Citizenship and taxes:
Agency decision
Ask Donna
What actually matters in this document.
Text
Citizenship and taxes:
Evaluating the effects of the U.S. tax system on
individuals’ citizenship decisions
Paul R. Organ+
University of Michigan
This draft: August 23, 2021
Abstract: The U.S. tax system applies to its citizens’ worldwide incomes and estates,
whether those citizens live in the U.S. or abroad. Fully escaping the U.S. tax system
requires renouncing or relinquishing U.S. citizenship, and in recent years a growing
number of individuals have done so. I use administrative tax microdata to answer three
questions: Who is renouncing their citizenship? Why are they renouncing? What are the
policy consequences? I show that the recent increase in renunciations is mainly driven by
those who have for many years lived abroad, rather than by individuals leaving the U.S.,
and that these renunciations are primarily a response to increased compliance costs, not
tax liabilities. I also present evidence that some individuals responded to tax law changes
affecting the costs of renunciation. I conclude by discussing how recent renunciations
relate to U.S. tax policy, including the revenue impacts, policy spillovers, and the implied
value of U.S. citizenship.
JEL codes: F22, H2, H24, H26, K34
+
prorgan@umich.edu. For helpful comments, suggestions, and support I thank my dissertation committee: Joel
Slemrod, Jim Hines, Ash Craig, and Ed Fox; my thanks also go to Katarzyna Bilicka, Sebastien Bradley, Dhammika
Dharmapala, Gabe Ehrlich, Jeff Hoopes, Daniel Reck, Max Risch, Molly Saunders-Scott, Bill Strang, and seminar
participants at the University of Michigan, the IRS, the U.S. Treasury Office of Tax Analysis, the 2020 National Tax
Association Annual Conference, and the 2021 IIPF Annual Congress. I am especially grateful to John Guyton, Anne
Herlache, Thomas Hertz, Pat Langetieg, Alicia Miller, Annette Portz, Alex Turk, and Carlos Zepeda at the IRS for
their support of this work. All data work for this project involving confidential taxpayer information was done on
IRS computers by IRS employees, and at no time was confidential taxpayer data ever outside of the IRS computing
environment. The author is a Student Volunteer with the IRS. The views and opinions presented in this paper reflect
those of the author and do not necessarily reflect the views or the official position of the Internal Revenue Service.
All results have been reviewed to ensure that no confidential information is disclosed.
1
1. Introduction and motivation
Renunciation of U.S. citizenship has risen sharply in the last decade, from roughly 500 a year
in the early 2000s to more than 4,000 each year from 2013-2018. What factors drove this recent
increase, and does it warrant a policy response? To answer these questions, I use administrative
tax microdata to explore who is renouncing and how that has changed over time. I show that
features of the tax system, in particular additional compliance costs brought on by increased
enforcement of offshore financial activity, explain a substantial part of the renunciations seen
over the last decade. I also show that most recent renunciations are likely to impose only a small
cost on the United States; any substantial impacts can be attributed to just a handful of wealthy
and high-income renouncers.
The U.S. is one of a handful of countries which tax their citizens’ worldwide income and
estates.1 As a result, policymakers have frequently raised concerns about U.S. citizens dropping
citizenship to avoid taxes. The first legislation intended to discourage tax-motivated expatriation
was passed in the 1960s. Several high-profile departures in the 1990s prompted new laws
requiring public disclosure of citizenship renunciation–since 1998 the names of those dropping
U.S. citizenship have been published in the Federal Register–as well as further substantial
changes to the expatriation tax system2 in 2004 and 2008. Since 2008 the relevant changes have
been in tax enforcement, starting with legal actions targeting Switzerland, and leading to a
broader increase in offshore financial enforcement under the Foreign Account Tax Compliance
Act (FATCA). Despite, or perhaps because of, these efforts, citizenship renunciations have
continued, with annual counts rising markedly in recent years. Between 2005 and 2018 more
than 35,000 individuals with at least $48 billion of combined reported net worth renounced their
U.S. citizenship.
1
Only two other countries, Eritrea and Myanmar, similarly tax their citizens regardless of residence. Eritrea levies a
flat income tax of 2% on its citizens living abroad; Myanmar applies the same rates to its citizens’ income, whether
derived at home or abroad.
2
I use the term “expatriation tax system” to refer to the laws and tax regulations which govern expatriation and
citizenship renunciation; these include filing and reporting requirements, and tax liabilities incurred at and after
renunciation. Following previous literature and the terminology of related legislation, I use the term “expatriation”
to mean giving up U.S. citizenship, rather than merely moving abroad.
