Where I'm more skeptical is the jump from "the data can't yet show effects" to "import limits are therefore necessary." Those are different claims.