The Court of Accounts has put it plainly in black and white: twenty territories labeled during the first wave of the French Impact program in February 2019, within the framework of the ESS Growth Pact, and no one knows how to measure what it has produced. France has a genuine capacity to launch local experiments, to mobilize actors, to bring new mechanisms to light. What it has not yet built is the chain that goes from experimentation to proof, and from proof to scaling. The gap between the two is the exact place where a social policy transforms into permanent subsidies without results, or disappears for lack of funding when it was actually working.
The Essential Points
- The Court of Accounts notes the inability to measure the impact of mechanisms supported under the French Impact program, despite twenty territories labeled on February 25, 2019 during a single wave of labeling.
- The structural problem is twofold: absence of standardized evaluation protocols and absence of a budgetary mechanism allowing for multi-year funding of what has proven itself.
- Methods exist, randomized evaluation, conditional generalization clauses, but they remain marginal in French administration.
- The long-term risk is that of social innovation trapped within its own scope: productive of experiences, sterile of transformations.
- Two trajectories are emerging by 2040: a chain of proofs that selects and scales, or indefinite labeling that subsidizes without deciding.
The Social and Solidarity Economy: A Sector That Counts Without Counting Itself
The social and solidarity economy represents, according to consolidated estimates, more than 10% of salaried employment in France and nearly 200,000 active structures. Associations, cooperatives, mutuals, foundations: this sector intervenes in domains that neither the State alone nor the market alone has been able to cover—occupational insertion, home care assistance, supported housing, local food systems. Its real weight in French social life is considerable.
This finding of utility, however, says nothing about the effectiveness of public mechanisms that support it. The Court of Accounts, in its evaluation published in September 2025, carefully distinguishes the two. That the social and solidarity economy is socially valuable is one thing. That public support—subsidies, labels, tax exemptions, calls for projects—produces measurable and lasting effects is another. On this second point, the findings are severe: measurement tools are lacking, objectives are vague, and the logic of labeling has developed without anyone previously asking how they would know if it was working.
The first 20 territories, all labeled on February 25, 2019 during one single wave within the French Impact program and the ESS Growth Pact, illustrate this gap. The intention was to constitute local ecosystems of social innovation, to concentrate resources there, to observe what emerged, then, implicitly, to draw generalizable lessons from it. But the observation phase was neglected. Monitoring indicators were defined after launch, often heterogeneously from one territory to another, making any comparison hazardous. Result: we do not know which of these territories produced effects, which simply benefited from a label without real transformation, and why some differ from others.
Labeling Without Measuring Has a Real Cost
It would be convenient to reduce this problem to an administrative question, boxes left unchecked in a dashboard. Reality is more serious. When a public mechanism is not evaluated, it does not remain neutral. It becomes permanent or disappears for reasons having nothing to do with its effectiveness: budgetary decisions, political affinities, the ability of those in charge to produce convincing reports, the media visibility of a project. Good mechanisms can die for lack of evidence; bad ones can survive for lack of equipped critics.
This dynamic is not unique to the social and solidarity economy. It runs throughout the entirety of French social policy. France often lets adjustment happen by default, lacking institutional mechanisms that force decisions. In the case of social innovation, the absence of active selection carries an additional cost: it discourages the most rigorous actors, those who know their results will be compared to those of less scrupulous structures receiving the same public funding.
The Court of Accounts does not merely note the absence of measurement. It points to a design problem: calls for projects are calibrated to finance startup phases, not to support growth. A mechanism that demonstrates its effectiveness at small scale faces a void: startup funding runs out, permanent funding does not exist for structures that have not reached the critical size needed to access public procurement or markets. The space between the pilot project and the established institution is a budgetary desert.
Three Obstacles to Scaling That France Has Not Yet Overcome
The first obstacle is methodological. Evaluating a social mechanism with rigor requires defining, before launch, what you seek to produce and how you will measure it. This presupposes a comparison group, a sufficient observation period, and indicators that resist manipulation. These protocols exist: randomized controlled evaluation, imported from development economics, has been successfully applied to insertion policies in France, notably in work conducted by J-PAL Europe. But its use remains the exception in French social administration, not the rule.
The second obstacle is budgetary. Multi-year funding is structurally difficult in a budgetary architecture where credits are voted annually. A mechanism that produces its effects over three or five years cannot be seriously evaluated with conventions renewed annually. This constraint is known. It has been partially lifted in other domains—State-region planning contracts offer a six-year horizon—but it has not been applied systematically to support for social innovation.
Project leaders adapt: they learn to survive in budgetary uncertainty, which mobilizes considerable energy to the detriment of the activity itself.
The third obstacle is political. Scaling what works implies abandoning what does not work. In a sector where each mechanism has its defenders—local elected officials, beneficiary associations, partner administrations—the decision to stop is politically costly. Labeling without evaluation is, from this perspective, a comfort solution: it allows support without ever having to choose. Rigorous evaluation, by making results comparable, forces arbitrations that no one wishes to assume.
