This study adapts Roger Myerson's game-theoretic model to examine democratic incentives through experimental design. The research tests four key predictions regarding leadership behavior:

• Corrupt leaders replaced more often in federal systems compared to centralized ones.
• In the centralized setting, subjects reached an equilibrium where responsive leadership led to reelection.
• Honest presidents receive similar reelection rates regardless of institutional structure.
• Decentralized power structures reduce presidential responsiveness. Beyond these main findings, robustness checks confirm consistent results across different experimental languages and methodological choices.