while player 2 prefers agenda 1 if
2λ1λ2α1r1r2 + r1λ2 (r1 + r2)(1 - α2) - (r1 + r2)2 + α2r1(r1 - r2)
> 0 (44)
(r1 + r2)2
In this case if r1 >r2 and λ1λ2 is sufficiently large players have different preferences
over agendas.
7) Player 1 obtains the interior demand x1 in agenda 1 and xe1 in agenda 2. Then,
at the limit for ∆ → 0, player 1 prefers agenda 1 if
r2[2λ1λ2α2r1 + λ1(r1 + r2)(l - aɪ) + aɪ(rɪ - r2)] > 0
(rɪ + r2)2
(45)
>0
(46)
while player 2 prefers agenda 1 if
-λM2((∏ + r2)2 + α2rι(r2 - rɪ)) + λιrι(rι + r2)(l - α2) + 2αιrιr2
(rɪ + r2)2λι
Then, for instance if rɪ <r2 player 1 always prefers agenda 1, while player 2 has the
same preference only if λ1λ2 is sufficiently small. ■
Proposition 3 establishes an intuitive result on the efficiency of sequential proce-
dures: when there is an important issue, this should be discussed first. However, this
result is not obtained in frameworks similar to ours (such as, Busch and Horstmann,
1997, 1999, Inderst, 2000 In and Serrano, 2002, 2003). In addition, proposition 3
shows that other incentives can be dominant. For instance, regardless of whether an
issue is the most important or not, a player may prefer either an agenda where he gets
a positive share in the initial agreement (see, e.g., case 4 in proof of Proposition 3) or
24