Adeola Bankole Advisory memo Decision Call · USD 280
Direct scheme membership, or stay with the provider?
60 minute call, 17 September. Reviewed and signed off by Adeola Bankole on 19 September.
You asked whether to become a direct scheme participant and drop your provider. On volume alone the case looks obvious at about 18 months. On the whole picture it is closer to four years, and the four year version is the one your board should see.
Direct participation saves roughly 0.9 basis points at your current volume, which is GBP 108,000 a year, against a build and run cost that is closer to GBP 400,000 in year one and GBP 240,000 a year afterwards. The break-even is not 18 months, it is around 40 million transactions a year. You are at 12 million.
The question
What was asked, and what was reviewed
As you put it: "Our provider takes 2.1 basis points. Direct is 1.2. Why are we not doing this?"
| Reviewed | Source |
|---|---|
| Provider contract and fee schedule | Supplied |
| 12 months of transaction volume and value | Supplied |
| Your build estimate | Supplied, from the engineering team |
| Published scheme participation requirements | Public |
| Your current settlement and reconciliation process | Walked through on the call |
| Treasury and liquidity position | Not supplied. See clause 4 |
The comparison
Both options on the same basis
| Stay with provider | Direct participation | |
|---|---|---|
| Per-transaction cost | 2.1 bps | 1.2 bps |
| Annual transaction cost at 12m | GBP 252,000 | GBP 144,000 |
| Build cost, year one | None | GBP 400,000 estimated |
| Annual run cost | None beyond fees | GBP 240,000 |
| Liquidity you must hold | None | Pre-funded, see clause 3 |
| Operational hours you must cover | Provider's problem | Yours, including settlement windows |
| Headcount required | 0 | 2 to 3, ongoing |
| Time to live | n/a | 12 to 18 months |
| Control over the roadmap | Limited | Full |
| Exposure if something breaks at 03:00 | Provider's | Yours |
The GBP 240,000 run cost is not a guess: it is two and a half people, the connectivity, the resilience testing regime and the annual assurance work, at the rates your own team quoted.
Hidden
The three costs that were not in your estimate
| Cost | Why it was missed | Scale |
|---|---|---|
| Pre-funded liquidity | Direct participation means funding your own settlement position rather than the provider funding it | Working capital, not P&L, and it is the largest item |
| Out-of-hours cover | Your team estimated build, not run. Settlement windows do not observe business hours | Roughly GBP 90,000 of the run cost |
| Assurance and testing regime | Annual, mandatory, and it is a real workload rather than a certificate | Roughly GBP 45,000 a year |
Not mine
The questions I am not answering
| Question | Who answers it |
|---|---|
| Whether Larksway meets the scheme's participation criteria | The scheme, on application |
| Whether your permissions cover direct participation | Your compliance function and your regulator |
| Whether safeguarding arrangements change | Your compliance function, with legal advice |
| Whether the provider contract can be exited when you want | Your solicitor. I read the fee schedule, not the termination clauses |
| How much liquidity you can afford to pre-fund | Your CFO. Treasury data was not supplied |
These are not caveats added at the end. Two of them could make the whole question moot, and neither is something a commercial adviser should opine on.
Next
What to do instead, this year
- Renegotiate with the provider now, from a position where you have done this analysis. At 12 million and growing you have more leverage than you are using, and 2.1 basis points is not a fixed price.
- Ask them explicitly for a volume tier at 20 million and at 40 million. Their willingness to offer one tells you a great deal about how they see you.
- Have the CFO size the pre-funding requirement. It is the number that decides whether this is ever a good idea for Larksway.
- Ask compliance the permissions question, unhurriedly, this quarter. It has a long lead time if the answer is complicated.
- Revisit the direct question at 30 million transactions, with a real build estimate rather than a first one.