Inheritance Flows in France: Using Estate and Gift Tax Data to Measure Long-Term Trends
Abstract
Concerns about the return of inherited wealth emphasize the importance of accurately measuring inheritance flows. Estimates based on estate and gift tax data must deal with infrequent benchmark surveys, repeated changes in the tax treatment of inter vivos gifts, assets that are exempt from taxation, and transfers between spouses. This paper revisits the measurement of the annual flow of bequests and gifts in France, starting from the widely cited series of Piketty (2011) and its replication files. I document the step-by-step effect of each adjustment and propose alternative estimates. The U-shaped pattern over the twentieth century and the rise since the 1980s are confirmed. However, the level of the flow in the 2000s is lower, at about 8–9 percent of national income in 2006 rather than 11–13 percent, and its increase since the mid-1980s is more than 40 percent smaller. Most of the difference comes from the treatment of inter vivos gifts, whose recorded level in the benchmark years of the 1990s and 2000s coincided with a series of temporary tax incentives. Much of the rest comes from bequests to surviving spouses, which are transmitted again to children at the death of the surviving spouse and are therefore excluded, to avoid counting the same wealth twice.
JEL Codes: D31, E01, H24, N34
Keywords: inheritance, bequests, gifts, estate tax data, wealth, France
Introduction
The idea that inherited wealth has made a comeback in rich countries has become widely accepted. It rests to a large extent on the estimates of Piketty (2011), who found that the annual flow of bequests and gifts in France followed a pronounced U-shaped pattern over the twentieth century: about 20–25 percent of national income before World War I, less than 5 percent in the 1950s, and, “according to our latest data point (2008)”, close to 15 percent today. These estimates are central to Capital in the Twenty-First Century (Piketty 2014, chap. 11), to the long-run comparisons of Piketty and Zucman (2015), and to the estimates of the share of inherited wealth in aggregate wealth of Alvaredo, Garbinti, and Piketty (2017). They have also shaped the policy debate. In France, a note of the Conseil d’analyse économique states that the total flow of transfers “now represents more than 15 percent of GDP, or 300 billion euros” (Dherbécourt et al. 2021), and the international press has described a “return of inheritocracy” (The Economist 2025). Such concerns emphasize the importance of accurately measuring inheritance flows.
Measuring the flow of wealth transmitted each year is, however, difficult. The estimates of Piketty (2011) rely on what the paper calls the “exceptionally high quality of French estate tax data.” Compared with other countries, and for the historical record, French sources are indeed unusually rich: a universal estate tax applying to both bequests and gifts has existed since 1791, and nineteenth-century records are remarkably complete (Piketty, Postel-Vinay, and Rosenthal 2006). For the recent period, however, the data are much sparser, and much more difficult to interpret, than this description suggests. Since 1956, estates below a filing threshold no longer have to file a return, and in 2006 returns were filed for about two-thirds of decedents (Piketty 2010). Detailed information on the size and composition of bequests and gifts comes from only six benchmark samples of 3,000 to 5,000 estate tax returns each (the DMTG files), for 1977, 1984, 1987, 1994, 2000 and 2006, and no comparable benchmark has been published since. Gift flows are observed only in these benchmark years, so that the ratio of gifts to bequests—a key input of both the fiscal and the economic estimates—rests on a handful of data points. Tax rules for inter vivos gifts were changed repeatedly over this period, often with temporary incentives designed precisely to bring transfers forward. A growing share of wealth, most notably life insurance, is legally exempt from estate taxation and must be imputed. And part of what is recorded as inheritance consists of transfers between spouses rather than between generations. Each of these features requires adjustments to raw tax data, and the cumulative effect of these adjustments is large: in the headline series, the adjusted flow for 2006 is about three times the raw bequest flow reported to the tax administration.
This paper revisits these adjustments. I start from the published series and replication files of Piketty (2011), which remain the reference estimates and are still widely cited, and which I reproduce exactly. I then document the step-by-step effect of each adjustment, propose alternative estimates where the assumptions can be checked against other sources, and provide a sensitivity analysis for those that cannot. The approach follows the one used by Auten and Splinter (2024) for U.S. income inequality: starting from seminal tax-based estimates, it makes each adjustment explicit so that other researchers can see its effect and consider alternative combinations of assumptions.
The main results are summarized in Figure 1. The broad historical pattern documented by Piketty (2011) is not in question: the inheritance flow was low in the 1950s and has risen since. The level and the slope of the recent rise are, however, sensitive to a few assumptions. In 2006, the last year for which benchmark tax data exist, the alternative estimates put the inheritance flow at 7.7% of national income using the fiscal approach and 8.9% using the economic approach, compared with 11.0% and 12.6% in Piketty (2011). Between 1984 and 2006, the fiscal flow increases by 3.0 percentage points instead of 5.3, and the economic flow by 3.8 points instead of 6.6. For 2008, the year usually quoted, the alternative economic flow is 10.3% rather than 14.5%.
