finally, self-selection effects may also play a role, with refugees more prone to return having chosen to settle in neighbouring countries.
References
[26] Alrababa’h, A., M. Casalis and D. Masterson (2020), Returning Home? Conditions in Syria, Not Lebanon, Drive the Return Intentions of Syrian Refugees, Innovations for Poverty Action, Washington D.C.
[25] Black, R. and K. Koser (1999), The end of the refugee cycle? Refugee repatriation and reconstruction,, Berghahn Books, Oxford.
[30] Bobrova, A. et al. (2025), Housing needs and the prospects for social housing in the Kalush Hromada, CEDOS: Ukraine, https://cedos.org.ua/wp-content/uploads/housing-needs-and-prospects-for-social-housing-in-the-kalush-hromada.pdf.
[37] CASE Ukraine (2025), Economic opportunities could help address Ukraine’s emigration problem.
[2] Centre for Economic Strategy (2025), Ukrainian Refugees After Three Years Abroad. Fourth Wave of Research, https://ces.org.ua/wp-content/uploads/2025/03/ukrainian-refugees.-fourth-wave.pdf.
[36] Centre for Economic Strategy (2025), Women in war: Motivations to stay and reasons to leave.
[10] Centre for Economic Strategy (2024), Ukrainian refugees. Future abroad and plans for return. The third wave of the research, https://ces.org.ua/wp-content/uploads/2024/03/research.-ukrainian-refugees.-third-wave.pdf.
[19] Constant, A. and D. Massey (2002), “Return Migration by German Guestworkers: Neoclassical versus New Economic Theories”, International Migration, Vol. 40/4, pp. 5-38, https://doi.org/10.1111/1468-2435.00204.
[18] Dustmann, C. and Y. Weiss (2007), “Return Migration: Theory and Empirical Evidence from the UK”, British Journal of Industrial Relations, Vol. 45/2, pp. 236-256, https://doi.org/10.1111/j.1467-8543.2007.00613.x.
[11] EUAA (2025), Ad hoc Report: Situation in Ukraine and Displacement to the EU+: Trends, Drivers and Future Prospects, https://euaa.europa.eu/sites/default/files/publications/2025-09/2025_09_EUAA_Ad_hoc_Report_Ukraine_EN.pdf.
[3] EUAA/OECD (2024), Voices in Europe: Experiences, Hopes and Aspirations of Forcibly Displaced Persons from Ukraine, OECD Publishing, Paris, https://doi.org/10.1787/ae33637c-en. (accessed on 9 April 2024).
[6] Factum Group (2023), YOUkraina. Special Edition: Research on Ukrainian Refugees., https://factum-ua.com/document/YOUkraina%20by%20Factum%20Group.%20UKRAINIANS%20ABROAD.pdf.
[39] Ghorbani, M. et al. (2024), “Flee 3: Flexible agent-based simulation for forced migration”, Journal of Computational Science, Vol. 81, p. 102371, https://doi.org/10.1016/j.jocs.2024.102371.
[45] ILO et al. (2025), High Employment Rates, but Low Wages: a Poverty Assessment of Ukrainian Refugees in Neighbouring Countries, https://data.unhcr.org/en/documents/details/115013.
[12] IMPACT (2024), Longitudinal Brief. Back, but not necessarily home: refugee experiences upon returning to Ukraine and becoming IDPs. Round 20., https://repository.impact-initiatives.org/document/reach/86bce150/IMPACT_Longitudinal_Situation_Overview_returnees_to_Ukraine_IDPs_Round21.pdf.
[34] IMPACT (2024), Unsafe returns: what makes refugees return to Ukraine and settle in the frontline areas of the country? | Longitudinal Survey of Ukrainian Returnees, Round 28 - August 2024, IMPACT Initiatives, https://repository.impact-initiatives.org/document/impact/361e9484/IMPACT_Longitudinal_Situation_Overview_R28_unsafe_returns.pdf.
