[论文解读] Overspend? Late? Failure? What the Data Say About IT Project Risk in the Public Sector

Alexander Budzier, Bent Flyvbjerg|arXiv (Cornell University)|Apr 16, 2013

Big Data and Business Intelligence被引用 16

一句话总结

本文分析了1,355个公共部门IT项目，以评估成本和进度表现中的风险，发现尽管规范性预期接近99%的合规性，仍有18%的项目成本超支超过25%。研究识别出项目持续时间和项目类型为关键风险驱动因素，并提出了四种基于证据的解决方案：基准比较、去偏见决策、降低复杂性，以及培养‘总建筑师’以改善项目成果。

ABSTRACT

Implementing large-scale information and communication technology (IT) projects carries large risks and easily might disrupt operations, waste taxpayers' money, and create negative publicity. Because of the high risks it is important that government leaders manage the attendant risks. We analysed a sample of 1,355 public sector IT projects. The sample included large-scale projects, on average the actual expenditure was $130 million and the average duration was 35 months. Our findings showed that the typical project had no cost overruns and took on average 24% longer than initially expected. However, comparing the risk distribution with the normative model of a thin-tailed distribution, projects' actual costs should fall within -30% and +25% of the budget in nearly 99 out of 100 projects. The data showed, however, that a staggering 18% of all projects are outliers with cost overruns >25%. Tests showed that the risk of outliers is even higher for standard software (24%) as well as in certain project types, e.g., data management (41%), office management (23%), eGovernment (21%) and management information systems (20%). Analysis showed also that projects duration adds risk: every additional year of project duration increases the average cost risk by 4.2 percentage points. Lastly, we suggest four solutions that public sector organization can take: (1) benchmark your organization to know where you are, (2) de-bias your IT project decision-making, (3) reduce the complexities of your IT projects, and (4) develop Masterbuilders to learn from the best in the field.

研究动机与目标

• 评估大规模公共部门IT项目在成本超支和进度延迟方面的实际风险特征。
• 识别显著增加成本超支和进度延迟可能性的因素。
• 将实际风险分布与规范性轻尾统计模型进行比较，揭示系统性偏差。
• 为公共部门组织制定可操作的、基于数据的建议，以降低IT项目失败率。

提出的方法

• 分析包含1,355个公共部门IT项目的数据库，平均支出为1.3亿美元，平均持续时间为35个月。
• 应用统计建模方法，将实际成本和进度结果与规范性轻尾分布模型进行对比。
• 按项目类型（如数据管理、电子政府）对项目进行分类，以评估不同领域间的风险差异。
• 对持续时间作为风险因素进行定量分析，测量项目每延长一年所增加的成本风险。
• 通过组织绩效基准比较，识别表现优异的机构。
• 基于实证发现和行为洞察，开发四种实用的风险缓解策略。

实验结果

研究问题

• RQ1 与规范性预期相比，大规模公共部门IT项目的实际成本超支和进度延迟分布如何？
• RQ2 哪些项目类型在成本超支和进度延迟方面风险最高？
• RQ3 项目持续时间与成本风险增加之间存在何种相关性？
• RQ4 标准软件项目与定制开发项目在风险特征上有多大差异？
• RQ5 哪些基于证据的组织实践可降低公共部门IT项目失败的可能性？

主要发现

• 18%的公共部门IT项目为异常值，其成本超支超过25%，远高于规范性轻尾分布模型所预期的1%。
• 平均而言，项目比最初计划延长24%，尽管典型项目并未出现成本超支。
• 数据管理项目风险最高，41%的项目成本超支超过25%，其次是办公管理（23%）和电子政府（21%）。
• 项目每延长一年，平均成本风险增加4.2个百分点。
• 标准软件项目的成本超支异常值率为24%，表明其风险高于预期。
• 本研究识别出四种实用解决方案：基准比较、去偏见决策、降低项目复杂性，以及培养‘总建筑师’以制度化最佳实践。