Governments routinely use tax incentives to encourage private contributions to activities perceived as socially valuable, such as (female) labour participation (Bastian and Lochner 2020) and research and development (Galindo-Rueda et al. 2020). Charitable donations typically benefit from generous tax credits, justified by the idea that private giving helps finance public goods and corrects for underprovision by markets or the state.

Political donations also benefit from tax relief in many countries, including France, Germany, Italy, and Spain (e.g. Cagé 2018). Yet, by contrast, their ‘social value’ is more debatable. For example, when the tax credit for political donations (to political parties and electoral campaigns) was enacted at the end of the 1980s in France, members of parliament effectively created a tax device from which they could directly benefit.

This asymmetry raises a fundamental policy question: should charitable and political donations be treated similarly by the tax system? Moreover, in countries like France, where they take the form of a non-refundable tax credit, fiscal incentives for political giving only benefit households subject to income tax, which excludes more than half of the population (in 2025, 45% of the fiscal households in France were liable to the income tax). Because higher-income households donate more, this system reinforces political inequalities (Cagé 2024).

In novel research (Cagé et al. 2026), we provide experimental evidence on how individuals respond to different fiscal incentives for charitable and political donations. Using a large-scale survey experiment with 12,600 respondents, we compare the effects of three different forms of tax relief – non-refundable tax credits, direct reimbursements, and matching subsidies – both for charitable giving and for political donations. Our study uniquely considers charitable and political donations within a unified experimental framework, allowing for a direct comparison of how donors respond to identical fiscal incentives. Our findings reveal marked differences in behavioural responses, with direct implications for the design of tax policy for giving.

A survey experiment on donation incentives

The (pre-registered) survey was conducted in 2022 on a sample representative of the French population (12,600 respondents). Individuals were randomly assigned to one of eight groups (approximately 1,575 individuals per group), designed to identify the effects of different tax-relief schemes on charitable and political giving.

We first provided all surveyed individuals with an information sheet explaining the existing non-refundable income tax credit system for donations, using 50/50 randomisation to allocate them either to the political giving group or to the charitable giving group. Second, we further randomised each sub-sample into four equally sized groups:

1. A control group reflecting the existing tax credit, equal to 66% of the amount of donation.
2. An ‘immediate reimbursement’ treatment group, where 66% of the donation amount would be immediately reimbursed to the donor, regardless of whether they pay the income tax (T1).
3. A matching treatment group, where each euro donated would be matched with 2 euros given directly to the organisation (T2).
4. A repeal of the existing tax credit treatment group (T3).

Participants were then presented with hypothetical donation decisions. They were asked whether they would donate, and how much, under different tax schemes.

Crucially, the control group and the first two treatments (T1 and T2) enjoy the same tax price of 33 cents paid per euro received by the charity, while in the case of the repeal treatment (T3), the price rises to 1 euro paid per euro received by the charity. The same incentive structures were applied separately to charitable donations and political donations, allowing for a clean comparison across these two groups.

Asymmetric effects for charitable donations and political donations

Figure 1 presents the average probability of donating and the average amount donated in the control and in each treatment group. In the control group, the probability of giving is seven times higher for charitable donations (42%) than for political donations (6%). This large gap persists across all treatments. Conditional on giving, average donation amounts are also substantially higher for charitable giving.

Figure 1 Average outcomes depending on the treatment status

a) Extensive margin (willingness to make a donation)

b) Intensive margin (amount given, conditional on giving)

Notes: The figure shows the average outcome by treatment status, at the extensive margin and at the intensive margin. The extensive margin is defined as the probability that the individual declares that they are willing to make a donation. For the subset of respondents who are willing to give, the intensive margin is defined as the amount they declare they are willing to give. The four groups are: (i) the control group (“control-“), (ii) the direct reimbursement treatment group (“direct reimburs.-“), (iii) the matching treatment group (“matching-“), and (iv) the repeal of the tax credit system treatment group (“repeal-“). The sample includes all individuals in the experiment. We consider separately the individuals who were shown the charitable-giving information sheet (shades of blue) and the individuals who were shown the political-giving information sheet (shades of orange). The error bars represent 95% confidence intervals.
Source: Cagé et al. (2026).

Two main conclusions emerge from the treatment effects. First, while a large body of literature in public economics has established that charitable giving is sensitive to its tax price (Bakija and Heim 2008, Fack and Landais 2010, Cagé and Guillot 2022), we similarly document that charitable donations react both at the extensive margin (the probability to give) and at the intensive margin (the amount contributed conditional on giving) to changes in the tax schemes. We indeed show that eliminating the tax credit leads to a substantial decline in both the willingness to give and the amount donated for charitable donations. Overall, the repeal leads to a drop of €35–40 (i.e. about 24%–31% of the mean) in the amount individuals are willing to contribute.

Political donations, in contrast, appear to be price inelastic. We find that removing the tax credit has little to no effect on both margins of political donations. This suggests that political donors are significantly less sensitive to the financial cost of giving than charitable donors. In other words, political contributions appear to be driven by motivations that are less easily altered by tax policy.

