The relationship between competitive restraints, R&D activity, productivity, and competition

Abstract

In this paper we empirically examine the effects of four types of competitive restraints
(non-disclosure, non-compete, non-solicitation of clients, and non-solicitation of
co-workers) on R&D activity, labour productivity, and industry competition. We use
data from an experimental survey by the ABS, linked with administrative tax and business
data. Because the survey measures restraint usage in a single year (2023), we impute
usage in other periods; this is the central limitation of our design and restricts us to crosssectional
variation. To gauge robustness to omitted-variable bias, we apply the bounding
method of Diegert et al. (2026), which is well suited to a setting with several correlated,
potentially endogenous restraints. Our most robust findings concern industry competition:
the use of non-disclosure agreements is estimated to raise market concentration,
while non-competes and non-solicitation of clients lower it. Effects on R&D and productivity
are smaller and more sensitive to selection assumptions. Overall, results are mixed
in terms of significance, direction, and robustness.