Blog · · EN
Where Swiss retail banking smells of implicit price coordination
Pattern-matching against the structural data, with a deliberately conservative read. Not a legal accusation — a research observation.
- Parallel competitive behaviour — each bank independently responds to identical inputs (cost of funds, SNB rate) and arrives at similar prices. Legal everywhere.
- Conscious parallelism / tacit collusion — each bank observes competitors and matches rather than undercuts, even when undercutting would be profitable. Hard to prove, often legal but suboptimal for consumers.
- Explicit collusion — banks agree (formally or informally) on prices. Illegal under Kartellgesetz Art. 5.
The strongest cluster: retail custody fees
1. 0.25 % is the Swiss anchor for depot fees, and it's hard to justify.
Across the 420 banks in the database, retail depot-fee distribution (excluding 0 and clamping to retail-relevant levels) looks like this:
| Bank type | Banks | Min | Average | Max |
|---|---|---|---|---|
| regional | 2 | 0.07 % | 0.24 % | 0.40 % |
| cantonal | 11 | 0.08 % | 0.26 % | 0.65 % |
| cooperative | 11 | 0.05 % | 0.30 % | 0.50 % |
| universal | 6 | 0.15 % | 0.34 % | 0.50 % |
| private | 3 | 0.10 % | 0.35 % | 0.65 % |
The cross-bank-type average is 0.25–0.35 % p.a. The actual marginal cost of holding a security at the central depository SIX SIS is a small fraction of that.
Schwab moved its US retail custody to 0 % in 2019; Fidelity, Vanguard and E*TRADE followed within months. DEGIRO and Trade Republic in the EU sit at 0.0–0.10 %. No comparable race-to-zero has happened in Switzerland. The new entrants (Swissquote, Saxo, IG, Alpian) sit at the low end and grow fast; the incumbents stay at the cluster average. That's the textbook signature of conscious parallelism: each bank knows that cutting starts a race, and nobody wants to start it.
Benign alternative reading: Swiss retail depot service is genuinely more expensive than US/EU (smaller scale, higher labor costs, more language overhead), and 0.25 % is a fair cost-plus margin. Possible — but the gap to US/EU is large enough to deserve a real cost study.
The second-strongest pattern: collective non-publication
2. ~56 banks all "happen" to not publish UHNW prices.
FINSA Art. 31 lets banks opt out of publishing prices to "qualified" clients. About 56 banks in the database use this opt-out and publish no price list at all. The atom-#275 publishers (Reichmuth, Bordier, Rahn+Bodmer, Bellerive, Baumann, Dreyfus) prove that publishing is commercially viable — they publish full schedules and continue to attract UHNW clients.
So why do 56 banks stay non-publishing? Either:
- They genuinely believe their clients prefer opacity (testable — but data is currently scarce);
- One bank publishing would normalize transparency and erode the comfortable industry posture;
- Or some combination.
This isn't illegal collusion. It's an industry-norm equilibrium: each individual bank rationally chooses to not publish because the others don't. No agreement needed — just the absence of any large player breaking ranks. The publishers exist but are small enough not to threaten the equilibrium. If UBS or Pictet published, the entire industry would flip within 18 months.
This is the most interesting pattern philosophically: it's coordination by omission, fully legal, but functionally identical (from a consumer-welfare perspective) to "we agree not to publish".
Three more patterns worth a closer look
3. Foreign-exchange markups (retail debit/credit cards) cluster at 1.5–2.0 %.
Wise entered at 0.4–0.5 % and grew fast among border residents and frequent travellers. Revolut, similar. No major Swiss bank has cut the FX markup on retail cards to defend against this. Same pattern as custody fees: everyone knows cutting starts a race, nobody starts it.
The card-issuer concentration helps maintain the cluster — Viseca, Swisscard and Cornèrcard handle issuance for most cantonal banks and several universals, and they set FX markups that propagate to every bank using them. That's structural rather than agreement-based, but the consumer-facing outcome is identical to a pricing cartel.
4. Credit card annual fees are uniform across issuing banks.
Silver ~CHF 80, Gold ~CHF 150–180, World/Platinum ~CHF 300+. The same tier structure repeats across nearly every Swiss issuing bank in our database, because Viseca / Swisscard / Cornèrcard set the underlying pricing for most issuers. This is structural concentration (three card processors serving 100+ issuing banks), not coordination by agreement, but the consumer-facing effect is the same. Weko has opened proceedings on interchange fees at various points; the issuer pricing has had less scrutiny.
