France's Protests: What Does the Damage Cost, and Could Reform Pay?

Two bills, two different questions

On 17 October, hundreds of thousands of people joined demonstrations in France over the cost of living and conditions in public services. Most marched peacefully; violent groups damaged shops and public property in several cities. Officials and organisers disagreed over the turnout, but the damage became visible almost immediately. By 20 October, Euronews reported a government tally of at least 3,000 affected shops. [1,2,3]

The insurance headline is already striking. A Morningstar analyst quoted by Le Monde expected the insurers' bill not to exceed €1.2 billion if the violence did not recur. But France Assureurs had not provided a quantified claims total in the reporting checked. An analyst's preliminary insured-loss forecast is neither a paid-claims amount nor a national economic-loss estimate. [4,5]

This leaves two economic questions. What resources and opportunities did the violence actually destroy? And, separately, would credible policies addressing education and household pressure produce benefits commensurate with their cost? The second question does not offset the first. Better policy cannot retroactively restore damaged property, and we cannot assume a reform prevents a future riot.

Why the insurance bill is not the economic cost

The French Senate documented €793 million in insured costs from the 2023 riots, covering 16,400 claims. That is a historical comparison, not a current measure. The Senate also found that the 2023 episode did not materially change France's aggregate economy beyond direct reconstruction, a warning against attaching an arbitrary GDP multiplier to each broken storefront. [6]

Insurance payouts largely decide who bears a loss. The destroyed property is the underlying loss; the reimbursement is not another destroyed asset. The same principle applies to commerce. Five days of closed shops do not mean five days of gross retail sales disappear from national welfare. Some sales move online or to nearby businesses, some are recovered later, and input purchases are avoided. What matters is the part of net activity that is genuinely lost.

We built a resource ledger with explicit assumptions about the severity of damaged shops, non-shop property, days closed, unrecovered net trading contribution, emergency hours and cleanup. Using the reported 3,000 shops as a fixed scenario denominator, selected parameter sets produce €157.1m (lower illustration), €643.8m (reference illustration) and €1,826.7m (higher stress) in illustrative 2026 euros.

Those are not estimates of what France actually lost, confidence bounds or reconciliations to the insurance forecast. For example, the reference case assumes 55% of the shops had minor damage, 30% moderate and 15% severe, with respective real repair-resource proxies of €20,000, €120,000 and €600,000. It further assumes €180.0m in other damaged assets, five closure-equivalent days per shop, 60% of net trading contribution permanently unrecovered, and 250,000 additional response hours. None is verified by a representative October inspection dataset.

The result is more informative than a dressed-up central estimate because readers can change exactly those assumptions. A final damage estimate requires inventories from insurers, municipalities and companies, not another decimal place.

What would reform have to achieve?

The people demonstrating did not submit a single unified budget. The CGT supported higher wages and pensions, lower prices and stronger public services, while education groups emphasised staffing and school conditions. We therefore model particular interventions rather than assuming either that every demand should be funded or that inaction has no consequences. [7,8]

Fiscal conditions narrow the choices. In May, the European Commission projected France's 2027 deficit at 5.7% of GDP and public debt at 120.2% under unchanged policy. These are forecasts only through 2027, not evidence of what the debt ratio will be in 2051. [9]

Our hypothetical education package adds 8,000 teachers at an assumed all-in resource cost of €80,000 a year for ten years, three annual €500 million school-renovation tranches and €40 million of extra maintenance each year for ten years. Each programme year is assumed to create a distinct cohort of 200,000 pupils who benefit. The model waits twelve years before their productivity benefits begin and then allows up to 30 working years.

At a hypothetical 0.4% increase in real annual output for employed beneficiaries, discounted education resource costs exceed modelled extra production: the restricted social balance is €-5.83bn through 2051. To break even over that 25-year horizon, the programme would need an output uplift of 2.08%. Capturing all the cohorts' projected working years through 2077 reduces the threshold to 0.80%.

These percentages are requirements imposed by this particular model, not measured returns to French education spending. The time horizon changes the answer because schools pay teachers now while benefits for today's pupils may accrue decades later.

