A related piece on this site already used OpenAIRE’s live index to track a sustainability and AI keyword surge in civil engineering doctoral theses. The same method, run against economics rather than civil engineering, produces a different shape of result: a sustainability-keyword share that has climbed steadily and substantially, and an AI-keyword share that is moving the same direction on too small a base to call a clean trend yet. Both are reported here with the exact query behind each number, queried live for this article on 2026-09-20, so a research office can reproduce or extend it against a fresh pull of the same index.
The query, stated exactly
Every figure below comes from the OpenAIRE Graph API’s publications search endpoint (api.openaire.eu/search/publications), filtered to instancetype=Doctoral thesis and a fromDateAccepted/toDateAccepted window covering each full calendar year, with keywords=economics as the base population and keywords=economics sustainability, keywords=economics artificial intelligence and keywords=economics climate change as three overlay queries. This is an index of what OpenAIRE has harvested and deduplicated from participating institutional and national repositories, not a national or global census of economics doctorates — a distinction worth stating plainly before the numbers, because the base population size itself does not move smoothly year to year, which the table below shows honestly rather than smoothing over.
| Year | Indexed economics doctoral theses | Sustainability-keyword count | Sustainability share | Climate-change-keyword count | Climate-change share | AI-keyword count | AI share |
|---|---|---|---|---|---|---|---|
| 2015 | 833 | 28 | 3.36% | 18 | 2.16% | 1 | 0.12% |
| 2018 | 919 | 27 | 2.94% | 34 | 3.70% | 4 | 0.44% |
| 2021 | 1,296 | 42 | 3.24% | — | — | 7 | 0.54% |
| 2024 | 1,033 | 84 | 8.13% | 60 | 5.81% | 14 | 1.36% |
The 2021 climate-change-specific query was not run for this article; that cell is left blank rather than estimated.
The sustainability share: a real and fairly large move
The sustainability-keyword share of indexed economics doctoral theses moved from 3.36% in 2015 to 2.94% in 2018 to 3.24% in 2021 to 8.13% in 2024. Read across the full four-point series rather than just the endpoints, the honest description is a roughly flat share through 2021 followed by a sharp rise by 2024 — not a smooth, steady climb across the whole decade. By 2024 the share is 2.4 times its 2015 level and 2.8 times its 2018 level. Unlike the smaller AI-keyword series discussed below, this move is built on a base large enough (27 to 84 theses across the sampled years) to support a genuine trend claim for the 2018–2024 period specifically, even if the earlier 2015–2018 window shows no clear movement at all.
The sustainability rise is substantially a climate-economics story
Running the narrower economics climate change query against the same yearly windows shows that climate-change-keyword theses grew even faster than the broader sustainability-keyword bucket: from 18 in 2015 (2.16% of that year’s indexed economics theses) to 34 in 2018 (3.70%) to 60 in 2024 (5.81%) — a 3.3-times increase in share from 2015 to 2024. Because climate-change-keyword theses are a subset of the broader sustainability-keyword count, this means climate change specifically accounts for a large and growing share of the sustainability figure itself: 60 of the 84 sustainability-keyword theses indexed in 2024 (71%) also carry the climate-change keyword. The practical reading is that “sustainability” as a rising theme in indexed economics doctoral research is, on this evidence, substantially a climate-economics phenomenon specifically, rather than an equally broad rise across sustainability’s other sub-themes (resource economics, circular-economy models, ESG-adjacent finance research, and so on, none of which this article queried separately).
The AI-keyword share: real direction, too small a base to trend confidently
The AI-keyword share moved from 0.12% in 2015 (1 thesis) to 0.44% in 2018 (4 theses) to 0.54% in 2021 (7 theses) to 1.36% in 2024 (14 theses) — more than eleven times the 2015 share by 2024, and roughly three times the 2018 share. The direction is consistent across all four sampled years, which is worth noting. But the absolute counts behind that share — 1, 4, 7 and 14 theses — are small enough that a single additional or missing indexed record in any one year would move the percentage meaningfully. This piece treats the AI-keyword series the same way the companion civil engineering piece treated its own noisy AI series: the direction is reported honestly, without dressing a one-to-fourteen-thesis move up as a confident statistical trend.
Why the base population itself is not a smooth series
A reader might reasonably ask why the total count of indexed economics theses went from 833 (2015) to 919 (2018) to 1,296 (2021) back down to 1,033 (2024), rather than rising steadily as global doctoral output has generally done over the same period. The answer is a property of the index, not of economics doctoral education: OpenAIRE’s coverage depends on which repositories it has harvested and when, and a repository’s own indexing lag, metadata completeness and deduplication behaviour all affect a given year’s indexed count independently of how many economics doctorates were actually awarded that year. A repository that joined OpenAIRE’s harvesting network in, say, 2022 can retroactively backfill older records inconsistently, and a repository that changes its metadata schema can temporarily drop out of a keyword search until its records are re-harvested — both plausible, unverified explanations for the kind of non-monotonic base count seen here, offered as illustration of why the index is not a census, not as a confirmed account of this specific series’ movement. This is the same caveat this site’s other OpenAIRE-based pieces apply, and it is why every percentage in this article is described as an index-wide share — the share of what OpenAIRE has indexed that carries a given keyword — rather than a national or global share of all economics doctorates awarded.