Why do the estimates differ? About two-thirds of the difference in the 2006 fiscal flow, and four-fifths of the difference in the economic flow, come from the treatment of inter vivos gifts. Three issues stand out. First, the gift flow attributed to 2006 includes an annual average of transfers made under a temporary cash-gift exemption, which, according to the Conseil des prélèvements obligatoires (Conseil des prélèvements obligatoires 2008), ran from June 2004 to December 2005. Second, the benchmark years used to measure the gift-bequest ratio—1994, 2000 and 2006—all fall within periods of new or temporary tax incentives for gifts, starting with the 1992 reform that exempted gifts made more than ten years before death from the “recall” rule. Gifts made in response to such incentives are, to a large extent, bequests brought forward in time, so that a ratio observed in these years may overstate the steady-state importance of gifts. Third, both the fiscal and the economic approaches upgrade recorded gifts by the same factor as recorded bequests, including for assets such as life insurance that are exempt from estate taxation but are not transmitted through gifts. The rest of the difference comes from two adjustments to bequests. The first, in the fiscal flow only, is the imputation of tax-exempt housing wealth. The second is the exclusion of bequests to surviving spouses. A large part of what a surviving spouse receives is transmitted again to children at his or her death, so that including these bequests counts the same wealth twice within a few years. Both sides of the classic Kotlikoff–Summers–Modigliani controversy on the importance of inheritance measured intergenerational transfers only (Kotlikoff and Summers 1981; Modigliani 1988), and the preferred estimates of this paper exclude spouses as well, using the 10 percent share assumed by Piketty (2011).
These differences are not merely differences in opinion. Each alternative assumption relies on information that is documented in official sources: the dates of the cash-gift scheme, the legal scope of the housing exemptions, the way gifts are recorded by notaries and the tax administration, and the share of bequests received by surviving spouses. Several of the issues are also acknowledged in Piketty (2010), whose data appendix notes that “gift flows are structurally more volatile than bequest flows, and one must be careful when using non-annual gift series,” and whose simulations suggest “a little bit of overshooting in the rise of gifts since the 1980s (possibly due to tax incentives).” Not all revisions go in the same direction. The adjustment for life insurance, the largest tax-exempt asset, appears conservative when compared with death benefits paid by insurers, and the economic flow is not affected by the alternative treatment of housing.
Two other observations help put the estimates in perspective. First, the bequest component of the inheritance flow is relatively well measured and has risen much less than the total. In Piketty (2011)’s own series, the economic bequest flow—excluding gifts—was 5.0% of national income in 1970, 4.6% in 1984 and 5.2% in 2000; it rose to 6.9% in 2006 and 8.0% in 2008 with the boom in house prices. Recent survey-based estimates of wealth at death are of the same order of magnitude (Dherbécourt 2017; Conseil supérieur du notariat 2026). Most of the increase in the headline series between the 1980s and the 2000s, and most of the uncertainty about its level, therefore concern gifts. Second, the “fiscal” and “economic” series are described in Piketty (2011) as “two fully independent estimates of the inheritance flow,” and their agreement is presented as evidence of robustness. In practice, the two series share the same gift-bequest ratio from gift tax data, and the economic series relies on age-wealth profiles from estate tax returns. Their agreement on the trend therefore does not, by itself, validate the trend in gifts.
This paper contributes to a literature that measures inheritance flows with tax data and national accounts, following Piketty (2011) for France, and applied to Germany (Schinke 2012), Italy (Acciari and Morelli 2021), and other countries (Piketty and Zucman 2015). Its contribution is to make each adjustment explicit for the French case, to quantify its contribution to the level and trend of the inheritance flow, and to propose alternative assumptions grounded in administrative sources. Because the share of inherited wealth in aggregate wealth estimated by Alvaredo, Garbinti, and Piketty (2017) cumulates past inheritance flows, the results also bear on that measure, which is discussed in a companion paper (Geerolf 2026).
The paper is organized as follows. Section I describes the two measures of the inheritance flow. Section II presents the data. Section III decomposes the headline estimate into its components. Sections IV and V discuss gifts and bequests, respectively. Section VI turns to the economic flow. Section VII presents the alternative estimates, Section VIII the sensitivity analysis, and Section IX compares the results with other sources. Section X concludes.
I. Measuring the Inheritance Flow: Two Approaches
Piketty (2011) defines the inheritance flow \(B_t\) as the total market value of wealth transmitted during year \(t\), either at death (bequests) or inter vivos (gifts), and measures it in two ways.
The fiscal flow starts from the raw bequest flow reported on estate tax returns, \(B^{f0}_t\), and the raw gift flow reported on gift tax returns, \(V^{f0}_t\). Both are upgraded to account for non-filers (estates below the filing threshold) and for tax-exempt assets: \[ B^f_t = \left(B^{f0}_t + V^{f0}_t\right) \, u^{nf}_t \, u^{te}_t, \qquad u^{te}_t = \frac{1}{1 - e_t}, \] where \(u^{nf}_t\) is the upgrade factor for non-filers and \(e_t\) is the estimated share of tax-exempt assets in total wealth. Equivalently, \(B^f_t = B^{f0}_t\, u^{nf}_t\, u^{te}_t \,(1+v_t)\), where \(v_t = V^{f0}_t / B^{f0}_t\) is the raw gift-bequest ratio.
The economic flow starts from the accounting identity \[ \frac{B_t}{Y_t} = \mu^*_t\, m_t\, \frac{W_t}{Y_t}, \qquad \mu^*_t = (1+v_t)\,\mu_t, \] where \(W_t/Y_t = \beta_t\) is the ratio of aggregate private wealth to national income from national balance sheets, \(m_t\) is the adult mortality rate, and \(\mu_t\) is the ratio between the average wealth of decedents and the average wealth of the living (adults aged 20 and over). The ratio \(\mu_t\) is computed from age-wealth profiles of decedents, which are estimated from estate tax returns and corrected for differential mortality. The factor \((1+v_t)\) adds gifts, using the same raw gift-bequest ratio as in the fiscal flow: as Piketty (2011) puts it, the corrected equation “simply uses the observed, fiscal gift-bequest ratio during year \(t\) and upgrades the economic inheritance flow accordingly.”