Beyond the tax price itself, the form of the fiscal incentives also matters. In a matching system, an entity – often the government, sometimes a firm – contributes an additional amount proportional to the individuals’ donations. Existing work mostly compares matching to standard tax credits, focusing on the intensive margin. This literature (see Epperson and Reif 2019 for a survey) shows that individuals give more under matching than under rebate schemes (Eckel and Grossman 2003, 2008, 2017). Consistently, we find that matching subsidies increase donation amounts, conditional on giving, both for political and for charitable giving.

However, we also show that matching reduces the probability of giving to a charity. This finding puts into perspective the rebate-versus-match literature, which largely focuses on the intensive margin. One possible explanation is that matching schemes are more complex and less transparent, particularly in the French context, where they are uncommon, which may discourage some potential donors from participating in the first place.

Heterogeneity in responses

Average effects can mask substantial heterogeneity. To explore this, we use a principal component analysis, which we complement with a generic machine learning approach (Chernozhukov et al. 2018) to identify patterns in how different groups respond to tax incentives.

In a nutshell, we identify three main components – defined with respect to income, education, profession, age, and political preferences – that capture the largest variations in the data. Following the categorisation of Cagé and Piketty (2023), we interpret them as: (i) the social-ecological left (left bloc), (ii) the progressive liberal centre (centre-right bloc), and (iii) the patriotic national right (far-right bloc).

We find that, at the extensive margin, repealing the existing income tax credit (T3) would mostly negatively affect far-right voters. In contrast, we document a negative effect of both T1 (direct reimbursement) and T2 (matching) for centre-right voters (who tend to be high-income), probably reflecting the use of charitable giving as a way to avoid paying income tax. There is no heterogeneity in treatment effects for political donations.

Implications for tax policy and the public funding of democracy

Our findings have important policy implications. First, they call into question the effectiveness of extending generous tax incentives to political donations. If political giving is largely inelastic, tax expenditures in this area may have a limited impact on aggregate contributions while disproportionately benefiting high-income donors who would have given anyway. In contrast, we calculate that the elasticity of charitable donations to the existing tax credit is equal to -0.18. Thus, from an efficiency perspective, while tax credit remains a powerful policy tool for charitable donations, it appears to be a pure windfall for political donations.

Second, matching schemes deserve careful consideration. If one is willing to further incentivise charitable giving, then a preferable policy would consist of replacing the existing tax credit with a matching system. Doing so, however, would increase the fiscal expenditures for the state, lead to a higher concentration of charitable giving, and subsidise the preferences of the richest and oldest citizens (given that, at the intensive margin, the centre-right bloc voters are those whose donations increase the most under the matching treatment).

References

Bakija, J, and B Heim (2008), “How does charitable giving respond to incentives and income? Dynamic panel estimates accounting for predictable changes in taxation”, NBER Working Paper 14237.

Bastian, J, and L Lochner (2020), “The EITC and maternal time use: More time working and less time with kids?”, NBER Working Paper 27717.

Cagé, J (2024), “Political inequality”, Annual Review of Economics 16: 455–90.

Cagé, J (2018), Le prix de la démocratie, Fayard (English version: The price of democracy, Harvard University Press, 2020).

Cagé, J, and M Guillot (2022), “Is charitable giving political? Evidence from wealth and income tax returns”, CEPR Discussion Paper 17597.

Cagé, J, M Guillot, and Y Huang (2026), “Should charitable and political donations benefit from similar treatments? Evidence from a survey experiment”, CEPR Discussion Paper 21132.

Cagé, J, and T Piketty (2023), Une histoire du conflit politique. Elections et inégalités sociales en France, 1789–2022, Paris: Le Seuil (English version: A history of political conflict. Elections and social inequalities in France, 1789–2022, Harvard University Press, 2025).

Chernozhukov, V, M Demirer, E Duflo, and I Fernández-Val (2018), “Generic machine learning inference on heterogeneous treatment effects in randomized experiments, with an application to immunization in India”, NBER Working Paper 24678.

Eckel, C C, and P J Grossman (2017), “Comparing rebate and matching subsidies controlling for donors’ awareness: Evidence from the field”, Journal of Behavioral and Experimental Economics 66: 88–95.

Eckel, C C, and P J Grossman (2003), “Rebate versus matching: Does how we subsidize charitable contributions matter? ”, Journal of Public Economics 87(3–4): 681–701.

Eckel, C C, and P J Grossman (2008), “Subsidizing charitable contributions: A natural field experiment comparing matching and rebate subsidies”, Experimental Economics 11(3): 234–52.

Epperson, R, and C Reif (2019), “Matching subsidies and voluntary contributions: A review”, Journal of Economic Surveys 33(5): 1578–601.

Fack, G, and C Landais (2010), “Are tax incentives for charitable giving efficient? Evidence from France”, American Economic Journal: Economic Policy 2(2): 117–41.

Galindo-Rueda, F, M Bajgar, C Criscuolo, and S Appelt (2020), “Effectiveness of R&D tax incentives in OECD economies”, VoxEU.org, 14 October.