Single-sourcing versus multisourcing: The roles of output verifiability on task modularity

This paper compares two modes for outsourcing the development of information services projects: singlesourcing (where one vendor handles all outsourced activities) and multisourcing (where multiple vendors handle those activities). We assess the relative efficacy of these two outsourcing modes by identifying the effects of three factors: task modularity, the extent of alignment between a (verifiable) performance metric and project revenue, and the extent to which project revenue is itself verifiable. We find that if tasks are modular then multisourcing strictly dominates single-sourcing-provided the verifiable performance metric and project revenue are not completely aligned. Yet if tasks are integrated, then the choice of sourcing mode is more nuanced: the best choice depends on trade-offs among the alignment between performance metric and project revenue, the verifiability of project revenue, and moral hazard. If the verifiable performance metric and project revenue are perfectly aligned, or if project revenue is completely verifiable, then firms prefer single-sourcing because it entails less moral hazard than does multisourcing. Comparative statistics for the effects of task interdependence costs and vendors' risk aversion reveal that multisourcing (single-sourcing) should be preferred when there are interdependence costs (/when vendors are strongly risk averse).