This paper is an attempt to eliminate LF subjacency effects and argues in particular that LF wh-island effects must be explained in terms of economy considerations rather than subjacency in LF or at S-structure. More specifically, I claim that what superficially looks like LF wh-island effects can be considered violations of the universal economy principle: It is not allowed to change relations that have been established at S-structure by further LF operations, unless this change is necessary. The paper first examines an LF subjacency analysis (Nishigauchi 1986, 1990) and an S-structute null operator movement analysis (Watanabe 1992). It next discusses the role of Q-morphemes in Japanese and suggests that wh-elements in this language must be "licensed" by a Q-morpheme at S-structure. It then extends Epstein's (1992) analysis of certain wh-elements in English with limited scope-taking ability to the syntactic behavior of Japanese wh-in-situ.