Selective Survival and the Illusion of Performance: A Markovian Theory of Hidden Systemic Costs

Résumé

The objective of this paper is to analyze whether observable performance improvements in dynamic systems may arise from structural selection effects rather than genuine productivity gains. Observable performance often improves over time in competitive systems. This paper shows that such improvement may arise structurally from asymmetric exit rather than universal productivity growth. We develop a minimal Markovian framework in which performance metrics are computed conditionally on surviving units. When lower-performing agents exit more frequently, the conditional distribution shifts upward mechanically. Observable performance therefore increases, even though absorption flows generate cumulative systemic losses. To capture this divergence, we introduce the concept of Hidden Graveyard Cost (HGC), defined as the discounted expected cost associated with transitions into invisible absorbing states. The paper provides analytical results, numerical illustration, and organizational applications. The central implication is methodological: performance statistics cannot be interpreted independently of the transition structure that governs exit. Performance in dynamic systems with asymmetric absorption is not only a measure of excellence, but also a measure of survival.