Presidential vetoes and coalition government in Brazil

Presidential systems are based upon the principle of the separation of powers. There two independent and usually viewed as conflicting powers, the Executive and the Legislative. Within this system, vetoes tend to be analysed as two actor model. The legislative approves a bill and the president vetoes if its content runs against his preferences. Executive centred models maintain the two actors model: the executive initiates legislation, the legislature amends and the executive reacts vetoing. To introduce legislative consideration of the veto does not alter the nature of the game: a basic two actor sequential game. In this regard, veto games type of analysis are compatible with the Linzean basic view of presidential systems, that is to say, as a conflictual system. The Executive and the Legislative are viewed as two different actors with conflicting preferences. In Brazil, Presidents do veto bills they introduce. From all bills the president proposes and Congress approves, 15% were vetoed (partially or totally). That fact seems to vindicate the two actor conflictual view. Why would a president veto a bill herself introduced? The only explanation is: the president is vetoing what congress added to the original bill. The president is vetoing amendments. We argue that veto game is more complex than the sequential two actor conflictual model. First, one has to consider that the Executive and not the president submits bill to Congress. The distinction is subtle but clear and consequential in coalition governments. The Executive is a collective actor whose members do not hold the same political preferences. Cabinet members come from different parties and the legislation they introduce and favour may not represent the preferences of the all coalition members. On the other hand, Congress cannot be considered as a single actor either. Legislatures are populated by political parties. Besides, parties within a legislature tend to divide themselves into two main groups, those that support and those that oppose the Executive. That is to say, the heterogeneous coalition that controls executive portfolios reproduces itself within the Legislature. Thus, the veto game may be seen not as the expression of the Executive-Legislative conflict but as the expression of the coalition politics. Coalitions, by definition, are politically heterogeneous. Each party holds or has a different preference over policies. Yet, if the coalition is to produce decisions, if bills are approved, these parties have to reach an agreement. The veto game allows us to observe the hammering of this internal agreement. In this paper, we explore this interpretation using an original data set we construct using the Brazilian case. We collected all legislation introduced by the Brazilian Executive that became law from 1995 to 2014. We identify more than 10.000 alterations introduced and approved by the Congress for each of these bills and distinguish the vetoed from the non-vetoed amendments. The analysis reveals the working of the coalition and the way the parties exercise control over coalition partners.