2
Who exactly is renouncing U.S. citizenship? Prior studies of this topic have been limited to
publicly available information, which in practice has meant only the quarterly counts compiled
from the names of renouncers published in the Federal Register. Using administrative tax
microdata, I provide more detailed information about the population of individuals dropping U.S.
citizenship from 1998-2018. The recent increase in renunciations has come mainly from those
who have long filed U.S. taxes from abroad (and thus likely lived abroad), rather than from
individuals who lived in the U.S. choosing to move abroad. Those renouncing citizenship are on
average higher-income and wealthier than the U.S. population: more than one-third of those
renouncing and reporting net worth are millionaires, compared with estimates of 5-10% for all
U.S. individuals. Renunciation is concentrated in relatively few destination jurisdictions, with the
top five (Canada, Switzerland, the United Kingdom, Germany, and Hong Kong) accounting for
more than half of the total.
Why are individuals renouncing U.S. citizenship, and lately in greater numbers? I study the
citizenship decision within an option value framework, arguing that maintaining U.S. citizenship
is akin to holding an option to return to live or work in the U.S. This framework motivates a
regression analysis to identify the characteristics associated with the decision to renounce
citizenship. One key finding is the significant positive effect of age on renunciation, as predicted
by the option value framework; as individuals age, the time value of their option decreases,
making renunciation relatively more favorable. Using jurisdiction-level analysis I also find
relationships consistent with the option value framework; U.S. taxpayers filing from jurisdictions
designated as tax havens, and with higher governance scores (measured using the World Bank’s
Rule of Law governance index), have relatively higher renunciation rates. I then use a differencein-difference analysis to test the effect of the Foreign Account Tax Compliance Act (FATCA) and
related offshore enforcement efforts which increased compliance costs for certain U.S. citizens
abroad. I find evidence supporting the claim that increased compliance costs caused an increase
in citizenship renunciations by U.S. citizens living abroad. I also discuss the connection between
recent expatriation tax law changes and the trends in renunciation. The data patterns suggest that
some very high-wealth and high-income individuals may have chosen to leave the U.S. and
renounce citizenship during the 2004-2008 period, in anticipation of the introduction of a mark3
to-market exit tax. The data also reveal a strong behavioral response to the notched design of a
net worth threshold embedded in the expatriation tax system.
What are the policy consequences of recent renunciations? I first use data on pre-renunciation
tax liabilities to consider the potential revenue impacts of recent expatriations. I find that for
most renunciations the revenue impact is probably negligible because individuals had no or little
tax liability in the years prior to expatriation. The distribution of liabilities is heavily skewed,
however, such that a handful of individuals’ renunciations have an outsize impact on revenues.
Considering the broader revenue impacts of the connection between citizenship and taxes, I
argue that if the effects of the tax system on renunciation decisions apply similarly to the much
larger group of individuals considering migration to the U.S., or naturalization once in the U.S.,
the corresponding revenue impacts could be significant. I then discuss the spillover effects of
FATCA and other enforcement actions on citizenship renunciation and the importance of timing
for the 2004 and 2008 expatriation tax law changes. I conclude by putting renunciations in a
broader context, considering the non-renunciation of most U.S. citizens and the in-migration of
newly naturalized citizens, and what these things imply about the value of U.S. citizenship.
The paper proceeds as follows: Section 2 sets up a conceptual framework for the costs and
benefits of renunciation and briefly describes expatriation-related tax law, offshore financial
enforcement, and related academic literature. Section 3 describes the data underlying the
subsequent analyses. Section 4 provides a description of who is renouncing citizenship. Section 5
explores what can explain the recent increases in renunciations. Section 6 discusses the policy
consequences, and Section 7 concludes.
2. Background and literature review
In this section I describe (1) the potential costs and benefits of citizenship renunciation, (2)
tax law related to expatriation and how that has changed over time, and (3) tax enforcement
related to offshore financial activity and how that has changed over time. Throughout the section
I highlight related academic literature. For additional details on the specific steps required for
citizenship renunciation, see Appendix B.
4
2.1.
Costs and benefits of citizenship renunciation
The specific costs and benefits of citizenship renunciation for any given taxpayer depend on a
variety of taxpayer characteristics3, but can generally be grouped into the categories shown in
Table 1: administrative costs and benefits (e.g., renunciation fee vs. removal of U.S. tax filing
obligation) and income- or wealth-dependent tax consequences (e.g., expatriation tax
consequences vs. lower future income or estate tax liabilities). This high-level framework allows
a consideration of how the net benefits of renunciation would change as any of the component
costs or benefits change. For example, consider one change which occurred in 2014, when the
State Department raised the fee for citizenship renunciation from $450 to $2,350. This change
uniformly lowered the net benefits of citizenship renunciation for all individuals considering it
by $1,900.