[4] IMPACT (2023), Longitudinal Brief: What do we know about Ukrainian refugees returning home since the full scale invasion? Round 18., https://repository.impact-initiatives.org/document/impact/a6a94e78/IMPACT_Longitudinal_Factsheet_R18.pdf.
[13] IMPACT Initiatives (2025), Ukriane Longitudinal Survey of Refugees, https://tinyurl.com/y9judz5r.
[44] IOM (2025), Returning Home From Abroad: Trends, Drivers, and Reintegration Challenges among Ukrainian Returnees from Abroad, IOM Ukraine, https://crisisresponse.iom.int/sites/g/files/tmzbdl1481/files/appeal/documents/IOM_UKR_Returning-Home-From-Abroad_July-2025_Updated.pdf.
[27] IOM (2025), Ukraine — Displacement and Return: Trends, Drivers and Intentions. Global Data Institute: Displacement Tracking Matrix.
[35] IOM (2025), Ukraine - Returning Home From Abroad. Global Data Institue, Displacement Tracking Matrix.
[9] IOM (2024), Conditions of Return Assessment Factsheet for Ukraine, Round 9 (December 2024), IOM, Ukraine, https://dtm.iom.int/fr/node/47191.
[43] IOM (2024), Ukraine — General Population Survey — Methodological Note (August 2024).
[31] IOM (2024), Ukraine — Housing Brief: Living conditions, rental costs and mobility factors —July 2024, IOM Global Data Institute Displacement Tracking Matrix, https://dtm.iom.int/fr/node/40466.
[15] IOM (2024), Ukraine — Returns Report — General Population Survey Round 17 (August 2024), IOM Global Data Institute Displacement Tracking Matrix, https://dtm.iom.int/fr/node/42456.
[23] IOM (2018), Family Matters - A Study into the Factors Hampering Voluntary Return of Migrants Residing at Family Locations, IOM, Geneva, https://publications.iom.int/system/files/pdf/family_matters.pdf.
[22] Konzett-Smoliner, S. (2016), “Return migration as a ‘family project’: exploring the relationship between family life and the readjustment experiences of highly skilled Austrians”, Journal of Ethnic and Migration Studies, Vol. 42/7, https://doi.org/10.1080.
[28] Miloserdov, V. (2024), Housing and housing conditions of Ukrainians: Survey results, CEDOS: Ukraine, https://cedos.org.ua/wp-content/uploads/zhytlo-ta-zhytlovi-umovy-ukrayin_ok_rezultaty-opytuvannya-1.pdf.
[5] MPI and IOM (2024), Exploring Refugees’ Intentions to Return to Ukraine: Data Insights and Policy Responses, https://www.migrationpolicy.org/sites/default/files/publications/mpie-iom_ukraine-return-intentions-2024-final.pdf.
[32] OECD (2025), OECD Economic Surveys: Ukraine 2025, OECD Publishing, Paris, https://doi.org/10.1787/940cee85-en.
[14] OECD (2024), Return, Reintegration and Re-migration: Understanding Return Dynamics and the Role of Family and Community, OECD Publishing, Paris, https://doi.org/10.1787/625fb5e6-en.
[16] OECD (2020), Sustainable Reintegration of Returning Migrants: A Better Homecoming, OECD Publishing, Paris, https://doi.org/10.1787/5fee55b3-en.
[33] OECD (2019), OECD Employment Outlook 2019: The Future of Work, OECD Publishing, Paris, https://doi.org/10.1787/9ee00155-en.
[24] OECD (2017), Interrelations between Public Policies, Migration and Development, OECD Publishing, Paris, https://doi.org/10.1787/9789264265615-en.
[20] Song, H. and E. Song (2015), ““Why Do South Korea’s Scientists and Engineers Delay Returning Home? Renewed Brain Drain in the New Millennium””, Science, Technology and Society, Vol. 20/3, pp. 349-368, https://doi.org/10.1177/0971721815597140.
[40] Suleimenova, D., D. Bell and D. Groen (2017), “A generalized simulation development approach for predicting refugee destinations”, Scientific Reports, Vol. 7/1, https://doi.org/10.1038/s41598-017-13828-9.
