Sort by *
DETECTION OF TAX
FRAUD BY INDIVIDUALS
Methods undoubtedly modernised, but results still
inadequate
Public thematic report
Executive Summary
November 2023
2
Executive Summary
The topic of detecting tax fraud by individuals was selected as part of the citizen
consultation launched by the Court of Accounts in 2022 to contribute to the work programme
of the financial jurisdictions.
Meanwhile, in June 2023, the government published a national plan to combat tax fraud,
structured around 35 measures. Apart from the fact that these measures are aimed more at
businesses than individuals, the plan does not specifically refer to the detection of fraud,
although the four measures designed to equip the tax authorities for the digital world are partly
covered.
However, detecting tax anomalies and irregularities is the first link in the overall policy to
combat fraud. It precedes - and determines the effectiveness of - the audit phase, at the end
of which taxpayers who are found to have committed fraud will have to pay tax arrears and
penalties.
In accordance with the request made to the Court, this report deals only with taxes paid
directly by individuals. These taxes accounted for more than €160
billion in 2022, or 30 % of
net tax revenues collected on behalf of all public administrations (the State, local and regional
authorities and other public bodies). These include income tax (€88.9 billion collected from
18.5 million households), tax on inco
me from transferable securities (€4.0 billion), property
wealth tax (€2.4 billion) and inheritance and gift tax (€18.6 billion) in taxes paid to the State,
as well as property taxes (€28.3 billion), council tax on second homes (€2.3 billion) and
property t
ransfer taxes (“notary fees”, €16.1 billion) paid to local authorities.
Major taxes such as value added tax (VAT) or even the generalised social contribution
(Contribution Sociale Généralisée, CSG) are legally payable by individuals, but in practice they
are assessed and collected without their involvement (by businesses in the case of the former,
by employers in the case of the latter), which explains why they have not been included in the
scope studied here.
Tax fraud: a phenomenon that is not measured and is often confused with
other methods of tax avoidance
The term ‘tax fraud’ is not always used rigorously, particularly when it is used generically
to describe all actions that have the effect of reducing the amount of tax paid. As the diagram
below illustrates, several phenomena (tax optimisation, unintentional irregularities, taxpayer
insolvency, etc.) have the same result without, however, constituting irregular conduct.
3
Optimisation, evasion, fraud, unintentional irregularities: the phenomena that reduce
tax yields
Source: Court of Accounts
Tax fraud: a phenomenon that has not been quantified and is often
confused
with other methods of tax avoidance
Tax fraud is defined by Article 1741 of the French General Tax Code (CGI) as a
deliberate violation of tax regulations, which implies an intentional omission on the part of the
taxpayer in their reporting obligations, a concealment of all or part of the sums subject to tax,
the fraudulent organisation of their insolvency or any other act intended to hinder the recovery
of tax.
Fraud is a criminal offence punishable by a fine of €500,000 and five years’
imprisonment.
Tax fraud, combined with unintentional irregularities and sums claimed by the tax
authorities but not recovered, corresponds to the “tax gap”,
which refers to the difference
between the amounts actually collected and those that would result from strict application of
tax regulations with no disruption.
A regrettable and persistent lack of estimates of tax evasion, with very
broad and imprecise measurement
Unlike many countries, France has no rigorous assessment of tax fraud or even of the
tax gap. The amount of each of these two phenomena, even when reduced to a rough
approximation, is unknown. The only known amounts are those claimed by the tax authorities
after an audit, i.e. a total of €14.61 billion in 2022, divided between €11.95 billion in evaded
taxes and €2.66 billion in penalties, around 1/5th of these sums being paid by individuals and
4/5ths by businesses and professionals.
A method as rudimentary as extrapolating to the French situation the tax gap observed
abroad (4.5 % in Estonia for the lowest, 16.6 % in the United States for the highest) would only
result in an extremely wide range, between €30 and €100 billion for all taxes
combined, or
between €7 and €27 billion for personal income tax alone, without it being possible to
distinguish fraud from involuntary irregularities and unrecovered sums.
This situation makes it impossible to relate the amounts detected or actually claimed to
the estimated amounts of fraud or of the tax gap in order to assess the effectiveness of the
tools used. The sums claimed as a result of tax audits can be traced and their evolution over
4
time measured, but it is impossible to state with any certainty the proportion of actual fraud that
they help to counter.
