The term's etymology is unclear. The proposed derivations from the isolated viewpoints of individual language families, do not do justice to the large area of distribution in German, Romance languages and Slovenian.
From an Italianist point of view, DELI suggests roa. baita, bait < Old High German wahta – but without considering the slv. bajta 'bad house' or the gsw. (Alemannic) Beiz, bar. Boazn, Beisl 'pub' (unfortunately, this widespread type is missing in the SDS, the Idiotikon and the BSA). The mentioned Germanic forms with ts, s can thus not be explained.
From a Germanistic point of view, Kluge (2011, 106 under Beiz(e) and Beisel) derives the gsw. (Alemannic) and bar. forms from Yiddish bajis 'house' < hbo. bajit 'house', which does not correspond to the roa. t (cf. EWD I, 203). Direct mediation from Hebrew (meaning without Yiddish mediation) is historically implausible given the large area and the connection to everyday life in Alpine farming. The semantics of the Romance forms ('hut, Alpine hut, stable', etc.) as well as the Slovenian records ('bad house') do not provide any justification for the assumption of a large-scale distribution of an originally adstratal borrowing from the Friulian-Slovenian contact area. A substratal borrowing from the pre-Slavic and pre-Germanic Old Romance of the Eastern Alps seems much more plausible. Ultimately, it appears to be a pre-Roman Alpine word.