Does economics take turns at the Nobel podium?

Reconstructing a hidden contest

Studying who wins a prize requires more than studying its winners. Researchers also need a comparison group: people who plausibly could have won but did not. For the Nobel Memorial Prize in Economic Sciences, that group is especially difficult to identify because nominations and committee deliberations are confidential.

Dolton and Tol assembled a roster they report as 374 candidates, drawing on other economics awards, historical biographical collections, citation-based lists and consultation with colleagues. They also included eventual Nobel recipients before their winning years. Their analysis covers the period described in the paper as 1969–2026.

The candidate pool changes annually. Economists must be over 40 and alive; they leave after winning or dying. Unsuccessful candidates otherwise remain eligible. That design recognises that each year’s winners emerge from a changing population, rather than competing against one fixed list throughout their careers.

The researchers collected publication and citation histories, education and employment records, and relationships involving supervisors, students, collaborators and journal editors. They also reconstructed committee membership and members’ research specialities, although they acknowledge uncertainty in that historical reconstruction.

Modelling fields before people

The analysis divides economics into 14 broad fields and proceeds in three stages.

First, it estimates transitions between fields: how often, for example, one research area follows another in successive awards. This is a Markov model—a statistical description in which the previous award’s field helps characterise the next.

Second, a probability model combines those transitions with field characteristics, including candidate numbers and previous prizes. Third, an individual-level model combines each candidate’s characteristics with the estimated probability that their field receives recognition.

This structure assumes that the committee chooses a field before choosing its recipients. It is a modelling assumption, not an observed sequence of confidential decisions. A particularly compelling candidate could instead influence the choice of field.

A recurring pattern, not a timetable

The authors’ main finding is that awards move between fields in a semi-regular pattern. Fields containing more candidates were also more likely to receive recognition.

Their preferred transition model separates the years when Assar Lindbeck served on the committee from those after his departure in 1994. That division describes the historical awards better than treating the whole period as one unchanged process. It does not establish that his departure caused the change.

At the individual level, age was a persistent predictor. The main model’s estimated probability peaked around 71–72, although the age pattern varied across specifications and periods.

Two academic relationships were comparatively robust: having a student who had already won, and having collaborated with a laureate, with some models specifically identifying an older collaborator. Other connections, including family ties and shared editorial work, were less consistent.

These results do not establish lobbying or favouritism. The same associations could arise through mentoring, shared research quality or talented people working together. The data cannot distinguish those explanations.

Citation results were more sensitive to modelling choices. In later years, the authors found a pattern consistent with recognising a small body of highly influential work. Some models associated stronger citation records for particular papers with winning, while total citations had a negative coefficient after accounting for other variables. That is not evidence that accumulating citations damages someone’s prospects.

Why the limits matter

The strongest caution concerns prediction. Several transition estimates use awards from across an entire historical period to explain outcomes within that period. One alternative even includes five subsequent years. Those models can describe history using information unavailable before an award; their fit is not evidence of successful advance forecasting. The authors report that an alternative updated only from information available through each award year did not produce a statistically significant transition predictor.

There are also few awards relative to the many relationships examined: 57 year-to-year transitions for 196 possible transitions between 14 fields. Results therefore depend on classification and modelling choices.

The comparison group is another constraint. Historical biographies dominate earlier candidate selection, while citation-based recognition becomes more important later. Web of Science omits books and some influential papers, potentially distorting comparisons across generations. Shared prizes are treated as individual wins rather than jointly selected combinations.

The models did not consistently detect an association with committee members’ research specialities or candidate gender. Neither finding proves an unbiased process: the analysis begins with an already highly selected group of prominent economists.

The study is an arXiv preprint. Its contribution is a structured account of historical associations—not access to the committee’s reasoning, a causal explanation of selection, or a validated system for naming the next laureates.