Progression and assessment rules are often treated as administrative details, yet they fundamentally shape who is allowed to remain in higher education, and on what terms. This article uses a calibrated agent-based model to examine how alternative progression regimes reconfigure dropout, time-to-degree, equity and students' psychological experience in a long, tightly sequenced engineering programme. Building on a leakage-aware longitudinal dataset of 1,343 students and a Kaplan-Meier survival analysis of time-to-dropout, we simulate three policy scenarios: (A) a historical "regularity + finals" regime, where students accumulate exam debt; (B) a direct-promotion regime that removes regularity and finals but requires full course completion each term; and (C) a direct-promotion regime complemented by a capacity-limited remedial "safety net" for marginal failures in bottleneck courses. The model is empirically calibrated to reproduce the observed dropout curve under Scenario A and then used to explore counterfactuals. Results show that direct promotion creates a "promotion wall": attrition becomes sharply front-loaded in the first two years, overall dropout rises, and equity gaps between low- and high-resilience students widen, even as exam debt disappears. The safety-net scenario partially dismantles this wall: it reduces dropout and equity gaps relative to pure direct promotion and yields the lowest final stress levels, at the cost of additional, targeted teaching capacity. These findings position progression rules as central objects of assessment policy rather than neutral background. The article argues that claims of improved efficiency are incomplete unless they are evaluated jointly with inclusion, equity and students' psychological wellbeing, and it illustrates how simulation-based decision support can help institutions rehearse assessment reforms before implementing them.

翻译：学业进度与考核规则常被视为行政细节，然而它们从根本上决定了谁有资格以何种条件留在高等教育体系中。本文采用基于校准的智能体模型，研究了在长周期、紧密衔接的工程教育项目中，不同进度制度如何重构辍学率、学位获取时间、公平性及学生的心理体验。基于包含1,343名学生的泄漏感知纵向数据集及针对辍学时间的Kaplan-Meier生存分析，我们模拟了三种政策情景：（A）历史性的“常规考核+期末考试”制度，学生在此过程中积累考试债务；（B）直接晋升制度，取消常规考核与期末考试，但要求每学期完成全部课程；（C）在直接晋升制度基础上，为瓶颈课程中边缘性不及格学生增设容量有限的补救性“安全网”。模型经实证校准，能复现情景A下观察到的辍学曲线，并用于探索反事实结果。研究表明，直接晋升制度形成了“晋升壁垒”：辍学现象在前两年急剧前置，总体辍学率上升，低心理韧性与高心理韧性学生间的公平差距扩大，尽管考试债务随之消失。安全网情景部分拆解了此壁垒：相较于纯直接晋升制度，它降低了辍学率与公平差距，并实现了最低的最终压力水平，但代价是需增加有针对性的教学资源。这些发现将进度规则定位为评估政策的核心对象，而非中性背景。文章指出，除非将包容性、公平性及学生心理健康纳入联合评估，否则关于效率提升的论断是不完整的，并阐释了基于仿真的决策支持如何帮助院校在实施考核改革前进行预演。