[21] Toomistu, T. et al. (2024), ““Determinants of return migration of Estonian young adults in transnational mobility””, Migration Studies, Vol. 12/1, pp. 42-67, https://doi.org/10.1093/migration/mnad033.
[41] UN Interagency Partners (2025), High employment rates, but low wages: a poverty assessment of Ukrainian refugees in neighboring countries, https://data.unhcr.org/en/documents/download/115013.
[46] UNHCR (2025), Ukraine population Movements Factsheet, https://data.unhcr.org/en/documents/details/114591.
[8] UNHCR (2024), Lives on hold: Intentions and Perspectives of Refugees, Refugee Returnees and IDPs from Ukraine #5 Summary Findings, https://data.unhcr.org/en/documents/download/106738.
[1] UNHCR (2024), Lives on hold: Intentions and Perspectives of Refugees, Refugee Returnees and IDPs from Ukraine #6 Summary Findings, https://data.unhcr.org/en/documents/download/112600.
[38] UNHCR and Brunel University of London (2025), Forecasting refugee return to Ukraine amid ongoing war and uncertainty, Policy Brief. Geneva, https://tinyurl.com/4n4vpj4m.
[7] Vox Ukraine (2024), Return or stay? What factors impact the decisions of Ukrainian refugees, https://voxukraine.org/en/return-or-stay-what-factors-impact-the-decisions-of-ukrainian-refugees#_ftnref8.
[17] Wahba, J. (2015), ““Who benefits from return migration to developing countries?””, IZA World of Labor, Vol. 123.
[42] World Bank et al. (2024), Ukraine: Fourth Rapid Damage and Needs Assessment (RDNA4), February 2022–December 2024, World Bank Group, Washington, D.C., http://documents.worldbank.org/curated/en/099022025114040022.
[29] World Bank et al. (2024), Ukraine - Third Rapid Damage and Needs Assessment (RDNA3): February 2022 - December 2023, World Bank Group, Washington, D.C., http://documents.worldbank.org/curated/en/099021324115085807.
The analysis of this chapter performed by the OECD Secretariat builds on three distinct micro datasets: IMPACT longitudinal data, IOM GPS Round 16, and IOM GPS Round 17.
The “IMPACT Initiatives” has been conducting a monthly Longitudinal Survey of Ukrainian Refugees – Ukrainians displaced abroad since the onset of Russia’s full-scale invasion. The objective of the survey is to regularly survey the same sample of people who left the country in the early days of the full-scale invasion, in order to understand mobility patterns, needs, integration trajectories, intentions to return, and returns.
Respondents were initially identified through convenience sampling among people who had crossed the border from Ukraine and were interviewed through a data collection initiative since 28 February 2022 in Poland, the Slovak Republic, Hungary, Romania and the Republic of Moldova at border crossings, transit sites, and reception centres, in partnership with UNHCR. From October 2022, additional sources were used, including Viber and Facebook dissemination campaigns. Surveys are conducted over the phone. Given the non-random sampling strategy, the results are not statistically representative and should be used as indicative only.
The sample includes individuals externally displaced, and returnees. Returnees are defined as individuals who were displaced abroad for at least one month, and have since returned to Ukraine (their original place of living, or another place within Ukraine). For the purposes of this report, the sample of returnees is further restricted to those returnees who also report an intention to continue staying in Ukraine.
While many waves of data collection exist, calculations in this report are done using four non-consecutive waves, in which all individual respondents are consistently re‑interviewed. This allows for a balanced longitudinal panel, where each person is present in all four rounds, enabling the reliable tracking of individual trajectories over time. Overall, 1 125 individuals are tracked, resulting in a total of 4 500 observations across rounds. The four waves of this longitudinal extract are:
Round 4 (conducted from 10 August to 3 September, 2022);
Round 14 (conducted from 7 June to 29 June 2023);
Pooled Rounds 25/26 (conducted from 8 May to 29 May 2024, and from 10 June to 24 June 2024);
Rounds 29/30 (conducted over from 13 September to 8 October 2024, and from 18 October to 12 November 2024).
