Social Entrepreneurship Index 2026: how the provisional SEI Diamond is built

How 40 indicators from public statistics, published indices and coded policy documents become comparable scores for 45 European countries, and how far those scores can be trusted.

This page is available in English only.

1 · What the index measures

The Social Entrepreneurship Index (SEI) measures the conditions a country offers social enterprises, not the size or performance of the social-enterprise sector itself. It asks whether the political, economic, legal, institutional and financial environment makes it easier or harder to start, run and grow an enterprise with a primary social or environmental mission.

The framework is the SEI Diamond. The political dimension sits at the centre as a circle, because political stability and political will condition everything else. The four corners are the economic, legal, institutional and financial environments. A larger, more even Diamond means a more supportive ecosystem; a lopsided one shows where a country is strong and where it is holding social enterprises back.

The full matrix has 63 indicators: 8 political, 20 economic, 12 legal, 16 institutional and 7 financial, grouped into sub-dimensions. They draw on three kinds of evidence: public statistics and published indices; a review of laws, strategies and programmes; and a survey of social entrepreneurs and ecosystem experts. This edition uses the first two. It scores the 40 indicators that can be filled for all 45 countries today; the 23 survey-based indicators complete the index once fieldwork is done.

2 · Countries and reference periods

The index covers 45 countries in five groups: the EU-27, the Western Balkans (Albania, Bosnia and Herzegovina, Kosovo, Montenegro, North Macedonia, Serbia), the Eastern Partnership (Armenia, Azerbaijan, Belarus, Georgia, Moldova, Ukraine), EFTA (Iceland, Liechtenstein, Norway, Switzerland) and the UK and Turkey. Regional figures are unweighted means of member countries.

Statistics use the latest year available, mostly 2024 or 2025. This edition takes each country’s most recent observation whatever its age, so a few values date from 2014 to 2018 (some GEM surveys, NEET for Ukraine, the UK’s statistics after it left EU-SILC); the year of every value is in the data, and version 1.0 will apply a maximum age. The policy layer describes the situation at end-2025. A few sources exist for one year only and are used as published: the Heritage Index of Economic Freedom 2020, the World Economic Forum Global Competitiveness Report 2019 (its last edition), the ease-of-getting-credit measure carried in the Global Innovation Index 2020, and the Open Budget Survey 2023.

3 · Indicators and sources

Table 1 lists every row of the SEI indicator register: the 40 live indicators with their sources, and the indicators still pending. Some pending rows group several survey indicators, which is why the register has fewer rows than the matrix has indicators. Full references for the data sources are listed at the end of this note (Cornell University et al., 2020; Eurostat, 2024a, 2024b, 2026a–2026g; Freedom House, 2025; Global Entrepreneurship Monitor, 2026; International Budget Partnership, 2024; Schwab, 2019; The Heritage Foundation, 2020; World Bank, 2026a, 2026b).

4 · From raw values to scores

Every indicator is turned into a score from 0 (least favourable) to 100 (most favourable), in one of three ways.

• Published indices already on a 0–100 scale (Worldwide Governance Indicators, Heritage, WEF, Open Budget Survey) are used as published.
• Statistics (poverty, inequality, unemployment, firm demography and so on) are rescaled between goalposts set at the 5th and 95th percentiles of all country-year values for that statistic across the 45 countries (the panel runs from 2014, 2015 or 2016 to 2024 or 2025, depending on the series). Investment and tertiary enrolment enter with one latest value per country, so their goalposts are the percentiles of those 44 values. Values beyond a goalpost are capped. Where lower is better, the scale is reversed. A country’s score uses its latest value.
• Policy codes are converted from rubric points: score = points ÷ maximum points × 100.

Percentile goalposts keep a single outlier from compressing everyone else. Because they are set on the whole panel, they stay fixed when the index is recomputed for earlier years, so a change in score reflects a change in the underlying value. Table 2 gives the goalposts.

5 · Aggregation

Scores are averaged in three steps, each an unweighted arithmetic mean: indicators into sub-dimensions, sub-dimensions into dimensions, and the five dimensions into the composite. Each sub-dimension therefore carries equal weight within its dimension, whatever its number of indicators, and each dimension carries a fifth of the composite.

