This paper investigates a hitherto less studied type of why-stripping, which we dub substitutive why-stripping (SWS). In regular why-stripping (RWS), (e.g., A: Taro-wa Hanako-ni gihuto-o katta yo 'Taro bought a gift for Hanako.' B: Naze Hanako-(ni)? 'Why (for) Hanako?'), the non-wh-remnant is a repetition of the correlate (Hanako-(ni)). In contrast, SWS (e.g., A: Taro-wa Hanako-ni gihuto-o katta yo 'Taro bought a gift for Hanako.' B: Naze imooto-(ni)? 'Why (for) (his) younger sister?') involves a paraphrase of the correlate as the non-wh-remnant. This paper shows that SWS has some significant implications for the nature of ellipsis that are not evident from studying RWS alone. Specifically, this paper addresses three controversial issues concerning the nature of ellipsis. First, this paper shows that SWS presents novel support for the view that the ellipsis involves a silent structure parallel to the antecedent clause (the structural approach) (Section 2). Second, this paper argues that SWS presents a piece of evidence that structural identity is required for licensing ellipsis (Section 3). Third, this paper demonstrates that the non-wh-remnant of SWS can stay inside the ellipsis site, supporting the in-situ approach to ellipsis (Section 4). It is further argued that SWS, unlike RWS, may involve an appositive structure (Section 5).