Securely implementable social choice rules with partially honest agents

We define secure implementation with partially honest agents in a social choice model and we show that strategy-proofness is a necessary and sufficient condition for it. This result offers a behavioral foundation for rectangularity ; and it remains valid even with only one partially honest agent. We apply the concept to a single-crossing voting environment, and we prove that it characterizes the family of augmented representative voter schemes. JEL Codes: C72, D03, D71, D82. Key-words: Secure implementation; partial honesty; strategy-proofness; singlecrossing preferences; representative (median) voter.