KOT scheme

Service Surcharges

Within the implementation of the KOT scheme, the various algorithms jointly contribute to, among other things, determining whether parents are entitled to the KOT scheme (the light test), checking data, determining whether parents are eligible for components (such as the break button or the guarantee letter) and monitoring if/when parents have been reached.

Last change on 8th of June 2026, at 12:06 (CET) | Publication Standard 1.0

• Publication category: Impactful algorithms
• Impact assessment: DPIA, IAMA
• Status: In use

General information

Theme

Social Security

Begin date

2021-01

End date

2027-12

Contact information

[e-mail]

Link to publication website

https://herstel.toeslagen.nl/

Link to source registration

(niet van toepassing)

Responsible use

Goal and impact

The purpose of deploying the algorithm is to make the process around the assessment of aggrieved parents as efficient and feasible as possible. The main goal is to make a timely and careful determination of whether a parent should be classified as a duped parent and to facilitate practitioners in doing so in a consistent and supportive manner. As sub-goals, the algorithm provides an alternative to full manual review of all files, which would not be feasible within a reasonable timeframe given the large number of filings (over 70,000). Algorithmic support allows processes to be significantly more efficient. In addition, the system contributes to better insight into the progress of files, strengthening process monitoring and reporting on processing. In doing so, the algorithms ensure that this is done in a consistent manner and that even exceptional groups, such as when criminal law is involved, are given the right assistance.

Considerations

Advantages:

1. Equivalence and equal treatment: The algorithm supports consistent assessment, ensuring similar cases are treated the same and reducing differences between individual practitioners.
2. Efficient and balanced distribution of cases: The initial allocation of applications to PZBs is more structured, contributing to a fairer distribution of work.
3. Legal certainty: As similar cases are supported in a uniform manner, the predictability and consistency of decision-making increases.
4. Transparency: The operation of the algorithm, the data used and the underlying choices are documented, making it clear why and how the algorithm is deployed and a parent is helped.
5. Respect for vulnerable groups: Specific situations, such as parents with criminal justice issues or other exceptionalities, remain identifiable and can be given extra attention where necessary.

Disadvantages:

1. Equity and equal treatment: Certain groups could be unintentionally disadvantaged, for example if they fell into the 'grey group' of the light test and therefore had to wait longer for treatment. This could lead to long(er) processing times and thus perceived injustice.
2. Timeliness of service provision: Parents who had to be assessed manually, such as at the initial stage or for specific exceptions (e.g. criminal fines), could face long waiting times. This affects the value of timely and careful support.
3. Trust in government: When parents had to wait a long time or when groups were inadvertently misclassified (e.g. persons in the green group who were not actually entitled to it), this could further strain trust in government, especially in a context where restoring trust is central.
4. Correct and careful assessment: In some cases, parents who were not entitled to a scheme still received an allowance, while others who were entitled had to wait a long time. This touches on the values of due diligence and proportionality. - Respect for vulnerable groups: Parents with complex situations (such as criminal problems or ambiguity about children) could be misclassified in the light test, leaving them without a decision for longer. This could lead to insufficient protection of vulnerable groups.

Human intervention

1. Employees play a central and indispensable role in the process. The algorithms only do a pre-selection (e.g. bin classification, fine/criminal law, duplicate applications). However, every case review is performed by an employee.
2. In addition, employees can deviate from the algorithm outcome when they find it necessary. They use the algorithm as a tool, not as a guiding decision. Parents can always contact UHT if they think something is incorrect or unclear.

Risk management

Risk management and safeguards:

The Surcharge Service has established conditions for algorithm development. These rules are contained in a quality framework. This contains rules and agreements to be followed when developing this algorithm. The conditions come from the National Audit Service and are guiding. The algorithm has business rules based on laws and regulations. These business rules are tested and maintained to remain compliant with laws and regulations and political wishes. When changes are made to the algorithm, the Surcharge Service tests whether the quality requirements are still met.

Equality, non-discrimination and privacy:

The use of data is tested against the General Data Protection Regulation (AVG). The AVG prescribes that we must not use more data than is necessary to achieve a particular purpose. This is called data minimisation. The Benefits Service examines which data is needed and may be used.

Profiling:

The use of data is assessed to avoid unlawful profiling.

Citizens' rights:

Citizens have the right to see the information collected about them. They can also apply to have this information amended.

Legal basis

Surcharge Recovery Act

- General Data Protection Regulation (Implementation) Act

- Citizen Service Number (General Provisions) Act

- Archives Act 1995

- Selection lists Benefits Agency

Links to legal bases

Laws government: https://wetten.overheid.nl

Link to Processing Index

(niet van toepassing)

Impact assessment

• Data Protection Impact Assessment (DPIA)
• Human Rights and Algorithms Impact Assessment (IAMA)

Operations

Data

The algorithm uses, among other things:

1. BSN, name and any dates of death
2. Contact details (phone, email)
3. Financial and legal data (previously paid amounts, criminal sanctions, misdemeanour fines)

Input data from TSEX-500 (Catshuis scheme):

- having children/childcare allowance

- misdemeanour fines or criminal convictions

- (former) partners already registered

- amount of recovery

- involvement in a CAF-comparable investigation or having faced the qualification of intentional misconduct/gross negligence (O/GS)

With the relevant data, we ensure that parents get the right letters in their journey at UHT.

Technical design

Explanation of the Catshuis scheme

Based on data, an initial test is conducted where fixed criteria are used to analyse which parents are likely to qualify for a €30 000 payment. Here, a classification is made into three groups: duped, possibly duped and probably not duped. A manual assessment follows for the last two groups.

External provider

(not applicable)

Link to code base

(niet van toepassing)