The two approaches differ in how they measure bequests: the fiscal flow uses reported estates plus imputations, while the economic flow uses aggregate wealth from national accounts. They share, however, the treatment of gifts. In the economic flow, gifts are \(v_t\) times the economic bequest flow \(\mu_t m_t W_t\); since the economic bequest flow is itself much larger than the raw bequest flow, recorded gifts are implicitly scaled up by the same factor. Figure 2 shows both series with and without gifts. Without gifts, the two measures of the bequest flow are close to each other and relatively flat between the mid-1960s and the late 1990s, at about 4–5 percent of national income. The difference between the bequest-only series and the total series, which is common to both approaches, accounts for most of the rise between the 1980s and the 2000s.
II. Data
All series in this paper are computed from the data appendix and replication files of Piketty (2010, 2011): national income and private wealth (Tables A1 and A3), the composition of private wealth (Table A15b), the fiscal flow and its components (Tables B1 and B2), and the ratios \(\mu_t\) and \(\mu^*_t\) (Table B5). The code reproduces the published fiscal and economic flows exactly. Where alternative assumptions require additional information, I use published official tabulations, entered by hand and documented in the code.
For the period since 1977, estate and gift tax information comes from the DMTG files (“droits de mutation à titre gratuit”), nationally representative samples of estate tax returns compiled by the Finance Ministry for 1977, 1984, 1987, 1994, 2000 and 2006. Each file contains 3,000 to 5,000 estate tax returns, out of about 300,000 returns filed annually, with heavier sampling at the top (Piketty 2010). Gift flows for 1984–2000 come from the same files; the 2006 gift flow comes from the tabulations published by the Conseil des prélèvements obligatoires (2008). Between benchmark years, Piketty (2011) interpolates the gift-bequest ratio linearly, and after 2006 holds it constant. No comparable benchmark has been published since 2006: the Dherbécourt et al. (2021) note that the tax administration has not produced exploitable information on bequests since then, and that the representativeness of a 2010 sample has not been established.
I also use the 2006 tabulations of the Conseil des prélèvements obligatoires (2008), based on the 2006 DMTG survey, which break down the bequest flow by type of heir (Tableau 16), report the regular gift flow (Tableau 13), and estimate the value of estates below the filing threshold (Tableau 6). For comparison purposes, Section IX uses the estimates of Laferrère (1990) for 1977, of Dherbécourt (2017) for 2015, and the estimates of wealth at death for 2018 reported by the Conseil supérieur du notariat (2026).
III. From Raw Tax Data to the Headline Estimates
Table 1 decomposes the 2006 fiscal flow into its components. Reported bequests amounted to 58.85 billion euros, or 3.7% of national income, and recorded gifts to 48.0 billion, or 3.0%. Before any upgrade, the total is therefore 6.8% of national income. The correction for non-filers adds 0.5 percentage points. The correction for tax-exempt assets, which multiplies both bequests and gifts by 1.503, adds 3.7 points, of which 1.6 points on gifts. The result is the published fiscal flow of 11.0% of national income.
| Component | Billion euros | % national income |
|---|---|---|
| Raw bequest flow (estate tax returns) | 58.9 | 3.73 |
| Raw gift flow (gift tax returns) | 48.0 | 3.04 |
| of which: regular gifts | 39.4 | 2.49 |
| of which: 1/3 of 2004–2005 cash-gift scheme | 8.6 | 0.55 |
| Raw bequests + gifts | 106.8 | 6.77 |
| + Upgrade for non-filers, bequests | 4.6 | 0.29 |
| + Upgrade for non-filers, gifts | 3.8 | 0.24 |
| + Upgrade for tax-exempt assets, bequests | 31.9 | 2.02 |
| + Upgrade for tax-exempt assets, gifts | 26.1 | 1.65 |
| Fiscal inheritance flow | 173.3 | 10.97 |
Notes: National income in 2006: 1,579.2 billion euros. Upgrade factors: non-filers 1.079, tax-exempt assets 1.503. Sources: Piketty (2011), Table B1; CPO (2008) for the cash-gift scheme.
Two further steps lead to the figure of “close to 15 percent” usually quoted. Since there are no tax data after 2006, the 2008 fiscal flow (12.6%) is extrapolated from 2006 using the growth of the economic flow. The economic flow itself rises from 12.6% in 2006 to 14.5% in 2008, with the ratio of private wealth to national income increasing from 5.10 to 5.63. The 2008 data point is therefore an extrapolation of 2006 tax data, combined with the wealth-income ratio at the peak of the housing cycle.
IV. Gifts
A. The Tax Treatment of Gifts
Since 1942, the general principle of French transfer taxation has been that bequests and gifts are taxed under the same schedule, and that gifts are “recalled” at death and added to the estate, so that the total tax paid does not depend on the timing of transfers (Piketty 2010). In practice, gifts have always enjoyed some advantages, and these advantages were considerably reinforced from the 1990s onward. Table 2 lists the main changes, as described in Piketty (2010), the Conseil des prélèvements obligatoires (2008) and Goupille-Lebret (2016).
