A Community Strategy Under Occupation: How Lysychansk Plans Its Recovery

The Lysychansk community is 17 settlements, 418 square kilometres, and approximately 115,000 people before the invasion. Since July 2022, its entire territory has been under occupation, and the administration operates from relocation: 41,359 residents are registered across Ukraine, with around 13,000 staying near four humanitarian hubs. Dozens of communities in Ukraine are in the same position, and no standard strategic planning methodology was designed for them. Here is how we built a strategy for Lysychansk when you can only plan half of the problem — and what this means for those who help displaced persons.

Why a community can exist without access to its own territory

Legally the community exists; physically it is unreachable. Local self-governance powers are exercised by the city military administration, operating from other cities.

What kind of community this is

Where the community is now

41,359 residents are registered on the territory of Ukraine. Around 13,000 are staying near four humanitarian hubs: Dnipro (approximately 6,029 people), Kyiv (4,754), Poltava (1,877), Zvyahel (359).

These are service platforms with a full range of services. The administrative services centre relocated to Dnipro as early as 13 July 2022, provides 40 services — 19 of them in the field of veterans' protection — and processed 5,705 applications and 1,173 notifications of damaged property in 2024. The relocated medical network and the Medical Rehabilitation Centre, opened in Dnipro on 1 March 2024, provided around 9,500 consultations in its first year.

What survived as an institution

Together with the functioning capacity of the administration itself, this is what the community has now instead of its territory: registers, staff, services, and an educational network operating away from home.

Why a standard community strategy does not work here

All conventional community planning methodologies were written for a different situation: territory under control, the object of planning in place, statistics accessible. In Lysychansk, none of these conditions hold.

The object of governance is split

• The community as a collective is accessible and measurable. Goals with indicators and budget programmes operate here.
• The occupied territory is unknowable until return. Neither the timeline of de-occupation, nor the actual scale of destruction, nor the composition of the population that will return are known.

Risk and deep uncertainty are different things

The distinction here is not terminological. Risk means that the distribution of possible states is known or at least estimable, and decisions can be optimised against it. Deep uncertainty means no such distribution exists, and parties do not even agree on a model of the world. This distinction comes from the paradigm of decision-making under deep uncertainty, developed in public administration by RAND Corporation and the group of researchers around Volodymyr Marsho and Warren Walker.

Under deep uncertainty, optimising for a single forecast loses its meaning. In its place, three types of decisions operate: robust ones, which maintain acceptable quality across many scenarios; adaptive ones, which reconfigure as information arrives; and contingent ones, which are tied to triggers in advance.

Five uncertainties compound one another

The complexity of this case lies not in a single factor but in how they reinforce each other:

1. Military. The de-occupation horizon the strategy works with is 2030–2032. Assets are frozen for years; there is no reconstruction starting point.
2. Fiscal. The budget is almost entirely subsidy-dependent, and in 2027 accumulated reserves will be withdrawn from the community.
3. Demographic. Approximately 45% of residents firmly plan to return, while back-migration to occupied territory due to hardship was at least 7,000 people.
4. Informational. 98.9% of residents cannot or consider it unsafe to visit their own home, and 53.7% have no information about its condition.
5. Environmental and engineering. Mining of agricultural land and forests, the risk of mineralised mine water rising into the Siverskyi Donets, and territories subject to landslides and undermined by mine workings.

Remote surveys cannot provide a reliable picture of damage and mine threats. A single-phase framework under these conditions leaves two exits, and both are bad: either plan the inaccessible territory on invented data, or exclude reconstruction from the strategy altogether. The first undermines the document's evidentiary basis; the second leaves the community unprepared for return.

What a two-phase strategy is: three goals now and five directions later

The strategy contains two classes of decisions with different legal status. This is the essence of the approach.

Five constructs on which the regime rests

1. Distinction of the object by phase. Phase A works with the relocated community; Phase B works with the de-occupied territory and the returning population.
2. The link between type of uncertainty and method. Risk leads to goals with indicators; deep uncertainty leads to contingent directions.
3. Pre-positioning. Phase A deliberately prepares Phase B: readiness indicators, a project pipeline, damage registers, and partnerships.
4. Trigger-based activation. Phase B directions are activated by a decree after de-occupation, and only then do numerical targets appear — based on damage and needs assessments.
5. Readiness instead of false precision. A leading indicator shows whether the community is ready to build, not how many buildings it promises to restore.

Phase A: three goals that are funded now

These have indicators and are implemented through annual action plans:

1. Support for displaced persons and preservation of the community. The hub network, the relocated administrative services centre, the medical network and rehabilitation, an educational network of 5,682 pupils.
2. Support for the Defence Forces. The largest expenditure item of the community; relocated municipal teams carry out preparatory fortification works, equipment repair, and metal processing.
3. Preservation of capable infrastructure outside the occupied territory. A temporary accommodation complex in Voinivka, Kirovohrad Oblast, is being expanded from 80 to 240 beds.