For more details on data collection, sample description, and descriptive statistics, see (IMPACT Initiatives, 2025[13]).
IOM has been conducting General Population Surveys in Ukraine since the onset of Russia’s full-scale invasion. The data, commissioned by IOM, were gathered by Multicultural Insights through screener phone‑based interviews using the computer-assisted telephone interview (CATI) method and a random digit dial (RDD) approach. The surveys are conducted among respondents based in Ukraine, in all oblasts, excluding the Autonomous Republic of Crimea and the areas of Donetsk, Luhansk, Kherson, and Zaporizhzhia under the temporary military control of the Russian Federation where phone coverage by Ukrainian operators is not available. The sample covers IDPs, returnees, and non-displaced individuals. It excludes individuals currently abroad. The sample frame is limited to adults aged 18+, who use mobile phones.
IDPs are defined as individuals who left their homes or are staying outside their habitual place of residence due to the full-scale invasion in February 2022, regardless of whether they hold registered IDP status. Returnees are defined as individuals who returned to their habitual place of residence after a significant period of displacement (minimum two weeks since February 2022), whether from abroad or from internal displacement within Ukraine. This definition excludes individuals who have come back to Ukraine from abroad but who have not returned to their places of habitual residence in the country. The terms “return” and “returnee” are used without prejudice to status. All remaining individuals are considered non-displaced.
In this report, rounds 16 and 17 are used separately.
Round 16 was conducted between 10 March and 11 April 2024. It covers 1 639 returnees, 1 428 IDPs, and 2 266 non-displaced individuals. This round also contains a special housing module. It enables the analysis of this chapter, showing how returnees, non-displaced individuals, and IDPs compare in terms of ownership and occupancy status, whether their dwellings have different conditions, whether they have documents to prove their current housing arrangement, as well as their awareness about state programmes for damaged housing and participation in such programmes.
Round 17 was conducted between 13 July and 12 August 2024. It covers 1 188 returnees, 1 428 IDPs, and 1 800 non-displaced individuals. This round also contains a special employment module, enabling the analysis of on employment outcomes in this chapter.
For more details on data collection, sample description, and descriptive statistics, see (IOM, 2024[43]; IOM, 2024[31]; IOM, 2024[15]).
Note: Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 17.
Note: Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 17.
Note: Tabulations are weighted to account for survey design and ensure representativeness.
Source: Computations based on IOM General Population Survey Round 17.
For the analysis, more than 60 surveys conducted in Ukraine and in host countries were reviewed to examine the extent to which three factors influence the return decisions of displaced Ukrainians. These factors are the availability and condition of housing (including ownership), access to employment, and access to social services in both Ukraine and host countries. Although these topics frequently emerge in discussions with displaced persons, many surveys do not address them explicitly.
The responses presented in this Annex draw on seven selected surveys. Several of these surveys were carried out in multiple rounds. Where surveys had several rounds, only the most recent responses are included. Earlier rounds are referenced only where they contained relevant questions that were later discontinued. In such cases, only these specific questions are included in the tables.
The selection of surveys was guided by the following criteria:
Country coverage: surveys of externally displaced persons include either multiple destination countries or the main host countries.
Thematic scope: surveys contain questions on housing, employment and social services, with responses that can be linked to return intentions.
Sample size: a minimum of 500 respondents.
Timing: surveys carried out after the first year of displacement, so as to avoid results that were strongly influenced by early expectations that the conflict would end quickly.
Note: For surveys conducted in several rounds, only the most recent responses are included. Earlier rounds are referenced only when they contained relevant questions that were later discontinued or not analysed; in such cases, only these specific questions are included in the table.
Note: For surveys conducted in several rounds, only the most recent responses are included. Earlier rounds are referenced only when they contained relevant questions that were later discontinued or not analysed; in such cases, only these specific questions are included in the table.
Note: For surveys conducted in several rounds, only the most recent responses are included. Earlier rounds are referenced only when they contained relevant questions that were later discontinued or not analysed; in such cases, only these specific questions are included in the table.
