The impoverishment risks framework and what it still explains about displacement outcomes

1. The problem the model was built to explain

Development projects that displace people have, with striking consistency, left those people poorer than they were before. This is not a claim about badly run projects; it is the central finding of the displacement literature across decades, geographies and sectors. The observation that demanded explanation was not that displacement caused hardship — that is unsurprising — but that it caused durable impoverishment even where compensation was paid, and even where the paying institution intended otherwise.

The impoverishment risks and reconstruction model was formulated to explain that gap. Its move was to decompose displacement into distinct risk domains rather than treating it as a single shock, on the reasoning that a compensation instrument aimed at one domain can leave the others entirely untouched. A household paid the market value of its land has been made whole against one risk and against none of the rest.

2. The eight risk domains

The model identifies landlessness, joblessness, homelessness, marginalisation, food insecurity, increased morbidity and mortality, loss of access to common property resources, and social disarticulation. Each is a mechanism, not a symptom, and each has its own path to impoverishment.

Two of the eight deserve particular attention because they are the least likely to appear in a compensation schedule. Loss of access to common property — grazing land, forest products, fishing grounds, water sources — removes income and subsistence that were never individually titled and are therefore invisible to a valuation exercise keyed to ownership. Social disarticulation describes the dismantling of the networks through which rural households actually manage risk: kinship, reciprocal labour, credit without collateral. Neither has an obvious market price, and both are routinely omitted from entitlement matrices for that reason.

The analytical consequence is that compensation adequacy cannot be assessed by asking whether the payment matched the asset's value. It can only be assessed by asking which of the eight risks the intervention addressed.

3. Why the framework has outlasted its critics

The model has been criticised as descriptive rather than predictive, as insufficiently attentive to agency and differentiation within displaced populations, and as constructed from a rural frame that transfers awkwardly to urban displacement. Each criticism has force. The framework does not tell you which households will be impoverished, it treats the displaced somewhat as a category, and its risk domains map more naturally onto land-based livelihoods than onto informal urban economies.

It has nonetheless outlasted its critics, for a reason worth stating plainly: no successor framework has improved on it as a specification of what to look for. Subsequent contributions have refined it — adding risks, differentiating by gender and vulnerability, extending it to host communities — rather than replaced it. A model that is repeatedly extended rather than discarded is doing analytical work.

4. From analytical frame to evidentiary requirement

The more interesting question for contemporary practice is not whether the model is correct but why projects that cite it in their planning documents still produce the outcomes it predicts. The answer is rarely that the risks were unknown. International standards now require livelihood restoration, attention to vulnerable groups, and consideration of common property. The requirements exist.

What is usually missing is measurement. A project can commit to livelihood restoration and have no instrument capable of establishing whether livelihoods were restored, because the baseline was captured as a census of assets rather than of income sources, and the monitoring captured payments made rather than outcomes achieved. Under those conditions the commitment is unfalsifiable, which in practice means unmet.

Read this way, the model's current utility is as a specification for data collection. Each of the eight risks implies a measurement: not only what a household owned but what it earned and from where, not only where it lived but who it lived among, not only titled holdings but access to resources held in common. A baseline instrument designed against the eight risks produces a monitoring system capable of detecting impoverishment. One designed against the entitlement matrix produces a record of disbursement.

5. Implications

Three implications follow for anyone designing a resettlement instrument. First, the socio-economic baseline is the binding constraint on everything that follows: risks not measured at baseline cannot be shown to have been mitigated at completion. Second, common property access and social network structure need to be captured explicitly, because no other part of the process will surface them. Third, indicators should be specified before enumeration rather than after, since indicators defined at closure necessarily measure whatever data happens to exist.

None of this is conceptually difficult. It is, however, expensive at the point in the project cycle where resettlement budgets are least defended, which is a reasonable summary of why the model continues to describe outcomes accurately three decades on.