In day labor markets, workers are particularly vulnerable to wage theft. This paper introduces a principal-agent model to analyze the conditions required to mitigate wage theft through fines and establishes the necessary and sufficient conditions to reduce theft. We find that the fines necessary to eliminate theft are significantly larger than those imposed by current labor laws, making wage theft likely to persist under penalty-based methods alone. Through numerical analysis, we show how wage theft disproportionately affects workers with lower reservation utilities and observe that workers with similar reservation utilities experience comparable impacts, regardless of their skill levels. To address the limitations of penalty-based approaches, we extend the model to a dynamic game incorporating worker awareness. We prove that wage theft can be fully eliminated if workers accurately predict theft using historical data and employers follow optimal fixed wage strategy. Additionally, sharing wage theft information becomes an effective long-term solution when employers use any given fixed wage strategies, emphasizing the importance of raising worker awareness through various channels.

翻译：在日结劳动力市场中，劳动者尤其容易遭受工资盗窃。本文引入委托代理模型，分析通过罚款机制抑制工资盗窃所需的条件，并建立了减少盗窃行为的必要与充分条件。研究发现，完全消除工资盗窃所需的罚款金额显著高于现行劳动法规定的处罚标准，这意味着仅依靠惩罚性措施很可能无法根治工资盗窃问题。通过数值分析，我们揭示了工资盗窃对保留效用较低的劳动者造成不成比例的影响，并观察到具有相似保留效用的劳动者——无论其技能水平如何——都会受到可比程度的影响。为突破惩罚性措施的局限，我们将模型扩展为包含劳动者认知的动态博弈模型。研究证明，若劳动者能基于历史数据准确预测工资盗窃行为，且雇主采用最优固定工资策略，则工资盗窃可被完全消除。此外，当雇主采用任意给定的固定工资策略时，共享工资盗窃信息将成为有效的长期解决方案，这凸显了通过多元渠道提升劳动者认知的重要性。