BACEN systematically commoditizes each profitable layer of Brazilian banking

Claim

BACEN's pro-competition agenda is not a series of one-off reforms — it is a standing policy of competitive commoditization, executed layer by layer over a decade and still running.

Action	Date	Profit pool attacked
PIX	2020	Payments, TED, boleto — annihilated. Free, instant, universal, central-bank-owned
Open Finance	full since Apr 2024	Data exclusivity and relationship lock-in. World's first comprehensive implementation
Law 14.690/2023	2024	Revolving card interest capped at 100%/yr; debt cannot exceed 2x principal
Prepaid interchange cap	in progress	Fintech debit/prepaid interchange — explicitly aimed at the instrument fintechs rely on
PIX Parcelado	adaptation Mar 2026	Card installment credit, brought to ~60M cardless Brazilians without a card
PIX Automático	mandatory Jan 2026	Recurring/subscription billing moved off cards
Drex	pilot	Collateralized lending and capital-markets access
Res. 495/2025 + RC 14/2025	phased to 2028	Capital minimums by activity — 339 fintechs (19%) non-compliant Jul 2026, 679 (39%) by 2028

The generalizable finding

Brazil has arguably the lowest structural switching costs in retail banking on earth, and this was deliberate policy. PIX removed rail lock-in; Open Finance made data portable by consent; salary and credit portability are legally mandated. Any Brazilian financial's stickiness is therefore behavioural, not structural — an execution advantage that must be re-won continuously, not a moat that compounds while management sleeps.

Consequence for modelling: do not extrapolate a Brazilian financial's current margin structure. Underwrite the franchise, not the margin. Terminal ROE should fade toward 1.3–1.5x cost of equity, not hold at current levels, because competing away extraordinary returns in banking is the regulator's explicit project.

⚠️ The asymmetry that matters: you cannot out-compete a central bank that builds free public infrastructure. A firm can only adapt faster than incumbents — and every adaptation cycle resets margins lower.

The two-sided effect — this is not uniformly bearish

The same agenda that compresses margins also consolidates the field toward scaled players:

• Will Bank liquidated Jan 2026, ~12M customers orphaned
• Res. 495/2025 capital minimums push 19–39% of fintechs out of compliance
• Big Tech has retreated — Meta discontinued WhatsApp business card payments in Brazil in Jan 2026 despite ~99% smartphone penetration. PIX left no payments profit pool worth capturing, which killed Big Tech's entry vector as thoroughly as it killed the incumbents' rail advantage
• Mercado Pago deferred its Brazilian banking licence (Feb 2026, "Não é o momento agora")

So the correct read is margin compression plus competitive-set contraction. A scaled, low-cost operator loses pricing power but gains a shrinking list of rivals. Model both.

Macro state — REFRESHED 2026-09-28

Selic 13.75% — five consecutive cuts, down from the 14.25% this note first recorded. Focus YE26 13.50% (was 14.00%). BCB's own odds of breaching the 4.5% inflation ceiling 90% (was 79%) — the easing is happening into a rising breach probability, which is the tension to watch. System delinquency 4.7% in June 2026, highest since the series began in 2011; household debt/income 49.9% (all-time high); debt service ex-housing 27.1% (record since 2005). BRL appreciated ~9.5% over 12 months.

🔁 SUPERSEDED 2026-09-28 — the election is NOT orthogonal to NU.

This note originally read: "Consensus election trade is state-linked names (BB, XP), not private-sector fintech — largely orthogonal to a name like NU." That is now false. Both presidential candidates are running on household-debt relief, which aims directly at the consumer-credit profit pool that is NU's entire business. The consensus trade may still favour state-linked names; the consensus policy does not spare private fintech.

Kept rather than edited, per SPEC: the original reading was a reasonable inference from the trade flow and a poor inference about the policy, and that distinction is the useful part.

General election Oct 4 / Oct 25, 2026.

The 2026 structural shift worth tracking

Consignado (payroll-deductible) lending went from 2% to ~6% of Brazilian non-mortgage consumer credit in a single year (R$133B+, 10M workers). This is a permanent transfer of the consumer lending profit pool from unsecured to secured, and it is decided by cost of funds — which favours incumbents (Itaú, BB, Caixa) over fintechs. Origination share mid-2026: Itaú 18%, BB 13%, Santander 10–11%, Nubank 0.4%.

Corollary worth remembering: "secured" in Brazil means "secured until Brasília says otherwise." NU became the largest FGTS provider in 18 months and the government restructured the product.

What would falsify this

A BACEN policy reversal toward protecting incumbent profit pools, or a sustained period (2+ years) with no new competitive-infrastructure initiative. Neither looks likely; the agenda has been consistent across multiple administrations and central bank governors.

Related

[[principle-currency-consistency-roe-vs-coe]] — the companion method note; any Brazilian financial's ROE is BRL-nominal and must be discounted accordingly.

[[principle-story-vs-revenue]] — LatAm fintech pitches lean heavily on profit-pool TAM ("7% of a >$100B pool"); require the revenue line.

History

• 2026-07-31 — established from the NU analysis, the agency's first work in Brazil/LatAm. kb.py find returned nothing on the field beforehand — this was a cold start, so confidence is medium rather than high: the pattern is well-evidenced but has been observed through a single company's lens.

Four new instances — 2026-09-28

The commoditization thesis predicted this pass's regulatory shape correctly, which is why confidence moves medium → high. Four concrete additions to the layer-by-layer table:

Action	Date	Profit pool attacked
Desenrola 3.0 / MP 1,393	signed 2026-09-25	Legacy bad debt — R$150B reverse auction at 90-95% discounts. Resolves NPLs but at the creditor's expense
BCB studying a revolving-credit ban inside PIX Parcelado	rules expected ~Nov 2026	Installment credit on the rails BACEN itself owns — the next layer, exactly as the pattern predicts
Galípolo macroprudential package	announced 2026	Cards and non-payroll personal loans — i.e. the unsecured consumer pool
Both presidential candidates on household-debt relief	election 10/04 · 10/25	Consumer credit pricing, from the political rather than the regulatory direction

The mechanism to carry forward: ~⅓ of NU's Q2 risk-adjusted NIM beat came from Desenrola and is ~80% spent. A regulatory action that resolves bad debt is a one-off earnings tailwind followed by a permanently smaller pool — read it as a level shift, never a run rate.

History

• 2026-07-31 — written from the NU baseline analysis.
• 2026-09-28 — refreshed on the NU re-analysis. Macro block updated (Selic 14.25 → 13.75, five cuts; ceiling-breach odds 79% → 90%). The "election largely orthogonal" reading superseded, not edited — kept in place with its correction, because the error was specifically about confusing the consensus trade with the consensus policy, and that is a reusable distinction. Four new instances added; confidence medium → high, since the note's predicted shape is what actually arrived.