Manufacturing Economics, Cost Curves and Operating Leverage

A manufacturer sells a physical unit and pays for it three times: once for the materials in it, once for the conversion that turns those materials into a product (labour, energy, machine time, scrap and rework), and once, in advance and whether or not the unit is ever sold, for the plant, tooling, engineering and people that make conversion possible. The third payment is what makes manufacturing different from trading. Fixed cost turns volume into margin on the way up and into losses on the way down, so a manufacturer's profit moves much more than its revenue. That is operating leverage. Cost also falls as a product is made more often, because workers, engineers and suppliers learn. That is the learning curve, first written down for aircraft by T. P. Wright in 1936 (from memory, verify) and visible in the Model T, solar modules and lithium-ion batteries. The United States still makes a lot: manufacturing value added was ~$2.95 trillion at an annual rate in Q3 2025, roughly 9.5% of GDP in current dollars (BEA via FRED; the share from a secondary summary, verify). But it employs fewer people each cycle. Revised Bureau of Labor Statistics data show ~103,000 manufacturing jobs lost between January 2025 and January 2026 (reported), and the Federal Reserve's G.17 release put total-industry capacity utilisation in the mid-75% range through mid-2026 (release of 18 Aug 2026), below its long-run average (from memory, ~79–80%, verify). Two current cases show both mechanisms. Boeing carried ~$13.9bn of 787 deferred production costs at the end of 2025: the excess of early units' cost over their average cost, booked as an asset to be recovered from later, cheaper units (10-K and Boeing's investor table). Ford took ~$19.5bn of charges in December 2025 when the electric-vehicle volume its plants were sized for did not arrive (Shared facts). This section explains how a manufacturer's cost stack is built, why fixed cost and scale dominate it, how learning curves work and fail, how mix and pricing power decide margins, and how to read the US manufacturing base in numbers. The companion worked example computes a learning curve and the cost of the first 1,000 units of a new product.