Two indicators combine sources. Where a country has no AROPE figure, indicator 2.1.2 uses the income-only at-risk-of-poverty rate. Indicators 4.2.3 (investment and property protection) and 5.2.3 (SME financing and financial freedom) average two sources. A missing indicator is left out of its sub-dimension’s average rather than imputed, and the number of indicators behind each dimension is reported with every score.

6 · The policy layer

Eight indicators come from coding policy documents against a common rubric (Table 3), with end-2025 as the reference date. For 26 EU member states the evidence is the national country reports prepared in 2025 for the mid-term review of the EU Social Economy Action Plan. Those reports are not public, so each EU code is tied to the documents the report cites for it: the law, strategy, programme, procurement framework or official page. Where a report names an instrument without citing it, the code points to the instrument’s official text, and where it records an absence, to the report’s statement. 184 of the 208 EU codes carry a source link; all are listed code by code in the policy-coding file. The other 19 countries (Latvia, the Western Balkans, the Eastern Partnership, EFTA, the UK and Turkey) were coded from laws, official sources and recent EU, OECD and UNDP reports. For the Western Balkans the coding also draws on Chichevaliev et al. (2023) and the synthesis of the European Commission’s 2018 country reports (Rosandić, 2018).

Two rules settle borderline cases. A law enacted but not yet in force scores 1. A strategy that has lapsed but whose measures continue scores 2. Each code carries the instrument it rests on, a source link, a confidence level and, where needed, a note for verification; these are published in the policy-coding file. The first coding was drafted with an AI model working from the cited documents and reviewed by the authors. In November 2026 it was checked twice. For the 19 countries coded from laws and official sources, each of the 152 codes was re-checked against primary sources: 142 were confirmed, 2 were wrong and 8 were judgement calls under the rubric. For nine of the 26 EU member states, the 72 codes were re-coded blind from the same country reports: 67 agreed (93.1%; Cohen’s κ = 0.90). The second coding used the same model family and rubric, so it measures how consistently the rubric is applied rather than agreement between independent judges. Six codes were corrected: Montenegro’s strategy (0 to 3, the 2025–2029 strategy having been adopted on 25 December 2025), definition (0 to 1) and institutions (0 to 1); national funding in Bosnia and Herzegovina (1 to 0) and Germany (2 to 1); and Italy’s strategy (3 to 1, its national action plan still being a draft at end-2025). Seven codes on rubric boundaries were settled by a ruling of the authors: EU funds in Armenia, Georgia, Moldova and Ukraine (2 to 1, matching the Western Balkans), Moldova’s institutions (2 to 1), Armenia’s strategy (1 to 0, a draft not pursued since 2021) and national funding in the Netherlands (0 to 1). Italy’s law code (2, Legislative Decree 112/2017) and the UK’s (1, the Community Interest Company within company law) were kept.

7 · The Diamond over time

8 · Weighting sensitivity

Equal weights are a choice, and the policy layer is the part of the index most open to dispute: it is coded from documents, it records what exists on paper rather than how well it works, and in some sub-dimensions it is the only evidence. Five of the 15 scored sub-dimensions consist only of policy codes, so the eight policy indicators, a fifth of the 40, carry about 43% of the composite weight in the median country (70% in Liechtenstein). We therefore recompute the index under two alternatives.

• Policy indicators at half weight. The eight policy-coded indicators count half within their sub-dimension, and each sub-dimension’s weight within its dimension falls with its share of policy codes, to one half where it consists only of policy codes.
• Without policy coding. Statistics and published indices only. Sub-dimensions with no other indicator drop out, and Liechtenstein, which has no other legal or economic data, cannot be scored.

9 · Provisional status and the field survey

Of the 23 pending indicators, three (legal forms, the pertinence of legislation and administrative procedures) need a further legal review; the other 20 cover what only social entrepreneurs and ecosystem experts can report: perceived political risks, access to public markets and tender support, staff and skills, experience of regulation, training and mentoring, sources and dependency of funding. The questionnaire is drafted. Until it is fielded, the provisional Diamond leans on general conditions and on the formal policy framework, and should be read as a first approximation.

