Replication in labor economics

Is there a reproducibility crisis in labor economics?

University of Canterbury, New Zealand

one-pager full article

Elevator pitch

There is growing concern that much of the empirical research in labor economics and other applied areas may not be reproducible. Correspondingly, recent years have seen an increase in replication studies published in economics journals. Despite this increase, there are many unresolved issues about how replications should be done, and how to interpret their results. Replications have demonstrated a potential for clarifying the reliability and robustness of previous research. Much can be done to encourage more replication research, and to exploit the scientific value of existing replication studies.

Number of replication studies published in
         duv               economics journals

Key findings


Replications can help to confirm that an original study was done correctly.

Replications can be used to determine if the findings of a study generalize to other, similar settings.

Replications have demonstrated their usefulness in assessing the results of previous, high-profile research.


There is no consensus about what a replication is, nor about what constitutes a “successful replication.”

Little is known about how frequently replications occur, or how often they confirm or disconfirm the original studies.

There is no standardized system to collect, categorize, and link replications to the respective original studies.

Author's main message

Empirical research inherently involves uncertainty. Across studies, this uncertainty is reflected in the fact that many studies investigating similar subjects report different results. Efforts to reduce this uncertainty can help solidify what is known, and unknown, in labor economics. Replication is one such effort. While the concept is seemingly straightforward, conducting and interpreting replications is difficult. Researchers and policymakers alike could benefit by improving access to original data sources, encouraging journals to publish more replications, and establishing a cataloging system to link replications to the original studies.

Full citation

Full citation

Data source(s)

Data type(s)