Under fallback bargaining, bargainers begin by indicating their preference
rankings over alternatives. They then fall back, in lockstep, to less and l
ess preferred alternatives - starting with first choices, then adding secon
d choices, and so on - until an alternative is found on which all bargainer
s agree. This common agreement, which becomes the outcome of the procedure,
may be different if a decision rule other than unanimity is used. The outc
ome is always Pareto-optimal but need not be unique; if unanimity is used,
it is at least middling in everybody's ranking. Fallback bargaining may not
select a Condorcet alternative, or even the first choice of a majority of
bargainers. However, it does maximize bargainers' minimum "satisfaction." W
hen bargainers are allowed to indicate "impasse" in their rankings - below
which they would not descend because they prefer no agreement to any lower-
level alternative - then impasse itself may become the outcome, foreclosing
any agreement. The vulnerability of fallback bargaining to manipulation is
analyzed in terms of both best responses and Nash equilibria. Although a b
argainer can sometimes achieve a preferred outcome through an untruthful an
nouncement, the risk of a mutually worst outcome in a Chicken-type game may
well deter the bargainers from attempting to be exploitative, especially w
hen information is incomplete. Fallback bargaining seems useful as a practi
cable procedure if a set of "reasonable" alternatives can be generated. It
leapfrogs the give-and-take of conventional bargaining, which often bogs do
wn in details, by finding a suitable settlement through the simultaneous co
nsideration of all alternatives.