1) One-page briefing

Submitter: Hleb Buziuk, independent public policy and governance researcher, writing in a personal capacity.

Three key messages

●        The Committee should judge HMRC’s Child Benefit anti-fraud-and-error intervention as an end-to-end control system, because the call covers strategy, governance, design, implementation, management and performance.

●        This evidence proposes that performance should not be judged on gross fraud/error yield alone. The Committee needs a contacted-case denominator, outcome funnel, stop-to-reinstatement data and recovery figures.

●        This evidence proposes that “lessons learned” should be accepted only where each lesson is linked to a named owner, changed control, implementation date and follow-up test.

Key findings

1. The first test should be whether HMRC can reconcile contacted cases to outcomes: response, stop, reinstatement, confirmed fraud/error and recovery. The HMRC Child Benefit statistics collection is a route to population context, but not the intervention-cohort denominator by itself [2].
2. This evidence proposes that the stop-to-reinstatement rate and median reinstatement time should be used as proportionality signals. A stopped claim later reinstated is not automatically evidence of fault; HMRC Charter reporting, customer evidence and Child Benefit entitlement sources provide context only [3][4][5][6].
3. Based on HMRC accountability reporting and cross-government counter-fraud standards, the Committee should ask who owned targeting, claimant contact, stop rules, quality assurance, recovery and lessons learned [1][7].
4. This evidence does not address individual cases. Anonymised aggregate evidence would enable the Committee to assess whether the intervention worked fairly and effectively at system level.

Key recommendations

●        Committee to HMRC: request, before oral evidence or within four weeks, an anonymised outcome funnel for the whole intervention cohort. Check that every outcome reconciles to the contacted-case denominator.

●        Committee to HMRC: request a governance map showing the Senior Responsible Owner, approval route, risk controls and quality-assurance sign-offs. Check against decision logs.

●        HMRC: publish or supply stop-to-reinstatement, resolution-time and complaint/escalation metrics before any repeat intervention, or within three months of HMRC’s response. Check by reason-coded samples.

●        Committee: seek a six-month follow-up on lessons learned, with each lesson tied to an owner, control change and assurance result.

What is new vs official sources

●        It combines fraud/error yield with claimant disruption and reinstatement measures.

●        It turns “performance” into a testable funnel rather than a narrative judgement.

●        It tests governance at operational decision points, not only at board level.

●        It treats lessons learned as auditable control changes.

Methods and limits: origin-first, no personal data, no unpublished outcome claims; where data is missing, this evidence proposes a data request. See section 3, Annex A, section 7 and section 9.

2) About the submitter

I am Hleb Buziuk, an independent public policy and governance researcher. I submit this evidence in a personal capacity. I have no role in HMRC’s Child Benefit operations and no individual Child Benefit case is discussed here.

The purpose of this evidence is to help the Committee frame checkable questions about the intervention’s strategy, governance, design, implementation, management, performance and lessons learned.

3) Methods, sources and limitations

This evidence uses an origin-first source ladder: legislation, HMRC official statistics routes, HMRC accountability material, HMRC Charter reporting routes, operational guidance and cross-government counter-fraud standards [1] to [7]. Committee reports should be treated as secondary where an origin source exists.

No intervention-specific audit or evaluation source is relied on here. If the Committee, NAO or HMRC holds such material, it should be treated as the primary outcome source. The HMRC collection pages cited at [2] to [4] are used as routes to official reporting or context, not as proof of intervention outcomes.

No headline outcome claim is made from secondary material. Claims about impact without intervention-cohort data are treated as hypotheses with a test plan in section 7.

4) Context and background

This evidence addresses HMRC’s Child Benefit anti-fraud-and-error intervention, including strategy, governance, design, implementation, management, performance and lessons learned. It avoids individual cases and identifying details.

Child Benefit operates within statutory and operational rules [5][6]. Those sources explain entitlement context, but they do not show how this intervention performed. The central question is whether the intervention can be assessed for targeting, proportionality, control and learning using anonymised aggregate data.

5) Evidence mapped to ToR/themes

5.1 Strategy and governance

Headline: The governance question is whether accountability was clear at the decision points where public money and claimant entitlement were affected. These include risk selection, claimant contact, stop/suspend rules, quality assurance, recovery action, escalation and post-implementation learning. HMRC accountability material and GovS 013 Counter Fraud provide relevant context, but the Committee needs intervention-level ownership evidence [1][7].

Evidence-based observations

●        Evidence gap: public sources cited here do not provide a complete intervention-level governance map.

●        The Committee should request the Senior Responsible Owner, decision logs, risk register entries and quality-assurance sign-offs.

●        The evidence should show who could pause, amend or escalate the intervention.

