Formalization policies that create new formal employment categories mechanically reduce measured informality even absent behavioral change. I propose a decomposition separating this mechanical channel from genuine formalization by re-estimating treatment effects excluding the mechanically formalized group. Applying this to Turkey's 2021 MESEM apprenticeship reform using a triple-difference design with Household Labour Force Survey data (2019--2023), I find 36% of observed formalization is mechanical, meaning that it is driven by apprentice registration rather than behavioral change. The statistically significant informality effect is -0.595, attenuating excluding apprentices. The decomposition is portable to any registration-based formalization policy.