Thirty-Two Dollars an HourNarrow moat

Costco employed 341,000 people at the end of fiscal 2025 — 223,000 in the United States, 55,000 in Canada, 63,000 elsewhere — with approximately 95% working in warehouses and distribution and around 5% represented by unions¹.

The pay is the notable part. In March 2025 the starting wage rose $0.50 to at least $20.00 an hour for all entry-level positions in the United States and Canada, the top of the wage scales rose $1.00, and the average American hourly rate reached approximately $32.00. The company targets at least 50% of its base being full-time and reports a retention rate of about 94% for employees with at least a year of service in the United States and Canada.

The conventional retail view is that this is a cost problem. The arithmetic says otherwise, and the arithmetic runs through the assortment. A warehouse with fewer than 4,000 items, goods displayed on pallets and inventory stored on racks above the sales floor needs a fraction of the labour hours per dollar of sales that a supermarket needs — the 10-K makes the point directly, that pallet display is "reducing labor required" and that operating hours are shorter than many other retailers'². Costco therefore employs fewer people per dollar and can pay each of them far more while still running SG&A at 9.25% of net sales³.

The retention figure is where the payback shows. Turnover in American retail routinely runs above fifty percent a year; every departure costs recruitment, training and a period of lower productivity, and in a warehouse where a handful of people move enormous volumes of goods, an experienced crew is worth a great deal more than in a store where the work is shelf-facing.

It is testable in one line: SG&A as a percentage of net sales excluding gasoline effects, 9.17% in fiscal 2025. If Costco can absorb a wage rise without that ratio deteriorating, the productivity argument is real.