A model of the firm's decision to adopt a payment by results system is
developed and tested with British establishment data. The model maint
ains that payment by results systems have larger set-up costs but lowe
r supervision costs than time rates, particularly for short-tenure wor
kers who are not well motivated by deferred compensation. The evidence
confirms the model's predictions by showing that payment by results s
ystems substitute for supervision and that larger establishments and t
hose with shorter-tenure workers are more likely to adopt payment by r
esults. In addition, both the presence of a payment by results system
and the new adoption of such a system are shown to exert a positive in
fluence on measures of establishment performance.