Detecting tax anomalies and irregularities, prior to tax audits
The detection of tax anomalies and irregularities refers to all the operations that make it
possible to detect them, whether they are the result of first-level checks carried out by tax
officials or, as is now mostly the case, automatic controls based on the mass processing of tax
data. In addition, there are the various methods of collecting and processing information
received by the tax authorities (through whistleblowing, investigations, information provided by
other French or foreign authorities, etc.).
This detection phase is likely to lead to a first level of exchanges between taxpayers and
the tax authorities and to the correction by the former of their initial declarations. Since the end
of the 2010s, it has also been used as the basis for scheduling the formal documentary and
on-site audits to be carried out by the specialist services.
Focusing on detection, this report deals only incidentally with the phases of audit and
recovery of evaded duties and taxes, which take place after the phase of detection of presumed
irregularities.
Marked improvement in detection tools during the 2010s
During the 2010s, new impetus was given to the fight against tax fraud, with the aim of
systematically detecting reporting anomalies and irregularities through the use of new
technologies.
In this respect, the Directorate General of Public Finances (DGFiP) has developed tools
to make massive use of the data held by it or provided by other administrations to identify
inconsistencies with the returns filed by individuals. In 2022, this mass cross-referencing of
data resulted in 155,000 proposals for audits of individuals, three times more than in 2018.
More generally, the scheduling of tax audits, which was historically based on the highest-
value cases - i.e. those involving the largest taxpayers - has changed radically. It is now largely
based on risk analyses fed by mass data processing.
In addition to
this “technological strategy”, the tax authorities have endeavoured to step
up the collection of information that will enable them to identify both sophisticated cases of
fraud, for which no declaration appears to be inconsistent in itself, and new cases of fraud that
have not yet been the subject of a significant risk.
This dual approach has been made possible by the acceleration of international
cooperation, by the continuous increase in the data available to the tax authorities (DGFiP)
and by the improvement in the processing capacity of the latter.
A consequence of the expansion and automation of detection techniques:
a system that better protects individual rights and freedoms
The technologies used to detect tax irregularities in individuals are potentially intrusive.
As a result, new provisions have been introduced by the legislature and the courts to protect
taxpayers’ rights.
The Constitutional Council has long recognised that the fight against tax fraud is an
objective of constitutional value, in respect of which the legislature is free to make adjustments
to individual rights and freedoms. However, in view of the challenges posed by technological
5
developments, it has been necessary to tighten the framework for the powers of the tax
authorities from the outset of the audit phase.
Today, the resources deployed by the tax authorities must comply with the principle of
proportionality, which ensures that the infringement of citizens’ rights and freedoms is not
disproportionate to the aims pursued, and with the personal data protection regime.
For example, mass data cross-referencing methods may only be used as tools to assist
with audits and may under no circumstances lead to the automatic identification of fraud. The
exercise of the right of communication for general detection purposes (for example through the
use of detailed records of calls made by private individuals, known colloquially as “fadettes”)
has been restricted. Finally, the automatic extraction and processing of the content of certain
social networks has been limited to pages made public by Internet users in order to respect
the right to privacy and freedom of communication.
Fraud detection methods whose effectiveness is difficult to assess due to
lack of quantification
In the absence of statistical estimates of tax fraud, it is impossible to establish what
proportion of it is detected, and whether this proportion has increased in recent years with the
implementation of more powerful tools. This is a major shortcoming that must be remedied: it
is the first of the priorities proposed by the Court below.
It is also difficult to assess the impact of the new detection tools on the effectiveness of
tax audits, and in particular on the relevance of the scheduling of audits.
The DGFiP’s activity indica
tors and benchmarks relate to the audits carried out and do
not establish a link between the reasons for scheduling these audits and their results. The main
IT systems used by the DGFiP (ALPAGE and ILIAD), which ensure the traceability of control
actions, were designed with a priority focus on monitoring the amounts to be adjusted following
audits, rather than monitoring their relevance. The reasons for audits are not sufficiently
detailed and are not linked to the breaches detected, hindering a rigorous assessment of the
effectiveness of the entire process (detection, audit, adjustment, recovery) related to tax
evasion. However, this gap is now being filled by the development of new IT systems that
should enable this link to be made.
Tax audit data show that the proportion of fraud cases or cases subject to reassessment
within the total number of cases audited has remained at around 55 % since 2018, which
prevents us from concluding that there has been a marked qualitative leap in the audit strategy.
