The MGA Micro-Business Book: Fraud, Underwriting Failure, or Both
You are head of financial crime at a specialty insurer. Eleven months ago an MGA launched a micro-business package; it now writes $18 million GWP across 2,300 policies with a 118% loss ratio. Seventy-four per cent of premium is funded by one premium finance company. Referral data shows 41 insureds sharing six addresses and three phone numbers, and 22% of policies bound within 14 days of the reported loss date. Nine policies name the same director. The MGA holds $250,000 binding authority and one delegated underwriter signs 80% of the business. Total MGA production including other lines is $60 million and the relationship is strategically important. Your agreement allows an on-site review on 10 days' notice. The premium finance company has written to say it may stop funding the MGA's book, which would collapse it. No policy has yet been declined.
1. Is this underwriting failure, distributor fraud, or both? What would you need to see to tell the difference?
Read the worked answer
Probably both, and the distinction matters because the remedies differ.
The facts fit three explanations:
- Genuine underwriting failure - a delegated underwriter writing micro-business without verification, with a premium finance channel generating volume.
- MGA-level fraud or collusion - the MGA, or individuals inside it, manufacturing policies or backdating cover.
- Insured-side fraud the MGA failed to detect - a cluster of small businesses, possibly one organiser, buying cover to claim.
Tests that separate them:
- Backdating. Compare bind timestamp, premium payment timestamp and loss date at policy level. A 14-day gap is unremarkable; a bind after the loss is not. Pull system audit logs, not formatted reports.
- Address and phone clusters. Are they formation agents, virtual offices, or genuinely one operator? Check the nine-director policy against corporate registries and against claims.
- The delegated underwriter. Sample his 80% and compare declinature, referral and documentation rates with peers.
- Premium finance. Are policies funded before or after binding, and who pays the first instalment?
Until the bind-timestamp evidence is in hand, do not label this fraud. It may be a bad book with an incentive problem, which calls for a remediation plan, not a termination.
2. Design the next 30 days: what you do, in what order, and how you avoid tipping off a possibly complicit MGA while not destroying a valuable relationship if it is innocent.
Read the worked answer
Days 1–10: quiet, at head office.
- Pull policy-level data and audit logs under the audit clause. Do not announce a fraud investigation.
- Suspend new binding above a threshold through a documented authority amendment, framed as a portfolio review.
- Brief your claims, legal and reinsurance teams; open a confidential file and take advice on privilege.
Days 11–20: on site.
- Exercise the 10-day notice. Sample across the delegated underwriter's book, not just the flagged policies.
- Interview MGA principals separately and ask for their own explanation of the clusters. Their reaction is evidence.
- Contact the premium finance company to coordinate. Do not let them learn from a regulator.
Days 21–30: decide.
- If MGA collusion is supported: suspend or terminate binding authority, invoke indemnity, report.
- If poor underwriting: remediation plan, tighter authority, file review of the delegated underwriter's cases.
- If insured-side fraud: refer to SIU, consider voiding, report.
The tension is real. If the evidence points to insureds rather than the MGA, tell the MGA chief executive early under confidentiality - you need their records and their cooperation, and a surprise raid on an innocent partner costs you the $60 million relationship for nothing.
3. What are your reporting, notification and recovery options once the facts are clearer?
Read the worked answer
Reporting depends on who did what.
- If the MGA or its staff are implicated: report to the regulator as a financial-crime and conduct matter. Check your own licensing obligations for appointed representatives in each territory.
- If insureds are implicated: refer claims to SIU, consider avoidance for misrepresentation, and report where the loss or the pattern meets local thresholds.
- Law enforcement: usually only worth engaging once you have a documented, defensible loss figure - intended loss where policies were backdated, actual loss where claims were paid.
Notification. The premium finance company is an affected third party; if funded policies are void, they need to know, and you need a lawful basis for sharing insured data. Your reinsurer will have fraud and void-policy notification clauses - check notice periods, because late notice can prejudice recovery.
Recovery. Options run in parallel: contractual indemnity from the MGA, errors-and-omissions claim against the MGA's professional indemnity cover, recovery of commission on voided business, and civil action against insureds where the sums justify it. Rank them by recoverable value against cost, not by moral satisfaction.
Remediation. Whatever the outcome, the finding is also about your own oversight: delegated authority monitoring, exception reporting on bind-versus-loss intervals, and a concentration limit tied to any single premium finance provider.