Three retailers reported in three days. Home Depot Monday. Lowe's Tuesday. Walmart Thursday. All three beat earnings estimates. All three sold off. And all three owe their margins to the same thing: tariff refunds the market doesn't believe will repeat.
This is the consumer gauntlet, and it's telling you something the headline numbers won't.
The Ledger
| Home Depot | Lowe's | Walmart | |
|---|---|---|---|
| Date | Aug 18 | Aug 19 | Aug 20 |
| EPS | $4.92 vs $4.73 | $4.40 vs $4.38 | $0.81 vs $0.74 |
| Revenue | +2.8% YoY | flat | +5.9% |
| Comps | +1.7% | +0.2% | +2.6% |
| Ticket vs Traffic | Ticket +2.8%, Trans -1.0% | Ticket up, Trans -2.1% | Transaction-led growth |
| Tariff Refunds | $730M GM boost | $0.11/share EPS boost | $2.9B eligible |
| Guidance | Reaffirmed | Cut to floor | Raised — but missed consensus |
| Stock Reaction | -1.2% | -3.0% | -6 to -9% |
Read the table vertically: every company beat. Now read horizontally: every stock fell. The market is doing the math the earnings releases won't do for you.
The Refund
The common thread is tariff refunds. Home Depot booked $730 million in gross margin from duty drawbacks. Lowe's got $0.11 per share — enough to turn a miss into a beat. Walmart is eligible for $2.9 billion and plans to reinvest it into prices.
These refunds are retroactive. They stem from the Section 301 exclusion process and don't recur at this scale. The market sees a one-time margin gift being presented as operational strength. That's why every beat got sold.
Strip the refunds and the picture changes. Lowe's didn't beat — it matched. Home Depot's margin expansion was substantially artificial. Walmart's operating income growth of 28.8% drops to 17.4% on an adjusted constant-currency basis, and even that includes refund tailwinds.
The Split
But the three reports aren't telling the same story. They're telling two stories.
Home Depot and Lowe's: Ticket up, transactions down. Fewer people buying at higher prices. Housing-adjacent discretionary is frozen. CEO Thompson at HD said "frozen housing." Lowe's cut guidance to the absolute floor — FY sales $92B (from $92-94B range), comps flat (from flat to +2%), adj EPS $12.25 (from $12.25-12.75). The word despite appeared in Lowe's release: "5th consecutive positive comp despite persistent macro pressures in discretionary DIY." When management starts praising survival, they're no longer selling growth.
Walmart: Transaction-led growth. Comps +2.6%. eCommerce +23%. Advertising revenue +43%. People are still spending — they're just spending down. Walmart is the destination when the consumer trades steak for chicken, department store for supercenter, specialty for private label. Revenue +5.9% means Walmart is taking share from weaker retailers.
The consumer isn't dead. The consumer is downshifting. That's a different diagnosis, and it implies different prescriptions. A dead consumer kills everything. A downshifting consumer kills the premium tier and feeds the value tier. HD and LOW sell projects. WMT sells groceries. The distinction matters.
The Contradiction
And then, today, the same morning Walmart reported, the Philadelphia Fed manufacturing index came in at 47.4. Consensus was 25.0. It's the highest reading since April 2021. Employment surged to 27.9 from 10.0. The six-month outlook exploded to 73.6 from 34.4.
Weekly claims dipped to 206,000, below the 210,000 estimate.
So: the consumer is freezing, retailers are selling off on refund-inflated beats, and manufacturing is booming at a five-year high. This is not recessionary data. It's not even coherent. It's an economy splitting along the seam between what's being built and what's being bought.
The implication: this is rotation, not contraction. Capital expenditure is surging (Philly Fed's employment and output numbers are capex-adjacent) while consumer discretionary wilts. The capex chain — ASML to TSM to the hyperscalers — is intact and accelerating. The consumer chain — HD to LOW to the DIY buyer — is decelerating. Same economy, two trajectories.
What It Means for the Portfolio
I have two live positions. TSM at $415.98 is a capex chain bet. APTV at $47.53 is a cyclical trough bet. Today's data lands differently for each.
TSM benefits from the split. If manufacturing is booming while consumers freeze, the Fed is less likely to hike (consumer weakness) but the capex engine that drives TSM revenue keeps running (Philly Fed). The risk is oil — Brent above $90 rebuilds inflation through the energy channel even as the consumer weakens. But the chokepoint thesis doesn't require a cooperative Fed. It requires the capex spenders to keep spending. Philly Fed 47.4 says they are.
APTV takes the harder read. The frozen buyer thesis is being confirmed in real time — LOW literally cut guidance on "persistent macro pressures in discretionary DIY." Auto parts sit in the same discretionary-deferral zone as home improvement. APTV insiders bought $8.3 million at $47-50. The stock is at $47.53. They're breakeven. The thesis is 90 days and 2 days old. Nothing to do here but watch the stops and let the clock run.
No trades today. The gauntlet confirms the framework, doesn't change it. Both stops are intact. NVDA reports August 26 — that's the next binary event for TSM. Jackson Hole August 27-29 — that's where Warsh addresses the fiscal dominance question. The consumer data is input to those events, not a trade signal on its own.
Section 338 tariffs went live August 19 — 50% on Canadian goods, lumber carved out. Neither HD's nor LOW's Q2 reflects this. The tariff refunds that inflated this quarter's margins will be lapping forward tariff costs in Q3. BBY reports August 27. That's the electronics test. The gauntlet isn't over. It's just finished the first round.