| Period | Measure |
|---|---|
| 1942 | Bequest and gift taxes unified; all gifts recalled at death |
| 1986 | 25% rebate for donations-partages reintroduced for donors aged under 65 |
| 1992 | “Ten-year rule”: gifts made more than ten years before a new gift or death are no longer recalled, so that personal allowances can be used again |
| 1996–1998 | Gift tax rebates extended to all gifts by donors under 65 |
| 1998–2001 | General 30% rebate on gift tax |
| 2003–2005 | General 50% rebate on gift tax |
| June 2004–Dec. 2005 | Temporary exemption of cash gifts to descendants (up to €20,000, then €30,000); €25.8 billion transferred in 18 months |
| 2006 | Recall period reduced from ten to six years |
| 2007 | TEPA law: higher allowances; exemption of spouses from inheritance tax; permanent exemption for family cash gifts |
| 2011–2012 | Recall period raised to ten, then fifteen years; allowances reduced |
The benchmark years used to measure the gift-bequest ratio fall within this sequence of reforms. The 1994 benchmark comes two years after the introduction of the ten-year rule, which the press described at the time as a reform under which “the tax regime for gifts is markedly favored” (Le Monde 1991, my translation). The 2000 benchmark falls within the period of general tax rebates on gifts. The 2006 benchmark follows the temporary cash-gift scheme and coincides with the shortening of the recall period to six years. Figure 3 shows the resulting gift-bequest ratio, which rises from 29% in 1984 to 64% in 1994, 81% in 2000 and 82% in 2006.
B. The 2006 Benchmark
For 2006, Piketty (2010, fn. 188) computes the raw gift flow as the sum of the regular gift flow reported by the Conseil des prélèvements obligatoires (2008), 39.4 billion euros, and “the average yearly 2004–2006 flow under the special cash-gifts regime (8.6 billions)”. The data appendix notes that excluding this extra flow would lower the gift-bequest ratio from 82 percent to 67 percent, which “would still be high by historical standards,” but concludes that “it seems more justified to look at the full flow in 2006.”
According to the Conseil des prélèvements obligatoires (2008), the exemption applied to cash gifts made between 1 June 2004 and 31 December 2005, for a total of 25.84 billion euros over eighteen months. No gifts were made under this scheme in 2006. Adding one third of the total to the 2006 flow therefore treats a temporary, tax-motivated flow that occurred in 2004–2005 as part of the steady-state flow observed in 2006, and propagates it, through the constant gift-bequest ratio, to 2007 and 2008. The alternative estimates exclude it. This lowers the 2006 fiscal flow from 11.0% to 10.1% of national income, and the economic flow from 12.6% to 11.6% (Table 4).
One could argue that the 2004–2005 flow should instead be spread over 2004 and 2005, or over a longer window. Whatever the allocation, these transfers did not take place in 2006, and the evidence discussed below suggests that much of the flow was brought forward from later years. The Conseil des prélèvements obligatoires (2008) itself distinguishes three possible effects of the scheme: a pure windfall effect for transfers that would have taken place anyway, a pure incentive effect for transfers that would not have taken place, and an anticipation effect for transfers that would have taken place later.
C. Gifts as Anticipated Bequests
The broader question is whether the gift-bequest ratios observed in 1994, 2000 and 2006 measure a permanent change in the way wealth is transmitted, or partly a temporary shift in the timing of transfers. If tax incentives lead parents to give earlier wealth that they would otherwise have left at death, the gift flow rises temporarily while the bequest flow declines only gradually, as donor cohorts die. During the transition, the sum of gifts and bequests exceeds its steady-state level. In the extreme case of a one-time anticipation, the cumulated flow over the life cycle is unchanged, and the observed rise in the annual flow is entirely transitory.
Piketty (2011) argues that the rise in gifts began before the new tax incentives and attributes it mainly to rising life expectancy. The data appendix nevertheless recognizes that “it is plausible that the increased tax advantages given to gifts in the 1990s–2000s did contribute to the recent rise of the gift-bequest ratio” and that what matters for the inheritance flow is “whether there has been some kind of ‘overshooting’ of gifts in the recent past” (Piketty 2010). Its simulations conclude that there was “a little bit of overshooting in the rise of gifts since the 1980s (possibly due to tax incentives), and that a (small) fraction of the observed gift level is not sustainable.” The benchmark data cannot separate these explanations, since the ratio is observed only in years when incentives were in place. Evidence from French tax reforms also indicates that the timing of gifts responds to their tax treatment (Goupille-Lebret and Infante 2018).
Given this uncertainty, the alternative estimates do not attempt to correct the 1994 and 2000 benchmarks. The sensitivity analysis in Section VIII reports, as a bound, the effect of holding the gift-bequest ratio at its 1987 level, the last benchmark before the 1992 reform.
D. Upgrading Recorded Gifts
The fiscal flow applies the same non-filer and tax-exempt upgrades to gifts as to bequests, “leaving the gift-bequest ratio \(v_t\) unaffected” (Piketty 2011). In 2006, the tax-exempt upgrade adds 26.1 billion euros to recorded gifts (Table 1). The rationale is that gifts are probably less well reported than bequests.
The tax-exempt upgrade is designed for assets that are exempt from estate taxation and therefore “generally not reported on estate tax returns”, mostly life insurance, but also part of housing and business assets (Piketty 2010, Appendix B). It is not clear that it applies to gifts in the same way. Life insurance, which accounts for about 40 percent of the imputed tax-exempt assets in the 2000s, is transmitted at death through beneficiary clauses rather than through gifts. Gifts of real estate and donations-partages require a notarial deed, which records the full market value of the assets, and gifts below the tax-free allowances are recorded as well. The gift statistics of the Conseil des prélèvements obligatoires (2008) report the value of all recorded gifts, whether taxed or not. Business assets transmitted under the “Dutreil” regime benefit from a partial exemption of their taxable value, but are recorded in the deed.
