DEA evaluation of a Y2K software retrofit program

When the final tally was in, the year 2000 (Y2K) software compliance issue had cost over a hundred billion dollars worldwide. The fact that essentially everyone was busy tackling the same problem provided a unique opportunity to use data envelopment analysis (DEA) to measure software team efficiency and productivity. The data set analyzed in this paper contained about 70 programs from a large Canadian bank. While there were about a dozen different programming languages and a number of hardware platforms involved, the work was very similar in nature as they were all fixing the Y2K "bug". We examined both team productivity and programmer efficiency when maintaining code where the maintenance objective was the same in all cases. DEA models were developed to measure software project efficiency focusing on the factors that affect software productivity, and we discuss how these findings could be applied to other projects. Suggestions are offered on how DEA could be combined with the bank's own ratio-based rating system to improve their software production metrics. Finally, potential management uses of these DEA results are presented.