Tellingly, the community's budget structure already matched these goals before the strategy was adopted: support for displaced persons accounts for around 28.6% of expenditure, the Defence Forces around 63.0%, and infrastructure and administration around 8.4%. The document formalised what had already taken shape, rather than introducing a new resource structure.

Phase B: five directions on standby

Demining and environmental rehabilitation; housing and municipal services; a new post-industrial economy; the return and reintegration of people; institutional capacity and digitalisation. Until activation, they create no financial obligations and are not assessed annually.

The sequence is set by order, not by sums of money. The logic of reconstruction here is the same as in the Build Back Better principle, but without timelines and cost estimates. Demining comes first because without it neither agricultural, nor residential, nor infrastructure recovery is possible. The economic direction is oriented not towards reviving industrial greatness but towards a diversified model: small-scale agro-processing, niche crops, small enterprise.

Phase B directions are aligned with the regional Luhansk Oblast Strategy to 2027 and with EU acquis standards, which set the environmental and engineering benchmark for future reconstruction before it begins.

How to measure what does not yet exist

Three forks that gave rise to this method

The approach is better shown not as a finished diagram but as a trajectory through the points where the conventional tool fails. At each one we chose not the "best in general" option but the one that wins given this specific problem structure. In the language of innovation economics, each such branch is a crisis innovation in method: a decision that would be redundant under normal conditions but becomes the only viable one under crisis conditions.

First fork: adapt the framework or build a separate regime

The default move is to take the classical methodology of local strategic planning and adapt it to wartime realities. For accessible communities — even front-line ones — this is justified. The alternative is to recognise that the object of governance has split into two entities with different status, and to build a separate regime for them.

It is the problem structure that decides, not preference. A single-phase framework forces one of two choices: either plan the inaccessible territory on invented data, or exclude reconstruction from the strategy. Both concessions are unacceptable, so the branch of a separate regime wins. Its cost is the rejection of a familiar, recognisable document form and the need to justify the non-standard approach to everyone who reads it.

Second fork: to set reconstruction figures now or not

A strategy by default has measurable targets, and reviewers, donors, and leadership expect them. The temptation to write plausible reconstruction indicators is strong.

Yet under deep uncertainty, when neither the de-occupation timeline nor the scale of destruction nor the composition of the returning population is known, any such figure is a stylistic device, not a commitment. The argument against is twofold: methodological, because a number without a probability distribution carries no information; and reputational, because invented targets destroy trust. The cost of this branch: the document looks less concrete to a superficial reader.

Third fork: money or connection

Intuition suggests the former: provide housing, payments, jobs, and people will stay connected to the community. This is precisely where empirical data entered the project decision and overturned it. Survey data showed that the full set of measured material factors explains about 1.6% of the intention to return — meaning the lever on which the obvious branch relied barely moves the target variable.

The practical cost of the chosen branch: informational resilience is harder to justify in a budget than square metres of housing.

Why this is a scientific claim, not a slogan

The three forks are not independent. The rejection of a single-phase framework makes it possible to reject false Phase B targets, and empirically grounded retention gives substance to the very Phase A that the first fork opened.

What is fundamentally different: each of the three decisions can be falsified. If there were a way to correctly assess the inaccessible territory, the first fork would collapse. If readiness did not correlate with subsequent speed of reconstruction, the second would be devalued. If in another community the intention to return proved to be materially driven, the third would resolve differently.

Where the money goes in a community that has lost its own economy

The community's own tax base has virtually disappeared along with its economy.

Budget structure

Dynamics over four years

What has been destroyed

The economic base that once generated taxes no longer exists. The LYNIK oil refinery was destroyed in 2022 and dismantled by the occupiers in 2024. The Lysychanskavuhillia mines — 11 subdivisions and around 3,200 miners — have been flooded, with wage arrears exceeding UAH 56 million. Up to 90% of economically significant assets have been destroyed or scrapped. In 2023 the community lost 760 individual entrepreneurs and 20 legal entities.

To this are added pre-war structural weaknesses that have not gone away: the territory's logistical dead-end character, a narrow domestic market concentrated around 70% in the administrative centre, an almost complete absence of agro-processing, and a negative foreign trade balance that amounted to UAH 262.8 million in 2021.

The 2027 cliff

For comparison, capital works have stalled in accessible Tsyrkuny for different reasons, and that is a separate story about a state concept. The critical point for Lysychansk is 2027, when accumulated reserves — around UAH 0.3 billion out of the UAH 1.3 billion accumulated since 2022 — will run out.

Under a pessimistic scenario, the community's fiscal space will shrink by 30–40%. The only lever the community can pull on its own is personal income tax from officially employed displaced persons. Because of this, employment policy for displaced persons ceases to be a purely social task and becomes a fiscal one.