What this means for a graduate school or economics faculty
Sustainability-adjacent economics, and climate economics specifically, is not a niche anymore. An economics faculty that has not yet named climate economics or the economics of the green transition as a recognised sub-field in its own thesis-topic guidance is behind a real and fairly steep shift in what candidates are already researching, at least among the subset OpenAIRE indexes. Supervision capacity is the practical constraint this tends to expose first: a faculty can confirm on paper that climate economics is an acceptable thesis area while still having only one or two supervisors genuinely equipped to examine it in depth, which is worth auditing against actual supervisor expertise rather than assuming the topic-approval process alone has kept pace.
AI-adjacent economics is early but moving the same direction. A faculty does not yet need a dedicated AI-in-economics track on this evidence alone — the 2024 count is still only 14 indexed theses — but the direction is worth watching against the faculty’s own thesis-title records over the next two or three intake cycles rather than dismissing it as noise indefinitely. A faculty already running a data science or computational-methods elective for economics candidates is better positioned to absorb this if the share continues climbing than one that has made no adjacent curricular provision at all.
A faculty can run this exact query against its own narrower sub-field vocabulary. Behavioural economics, development economics, labour economics and monetary economics are all viable keywords= substitutions for a faculty that wants a sub-field-specific version of this same analysis, using the identical endpoint and parameters stated above — and the climate-change-versus-sustainability breakdown run here is itself a reusable pattern: querying a broad theme keyword alongside a narrower, more specific one inside it to establish what is actually driving an apparent trend before acting on the headline figure alone.
How this differs from the site’s other data pieces
This is the second time this site has run a keyword-surge analysis against OpenAIRE, after the civil engineering piece — deliberately, since the method generalises cleanly across fields and the comparative value increases with each additional field covered. It is a different question from how many doctoral theses are produced each year, which sizes the overall global count rather than a keyword-level share within one field, and from postgraduate enrolment by field of study, which uses EU-27 ISCED-F enrolment statistics rather than a repository-index keyword search. Each of the three answers a genuinely different institutional question, and none substitutes for the others. It is also worth distinguishing from the named-institution benchmarking pieces on this site, such as this site’s TU Munich doctoral-regulations benchmark: those pieces document what individual institutions’ own published rules require, while this piece and its OpenAIRE-based companions document what candidates across many institutions are actually researching, which is a demand-side question rather than a policy-compliance one.
Where Tesify fits
Topic-selection guidance for economics doctoral candidates is a faculty and supervisor decision this piece does not attempt to replace. Where Tesify for Institutions is relevant is once a candidate has chosen a sustainability-, climate- or AI-adjacent topic and is drafting: the platform gives supervisors visibility into a thesis’s progress against the institution’s own formatting and citation conventions, whatever the sub-field. A free departmental pilot lets one economics cohort test that visibility before any procurement decision.
Frequently asked questions
What data source underlies this keyword-share analysis of economics doctoral theses?
The OpenAIRE Graph API’s publications search, filtered to instancetype=Doctoral thesis and a fromDateAccepted/toDateAccepted window per year, queried live for this article. It is an index of what OpenAIRE has harvested and deduplicated from participating repositories, not a national or global census of economics doctorates.
How much did the sustainability-keyword share of economics theses actually change?
From 3.36% of indexed economics doctoral theses in 2015 (28 of 833) to 8.13% in 2024 (84 of 1,033) — but the movement was concentrated in the 2018–2024 window; 2015 to 2018 showed no clear change.
Is the sustainability rise spread evenly across sustainability’s sub-themes, or concentrated in one area?
Concentrated. The narrower economics climate change query accounts for 60 of the 84 sustainability-keyword theses indexed in 2024 (71%), meaning the broader sustainability rise is substantially a climate-economics phenomenon on this evidence, not an equally broad rise across every sustainability-adjacent sub-theme.
Is the AI-keyword count in economics theses trending as clearly as the sustainability one?
The share rose from 0.12% (2015) to 1.36% (2024), but the underlying counts are small — 1, 4, 7 and 14 theses across the four years sampled — which is too small a base for a confident trend claim, even though the direction is consistent across every year sampled.
Why did the total number of indexed economics theses go up and down rather than rise smoothly?
OpenAIRE’s index reflects what has been harvested from participating repositories at query time, not a complete or smoothly-growing census — repository coverage, harvesting timing and deduplication all affect the indexed total for a given year independently of how many economics doctorates were actually awarded.
Can an institution reproduce this query itself?
Yes. The exact endpoint, parameters and date windows used are stated in this article, so a research office or graduate school can re-run the same query against its own field or sub-field terms.
Should a faculty base curriculum or supervision-capacity decisions on this data alone?
No. It is one input, best combined with the faculty’s own thesis-title records and enrolment data over several intake cycles — particularly for the AI-keyword series, where the sample sizes are still too small to carry a decision on their own.