91ɫ��Ƭ

Credibility in Economics Publications: What Z-Curve Predicts vs What Actually Replicates

How well do published economics results hold up—and can we forecast which ones will? Using the 2025 Institute for Replication (I4R) meta‑database, I document that around three‑quarters of reproductions succeed (76%), while replication is 57%-70% for recent papers. I show that methodological and subfield-related heterogeneity in replication performance is substantial. 

I then bring Z‑curve—a mixture‑model will-known in psychology replicability meta-analysis,  diagnostic of publication bias and power—into economics and compare its expected replication rates (ERR) with the actual I4R replication outcomes. 

Z‑curve’s result closely matches the actual replications in most cases. The implied overall false discovery rate is low—about 7% (with 95% Confidence Interval [2%,12%]), indicating relatively few outright false positives under model assumptions. Together, actual replications and Z‑curve diagnostics provide complementary credibility checks. I close by highlighting publication policy relevant patterns and practical guidance on where ex‑ante diagnostics can help target future replication resources.

If you are interested in attending, please fill out the expression of interest form below.

Zhuoran is a current PhD student in the School of Economics at the University of New South Wales. He has two main research focuses. One is on the credibility of economic research through meta-analysis. Another one is non-monetary incentives in behavioural contract design. Zhuoran also works closely with industry friends to test his hypotheses.

Learn more about him on his .