10 · Limitations

• Dated, one-off sources. Heritage 2020, WEF 2019 and GII 2020 are the latest comparable figures for 45 countries on these topics, but they predate the pandemic and the war in Ukraine. The Doing Business data behind the credit indicator has been discontinued.
• Coverage gaps. Eurostat does not cover Bosnia and Herzegovina, Kosovo or the Eastern Partnership for several social statistics; World Bank and ILO modelled estimates fill some gaps and are flagged. Liechtenstein, Kosovo, Azerbaijan and Belarus have the thinnest coverage. Liechtenstein’s legal score rests on the policy codes alone.
• De jure, not de facto. The policy layer records whether strategies, laws, bodies and funding lines exist, not how well they work. The field survey is designed to capture implementation.
• Relative goalposts. Statistic scores are relative to the 45 countries. A score of 100 means “at or above the 95th percentile of all country-year values in this group”, not an absolute standard.
• Shared policy codes. The political dimension correlates with the legal and financial dimensions (r = 0.58 and 0.69 across the 45 countries), but mainly through the policy codes all three contain: political will correlates with them at 0.62 and 0.74, political stability at only 0.14 and 0.16. Otherwise the dimensions carry distinct information: the legal and institutional dimensions are uncorrelated.
• EU funds outside the EU. Indicator 5.1.1 records EU external assistance in candidate and neighbourhood countries and scores EFTA countries and the UK at zero, largely because they are not eligible. As the only indicator in its sub-dimension, it makes up half of the financial dimension, which lowers EFTA’s and the UK’s financial scores and raises the Eastern Partnership’s. Treating it as not applicable was considered and not adopted: with sub-dimension 5.1 dropped, these countries’ financial scores would rest on access to funding (sub-dimension 5.2) alone, and the UK would move from 11th to 1st and Norway from 28th to 21st.
• Measures narrower than the literature. Taxation is scored on the overall tax burden, although the literature stresses the design of the tax system; the efficiency indicator measures the legal framework rather than public spending.

11 · Data, citation and versions

All scores, raw values, policy codes, time series and the sensitivity results are published as an open dataset (CSV and JSON, with a Frictionless data package) on the data page, under a Creative Commons Attribution 4.0 licence for the SEI’s own scores and codes. Third-party values remain under their providers’ terms.

Version history

0.9.1 (3 November 2026) — corrections after a full check of every value against its source and of every policy code. Gini: Eurostat series for Albania and Montenegro (previously World Bank estimates) and 2025 values added. Youth unemployment for Bosnia and Herzegovina: Eurostat from 2021 (previously ILO modelled estimates). Three-year survival rates for 2021–2023 added for Finland, Iceland and Turkey. Ease of getting credit: Global Innovation Index 2020 values (nine countries had carried the 2019 edition’s values). Insolvency framework: the two WEF 2019 scores averaged before rounding (seven countries change by 0.1). Freedom House rescaling: halves rounded up. Provisional flags added for Lithuania. Policy codes: six corrected and seven settled by author ruling (section 6); each EU code now links to the documents its country report cites. Spearman’s ρ now allows for tied ranks. Section 4 now states correctly that goalposts are percentiles of the whole panel. The largest changes in the composite are Montenegro (39.8 to 52.5), Moldova (56.8 to 49.3), Armenia (46.0 to 39.3), Ukraine (44.1 to 39.1), Georgia (43.0 to 38.0), Italy (55.3 to 52.1), Bosnia and Herzegovina (49.8 to 47.0), Germany (67.8 to 65.3) and the Netherlands (59.7 to 62.2).

0.9.0 (November 2026) — first public provisional edition: 45 countries, 40 live indicators, Diamond over time 2015–2024, weighting sensitivity. 1.0 (planned, SECon 2027, Skopje) — SEI 2026 with field-survey indicators for the pilot countries.

References

APA 7th edition. Data sets are listed with the release used; values were last checked against each source on 3 November 2026.

