Algebraic effects and handlers are a popular language feature among programming language researchers and they also start to gain traction in more industrially used languages. This increasingly raises the need for efficient implementation techniques that eliminate the overhead incurred by the powerful abstraction. One promising approach for lexical effect handlers is to use capability-passing style. Translating lexical handlers into iterated continuation-passing style has been shown to open up large potential for optimizations, potentially removing the overhead of the added abstraction altogether. However, to avoid a runtime search for the correct handler, the translation requires knowledge about the nesting structure of handlers. We show how to infer this information for $\mathsf{Effekt}$, a language with a second-class restriction on functions. We present a translation from $\mathsf{System};\Xi$ (the core calculus of $\mathsf{Effekt}$) to $\Lambda_{\mathsf{\textbf{\textsf{cap}}}}$, a language with a region system and explicit subregion evidence, inferring the missing information. This allows us to immediately reuse the existing iterated CPS translation from $\Lambda_{\mathsf{\textbf{\textsf{cap}}}}$ and unlock the optimization potential.