← 1. The survey also contains other questions, such as intentions to return within one month, and intentions to return one day. However, the questions were not asked in all periods, precluding from a longitudinal type of analysis of the answers. Here, preference is given to the analysis of answers to the question “intentions to stay in current place”, because it was asked in a greater number of rounds and contains more non-missing observations; although the opposite of this intention is not necessarily “return”, it may also be “moving forward”.
← 2. The relatively high share of individuals reporting an intention to stay may be partly explained by the way the question is framed: it refers only to plans for the next six months.
← 3. Unless stated otherwise, here and further in this section: computations of OECD staff based on IOM General Population Survey, rounds 16 or 17.
← 4. Full data description, survey methodology, and socio‑economic characteristics of returnees can be found in IOM publications, for example, in (IOM, 2024[15]).
← 5. Data are collected among individuals aged 18 and over.
← 6. These results correspond to findings of other studies, such as (IMPACT Initiatives, 2025[13]), which reports that, in the same period (August 2024), 27% of returnees have settled in frontline areas, comprising Kharkiv, Mykolaiv, Zaporizhzhia, Donetsk, Sumy, Kherson, Dnipropetrovsk and Chernihiv oblasts in that study.
← 7. In subsequent rounds of IOM General Population Surveys, the response option was augmented to include “hosted for free or paying utilities only”. The majority of individuals in fact report paying for utilities, that is, not being hosted entirely for free.
← 8. “Employment status” refers to an individual’s position in the labour market, categorising them as employed, unemployed, or outside the labour force. This classification is based on the person’s engagement in economic activities during a specified reference period. “Status in employment”, according to the most recent International Classification of Status in Employment (ICSE), developed by the ILO, classifies employed individuals based on the type of authority exercised by the workers in their job as “dependent” and “independent”. The IOM survey data provides only categorisation “employee” and “self-employed”, which can be considered as broadly corresponding to the “dependent worker” and “independent worker” current international statistical standard, at least for the purposes of this report.
← 9. Informal employment can be permanent or temporary; conversely, permanent employment in principle can sometimes be informal. The data at hand singles out “permanent”, “temporary”, and “informal” as independent categories, with an implicit assumption that “informal” is used synonymously to casual. For this reason, and also to allow sufficient observations to enable the analysis, in this section, “temporary” and “informal” are grouped and are contrasted with “permanent”.
← 10. In this section, authored by UNHCR in collaboration with Brunel University London, the term “refugee” is used to refer to individuals displaced from Ukraine, reflecting international refugee law and UNHCR’s mandate. Elsewhere in this report, a broader analytical term “externally displaced Ukrainians” is used, which encompasses refugees as per international refugee law, and additionally includes persons who left Ukraine to de facto seek international protection abroad, but who were admitted by another country because they were already in possession of other documents authorising travel and stay, for example on grounds such as tourism, study, employment.
← 11. As highlighted, for example, in Ukraine’s Strategy for Demographic Development up to 2040.
← 12. The model considers only the Ukrainian refugee population in EU27 + Moldova (4.38 million). Total returns indicated are calculated based on the total refugee population in Europe (5.19 million) as of October 2025.
← 13. See UNHCR (2025[46]).
← 14. See ILO (2025[45]).
← 15. Insights from the forthcoming UNHCR Skills Survey on Ukrainian refugees.
← 16. The macro-regions classification of oblasts referred to in the analysis of the UNHCR-BUL ABM results is as follows: Centre (Cherkasy, Kirovohrad, Poltava, Vinnytsia), East (Dnipropetrovsk, Donetsk, Kharkiv, Luhansk, Zaporizhzhia), Kyiv (Kyiv City), North (Chernihiv, Kyiv oblast, Sumy, Zhytomyr), South (Kherson, Mykolaiv, Odesa), West (Chernivtsi, Ivano-Frankivsk, Khmelnytskyi, Lviv, Rivne, Ternopil, Volyn, Zakarpattia).