With regard more specifically to the detection of tax irregularities in individuals through
the cross-referencing of mass data, 27 % of the audits carried out will result from this in 2022.
Carried out by a team of seven agents, the productivity gains associated with the cross-
referencing of mass data are likely, although they cannot be quantified, due to the lack of
precise data on the jobs allocated to detection over the period 2017-2021.
Finally, the increased use of “tax intelligence”, which consists of using
information
communicated by third parties, is showing clear results. This is particularly true of tax
informants, the term used to describe people who report potential fraud to the DGFiP in return
for remuneration. They have made it possible to recover €11
0 million in four years since 2018,
compared with the €1.8 million paid out to them in compensation.
6
A formalised and better-structured strategy for detecting tax irregularities
among private individuals, to be implemented around six priority areas
Steps in risk management
The national anti-fraud plan presented by the French government in June 2023 is based
on 35 operational measures covering a range of issues. However, this ambitious plan does not
deal with the detection of tax irregularities as such. Nor does it clarify the principles and
resources mobilised by the tax authorities, since it returns to the forefront a number of practices
(targeting the largest taxpayers) or concerns (the number of tax audit staff) that the strategy of
the 2010s had replaced with other, more relevant matters.
This plan, provided that it is further defined and extended, is nevertheless an opportunity
to structure and formalise a genuine strategy for detecting tax irregularities, serving as the
basis for and initial phase in the policy of combating tax fraud. This strategy must be
comprehensive and structured, and must remedy the piecemeal progress and opportunities
that have characterised recent years. It could be based on six priorities, which are reflected in
the six recommendations made by the Court of Accounts.
Transparent presentation of tax audit choices and practices, ensuring equal
treatment of taxpayers
There are several possible approaches to detecting tax irregularities. It may target major
taxpayers, taking the view that they account for the majority of the budgetary implications, even
if their numbers are small; this was the approach that prevailed in the past, with the systematic
three-yearly audit of so-
called “high or very high stakes” cases. Alternatively, it may
be based
on risk analysis to maximise the yield from tax audits; this is the approach that has prevailed
with the mass cross-referencing of data since the middle of the 2010s. Lastly, the audit strategy
may leave room for random selection, which would reflect a form of equal probability of being
audited; this is not currently the case, as the tax authorities believe that such a rationale would
mean devoting resources to files that are predominantly compliant, at the cost of reducing the
audit yield.
Formalising a strategy for detecting tax irregularities would make it possible to clarify the
place reserved for each of these approaches and to allocate the resources assigned to the
audit according to transparent, clearly formulated and stated objectives. It would also be the
vehicle through which the tax authorities would present the conditions for the deployment and
use of their technological tools throughout the country, in particular by eliminating the
exceptions that persist today (for example in Corsica and the overseas territories with regard
to the detection of undeclared buildings using aerial views).
A need for statistical estimates of tax evasion
The tax authorities must catch up with many of their foreign counterparts by estimating
the amount of tax evasion, or the tax gap, for each of the major taxes.
The DGFiP plans to complete such estimates by 2027, using rigorous modelling based
on the VAT experiment that was successfully completed in 2021 and 2022. The complexity of
personal taxation must not lead to delays in these estimates.
7
More cross-functional IT systems to improve data flow within the administration
The siloed design of the IT systems used by the tax authorities, specialised by tax or by
function (management of returns, audit, collection, etc.), is a weakness that has been identified
for several years. The tax administration is working hard to remedy this, with several projects
to modernise or overhaul its major applications due to be completed in the short to medium
term.
In this regard, the Court’s recommendations focus on two aspects: ensuring the
traceability of the administration’s actions, from the receipt of taxpayers’ returns (which is the
responsibility of the tax management departments) to the tax audit and its follow-up, in order
to monitor and evaluate the relevance and effectiveness of these actions on an ongoing basis;
and introducing more cross-functionality into the applications to facilitate an overall
understanding of the past and present situation of taxpayers, which is currently still broken
down by tax or territory.
Expand the tax intelligence policy
Effective detection also requires better gathering and processing of tax intelligence
(reports from judicial or police departments, whistleblowing, information from foreign
administrations, etc.) to identify fraud schemes as early as possible. However, the involvement
of these services and the procedures and tools for sharing intelligence within the tax authorities
appear to be inadequate. The anti-fraud plan presented in 2023 makes general provision for
increased use of tax intelligence, mainly from an international perspective, without specifying
how this is to be achieved.
