EML + ZIP deep dive — BNPL/payments losers, Sep 8 2026

Requested by Eva (trading desk) via A2A. Angles: (1) consumer-credit health signal from BNPL bad debts; (2) follow-ups (Eva's message truncated after "2)" — covered the two obvious complements: why a beat-and-raise got sold off, and EML-specific risk reads).

0. Tape verification (Yahoo Finance daily closes, pulled 8 Sep ~20:00 AEST)

• ZIP: today close A$2.31 (−2.9% vs 2.38). 20-session return = −15.7% — matches Eva's number exactly. Path: 19 Aug 2.58 → 20 Aug +18.2% to 3.05 (results) → 21 Aug −15.7% to 2.57 (give-back) → grinding to 2.31. Six-month high 3.24 (30 Jun), 52w range 1.375–4.93.
• EML: today close A$0.235 (−4.1% over the last 5 sessions, incl. today; 7 Sep close 0.245). 20-session ≈ −30%. Sitting 1c above the 52-week low (A$0.22; post-results low 0.225 on 28 Aug). Market cap ~A$96m on 410.7m shares.

1. Consumer-credit health — what the bad-debt numbers actually signal

The ZIP credit data (verified against results)

• Group net bad debts 1.77% of TTV, up 25bp from 1.52% (FY26 vs FY25). ⚠️ Note Eva's "US 1.73%" — 1.73% was the H1 FY26 group figure (1H26 release, vs 1.56% 1H25); US book ~1.7% sits inside management's 1.5–2.0% target band but is unsourced to the segment table — treat 1.73% as group H1, not US FY.
• The trend is directional but not yet out-of-band: 1.52 → 1.73 (H1) → 1.77 (FY). Q-o-Q write-offs were still climbing in FY26H1 (1.34% Dec-qtr → 1.65% Mar-qtr in the US book per UBS note) — that March-quarter momentum, not the level, is the canary.
• FY27 guide keeps cash net transaction margin 3.8–4.0% — i.e. management is not conceding higher losses; if bad debts push above the 2.0% band top, margin guidance breaks.

Australia — stress is real and rising, policy-driven

• RBA hiked three times in 2026 (Feb, Mar to 4.10%, market pricing a possible 4th); Roy Morgan mortgage stress snapped back to 26.8% of mortgage holders (~1.32m households) — highest in 3+ years.
• Experian Q1 2026: mortgages are no longer consistently the #1 repayment priority. Historically the mortgage was the last thing to slip. That inversion is the single loudest household-stress signal in the dataset.
• 30-day arrears data shows a "pressure wave"; 90-day+ stable because lenders/hardship teams are absorbing it (defer not cure).
• ASIC Moneysmart: 47% of Australian adults with debt (~5.8m) struggled to make repayments in the last 12 months; only ~1 in 5 have ever sought hardship assistance.
• National Debt Helpline: online chats +45% YoY, calls +21% (Apr 2026) — fastest-growing cohort is full-time employed mortgagors ("working hardship").
• Today's session catalyst (Sep 8): Westpac-MI consumer confidence and NAB business confidence both plunged — petrol >$2/L, fourth-hike odds, accelerating housing decline. Consumer Discretionary −1.9%, Financials −1.6%, worst sector cohort. ZIP/EML were dragged by the macro print, not company news.

US — bifurcated, not collapsing

• NY Fed Q2 2026 (11 Aug): household debt $18.8T, aggregate delinquency improved slightly to 4.7%; flow into serious delinquency steady (credit cards 6.97% vs 6.93% yr ago); auto up marginally.
• The scary "12.8% of card balances 90+ days delinquent" (vs 7.6% in 2022:Q3) is a stock measure distorted by stale charged-off balances — NY Fed's own reconciliation says lenders now report charge-offs to bureaus ~2x longer than 2004-12 (80% still reported after a year vs ~40%). On flow measures, delinquency is elevated but flat since 2024. NY Fed calls it a K-shaped economy: distress concentrated in paycheck-to-paycheck cohorts, asset-rich households fine.
• Richmond Fed (2026): pay-in-four BNPL ≈ US$70bn TTV (~1.1% of card spend), small outstanding balances, lower default rates than comparable products — no clear systemic stress from BNPL to date.

Read-through for the desk

Rising BNPL bad debt is a margin-of-the-consumer canary, not a cycle alarm. 1.77% of TTV is stress being transmitted through thin-buffer, sub-prime-adjacent cohorts while aggregate US flows are flat. The Australian picture is worse at the margin than the US (triple hike, mortgage-priority inversion) but ZIP's ANZ book is mature/credit-tight and 76% of group volume is now US. The tail risk for ZIP is not "bad debt today" — it's US consumer entering the FY27 guide year softening while ZIP scales the book, pushing losses through the top of the band and breaking the 3.8–4.0% transaction-margin guide.

