Semi-Presidential Systems: Dual Executive And Mixed Authority Patterns

Introduction

Since it was first proposed by Maurice Duverger, the notion of a democratic regime type that can be characterized as semi-presidential has become well established. Given the substantial growth in the past quarter century of a distinct subfield devoted to the analysis of executive-legislative relations,¹ and the proliferation of new constitutions after the fall of the Communist bloc, it is worthwhile to take stock of the theory and practice of Duverger's 'new political-system model.' Many new constitutions -- not only in the region extending from Central Europe to Central Asia, but also in Africa -- could be characterized as semi-presidential, yet there are important distinctions among them that need to be understood precisely as further analytic work proceeds. My goal in these pages is to return to Duverger's original conception and ground it more firmly in theoretical traditions in comparative constitutional design stretching back to theorists such as James Madison and Walter Bagehot, and then to turn to the contemporary empirical record and see how these traditions are reflected in the practice of semi-presidential systems.

Adapting from Duverger's (1980) original and influential definition, semi-presidentialism may be defined by three features:

A president who is popularly elected;

The president has considerable constitutional authority;

There exists also a prime minister and cabinet, subject to the confidence of the assembly majority.

These features define a dual executive (Blondel, 1984), in that the elected president is not merely a head of state who lacks political authority,² but also is not clearly the 'chief' executive, because of the existence of a prime minister who may not be strictly a subordinate of the president. The precise relationship of the president to the prime minister (and cabinet), and of the latter to the assembly vary widely across regimes that fit the basic Duvergerian conception of semi-presidentialism, and these formal institutional variations are likely to have significant consequences for the behavioral performance of different systems.

By virtue of its being called semi -presidential, the regime type in question is clearly identified as a hybrid that is neither presidential nor parliamentary. If we consider 'presidential' and 'parliamentary' to be terms denoting pure types, from both of which semi-presidentialism draws certain characteristics, we need clear benchmarks as to the features of...