The importance of information for tax design: Evidence from the French wealth tax

Many countries are considering introducing or reforming their wealth tax policies. This column studies the effects of a French wealth tax reform, which reduced reporting requirements, on taxpayer behaviour. The scaling back of reporting requirements had large negative effects on tax compliance, with 35% of wealth taxpayers missing in affected brackets and evading 10% of their total wealth tax payments each year. The findings indicate that taxpayers are more likely to under-report the value of less easily verifiable assets such as housing, whereas financial assets are more transparent and thus harder to manipulate.

The taxation of household wealth has been at the forefront of global tax policy debates in recent years. In 2019, two presidential candidates in the US, Bernie Sanders and Elizabeth Warren, proposed introducing a progressive federal wealth tax. In 2024, under the presidency of Brazil, the G20 started working on an internationally coordinated minimum tax on ultra-high-net-worth individuals (Zucman 2024).

These policy initiatives have led to renewed interest in the historical experience with wealth taxation, and a nascent body of work studies behavioural responses to wealth taxes (Seim 2017, Duran-Cabré et al. 2019, Jakobsen et al. 2020, Agrawal et al. 2020, Brülhart et al. 2021, Londoño-Vélez and Avila-Mahecha, forthcoming). These are critical for formulating empirically grounded optimal tax policy and informing the public debate on core policy issues such as the feasibility and desirability of taxing wealth.

But it is difficult to use empirical tax elasticities for policy because tax base elasticities are not structural parameters – they are shaped by tax design. Policy features such as enforcement and reporting requirements influence how elastic the tax base appears. Because such features vary across contexts, elasticities estimated in one setting often don’t generalise to others.

To address this issue, isolating the causal effect of specific tax design elements is essential – but difficult. There are three main challenges. First, design changes are rarer than rate changes. Second, when they occur, they’re often bundled with rate changes, complicating identification. Third, in the case of wealth taxation, only a few countries levy such taxes and collect detailed microdata.

In Garbinti et al. (2024), we attempt to address these challenges by studying the effects of changes in information reporting requirements, a key element in tax design, on taxpayer behaviour. Specifically, we examine a unique French wealth tax reform that significantly relaxed reporting obligations. Before 2011, all wealth-tax filers had to report detailed asset information. A 2011 reform simplified reporting for taxpayers whose wealth was below €3 million (further reduced to €2.57 million in 2013), allowing them to report only total gross and net taxable wealth, without asset breakdowns.

This change drastically reduced the information provided to the tax authority, while leaving tax rates and thresholds largely unchanged. It allows us to cleanly identify the behavioural effects of reporting requirements, independent of other tax features, by leveraging rich panel data on the universe of French wealth taxpayers linked to their income tax returns.

We also develop a new empirical method: dynamic bunching. Rather than fitting smooth counterfactuals, we study discontinuities in the distribution of normalised wealth growth rates, comparing treated and control groups. Control distributions are constructed from unaffected taxpayers. This method is broadly applicable to any setting with panel data and measurable growth distributions.

Three main findings emerge.

First, reporting requirements have a first-order effect on taxpayer behaviour. There is sharp bunching at thresholds where reporting rules change – but no detectable bunching at pure marginal tax-rate kinks, even where tax rates jump.

Figure 1 Bunching at the €2.57 million simplification threshold

Notes: This figure shows the distribution of taxpayers by net taxable wealth around the simplification threshold of €2.57 million introduced in 2013. There is sharp and persistent bunching over time. Before 2013, when €2.57 million corresponds to a mere kink in the tax schedule (but not to information reporting notch), there is no bunching.

For example, estimated elasticities at the second and third bracket kinks are small and insignificant (0.08–0.14). In contrast, bunching is large and persistent at thresholds combining tax and information changes (e.g. exemption and simplification thresholds). This result highlights the central role of reporting in shaping responses and cautions against interpreting tax elasticity estimates from thresholds that combine rate and reporting changes.

Second, using our dynamic bunching method, we estimate heterogeneous behavioural responses to the 2013 simplification for taxpayers with wealth below €2.57 million. Using unaffected high-wealth taxpayers as a control, we construct counterfactual wealth growth distributions. Treated taxpayers reduced their reported wealth growth by 0.5 percentage points on average – about 20% of the pre-reform growth rate. These effects are driven by a small share of compliers (roughly 15%) who substantially reduced their reported growth (by 3 points or more).

Figure 2 Wealth growth rates decline as one approaches the simplification threshold

Notes: This figure shows the distribution of yearly wealth growth rates by wealth bins, pooled for different periods, around the simplification threshold (vertical solid line). The figure pools all years for the period during which the simplification threshold was €2.57 million (2013–2017). The shaded areas depict 95% confidence intervals.

Third, an inspection of the mechanisms behind taxpayer responses reveals that individuals who report lower wealth growth after the reform show no corresponding change in their third-party reported labour or capital income – providing strong evidence that the response reflects wealth misreporting or evasion, not real changes in savings or investment.

Overall, our findings indicate that taxpayers are more likely to under-report the value of less easily verifiable assets such as housing, whereas financial assets are more transparent and thus harder to manipulate.

By spurring tax evasion, the scaling back of reporting requirements had large negative effects on tax compliance. By 2017, 35% of wealth taxpayers are missing in the affected bracket. These individuals evade 10% of their total wealth tax payments each year. The magnitude of the behavioural responses prompted by the change in reporting requirements is noteworthy, given how trivial the simplification reform appeared to be.

This episode highlights that past wealth taxes often had major design flaws. These poor tax design choices had significant implications for tax enforcement. But if specific design choices can contribute to increasing tax elasticities, other choices can contribute to reducing them, such as mandating pre-populated returns, collecting and using information that is automatically transmitted by domestic and foreign third parties, or taxing non-residents.

Tax evasion is not a law of nature and can be increased or reduced through tax design choices. A full cost-benefit analysis of different elasticity-reducing design features of wealth taxes constitutes a fruitful avenue for future research.

Source : VOXeu