A 50% Canadian car tariff sends most of its bill to Tokyo

Fifty per cent is the headline rate; the actual bill is smaller and murkier. The April 2025 order taxed only the non-US content of a Canadian-built vehicle ^[3], so the cash a CR-V owes at the bridge is half of whatever share of it was not made in America ^[20], and this source never states that share ^[20]. Until someone publishes the ratio, the "thousands of dollars on the sticker" figure ^[8] has no stated basis behind it. It also understates the parts problem, since a single engine block can cross between Michigan and Ontario three or four times before assembly ^[11], and a levy applied at each crossing compounds in ways a single percentage does not capture ^[11].

What is arithmetically clean is the increment: moving from 25 to 50 bolts on a second levy exactly the size of the first ^[15], and by the start date the original Section 232 duty will have been running about 21 months ^[16], long enough for Ontario's cost base to have absorbed one round, short enough that nobody has yet moved a line.

Then the ordering, which is the part I would trade on. The announcement, dated August 24 in a source whose other dates sit in 2026, falls 130 days before the rate begins ^[17], while Carney's retaliation lands 15 days after the announcement ^[18]. A threat dated four months out against a response dated next week looks like a bargaining chip timed to extract concessions before the rate ever takes effect.

Honda's 24% against Toyota's 17% ^[7] makes Honda's exposure roughly 1.4 times its rival's ^[19], and neither figure yields to pricing alone. The two of them build more than three-quarters of Canada's vehicles ^[2], which is why the source's blunt reading, a tariff aimed at Ottawa and paid in Tokyo ^[9], holds. The menu is absorb, pass through, or relocate ^[8]. Relocation costs years and billions ^[8] to produce a plant that adds no unit of capacity anywhere, because the Ontario line it replaces already works, and that is capital which buys tariff avoidance and nothing else.

Separate the branches. The date slips or gets traded away in the autumn, in which case the correct 2026 capex decision was to do nothing. It holds and the first year is eaten in margin and on the window sticker ^[8], because a quarter or more of North American production sits behind the wall ^[21] with no time to move it. Or it holds and migration starts, and Canada learns that 75% of its auto output belongs to two firms whose home government was never the counterparty ^[2].

This is probably wrong, but I would price the first branch highest, partly because the publisher's own headline calls the 50% a proposal while its text calls it announced ^[13]. What would falsify me is a Toyota or Honda US investment naming the CR-V or RAV4 lines before January ^[8], because capex converts an option into a sunk cost, and firms do not sink cost against threats they expect to expire.