IT Outsourcing Contracts and Performance Measurement

Abstract

Companies that outsource information technology (IT) services usually focus on achieving multiple objectives. Correspondingly, outsourcing contracts typically specify a variety of metrics to measure and reward (or penalize) vendor performance. The specific types of performance metrics included in a contract strongly affect its incentive content and ultimately its outcome. One specific challenge is the measurement of performance when an outsourcing arrangement has a mix of objectives, some that are highly measurable and others that are not. Recent advances in contract theory suggest that the design of incentives for a given objective is affected by the characteristics of other objectives. However, there is little empirical work that demonstrates how relevant these “multitask” concerns are in real-world contracts. We apply contract theory to examine how objectives and incentives are related in IT outsourcing contracts that include multiple objectives with varying measurement costs. In our context, contracts generally share the objective of reducing IT costs but vary in the importance of increasing IT quality. We establish empirical results about performance measurement in IT outsourcing contracts that are consistent with recent theoretical propositions. We find that the use of strong direct incentives for a given measurable objective is negatively correlated with the presence of less measurable objectives in the contract. We show that outsourcing contracts that emphasize goals with high measurement costs employ more performance metrics than initiatives whose objectives have a lower measurement cost profile. Surprisingly, as the number of performance metrics increases, satisfactory outcomes decrease, which we explain within a multitask theory framework. Overall, our results provide empirical support for multitask principal-agent theory and important guidance in designing outsourcing contracts for complex IT services.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 92 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.