1. Chichevaliev, S., Debarliev, S., & Janeska Iliev, A. (2023). How is social entrepreneurship pursuing the path of development? Regional perspectives in the Western Balkans. In V. Ramadani, S. Kjosev, & B. S. Sergi (Eds.), Entrepreneurship development in the Balkans: Perspective from diverse contexts (pp. 33–68). Emerald Publishing Limited. https://doi.org/10.1108/978-1-83753-454-820231003
2. Cornell University, INSEAD, & WIPO. (2020). The Global Innovation Index 2020: Who will finance innovation? (S. Dutta, B. Lanvin, & S. Wunsch-Vincent, Eds.). https://www.wipo.int/edocs/pubdocs/en/wipo_pub_gii_2020.pdf
3. Eurostat. (2024a). Business demography by legal form and NACE Rev. 2 activity (2004–2020) (BD_9AC_L_FORM_R2) [Data set]. https://doi.org/10.2908/BD_9AC_L_FORM_R2
4. Eurostat. (2024b). Business demography by size class and NACE Rev. 2 activity (2004–2020) (BD_9BD_SZ_CL_R2) [Data set]. https://doi.org/10.2908/BD_9BD_SZ_CL_R2
5. Eurostat. (2026a). At-risk-of-poverty rate by poverty threshold, age and sex – EU-SILC and ECHP surveys (ILC_LI02) [Data set]. Retrieved November 3, 2026, from https://doi.org/10.2908/ILC_LI02
6. Eurostat. (2026b). Business demography by size class and NACE Rev. 2 activity (BD_SIZE) [Data set]. Retrieved November 3, 2026, from https://doi.org/10.2908/BD_SIZE
7. Eurostat. (2026c). Gini coefficient of equivalised disposable income by age (ILC_DI12) [Data set]. Retrieved November 3, 2026, from https://doi.org/10.2908/ILC_DI12
8. Eurostat. (2026d). Persons at risk of poverty or social exclusion by age and sex (ILC_PEPS01N) [Data set]. Retrieved November 3, 2026, from https://doi.org/10.2908/ILC_PEPS01N
9. Eurostat. (2026e). Severe material and social deprivation rate by age and sex (ILC_MDSD11) [Data set]. Retrieved November 3, 2026, from https://doi.org/10.2908/ILC_MDSD11
10. Eurostat. (2026f). Young persons neither in employment nor in education and training (NEET) by sex (SDG_08_20) [Data set]. Retrieved November 3, 2026, from https://doi.org/10.2908/SDG_08_20
11. Eurostat. (2026g). Youth unemployment rate by country of birth (YTH_EMPL_100) [Data set]. Retrieved November 3, 2026, from https://doi.org/10.2908/YTH_EMPL_100
12. Freedom House. (2025). Freedom in the World 2013–2025 raw data [Data set]. https://freedomhouse.org/report/freedom-world
13. Global Entrepreneurship Monitor. (2026). Entrepreneurial behaviour and attitudes: Adult Population Survey key indicators [Data set]. Global Entrepreneurship Research Association. Retrieved November 3, 2026, from https://www.gemconsortium.org/data/key-aps
14. The Heritage Foundation. (2020). 2020 Index of Economic Freedom. https://www.heritage.org/index/
15. International Budget Partnership. (2024). Open Budget Survey 2023. https://internationalbudget.org/open-budget-survey/open-budget-survey-2023
16. Rosandić, A. (2018). Social economy in Eastern Neighbourhood and in the Western Balkans: Synthesis report. AETS Consortium for the European Commission, Directorate-General for Neighbourhood and Enlargement Negotiations. https://ec.europa.eu/docsroom/documents/29642/attachments/15/translations/en/renditions/native
17. Schwab, K. (Ed.). (2019). The Global Competitiveness Report 2019. World Economic Forum. https://www3.weforum.org/docs/WEF_TheGlobalCompetitivenessReport2019.pdf
18. World Bank. (2026a). World Development Indicators [Data set]. Retrieved November 3, 2026, from https://databank.worldbank.org/source/world-development-indicators
19. World Bank. (2026b). Worldwide Governance Indicators (data update of September 25, 2026) [Data set]. Retrieved November 3, 2026, from https://www.worldbank.org/en/publication/worldwide-governance-indicators