The Method of Those Who Have Solved the Problem
France is not the only democracy to have had this problem. Nor is it the only one seeking a way out. A few foreign experiences deserve close examination, without idealizing them.
In the United Kingdom, the What Works Network, created in 2013 at the instigation of the Cabinet Office, today brings together about ten sectoral centers responsible for synthesizing evidence of effectiveness in domains ranging from education to well-being through crime prevention. The idea is not to produce additional evaluations, but to constitute bases of evidence accessible to local and national decision-makers, with explicit confidence levels. The mechanism is not perfect, adoption by administrations remains uneven, but it has institutionalized the demand for evidence in the public policy cycle.
In Canada, the Impact and Innovation Unit of the Privy Council Office has developed conditional funding tools: resources are allocated in several tranches, the next being conditional on the achievement of contractually defined intermediate objectives. This architecture forces mid-course evaluation and allows mechanisms to be reoriented or stopped without waiting for their term.
In France, initiatives exist but remain marginal. The Experimentation Fund for Youth, created in 2009, financed projects with an obligation to evaluate and produced useful syntheses. The social innovation lab of the Bank of Territories experimented with methods of co-construction and evaluation. But these initiatives are dispersed, without common doctrine, and without systematic translation into national policy. The capacity to innovate locally is real, forms of social innovation that manage to grow show that the business model can precede public aid, but the connection with central government remains fragile.
Two Paths by 2040, and What Distinguishes Them
The question raised by the Court of Accounts’ report will not be resolved in the coming months. It draws a bifurcation whose effects will be measured on the scale of a generation.
In a first scenario, the State effectively builds a chain of proofs. This would require several things simultaneously: standardized evaluation protocols, defined before the launch of any supported mechanism, with comparable indicators from one territory to another; a multi-year financing mechanism reserved for mechanisms that have demonstrated their effectiveness following an initial phase; and a conditional generalization clause, automatic or quasi-automatic, for mechanisms whose evaluation is positive and whose model is transferable. This scenario requires a profound change in administrative culture. It demands accepting that certain mechanisms will be stopped, that certain labelings will not be renewed, that public support will be explicitly selective. The signal to watch in the coming years is simple: what is the number of mechanisms generalized after positive evaluation, and what is the share of multi-year funding in total support for the social and solidarity economy?
If these indicators progress, the bifurcation is beginning.
In the second scenario, current logic continues. New labels are created, new calls for projects launched, new territories designated as experimentation sites. Social innovation remains active, visible, politically valued. But it remains confined to its own scope, incapable of irrigating institutions of common law—schools, hospitals, public employment services—which are the only vectors of large-scale change. The risk, in this case, is that the social and solidarity economy becomes a space for absorbing institutional failures rather than a laboratory whose results transform the institutions themselves.
Social innovation would be productive of experiences and sterile of transformations.
What distinguishes the two trajectories lies in concrete institutional capacity: the State must have the tools to evaluate, compare, select, and provide multi-year financing. The INSEE, whose growing number of publications examine methods for evaluating public policies, and the Committee for the Evaluation and Control of Public Policies of the National Assembly have the technical skills. These skills must be mobilized systematically rather than sporadically.
What Must Be Built, Not Just Financed
The needs are known. They are not beyond reach.
A randomized evaluation mechanism for social policies requires modest means compared to the amounts committed in the mechanisms themselves. J-PAL Europe, based in Paris, has shown that this type of evaluation is feasible in the French context. The issue is making it mandatory for any mechanism supported beyond a certain threshold, and constituting a common database allowing inter-territorial comparisons.
A budgetary mechanism for multi-year financing of positively evaluated mechanisms could draw inspiration from existing planning contracts, applying them explicitly to social innovation. This would presuppose a dedicated budget line, distinct from annual credits, with governance associating the State, regions, and representatives of the sector.
Finally, a generalization clause, committing the State to take charge or to integrate into common law the mechanisms whose effectiveness is established, would fundamentally change overall logic. Project leaders would know that their evaluation work has institutional value, not merely academic. Administrations would be incentivized to design scalable mechanisms from the start, not projects optimized for the next call for projects.
France knows how to make things germinate. The question that is opening is whether it will learn to harvest.
Sources
- Court of Accounts, Public Support for the Social and Solidarity Economy, September 2025
- INSEE, Courier of Statistics, methods for evaluating public policies
- J-PAL Europe, work on randomized evaluation of insertion policies in France (Abdul Latif Jameel Poverty Action Lab, Paris School of Economics)
- What Works Network, Cabinet Office, United Kingdom (British government, www.gov.uk/guidance/what-works-network)
- Court of Accounts Report - Public Support for the Social and Solidarity Economy (September 2025)
- Labeling of 20 first French territories to receive the French Impact Territory label (February 25, 2019)
- J-PAL Europe - Paris School of Economics
- What Works Network - GOV.UK
- Experimentation Fund for Youth - INJEP
- Impact and Innovation Unit - Privy Council Office of Canada
- Social and Solidarity Economy France - key figures (economie.gouv.fr)