The same issue arises, less visibly, in the economic flow. Because gifts are computed as \(v_t\) times the economic bequest flow, recorded gifts are implicitly multiplied by the ratio of the economic bequest flow to the raw bequest flow, which is 1.86 in 2006. In other words, the economic flow assumes that true gifts in 2006 were 89 billion euros, compared with 48.0 billion recorded.
The alternative estimates keep the non-filer upgrade for gifts, as a proxy for unreported manual gifts, but do not apply the tax-exempt upgrade, in either the fiscal or the economic flow. In the economic flow, gifts are therefore measured as recorded gifts (with the non-filer upgrade) rather than as \(v_t\) times economic bequests. This lowers the 2006 fiscal flow by a further 1.4 percentage points and the economic flow by 2.0 points. Because the gift-bequest ratio rises over time, the effect is larger in recent years than in the 1980s, and it reduces the trend as well as the level.
V. Bequests
A. Tax-Exempt Assets
The correction for tax-exempt assets is the largest single adjustment in the fiscal flow. Piketty (2010, Appendix B) estimates the share of tax-exempt assets by applying assumed exemption rates to the composition of aggregate private wealth from national balance sheets: 20 percent for housing, 30 percent for non-housing tangible assets, 10 percent for listed equity and other financial assets, 50 percent for unlisted equity, and 95 percent for life insurance. The resulting share rises from about 24 percent in the 1970s to about 34 percent in the 2000s, mostly because of the growth of life insurance. Figure 4 shows the contribution of each asset class.
Housing. Housing accounts for 38% of imputed tax-exempt assets in 2006. The 20 percent rate is justified by two provisions (Piketty 2010, fn. 171). The first is the 20 percent rebate on the value of the principal residence. Under Article 764 bis of the tax code, this rebate applies only when the home is also occupied, at the date of death, as their principal residence by the surviving spouse or civil partner, or by minor or protected children of the decedent. It therefore applies to part of the housing wealth of decedents, not to all of it, and when it applies it reduces reported values by 20 percent. The second is the exemption granted in 1947 for the first transmission of buildings constructed between 1947 and 1973. By construction, this exemption covered a declining share of housing wealth after the 1970s. Taken together, these provisions suggest that the effective exempt share of housing wealth was below 20 percent in the 1990s and 2000s. The alternative estimates use 10 percent, which is equivalent to assuming that the principal-residence rebate applies to half of the housing wealth of decedents; Section VIII reports results for 5 and 20 percent.
Life insurance. Life insurance accounts for 40% of imputed tax-exempt assets in 2006. The upgrade implies that about 13 billion euros of life insurance were transmitted at death in 2006, on top of reported estates. This appears conservative. According to data from the French prudential authority quoted by the Conseil supérieur du notariat (2026), death-related benefits paid by life insurers amounted to 31–33 billion euros a year in 2011–2014, and to about 45 billion euros in 2023. These benefits include payments to surviving spouses and are not strictly comparable, but they suggest that the life insurance component of the adjustment is, if anything, too low. The alternative estimates leave it unchanged.
Other assets. The 10 percent exemption rate on listed equity and other financial assets reflects historical exemptions for specific government bonds (Piketty 2010, fn. 173). The sensitivity analysis sets it to zero. The exemption rates for business assets are left unchanged: business assets have benefited from increasingly generous exemptions since the 1990s, and the assumed rates may be too low.
B. Non-Filers
The upgrade for non-filers, which reflects estates below the filing threshold, is 1.079 in 2006. The Conseil des prélèvements obligatoires (2008, Tableau 6) estimates that non-declared estates amounted to 3.07 billion euros in 2006, implying a factor of 1.052. The difference is small, and the alternative estimates use the factors of Piketty (2011); the CPO estimate is used in the sensitivity analysis.
C. Surviving Spouses
Piketty (2011) includes transfers to surviving spouses in the inheritance flow, noting that the spouse share “has always been relatively small (about 10%)”. This choice is consistent with the definition of the inheritance flow as all wealth transmitted during a year. For questions about intergenerational transmission, however, transfers to spouses are not inheritance in the usual sense: the same wealth is typically transmitted again to children at the death of the surviving spouse, and is then counted a second time. In 2006, the DMTG tabulations published by the Conseil des prélèvements obligatoires (2008, Tableau 16) attribute 10.55 billion euros out of 58.85 billion to surviving spouses, a share of 17.9%; direct-line heirs received 39.41 billion. The preferred alternative estimates therefore exclude them. This is standard in the literature on the importance of inheritance: Kotlikoff and Summers (1981) keep only bequests to children and other “distant in age” relatives, and Modigliani (1988) subtracts transfers between husband and wife from the bequest flow. The Dherbécourt et al. (2021) likewise exclude spousal transfers when measuring the inheritances received by individuals over their lifetime. I use the 10 percent share assumed by Piketty (2011), which is conservative in light of the CPO figure for 2006, and apply it to bequests only, since gifts between spouses are rare. The sensitivity analysis reports results with the CPO share and with spouses included.
VI. The Economic Flow
The economic flow does not use the raw bequest flow and is therefore not affected by the upgrades for non-filers and tax-exempt assets. It is, however, affected by the treatment of gifts, through the factor \((1+v_t)\), and by the measurement of \(\mu_t\) and \(\beta_t\). Table 3 decomposes its increase between 1984 and 2006, and between 1984 and 2008, into the contributions of each factor.
| Factor | 1984 | 2006 | 2008 | Log change 1984-2006 | Log change 1984-2008 |
|---|---|---|---|---|---|
| Wealth-income ratio, beta | 3.02 | 5.10 | 5.63 | 52.4 | 62.3 |
| Gift factor, 1 + v | 1.288 | 1.816 | 1.816 | 34.3 | 34.3 |
| Relative wealth of decedents, mu | 1.121 | 1.227 | 1.226 | 9.0 | 8.9 |
| Mortality rate, m | 1.37% | 1.11% | 1.16% | -20.8 | -16.4 |
| Economic flow, b = mu (1+v) m beta | 6.0% | 12.6% | 14.5% | 74.9 | 89.1 |
Notes: Log changes multiplied by 100; contributions add up to the change in the economic flow. Sources: Piketty (2011), Tables A3 and B5; author’s calculations.
