Risicoscan Verwijtbare Werkloosheid triples the hit rate of UWV checks

The Netherlands Court of Audit found that UWV caseworkers find three times as much culpable unemployment in files flagged by the agency's risk scan as in a random sample.

  • 2 an organization

    Caseworkers found culpable unemployment in 17% of flagged files against 5.5% of random files, with far fewer staff than checking every application would need.

  • 1 a person

    The Court found that the scan made the process less intensive for applicants as well as for UWV, though it gives no measure.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Reported benefit.
Evidence

foundstudiedreportedclaimed

An independent audit report supports the benefit.

Compared with

Unemployment benefit applications drawn at random and handled blind by the same caseworkers. Measured.

Without the system
In 5,5% van de dossiers is verwijtbare werkloosheid gevonden door behandelaren
With it
In 17% van de dossiers is verwijtbare werkloosheid gevonden door behandelaren
The system’s part
The Court compares files that the scan flagged with a random sample that caseworkers handled blind, which isolates the scan's effect on the hit rate. It describes a model that estimates the probability of culpable unemployment from applicants' characteristics, with each variable's contribution justified statistically. It does not name the technique. UWV writes that it checks whether the population used to develop the scan is still representative.
Limits
Not yet checked.
Costs and harms
Not yet checked.
Later
Not yet checked.

Download data Cite Suggest a correction

Benefit type
Improved Task Accuracy
System
Risicoscan Verwijtbare Werkloosheid (risk scan for culpable unemployment)
Made by
UWV (Uitvoeringsinstituut Werknemersverzekeringen)
Used by
UWV, for unemployment benefit (WW) applications
Where
Netherlands (national)
Stage
In routine use
Date
, when the source was published
Kind
Aggregate result
Broad benefit
Analyzing Data or Information, Organizing, Planning, and Prioritizing Work
Collections
Scrutiny and findings
Last revised
31 October 2026

Sources

  1. Audit report, by authors with no tie to the system.

    Algemene Rekenkamer. Resultaten verantwoordingsonderzoek 2024 ministerie van Sociale Zaken en Werkgelegenheid (XV), Rapport bij het Jaarverslag 2024, §5.1 Algoritme Risicoscan Verwijtbare Werkloosheid bij UWV, pp. 56-59, Figuur 8. Published 21 May 2025.

    De risicoscan blijkt een effectieve manier om verwijtbare werkloosheid te onderzoeken. Na signalering door de risicoscan vinden behandelaars 3 keer zo veel verwijtbare werkloosheid vergeleken met een willekeurige steekproef.

  2. Other, by the organization that used it. Context or contrary evidence.

    UWV. Risicoscan Verwijtbare Werkloosheid. Web page on uwv.nl, read on 31 October 2026.

    Zo wordt bijvoorbeeld gecontroleerd of de populatie die bij de ontwikkeling van de risicoscan is gebruikt nog representatief is.

Details

AI system

Workflow

The scan estimated each application's chance of culpable unemployment from characteristics such as earlier benefit applications and work history. Caseworkers investigated the flagged files and a 30% random sample, hearing both the applicant and the employer.

Money

Kind of value

Potential yearly benefits of working with the risk scan, by UWV's estimate

Amount

Estimate: tussen de €29 en €44 miljoen per jaar

Recipients and payers

UWV

Follow-up

Conditions for replication

Caseworkers must not know whether a file was flagged or drawn at random, which prevents biased assessment. Each variable must be shown to help predict culpable unemployment, as UWV did to rule out direct discrimination.

Entry details

Sector

Social security (unemployment benefit control).

Purpose

UWV built the Risicoscan Verwijtbare Werkloosheid to flag unemployment benefit applications with a high chance of culpable unemployment for investigation by caseworkers.

Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, beneficiaries, reach, distribution and access, duration, every measurement question, currency and prices, period and unit, realization, gross or net, included costs, comparison, method, return and payback, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.

Cite this record

Peter H. Vartanian. (2026). Risicoscan Verwijtbare Werkloosheid triples the hit rate of UWV checks (AIBD-0917). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0917.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)

BibTeX Data Field definitions