Technology Adoption and the Design of Incentives

Département: Finance

Intervenant: Robert Marquez (UC Davis) 

Salle: T022

Abstract

We develop a dynamic principal–agent model in which multiple agents use a technology
whose contribution to performance is uncertain. Because realized output is both
the basis for compensation and a signal about technology quality, agents anticipate
that strong early performance will make the technology appear more productive and
can reduce their future compensation. Agents therefore shade effort, while the principal
responds with stronger first-period incentives. This creates an endogenous cost of
adoption, so that a technology can be free to acquire yet costly to adopt. Broader deployment
mitigates the distortion by improving learning about the common technology
and reducing the influence of any one agent on the posterior, which raises per-agent
profitability and lowers the adoption threshold. Contractual commitment eliminates
effort shading but sacrifices the ability to adapt compensation to what is learned. We
endogenize technology architecture and show that cross-agent learning can make partial
standardization optimal even when personalized technology is more productive in
expectation. Our analysis has implications for settings where compensation is tied to
performance, such as sales, credit origination and risk assessment, or procurement, to
name a few.