# Committee brief — Flood reinsurance (£5m cap)

**To:** Risk Committee, Horizon Risk Consulting  
**From:** Horizon trainee (worked example)  
**Subject:** Buy reinsurance capping annual flood losses at £5m for £400,000?

## Executive summary

Under the Module 4 surface-water stress, gross portfolio loss is about **£7.58m**. A treaty that caps retained loss at **£5m** for an annual premium of **£400k** leaves a net cost of **£5.4m** in that scenario — about **£2.2m** better than standing uncapped. **I recommend buying** the cover if the premium is available on these terms and the stress remains a planning scenario the board takes seriously. The recommendation is about capping a bad year, not beating expected value in an average year.

## Background

In 2019–style ERM work we showed that capital equal to 1.5× normal-year flood loss (~**£4.4m**) does not cover the stress (~**£7.6m**). The committee asked whether a simple stop-loss / cap at **£5m** for **£400k** is worth buying. That is a capital and continuity decision for a rare concurrent flood season, not a motor pricing tweak.

## Method

- **Data:** `flood_portfolio.csv` — 500 synthetic household policies (sum insured, flood score 0–100), same generator as Actuarial 201 Module 4 (seed 77).  
- **Stress:** scores ≥70 → 15% of sum insured; 50–69 → 5%; below 50 → 0 (deterministic scenario).  
- **Structure:** retained = min(gross stress, £5m); recovery = max(gross − £5m, 0); net = retained + £400k premium.  
- **Chart:** `outputs/charts/actuarial301_main.png`.

## Results

| Item | £ |
| --- | ---: |
| Gross stress loss | **7,580,350** |
| Retained with £5m cap | **5,000,000** |
| Reinsurance recovery | **2,580,350** |
| Premium | **400,000** |
| Net cost (retained + premium) | **5,400,000** |
| Saving vs uncapped stress | **2,180,350** |

In this stress year the cover attaches: recovery exceeds the premium by a wide margin. Regional stress remains spread across East / North / South / West; the treaty is portfolio-level, so region mix matters less than total severity.

## Recommendation

**Buy** the £5m cap for £400k, subject to contract wording (what triggers the year aggregate, exclusions, reinstatements). Pair it with the earlier advice not to rely on 1.5× normal-year capital alone. If the premium rose sharply or the stress model were softened below the cap, revisit — the value is the **£2.6m** layer above £5m in a bad year.

## Limitations

Synthetic portfolio and a blunt score-based stress, not a hazard-map catastrophe model. One scenario, not a return-period distribution. Premium is treated as certain; no basis risk, disputes, or counterparty default. Motor and other lines ignored. Buying for this stress does not prove the cover is cheap relative to the *probability* of the stress — only that *if* the stress hits, net cost falls.

## Personal reflection

Choosing the question (reinsurance vs another pricing or pension option) was harder than the arithmetic. The actuarial bit was stating the assumption list before calculating and recommending despite uncertainty. Enterprise risk / capital decisions feel more like the job I want than pure coding.