Donald Horowitz, Democratic South Africa? Constitutional Engineering in a Divided Society (1991).

Main Argument: Horowitz draws on a wide comparative knowledge of democratic transitions in developing countries to demonstrate that constitutional design does matter, and that it ought to be correct at the outset rather than left to the exigencies of constitutional negotiation. Incentives both during and after transition must appeal to the self-interest of parties and politicians; therefore institutions and processes have to be designed to reward moderation and undercut polarisation, and, if done successfully, politicians will respond moderately in their own self-interest. Horowitz argues that there is a type of constitutional engineering that can render the realities of South Africa’s cultural divisions compatible with competitive elections based on majority rule in a unitary state.
Method: Horowitz draws on comparative knowledge of problems of democratisation experienced in other divided societies (concentrates on the ethnic, racial and ideological cleavages.)
== Notes ==
– Uses Path dependency arguments: once you start off on the wrong track it is almost impossible to undo the damage later on
Horowitz’s Design:
– A presidential (with the requirement that the presidential majority be distributed in a particular way rather than simply arithmetical), federal system of government for South Africa in which there has also been a ‘compositional change in the instruments of force’ to make them ‘coup-proof’
– Electoral system should build in incentives for moderation
– The specific electoral mechanism he favours is coalition building based on party proliferation, proportional representation, and particularly ‘vote pooling’ [167] across ethnic or racial lines as solutions and incentives to the “moderate middle” [197]
To induce vote pooling in the electoral system there has to have:
(i) a proliferation of parties;
(ii) heterogeneous constituencies; and
(iii) electoral incentives to make vote polling profitable