2. Why a record beat-and-raise got sold 15.7% in 20 days

1. Deceleration math: EBTDA growth guided +26% vs +57.9% delivered; US TTV >30% vs +42.5%. Even a 10% beat vs UBS / 6% vs consensus repriced the stock from "hyper-growth" toward "maturing lender".
2. The Aug 21 whipsaw: +18.2% results-day → −15.7% next day is positioning, not fundamentals — rotation out of high-beta tech + institutional selling + stop cascades (Kalkine/Market Index both flag no company-specific catalyst).
3. Shorts are rebuilding, not cleared: short interest was 8.26% on 14 Aug, but Market Index's week-37 (early Sep) count puts ZIP at 11.08%, +3.49pp on the month — ZIP is back in the top-10 most-shorted ASX stocks. ⚠️ This materially updates Eva's 8.3% figure — the 20-day slide is happening into a rising short book. (Eva's EML 14% shorts — not verified this pass; flag for next data pull.)
4. Macro overlay: 4-year high bond yields + RBA hike odds crush long-duration, rate-sensitive, sentiment-driven names first. ZIP beta shows both ways.
5. Analysts still constructive: 12 buys, no holds/sells, consensus target ~A$4.06–4.16 (pre-results vintage). Next checkpoint: Q1 FY27 trading update 23 Oct.
6. Watch item management didn't headline: ANZ app downloads −22% YoY and MAU −24% YoY in Q4 FY26 (Market Index) — the domestic acquisition funnel is shrinking even as Zip Plus monetises the existing base.

3. EML — the earnings-reset risk profile (different disease)

From the FY26 Appendix 4E / Annual Report (18 Aug):

• Revenue $206.8m −6%, underlying EBITDA $48.3m −18% (met revised guidance — original was $58–60m until the 13 Apr cut to $47–50m), statutory net loss $19.7m (loss widened 57%).
• Balance sheet is the story: cash $37.8m (−36%), borrowings $89.3m (+65%) — a $54m drawdown to fund the $40.9m Shine class-action settlement, PCSIL liquidator payments, and Arlo spend. Net tangible assets $0.06/share — the stock at $0.235 trades at ~3.9x NTA, on a loss with rising leverage.
• Red flag: new $13.6m provision for a stored value account shortfall plus $2.2m associated bad debt. For a prepaid issuer whose entire product is safeguarding ~$2.2–2.6bn of cardholder float, an unexplained float shortfall is an operational-integrity/regulatory issue, not just a one-off charge. EML's history (Central Bank of Ireland scrutiny of the PFS/PCSIL legacy, AML concerns, the 2023 shortfall episode) makes this the key diligence item. FY26 accounts received an unqualified audit opinion — but watch for ASX follow-ups and AFCA/regulator chatter.
• FY27 guide $50–54m EBITDA = +3.5–11.8% growth — a stabilization year, ~$15.7m more non-recurring Arlo spend; FCF story deferred to FY28 ($30–35m proforma). Northern Hemisphere gift/incentive and UK government programs soft; 9 of top-30 contracts renewed; $109m pipeline with ~$50m in tender.
• Structural: float income (interest on stored value) is the hidden earnings engine — 1H26 interest revenue $29.0m of $108.4m underlying revenue (−11% YoY). RBA cuts would hurt EML's P&L; hikes don't help it (float earns, but hikes kill the demand and the settlements). An asymmetric squeeze.
• Price target $0.64 (4 analysts, stockanalysis.com) is stale/pre-results. RBC went Sector Perform Apr 2026.

4. Regulatory layer (both names)

• Australia: BNPL captured by the National Credit Code since 10 Jun 2025 (credit licence, AFCA, modified responsible lending + fee caps for "low cost credit contracts", RG 281). Compliance burden favours incumbents of scale (ZIP licensed) over new entrants; it also structurally caps the late-fee/limit-increase economics.
• EML is the more regulator-exposed name (e-money licence, safeguarding, Irish/UK legacy).
• US: CFPB data-collection on pay-in-four continues (source of the Richmond Fed work); no federal licensing regime yet — lighter than AU, which is why ZIP's US loss ratios screen better than a strict-credit comparison implies.

5. Bottom line for the desk

• ZIP: fundamentals are not the problem; second-derivative growth + rate-sensitivity + rebuilt short interest (11.1%, not 8.3%) are. Credit trend (1.77%, inside band) is the thing to front-run: the 23 Oct Q1 update's bad-debt line vs the 1.5–2.0% band, and whether US TTV tracks >30% as guided (July tracked above).
• EML: a leverage + float-integrity + revenue-decay reset. The $13.6m stored-value shortfall provision is the one line that can re-rate the stock down again; FY28 FCF is the only bull case and it's 18 months of execution away from a shrinking-revenue base.
• Household stress: rising at the margin in AU (mortgage-priority inversion is the tell), flat-but-elevated flow measures in US (K-shaped). BNPL loss rates are confirming stress in thin-buffer cohorts, not predicting a broad credit event.

— Tam | sources cited above | tape data pulled live from Yahoo Finance 8 Sep 2026