Added value: This turns governance from a general assurance claim into a decision-point ownership test.

Quotable points

●        “Governance should be tested where entitlement decisions were operationally changed.”

●        “A named owner is not enough without stop/go thresholds and escalation evidence.”

●        “Lessons learned should trace back to the decision points that created risk.”

Counterpoint/trade-off: Some targeting detail may be sensitive. The Committee can request sensitive detail privately while still seeking public assurance on ownership, thresholds and escalation.

Options: Option A: summary governance note. Option B: responsibility map and sampled decision logs. Preferred: Option B, because it is checkable.

5.2 Design and targeting

Headline: The design question is whether HMRC translated entitlement rules and risk information into a proportionate intervention. Child Benefit entitlement is governed by statutory and operational material [5][6]. HMRC Child Benefit statistics can give population context, but they are not the same as the contacted intervention cohort [2]. The Committee should ask how HMRC defined the target group, what exclusions were applied, how equality and vulnerability risks were assessed, and how possible false positives were monitored.

Evidence-based observations

●        Evidence gap: the contacted-case denominator and target-selection rules are not visible in the sources used here.

●        Evidence gap: without reinstatement and reason-code data, the Committee cannot assess possible false-positive burden.

●        The intervention population should not be confused with the wider Child Benefit population.

Added value: This separates the outer population denominator from the actual intervention denominator.

Quotable points

●        “A broad population statistic cannot prove a targeted intervention was proportionate.”

●        “The Committee needs the contacted cohort, not only Child Benefit totals.”

●        “Targeting quality is visible in stop, correction and reinstatement data.”

Counterpoint/trade-off: Narrow targeting can miss some fraud or error. Over-broad targeting can create avoidable payment disruption and administrative burden.

Options: Option A: general explanation of targeting criteria. Option B: anonymised cohort counts, exclusions and quality-assurance results. Preferred: Option B, with sensitive details handled privately.

5.3 Implementation, claimant contact and service management

Headline: Implementation should be assessed through what claimants were asked to do, response times, channels available, vulnerable claimant handling, and what happened when a claim was stopped or reinstated. HMRC Charter reporting and customer evidence provide service context only [3][4]. The technical and legal framework provides entitlement context [5][6]. The Committee should request intervention-specific contact, response, resolution and escalation data.

Evidence-based observations

●        Evidence gap: response rate, stop rate, reinstatement rate and median reinstatement time are needed.

●        Evidence gap: vulnerable claimant handling should be measured by outcomes, not only by guidance.

●        Complaints, MP referrals and formal challenges should be reported per 1,000 contacted cases.

Added value: This links operational contact quality to measurable outcome and burden indicators.

Quotable points

●        “Implementation quality is visible in response, reinstatement and complaint data.”

●        “A claimant safeguard is only meaningful if its outcomes are recorded.”

●        “Service pressure should be measured against the intervention cohort.”

Counterpoint/trade-off: Short response windows can protect public money. They also increase risk if contact routes or evidence requirements are unclear.

Options: Option A: improve letters and scripts. Option B: add escalation, pause and reinstatement rules with metrics. Preferred: Option B, because it tests whether safeguards worked.

5.4 Management, performance and lessons learned

Headline: Performance should be measured as net assurance, not gross yield alone. The Committee should ask HMRC to distinguish confirmed fraud, claimant error, official error, prevented loss, recovered overpayment, reinstated claims and unresolved cases. Lessons learned should then be linked to changed controls, named owners, dates and assurance tests. This is consistent with a counter-fraud control approach [7].

Evidence-based observations

●        Evidence gap: public sources cited here do not provide a complete outcome funnel for this intervention.

●        Evidence gap: value for money cannot be judged without cost, yield, burden and false-positive proxy measures.

●        A lesson learned should be evidenced by a changed control, not only a narrative statement.

Added value: This creates a single test for performance, proportionality and learning.

Quotable points

●        “Gross savings should not be the only measure of performance.”

●        “Every outcome claim should reconcile to the contacted-case denominator.”

●        “A lesson learned is credible only when it changes a control.”

Counterpoint/trade-off: Some outcome data may still be live or sensitive. Aggregated cohort reporting should remain possible without naming individuals.

Options: Option A: written HMRC update on lessons learned. Option B: owner-dated action tracker with six-month follow-up. Preferred: Option B, because it creates accountability.

