A Field Study of Scale Economies in Software Maintenance
通过分析一家大型金融服务机构的软件增强项目，使用数据包络分析发现软件维护中存在显著的规模经济，将小修改项目合并为大型计划发布可降低36%的成本，并探讨了管理者未利用规模经济的原因。
Software maintenance is a major concern for organizations. Productivity gains in software maintenance can enable redeployment of Information Systems resources to other activities. Thus, it is important to understand how software maintenance productivity can be improved. In this study, we investigate the relationship between project size and software maintenance productivity. We explore scale economies in software maintenance by examining a number of software enhancement projects at a large financial services organization. We use Data Envelopment Analysis (DEA) to estimate the functional relationship between maintenance inputs and outputs and employ DEA-based statistical tests to evaluate returns to scale for the projects. Our results indicate the presence of significant scale economies in software maintenance, and are robust to a number of sensitivity checks. For our sample of projects, there is the potential to reduce software maintenance costs 36% by batching smaller modification projects into larger planned releases. We conclude by rationalizing why the software managers at our research site do not take advantage of scale economies in software maintenance. Our analysis considers the opportunity costs of delaying projects to batch them into larger size projects as a potential explanation for the managers' behavior.