Statistical analysis of US government supply chain contract breaches

Approximately half of US government supply chain procurement contracts failed to meet all objectives. In 2018 the US government spent almost US$2bn on procurement contracts. Failed procurement projects represent a lot of public tax-payer dollars which are wasted each year. The US government does not necessarily intend for its government procurement contracts to fail, but when they are awarded on a fixed firm price basis, the entire project management is transferred to civilian contractors for a specific cost. When something goes wrong, contractors are forced to adjust the other parameters such as time, scope and/or quality. It appears many government supply chain procurement contracts are failing despite substantial analysis of project data by academic scholars and practitioners. In the literature, most researchers could not statistically identify the underlying causes of government procurement contract failures. In fact, the sample sizes surpassed 10,000 and in one case almost 60,000 cases were examined without generating a conclusive statistical result. It was clear a new study like ours was warranted. Thus, in this study, we collected a large sample of procurement project data from US government contractors. We tested several hypotheses in an attempt to identify which conditions affected government procurement contract breaches. We applied robust parametric statistical techniques, namely logistic regression, to test the hypotheses. We developed a statistically significant model which adequately explained US government supply chain procurement contract performance. This paper is in a scholarly format using statistical techniques which are primarily aimed at an academic audience rather than practitioners. We followed the well-known American Psychology Association conventions for writing style which resulted in this paper having a strong academic tone. Nonetheless, the results ought to be of interest to scholars and practitioners alike. Scholars will be able to replicate our research design with new sample sizes of at least 1,000 in other states or countries and cite our findings. Practitioners and government procurement programme managers should find it relevant that our statistical model correctly classified 67 per cent of US government procurement contract performance, with an 11 per cent effect size using only two predictors (project manager certification and ISO quality registered). We found that if a project manager was professionally certified, this increased the likelihood of contract success by 2.3 times. Decision makers could use this model for procurement policy revision or to inform selection criteria in government contract awards.