A more structured approach needs to be put in place to improve the sharing of tax
intelligence, to ensure that it is followed up and to draw conclusions from it. One of the aims of
this approach should be to develop cooperation between tax management departments and
tax audit departments in order to monitor and exploit the mass of information at their disposal.
A proactive prevention approach aimed at taxpayers
The current fraud prevention approach is primarily aimed at businesses. It takes the form
of information and support systems, based on a map of abusive practices and arrangements
that is kept up to date and published on the impots.gouv.fr website.
The tax authorities have other opportunities to provide taxpayers, particularly private
individuals, with information and prevention messages. As in most OECD countries, the tax
authorities are taking advantage of the growing dematerialisation of their dealings with
taxpayers, particularly when they file their tax returns online, to develop tools for raising
awareness and analysing risks in real time, in the form of pop-up windows. These pop-up
windows alert taxpayers to potential inconsistencies or breaches that could give rise to a
subsequent audit. Strengthening this approach would increase the number of alerts issued to
taxpayers acting in good faith, deter certain minor frauds (e.g. tax reductions and credits) and
improve the quality of relations with the public.
Greater recognition of the specific skills and professions involved in combating fraud
required in human resources management
The technological strategy deployed in the 2010s was not accompanied by a precise
identification of the specific skills required to detect tax irregularities. There is a lack of data to
assess training efforts in this area. In addition, the DGFiP’s human resources management,
8
from recruitment to career management, consistently favours cross-functional skills, without
clearly indicating the place given to specific skills related to the fight against fraud and, in
particular, to detection, for which there is no recognised role within the tax administration.
Implementing a policy of attracting and retaining staff specialised in detecting tax fraud
is therefore a major challenge if the strategy that the Court is calling for is to be formalised over
the long term.
***
In the 2010s, the tax authorities took advantage of the possibilities offered by the
dematerialisation of tax management to develop powerful tools for detecting anomalies in tax
returns and tax irregularities. These tools have profoundly changed the way tax audits are
planned, moving away from an approach that targets the largest taxpayers to one based on
risks identified by the mass processing of data collected. However, these tools have not been
integrated into an explicit fraud detection strategy. This is evidenced by the fact that fraud in
the various taxes is still not statistically estimated using tried and tested methods, making it
impossible from the outset to decide on the best allocation of resources and the effectiveness
of the systems deployed. In fact, this report notes that the relevance of tax audits has improved
over the last five years, a sign of more effective upstream detection, but there is a lack of data
to support and quantify this progress more precisely.
The national anti-fraud plan presented by the French government in June 2023 is an
opportunity to specify and formalise a detection strategy, although none of its 35 measures
deals directly with this issue. The Court’s insistence on the need to formalise this strategy and
make its main aspects public is a response to a public issue that goes beyond questions of
efficiency and the appropriate allocation of resources. At a time when public debate sometimes
leaves room for suppositions, approximations and even untruths, it is the responsibility of the
tax authorities to explain who they are auditing and why, on the basis of what risks, and with
what balance between the power of the technological tools at their disposal and the protection
of taxpayers’ rights. Only then will it be possible to strengthen tax compliance, which is the
cornerstone of the Republican Pact.
9
Recommendations
The Court of Accounts recommends that the ministry of the economy, finance and
industrial and digital sovereignty formalise, by the end of 2024, a national strategy for detecting
tax irregularities to complement the June 2023 anti-fraud plan, structured around the following
six priorities and recommendations:
1.
present the strategic choices for scheduling tax audits in a transparent manner and ensure
that the tools and methods deployed for this purpose guarantee equality for taxpayers by
eliminating unfounded customary exceptions;
2.
establish a methodology and schedule for estimating personal income tax fraud by the end
of 2024, in partnership with Insee and on the basis of the lessons learned from the estimate
made of VAT fraud;
3.
ensure that the PILAT project enables a link to be made between suspected, detected and
established fraud, and thus assess the relevance of the reasons for scheduling audits;
4.
on the basis of experiments and local best practice, create a national IT tool for mobilising
internal intelligence that is simple, quick to use and includes systematic feedback to the
issuing agent;
5.
step up preventive action in the form of real-time alert messages when data provided during
online declarations seems inconsistent or missing, warning taxpayers of the checks that
could ensue;
6.
ensure that recruitment and transfer procedures enable the DGFiP to make the most of the
professions and skills specific to the fight against tax fraud.