In his syntactical grammar (p. 334), Daniel B. Wallace has a pretty good discussion on Eph. 2:8 and its utilization of τοῦτο. The question: what is the antecedent of τοῦτο in Eph. 2:8? Wallace lists four standard interpretations: (1) χάριτί is the antecedent, (2) πίστεως could be the antecedent of τοῦτο, (3) the concept of "grace by faith" is the antecedent of the demonstrative pronoun τοῦτο, (4) καὶ τοῦτο has an adverbial force with no antecedent.
Wallace then discusses the shortcomings of options (1) and (2) because τοῦτο is neuter and χάριτί and πίστεως are feminine. He then refers to an article by R.H. Countess ("Thank God for the Genitive!") in which Countess argues that gender shift is common in Attic Greek, so there is no problem in construing a neuter demonstrative back to "a noun of a different gender" (Wallace 334). Needless to say, Wallace is critical of this approach because he thinks that Countess' examples from Attic literature are debatable and he thinks that they are best understood as referring "to a concept [rather] than a noun" (Wallace 334).
Wallace thus concludes that while "on rare occasions there is a gender shift between antecedent and pronoun, the pronoun is always caught between two nouns of a different gender," Eph. 2:8 is not an example of this phenomenon, however (Wallace 334-335). Wallace, therefore, goes on to discount options (1) and (2) as possible antecedents for τοῦτο and instead chooses option (3). That is, he argues that the antecedent of τοῦτο in Eph. 2:8 is "the concept of grace-by-faith salvation" (335). But I am skeptical about Wallace's proposal by virtue of the fact that he opts for a "concept" as the antecedent over against a substantive such as χάριτί or πίστεως. But there
are some reasons why I'm inclined to question Wallace's construal of τοῦτο as well (based on Attic and NT literature). Yet one thorough commentary written by Harold Hoehner also takes the position that the demonstrative pronoun refers to the entire preceding section, as opposed to just one noun or participle. He writes that it refers to "the concept of salvation by grace through faith" (Ephesians, page 343).
I conclude by saying that Wallace's decision regarding the antecedent of τοῦτο in Eph. 2:8 is based on grammatical parallelism (not attraction). The professor even writes that the issues surrounding Eph. 2:8 "are complex and cannot be solved by grammar alone. Nevertheless, syntactical considerations do tend toward" choices (3) or (4).