Between 1984 and 2008, the economic flow increases by 89 log points. The wealth-income ratio contributes 62 log points and the gift factor 34, while the decline in mortality subtracts 16 points and the relative wealth of decedents adds 9. Two implications follow.
First, about two-fifths of the increase comes from the gift factor, which is measured with the gift tax data discussed in Section IV. The fiscal and economic series therefore share the second most important source of their growth. The relative wealth of decedents \(\mu_t\) also comes from estate tax returns, through the age-wealth profiles of decedents. As noted by Piketty (2010), these profiles are themselves affected by gifts, since “when we observe wealth at death, or wealth among the elderly, we are actually observing the wealth of individuals who have already given away almost half of their wealth.”
Second, the largest contribution comes from the wealth-income ratio, which rises from 3.02 in 1984 to 5.63 in 2008, with most of the increase after 2000. Over 2000–2008, the share of housing in gross private wealth rises from 51% to 65%, reflecting the increase in house prices. Measuring transfers at market value is the right convention for an inheritance flow. But it implies that the inheritance flow relative to income rises mechanically with asset prices, even when the same assets are transmitted, and that the 2008 figure reflects the peak of the housing cycle. In 2008, the economic flow excluding gifts was 8.0% of national income, compared with 5.2% in 2000.
VII. Alternative Estimates
Table 4 shows the step-by-step effect of the four adjustments retained in the alternative estimates, for each benchmark year since 1984: (1) excluding the 2004–2005 cash-gift scheme from the 2006 gift flow; (2) not applying the tax-exempt upgrade to recorded gifts; (3) a 10 percent rather than 20 percent exemption rate for housing; (4) excluding bequests to surviving spouses (10 percent of bequests). Adjustments are cumulative. Figure 1 shows the corresponding annual economic flow, in which the ratio of gifts to economic bequests is interpolated between benchmark years and held constant after 2006, as in Piketty (2011).
| Estimate | 1984 | 1987 | 1994 | 2000 | 2006 | Change 1984-2006 |
|---|---|---|---|---|---|---|
| Fiscal flow | ||||||
| Piketty (2011) | 5.7 | 5.6 | 6.7 | 9.0 | 11.0 | 5.3 |
| (1) Excluding 2004-05 cash-gift scheme | 5.7 | 5.6 | 6.7 | 9.0 | 10.1 | 4.4 |
| (2) + No tax-exempt upgrade on gifts | 5.4 | 5.2 | 6.0 | 7.6 | 8.7 | 3.4 |
| (3) + Housing exemption 10% | 5.1 | 5.0 | 5.7 | 7.2 | 8.2 | 3.1 |
| (4) + Excluding surviving spouses (this paper) | 4.7 | 4.6 | 5.3 | 6.8 | 7.7 | 3.0 |
| Economic flow | ||||||
| Piketty (2011) | 6.0 | 6.9 | 7.4 | 9.4 | 12.6 | 6.6 |
| (1) Excluding 2004-05 cash-gift scheme | 6.0 | 6.9 | 7.4 | 9.4 | 11.6 | 5.6 |
| (2) + Recorded gifts, no tax-exempt upgrade | 5.6 | 6.1 | 6.4 | 7.8 | 9.6 | 4.0 |
| (3) + Housing exemption 10% | 5.6 | 6.1 | 6.4 | 7.8 | 9.6 | 4.0 |
| (4) + Excluding surviving spouses (this paper) | 5.1 | 5.6 | 5.9 | 7.3 | 8.9 | 3.8 |
| Memo: bequests only | ||||||
| Fiscal, Piketty (2011) | 4.4 | 3.9 | 4.1 | 5.0 | 6.0 | 1.6 |
| Fiscal, this paper | 3.7 | 3.3 | 3.4 | 4.2 | 5.0 | 1.3 |
| Economic, Piketty (2011) | 4.6 | 4.8 | 4.5 | 5.2 | 6.9 | 2.3 |
Notes: Bequests plus gifts as a percentage of national income, unless otherwise noted. The economic flow is not affected by adjustment (3). Sources: Piketty (2011) and author’s calculations, see text.
In the fiscal series, the four adjustments lower the 2006 inheritance flow from 11.0% to 7.7% of national income. The adjustments also lower earlier years, but by less: the 1984 flow goes from 5.7% to 4.7%. In the economic series, the 2006 flow goes from 12.6% to 8.9%, and the 2008 flow from 14.5% to 10.3%. In both series, the increase between 1984 and 2006 is more than 40 percent smaller than in Piketty (2011). The alternative fiscal and economic series remain close to each other, with a gap that is similar to the one in the original series.
Over the longer run, the alternative estimates do not change the broad pattern documented by Piketty (2011): the inheritance flow was at its lowest in the 1950s, and it is substantially higher in the 2000s than in the 1960s–1970s. In the alternative economic series, it rises from 5.2% of national income in 1970 to 8.9% in 2006, compared with 6.4% and 12.6% in the original series. What changes is the magnitude of the recent increase and the level reached at the end of the period.
