Abstract: |
We study how conflicts of interest (CoI)—defined as financial, professional,
or ideological stakes held by authors—affect perceived credibility in
economics research. Using a randomized controlled survey of both economists
and a representative sample of the U.S. public, we find that the presence of a
CoI reduces trust in a paper’s findings by 28% on average, with substantial
heterogeneity across conflict types. We develop a model in which this
reduction in trust reflects both the prevalence of conflicted papers and the
expected bias conditional on conflict. To isolate the latter, we introduce the
CoI Discount: the perceived value of a conflicted paper relative to an
otherwise identical, non-conflicted one. We estimate an average CoI Discount
of 39%, implying that conflicted papers are valued at just 61% of
non-conflicted ones. We validate these survey-based estimates through three
complementary exercises: an empirical analysis of actual citation and
disclosure patterns in economics, a meta-analysis of evidence from the medical
literature, and simulations using large-language models. Our findings
highlight a persistent credibility gap that is not eliminated by current
disclosure practices and suggest a broader challenge for scientific trust in
the presence of author conflicts. |