What the survey of 1,415 displaced persons showed about intention to return

The survey was conducted in February 2024 among registered displaced persons of the community. The result was the opposite of intuition.

How it was measured

1,415 individual responses, online, non-representative sample. An important methodological detail: individual responses were collected, not consolidated household positions. This made it possible to move from shares to relationships between factors. Descriptive statistics, Spearman rank correlation for ordinal variables, Cramér's V for categorical variables, and binary logistic regression were applied to assess the aggregate effect of the material block on the dichotomous intention to return.

What determines the intention to return

Overall willingness to return after de-occupation is approximately 45%. Two factors proved significant: age and housing condition. Age is positively associated with willingness, with a rank correlation of approximately +0.15.

Two results here are atypical. The first atypical result is that the set of material factors barely moves the target variable, even though logic suggests the opposite. The second is that among owners of fully destroyed housing, willingness to return is lowest. A lost home acts as a severance from place.

Social profile of the community

This is an educated and predominantly working-age core, yet structurally vulnerable, with an agenda shifted from development to survival.

What this means for Phase A

The community can be lost not through combat but through the slow dissolution of people into their relocation sites. Retention must be symbolic, informational, and targeted, with particular attention to young people who have lost their physical attachment to place. We call this active informational resilience, not passive maintenance of contact.

The same two-phase logic appears in the community's consolidated SWOT analysis: each factor there is labelled with the phase in which it operates, and the security dimension is integrated into weaknesses and threats, rather than being placed in a separate block. The social lens is inverted as well. For a community on controlled territory, a strength is cohesion in place; for a relocated one, the question is different — whether it will be possible to preserve connection with a dispersed collective.

Why Tsyrkuny plans in cost estimates while Lysychansk plans in readiness

Both communities were severely affected, I helped develop both strategies, but the planning regimes turned out differently. The reason is one: territorial accessibility.

Comparison of the two communities

Why one community can set figures and the other cannot

The Tsyrkuny-2030 Strategy was built in a different situation: the community has access to its territory and assessable destruction, so specific sums and timelines can be set. Lysychansk cannot do this: any numerical reconstruction targets would be invented.

This is a typology in numbers. With an accessible territory, numerical targets are possible and appropriate. With an inaccessible one, they are impossible honestly.

Different communities require different sociology

In Tsyrkuny, 134 households — about 10.2% of those present — were surveyed with a 50-question questionnaire in July–August 2025, and consolidated household positions were recorded. This fits a community where around 3,150 people remained in place and where a family lives together.

In Lysychansk, individual responses from dispersed people were needed: only they allow statistical analysis of connections, as required by a fully dispersed community.

What this means for foundations and donors

Lysychansk data are useful to everyone who distributes assistance among displaced persons, and several conclusions directly change practice.

Four conclusions that change the work

• Volume transferred does not measure outcome. If material factors explain 1.6% of the intention to return, then aid addresses the family's survival task, not the community preservation task. These are different goals, and measuring them with one indicator is inadvisable.
• Risk groups are visible in advance. Young people with destroyed housing become detached from the community two to three times more often than the average displaced person. Targeted work with them yields more than an even distribution.
• Information carries the weight of aid. More than half of those surveyed do not know the condition of their home. Reliable information about one's own property functions here as a distinct service.
• Documents unlock money. Having valid urban planning documentation is a condition for access to reconstruction funds and protection against alternative master plans. Supporting communities in preparing documents pays dividends for years ahead.

Where our model's limits lie

In its recovery direction, the International Unity Movement Foundation works with communities in Kharkiv Oblast that are already de-occupied and accessible — that is, in the field where the classical planning regime operates. The Lysychansk case shows the boundary beyond which our model does not work, and it is useful to know this before offering it to the next community.

For communities in relocation, the toolkit is different: information work instead of capital projects, targeting by risk group instead of even distribution, preparing documents instead of construction.

What this research does not prove

The restraint in formulations is deliberate, and it is stated in the article itself.

What limits the result

Owing to occupation, field and engineering surveys are impossible, and state statistics broken down by community are lacking, so damage estimates are interval-based and rely on secondary sources.

When the approach is appropriate and when it is not

It makes sense only where three conditions hold simultaneously: the territory is inaccessible, the community is relocated, and uncertainty about return is deep. If even one condition is lifted, the classical regime adapted for war becomes more appropriate.

The findings are applicable to dozens of communities in eastern and southern Ukraine with temporarily occupied territory. The next step we are planning is replication of the approach in two or three communities and external validation, without which the framework cannot become a methodology.

This material summarises our research published in a peer-reviewed journal and is for informational purposes. The full text with tables and methodology is available at the link in the sources. Legal provisions are cited as of the date of publication of the research and may change.