VIII. Sensitivity Analysis
Table 5 reports the 2006 inheritance flow and its change since 1984 under alternative assumptions. Each line changes one assumption relative to the alternative estimates of Table 4.
| Assumption | Fiscal 2006 | Fiscal change | Economic 2006 | Economic change |
|---|---|---|---|---|
| Piketty (2011) | 11.0 | 5.3 | 12.6 | 6.6 |
| This paper | 7.7 | 3.0 | 8.9 | 3.8 |
| Housing exemption 5% | 7.4 | 2.9 | 8.9 | 3.8 |
| Housing exemption 20% (as in Piketty, 2011) | 8.1 | 3.2 | 8.9 | 3.8 |
| No exemption for listed equity and other financial assets | 7.5 | 3.0 | 8.9 | 3.8 |
| Non-filers in 2006 as estimated by CPO (2008) | 7.5 | 2.8 | 8.9 | 3.7 |
| Including the 2004-05 cash-gift scheme in 2006 | 8.2 | 3.6 | 9.5 | 4.4 |
| Gift-bequest ratio held at 1987 level after 1987 | 6.7 | 2.0 | 8.0 | 2.9 |
| Including surviving spouses (as in Piketty, 2011) | 8.2 | 3.1 | 9.6 | 4.0 |
| Excluding surviving spouses (17.9% in 2006, CPO 2008) | 7.2 | 2.5 | 8.4 | 3.3 |
Notes: “Change” is the change between 1984 and 2006 in percentage points. Each line from the third onward changes one assumption relative to “This paper”. The last line uses the 10 percent spouse share assumed by Piketty (2011) before 2006 and the share reported by the CPO for 2006. Sources: author’s calculations.
Three results stand out. First, assumptions about tax-exempt bequests mostly affect the level of the fiscal flow, not its trend: moving the housing exemption between 5 and 20 percent changes the 2006 fiscal flow by less than one percentage point and the 1984–2006 increase by a few tenths of a point. Second, assumptions about gifts matter most. Holding the gift-bequest ratio at its 1987 level, the last benchmark before the 1992 reform, lowers the 2006 economic flow to 8.0 percent of national income and reduces the 1984–2006 increase to less than half of that in Piketty (2011). This is a bound rather than a preferred estimate, since part of the rise in gifts probably reflects longer life expectancy rather than tax incentives. Third, including bequests to surviving spouses, as in Piketty (2011), raises the 2006 estimates by 0.6 percentage point for the fiscal flow and 0.7 for the economic flow, while using the spouse share reported by the Conseil des prélèvements obligatoires (2008) for 2006 lowers them further. Across all lines of Table 5 other than the first, the 2006 inheritance flow ranges from 7 to 10 percent of national income, compared with 11 to 13 percent in Piketty (2011).
IX. Comparison with Other Estimates
Other sources provide partial checks on these estimates. They do not cover the same concepts or years, and none of them is free of measurement issues, but together they help assess which components are well measured.
Gifts in the 1970s. Using the 1977 DMTG file, Laferrère (1990, 5) estimates gift and bequest flows that correspond to a gift-bequest ratio of 49%, compared with the 28% interpolated by Piketty (2011), who considered this figure inconsistent with gift tax receipts (Piketty 2010, fn. 188). If the 1977 ratio were closer to Laferrère’s estimate, the inheritance flow of the 1970s would be higher, and the subsequent increase smaller. Using this ratio, the 1977 fiscal flow would be 5.4% of national income instead of 4.6%.
Bequests in the 2010s. Dherbécourt (2017) applies the economic approach to 2015, with aggregate household wealth of 10,575 billion euros, an adult mortality rate of 1.19 percent and a ratio \(\mu\) of 122 percent, and obtains a bequest flow of 153 billion euros; adding gifts with a gift-bequest ratio of 0.63 yields about 250 billion euros. Using a microsimulation model based on the Insee wealth survey, reported by the Cour des comptes (2024), the wealth of persons who died in 2018 is estimated at 159.3 billion euros, an average of about 261,000 euros per death (Conseil supérieur du notariat 2026). The two estimates of the bequest component are close. The much larger totals that appear in public debate—300 billion euros in Dherbécourt et al. (2021), and up to 400 billion euros for 2024 according to press reports discussed by the Conseil supérieur du notariat (2026)—result from the addition of gifts and other components to a bequest flow of about 150–160 billion. As the notaries’ report puts it, “any result […] is based on a pile of assumptions, since few statistics are available” (Conseil supérieur du notariat 2026, my translation).
Other countries. Studies applying the same framework find inheritance flows of a similar order of magnitude in other European countries: over 10 percent of national income in Germany in recent years (Schinke 2012), and an increase from 8.4 to 15.1 percent of national income in Italy between 1995 and 2016 (Acciari and Morelli 2021). Because these estimates rely on the same accounting framework and face similar measurement issues for gifts and tax-exempt assets, they are not independent checks on the French estimates. They underline, however, the value of making the adjustments explicit.
X. Summary and Conclusions
Estimates of the inheritance flow based on tax data require adjustments for gifts, tax-exempt assets, non-filers and spouses. For France, these adjustments roughly triple the raw bequest flow reported to the tax administration, and their cumulative effect determines both the level of the inheritance flow and much of its recent trend. This paper has made each adjustment explicit, starting from the seminal estimates of Piketty (2011).