6) Recommendations

Quick wins

1. Committee lever: outcome funnel
   Who: Committee to HMRC. What: request contacted, responded, stopped/suspended, reinstated, confirmed fraud, confirmed error, recovery and open-case counts. When: before oral evidence or within four weeks. Check: totals reconcile to the contacted-case denominator.
2. Committee lever: governance map
   Who: Committee to HMRC. What: request Senior Responsible Owner, responsibility map, approval records, risk register and quality-assurance sign-offs. When: before oral evidence. Check: sample three decision points: targeting, stopping and reinstatement.
3. Delivery action: proportionality dashboard
   Who: HMRC. What: report stop-to-reinstatement rate, median reinstatement days and complaint/escalation rate. When: before any repeat intervention, or within three months of HMRC’s response, whichever is earlier. Check: reason-coded sample audit.

Structural

4. Delivery action: lessons tracker
   Who: HMRC. What: maintain an owner-dated tracker linking each lesson to a changed control and assurance result. When: within three months of Committee recommendations. Check: closed actions have evidence.
5. Committee lever: follow-up review
   Who: Committee, with NAO or internal audit support if appropriate. What: revisit the data pack and lessons tracker. When: six months after HMRC’s formal response. Check: dashboard fields are in routine use.

7) Quant-lite metrics and Minimum Data Request Pack

The core derived metrics cannot be computed from the published material cited here because the contacted-case intervention denominator and outcome counts are not available. Published population statistics do not provide the intervention denominator. These metrics should not be estimated.

Ready-to-compute metrics

●        Stop-to-reinstatement rate: reinstated claims divided by stopped/suspended claims. Tests proportionality. Requires HMRC intervention-cohort data and reason codes.

●        Confirmed yield per contacted case: confirmed fraud/error value divided by contacted cases. Tests targeting efficiency. Requires HMRC intervention-cohort data and must not merge prevented loss and recovered overpayment.

●        Complaint/escalation rate: complaints, MP referrals or formal challenges divided by contacted cases, multiplied by 1,000. Tests burden adjusted for scale. Requires HMRC intervention-cohort data.

8) Impact, performance and VFM artefact

This evidence proposes a minimal dashboard that separates inputs, outputs, outcomes, burden and assurance.

This template would let the Committee test whether the intervention delivered value for money without overlooking avoidable disruption.

9) Committee may wish to ask

To HMRC

1. What was the contacted-case denominator for the intervention?
2. How many cases were stopped, reinstated, confirmed as fraud, confirmed as claimant error, or left unresolved?
3. What was the median time from stop to reinstatement?
4. What reason codes explain reinstated cases?
5. Who approved targeting, claimant letters, stop rules and recovery rules?
6. What vulnerable claimant safeguards were used, and how were their outcomes measured?

To HM Treasury or departmental oversight witnesses

7. How did oversight bodies assure themselves that gross yield did not mask avoidable disruption?
8. What follow-up will test whether lessons learned have changed controls?

To NAO or internal audit, if involved

9. Can a sample of stopped and reinstated cases validate HMRC’s reason coding?
10. Are the Committee’s requested dashboard fields already available in HMRC systems?

10) Risks and further work, references and Annex A

Risks and further work

●        Do not overclaim performance without intervention-cohort data.

●        Do not merge fraud, claimant error, official error, prevented loss and recovery.

●        Do not include personal or identifying case material.

●        Treat complaints and lived-experience signals as context unless representative.

●        Audit/evaluation evidence is not cited because none has been verified for this draft.

●        Exclude High Income Child Benefit Charge material unless the intervention directly involved it.

References

[1] HM Revenue & Customs, Annual Report and Accounts 2024 to 2025, GOV.UK, https://www.gov.uk/government/publications/hmrc-annual-report-and-accounts-2024-to-2025

[2] HM Revenue & Customs, Child Benefit statistics collection, GOV.UK, used here as a route to official population statistics and not as intervention-outcome evidence, https://www.gov.uk/government/collections/child-benefit-statistics

[3] HM Revenue & Customs, HMRC Charter annual reports collection, GOV.UK, used here as service and Charter context only, https://www.gov.uk/government/collections/hmrc-charter-annual-reports

[4] HM Revenue & Customs, HMRC customer survey collection, GOV.UK, used here as customer-service context only, https://www.gov.uk/government/collections/hmrc-customer-survey

[5] HM Revenue & Customs, Child Benefit Technical Manual, GOV.UK, https://www.gov.uk/hmrc-internal-manuals/child-benefit-technical-manual

[6] Child Benefit and Guardian’s Allowance Administration Regulations 2003, legislation.gov.uk, https://www.legislation.gov.uk/uksi/2003/492/contents

[7] Cabinet Office and HM Treasury, Government Functional Standard GovS 013: Counter Fraud, GOV.UK, https://www.gov.uk/government/publications/government-functional-standard-govs-013-counter-fraud

Annex list

●        Annex A: Evidence Ledger.

●        No vignettes or synthetic scenarios are included.

Annex A: Evidence Ledger