5. Kassenobligation rates across small banks.
Small banks fund Kassenobligationen primarily via Pfandbriefzentralen — joint funding vehicles owned by the cantonal banks (Pfandbriefzentrale der schweizerischen Kantonalbanken) and the mortgage institutes (Pfandbriefbank schweizerischer Hypothekarinstitute). Both are legal joint structures that pool funding. The result is that the marginal cost of issuing a Kassenobligation is essentially identical across all participating banks. Add a roughly similar margin and the published rates cluster tightly. This is cartel-by-cost-equivalence — fully legal because the joint vehicles are themselves legal — but functionally indistinguishable from rate coordination.
What I initially suspected but the data actually disproves
6. Cantonal bank Privatkonto base fees — NOT clustering.
I initially claimed cantonal banks cluster at CHF 3–5/month for private current accounts. The data says otherwise. Of the 17 cantonal banks for which the database has explicit monthly base-fee data:
| Monthly fee | Cantonal banks |
|---|---|
| CHF 0 | 12 |
| CHF 3 | 2 |
| CHF 5 | 2 |
| CHF 6 | 1 |
Zürcher Kantonalbank explicitly markets: "ZKB Banking ist für Privatkundinnen und Privatkunden kostenlos." AGKB, BSKB, FRKB, and several others sit at CHF 0 as well. The market has already raced toward zero for the headline retail account. The remaining banks at CHF 3–6/month (BLKB, BEKB, Bancastato among them) are laggards, not coordinators. The competitive process here is visible and ongoing.
This is a useful corrective: pattern-matching from memory or general assumptions is unreliable; the data has to do the work. The cantonal-bank base-fee market shows real competition, not tacit coordination.
7. Säule 3a cash rates — clustered, but at 0.20 %, not "irrationally high".
The actual current distribution of Säule 3a CHF cash rates across retail-bank providers in the database:
| Rate | Banks |
|---|---|
| 0.20 % | 14 |
| 0.10 % | 9 |
| 0.15 % | 8 |
| 0.30 % | 3 |
| 0.25 % | 3 |
| 0.60 % | 2 |
| Other (0.05 – 0.85 %) | ~10 |
There IS clustering — 14 banks at exactly 0.20 % is suspiciously many. But the cluster sits where commercial CHF rates are, not above. The competitive heat in 3a has moved from cash rates to fund-direct TER (VIAC, finpension and Frankly at 0.40 %, traditional banks at 1.0–1.5 %), and that gap is where the visible price competition is happening today.
So the right read is: the 0.20 % cluster looks like price-following behaviour, but the welfare cost is small. The cluster that matters is in 3a-Vorsorgedepot fees, where traditional banks are still 2–3× the VIAC/finpension cost and aren't being forced to follow yet.
Patterns that are NOT suspicious
For analytical balance:
- TWINT P2P fee = free. Consumer-side coordination would have triggered immediate Weko action. Merchant fees were investigated (2018–2021 proceedings, settled with conditions).
- esisuisse fees and structure. Industry-mutual deposit insurance is a regulated, supervised structure — not a cartel.
- SIX Interbank Clearing fees. Central infrastructure, regulated tariffs.
- Pfandbrief rates themselves. Determined by transparent auction market.
- Atom-#275 publisher growth. Real competitive entry, real customer attraction, real pricing differentiation. The market IS contestable when someone tries.
- Cantonal bank Privatkonto base fees. Race to zero has already happened for most. (Corrected from my initial framing — see section 6.)
The structural picture
The list:
- Schweizerische Bankiervereinigung (SBA) — active guideline-setting
- esisuisse — mutual deposit insurance
- Pfandbriefzentralen × 2 — joint mortgage funding
- SIX Group — joint clearing, settlement, card issuance
- TWINT AG — joint P2P payment system
- Viseca + Swisscard + Cornèrcard — concentrated card issuance
- Verband Schweizerischer Kantonalbanken — cantonal-bank coordination
- Verband schweizerischer Vermögensverwalter, Verband Schweizer Privatbanken, etc.
Swiss banks meet each other in 7+ different committees every quarter to discuss legitimate infrastructure questions. Whether anything specifically anti-competitive happens in those rooms is a Weko question. The opportunity for tacit coordination is structurally embedded in a way that's not true in most other developed banking markets.
Honest analytical caveat
This document is pattern-matching on price clusters and structural opportunity. Real competition-law analysis requires showing prices are above competitive equilibrium — which needs cost-of-funds modeling, capital cost, and an acceptable-margin baseline per product. None of that is in the bankproducts.info dataset.
The most defensible statement is: these clusters look like the kind of patterns that would prompt a competition authority to investigate, and several already have been investigated. Weko has the legal standing and the data access to actually answer the question. They've used both, repeatedly. Resource constraints, not the absence of suspicious patterns, are likely why they don't investigate more.
A useful follow-up: a credible journalist or researcher asking Weko whether the visible price clustering in custody fees, FX markups and credit card annual fees warrants an industry-wide enquiry. That's how the German Bundeskartellamt forced the financial-sector enquiry in 2017–2019.