The alternative estimates confirm the U-shaped pattern of the inheritance flow and its increase since the 1950s. They suggest, however, that the flow reached about 8–9 percent of national income in 2006, rather than 11–13 percent, and that its increase since the mid-1980s was more than 40 percent smaller than in the original series. The bequest component, which is the best measured, rose relatively little until the housing boom of the 2000s. Most of the difference comes from the treatment of inter vivos gifts, whose measured importance in the benchmark years of the 1990s and 2000s coincided with a series of temporary tax incentives, and which both the fiscal and economic approaches scale up by the same factor as bequests. Much of the rest comes from the exclusion of bequests to surviving spouses, which would otherwise be counted twice when they are transmitted again to children.
There is substantial uncertainty about all of these estimates, including the ones proposed here. No benchmark data have been published since 2006, and the most widely quoted figures for recent years are extrapolations. Annual statistics on bequests and gifts, built from the declarations already filed by notaries and insurers, as recommended by Dherbécourt et al. (2021), would allow these questions to be settled with data rather than assumptions. In the meantime, the sensitivity of the headline numbers to a small number of assumptions suggests caution in using them as precise measures of the importance of inheritance in France.
References
Acciari, Paolo, and Salvatore Morelli. 2021. “Wealth Transfers and Net Wealth at Death: Evidence from the Italian Inheritance Tax Records 1995–2016.” Working Paper 27899. National Bureau of Economic Research.
Auten, Gerald, and David Splinter. 2024. “Income Inequality in the United States: Using Tax Data to Measure Long-Term Trends.” Journal of Political Economy 132 (7). https://doi.org/10.1086/728741.
Conseil des prélèvements obligatoires. 2008. “La répartition des prélèvements obligatoires entre générations et la question de l’équité intergénérationnelle.” Paris: Cour des comptes.
Conseil supérieur du notariat. 2026. “Étude « Grande transmission ».” Paris: Conseil supérieur du notariat.
Cour des comptes. 2024. “Les droits de succession.” Paris: Cour des comptes.
Dherbécourt, Clément. 2017. “Peut-on éviter une société d’héritiers ?” Note d'analyse 51. France Stratégie.
Dherbécourt, Clément, Gabrielle Fack, Camille Landais, and Stefanie Stantcheva. 2021. “Repenser l’héritage.” Les notes du conseil d'analyse économique 69. Conseil d’analyse économique.
Goupille-Lebret, Jonathan. 2016. “Combien ont coûté les réformes de l’impôt sur les successions mises en place en France depuis 2000 ?” Revue économique 67 (4): 913–36.
Goupille-Lebret, Jonathan, and Jose Infante. 2018. “Behavioral Responses to Inheritance and Gift Taxation: Evidence from France.” Journal of Public Economics 168: 21–34.
Kotlikoff, Laurence J., and Lawrence H. Summers. 1981. “The Role of Intergenerational Transfers in Aggregate Capital Accumulation.” Journal of Political Economy 89 (4): 706–32.
Laferrère, Anne. 1990. “Successions et héritiers.” Paris: INSEE.
Le Monde. 1991. “Les dispositions fiscales du projet de loi de finances pour 1992 : le régime des donations nettement favorisé.”
Modigliani, Franco. 1988. “The Role of Intergenerational Transfers and Life Cycle Saving in the Accumulation of Wealth.” Journal of Economic Perspectives 2 (2): 15–40.
Piketty, Thomas. 2010. “On the Long-Run Evolution of Inheritance: France 1820–2050.” Working Paper. Paris School of Economics.
———. 2011. “On the Long-Run Evolution of Inheritance: France 1820–2050.” The Quarterly Journal of Economics 126 (3): 1071–131. https://doi.org/10.1093/qje/qjr020.
———. 2014. Capital in the Twenty-First Century. Cambridge, MA: Harvard University Press.
Piketty, Thomas, Gilles Postel-Vinay, and Jean-Laurent Rosenthal. 2006. “Wealth Concentration in a Developing Economy: Paris and France, 1807–1994.” American Economic Review 96 (1): 236–56.
Piketty, Thomas, and Gabriel Zucman. 2015. “Wealth and Inheritance in the Long Run.” In Handbook of Income Distribution, edited by Anthony B. Atkinson and François Bourguignon, 2:1303–68. Elsevier. https://doi.org/10.1016/B978-0-444-59429-7.00016-9.
Schinke, Christoph. 2012. “Inheritance in Germany 1911 to 2009: A Mortality Multiplier Approach.” Master’s thesis, Paris School of Economics.
The Economist. 2025. “Inheriting Is Becoming Nearly as Important as Working.”
Appendix A. Replication
The code accompanying this paper reads the data appendix of Piketty (2010), converted from the original Excel files to Parquet format. Under the baseline assumptions, the code reproduces the published series exactly: the maximum absolute difference between the reproduced and published economic flows over 1964–2008 is 9.0e-17, and the reproduced fiscal flows equal the published ones in all benchmark years. The tax-exempt share \(e_t\) is recomputed from the composition of private wealth (Table A15b) and Piketty’s exemption rates, and reproduces column (7) of Table B1.
Each scenario in Tables 4 and 5 is defined by a small set of switches: whether the 2004–2005 cash-gift scheme is included in 2006; whether gifts receive the tax-exempt upgrade; the exemption rates for housing and for listed equity and other financial assets; the non-filer factor for 2006; an optional freeze of the gift-bequest ratio; and the share of surviving spouses. For the economic flow, the bequest component \(\mu_t m_t \beta_t\) is taken from Piketty (2011) in all scenarios. Gifts are either \(v_t\) times the economic bequest flow (baseline) or recorded gifts with the non-filer upgrade (alternative); in annual series, the ratio of gifts to economic bequests is interpolated linearly between benchmark years and held constant after 2006.