The research has a valuable reality check. A French DEPP study published on 13 October reported persistent but modest sixth-grade maths-score gains in later cohorts exposed to smaller priority-school classes, while durable French-language gains were not statistically detected. Earlier findings for the first cohort had been less encouraging. Neither study translates directly into an earnings premium, so we do not pretend that a test-score change identifies our 0.4% or the break-even parameter. [10,11]

Household support is a different kind of policy

The alternative package gives households a hypothetical €2 billion per year for three years, plus an assumed 1.5% administrative cost. The payments are government expenditure and household income, not national resource destruction of the same amount. Its real economic effect depends on recipients' spending, domestic production capacity, price response and the taxes required eventually to finance it.

Our illustrative demand channel assumes that 75% of relief is spent, 65% is spent on domestically produced goods and services, and half of that domestic spending represents genuinely additional production. This produces a positive restricted, conditional monetary result for relief, but it is not a welfare endorsement. With no genuinely new production, the modeled real resource balance is negative by the administrative costs. Neither case values the benefit of alleviating hardship or the distortion from eventual financing.

A third scenario adds education and relief together, but it assumes away interaction and crowding-out. The fiscal model keeps government cash flows and real social resource flows apart, and a separate gross-debt overlay illustrates borrowing exposure. It does not turn 2027 official fiscal forecasts into a fifty-year government forecast.

The investment question is what survives better evidence

For insurers and retailers, the immediate decisions concern repairs, liquidity, exclusions and the shape of claims. For public authorities, the important unknown is not whether education can produce returns in principle, but which programme, for whom, at what marginal cost, with what persistent effect and how it will be financed. Both are legitimate capital-allocation questions; neither is answered by treating peaceful protesters and criminal looters as one economic actor.

The strongest defensible conclusion today is a threshold, not a promise: under an explicitly costed hypothetical package, substantial long-term productive benefits would be needed for education expenditure to repay its real resource costs. The exact threshold depends on beneficiaries, timing, implementation, and the time horizon. If those inputs change, the conclusion changes. No avoided-future-riot benefit is added because there is no credible causal estimate that the proposed packages would produce it.

Research note: This is a scenario-based analytical draft with an evidence cut-off of 20 October 2026. Reported shop counts, analyst insurer estimates, school study results and European Commission forecasts are cited external evidence; all repair-severity, closure, teacher-cost, programme-effectiveness and output-conversion parameters are Rokin illustrative assumptions. The low/reference/high event amounts are not confidence intervals or measured losses. The long-term monetised social balance omits material distributional and financing effects.

Explore the research: The accompanying technical report contains event accounting boundaries, education-cohort formulas, fiscal and resource NPVs, source classifications, alternative parameters and limitations. The editable Excel model contains inputs, annual paths, break-even thresholds and internal checks. Both are working research attachments, not an approved public release.

Sources

1. Associated Press / Euronews (17 October 2026). 'No room for violence,' says Macron after rioters and looters hijack peaceful demonstrations. Euronews.

2. Euronews (20 October 2026). France's riot damage doubles: Who will pay for thousands of wrecked shops?. Euronews.

3. Le Monde (English edition) (18 October 2026). October 17 French protests: Crowds marched through the streets before violence, looting and chaos broke out. Le Monde; turnout comparisons.

4. France Assureurs (19 October 2026). Violences urbaines : les assureurs se mobilisent pour accompagner les assurés et accélérer l’indemnisation. Communiqué de presse.

5. Le Monde (20 October 2026). Après les violences urbaines du 17 octobre, assureurs et État se disputent le partage des responsabilités. Le Monde; Morningstar analyst cited (subscriber content partially accessible in original research).

6. Sénat, Commission des lois (rapporteur François-Noël Buffet) (9 April 2024). Émeutes de juin 2023 : comprendre, évaluer, réagir. Rapport d’information n° 521 (2023–2024).

7. Confédération générale du travail (CGT) (16 October 2026). Il est urgent de répondre aux revendications de celles et ceux qui sont l’avenir du pays. CGT mobilisation statement.

8. SNES-FSU (16 October 2026). Poursuivre et amplifier la mobilisation. Union communiqué.

9. European Commission (21 May 2026). Economic forecast for France. Spring economic forecast; France country page and indicator table.

10. Direction de l’évaluation, de la prospective et de la performance (DEPP; El Bah) (3 February 2026). Les effets à long terme de la réduction de la taille de classe en éducation prioritaire sur les performances des élèves. Document de travail, série études E2026-01 (first-cohort study).

11. Direction de l’évaluation, de la prospective et de la performance (DEPP; El Bah) (13 October 2026). Les effets à long terme de la réduction de la taille de classe en éducation prioritaire sur les performances des élèves. Follow-up analysis of later cohorts; October 2026 summary.