Table 1: Costs and benefits of tax-informed citizenship renunciation
Type
General
Costs
Benefits
Administrative costs of act of expatriation
(e.g., time, renunciation fee)
Reduction of ongoing administrative burden
(e.g., banks wary of U.S. citizens)
Loss of benefits of U.S. citizenship
(e.g., visa-free travel to many countries)
Reduction of yearly administrative burden
(e.g., U.S. tax filing)
Income-dependent Expatriation tax consequences
Lower future income tax liabilities
Wealth-dependent Expatriation tax consequences
Lower future estate and gift tax liabilities
Some of these costs and benefits are simple to value (the renunciation fee is known and is
exactly $2,350) while others are longer-term and more uncertain (e.g., comparing expected U.S.
income tax liability vs. foreign income tax liability on the next 10 years of income). However,
even when exact values are unavailable, as long as one can characterize the sign of the change, it
is possible to elicit a prediction about the effect of a policy change on the incentive to expatriate.
In later sections I will discuss several changes to expatriation tax law or offshore financial
3
For example, whether a taxpayer already lives or holds citizenship abroad; the amount and type of income a
taxpayer receives currently and expected to receive in the future; the amount and type of assets a taxpayer holds
currently and expects to bequeath in the future; the tax system of the anticipated destination country; and whether a
taxpayer is currently compliant on their U.S. taxes.
5
enforcement and consider how these policy changes would be expected to affect incentives for
certain types of taxpayers considering citizenship renunciation.
2.2.
Citizenship and U.S. tax law
The U.S. tax system has attempted to discourage tax-motivated expatriation for several
decades. The Foreign Investors Tax Act of 1966 introduced §877 of the Internal Revenue Code
(IRC), requiring taxation of former citizens for ten years following expatriation if tax avoidance
was a “principal purpose of the expatriation” (Craig 2012). Thirty years later, a formal test for
tax-motivated expatriation was introduced, as part of the Health Insurance Portability and
Accountability Act of 1996 (HIPAA). Under the new objective standards, expatriating
individuals were deemed “covered expatriates” if either past-five-years average net income tax
liability exceeded a certain threshold, or if net worth exceeded a different threshold. 4 Taxpayers
also had to certify that they were compliant on all federal tax obligations for the five tax years
preceding expatriation. As before, designation as a covered expatriate meant a taxpayer was
liable for U.S. taxes on U.S.-source income and on income effectively connected with a trade or
business in the U.S., at the same progressive rates faced by U.S. citizens, for the ten years
following expatriation. In practice, even if a taxpayer was deemed a covered expatriate under the
objective tests, one could appeal this designation and most who did so were successful. 5 Also of
note, in an attempt to further discourage tax-motivated expatriation, HIPAA required the names
of expatriating individuals to be published in the Federal Register (Internal Revenue Code,
§6039G).
The American Jobs Creation Act (AJCA) of 2004 brought additional changes: (1) the removal
of expatriates’ ability to challenge their designation as tax-motivated, (2) an increase in the net
4
The thresholds during 2019 were $168K (average past-five-years income tax liability) and $2M (net worth). Figure
26 in the Appendix shows how these have changed over time. Note that the income tax liability threshold is applied
to tax liabilities, not incomes; to have an income tax liability of $168K in 2019 would have required income of more
than $500K. This distinction is sometimes missed in discussion of the expatriation tax system, with some suggesting
that the threshold applies to income itself (and thus implying that many more individuals would be treated as
covered expatriates according to this threshold than is truly the case).
5
Between 1997 and July 2002, 270 applications for private letter rulings overturning the presumption of taxmotivated expatriation were made to the IRS. Of these about half received favorable responses, and all but 11 of the
remainder received neutral responses. Favorable and neutral responses meant that applicants could proceed without
fear of further IRS enforcement under the expatriation tax regime. This suggests that roughly 96% of appeals were
successful (259/270 = 0.959) (Kwong 2009, 421).
6
worth threshold from $622K to $2M; and (3) requiring the filing of Form 8854 to complete
expatriation for tax purposes.6 The next changes were introduced in the 2008 Heroes Earnings
Assistance and Relief Tax (HEART) Act, which created IRC §877A and changed the
consequences for covered expatriate designation to now include a mark-to-market exit tax, rather
than the taxation of next-ten-years’ U.S.-source income. Under the new regime, gains on all of a
covered expatriate's assets (with a few minor exceptions 7) are deemed realized as of the day prior
to the expatriation date, and taxes owed on deemed gains above a certain exempted amount. 8 The
2008 bill also removed the requirement that Form 8854 be filed to complete expatriation for tax
purposes.9 Selected aspects and changes to the expatriation tax system are shown in Table 2.
Table 2: Selected aspects of and changes to the U.S. expatriation tax system
Expatriation date
On or before
June 3, 2004
June 4, 2004 to
June 16, 2008
On or after
June 17, 2008
Test for tax-motivation
Tax consequences
Other consequences
Net worth > $622K (2004);
Avg. inc. tax liability > $124K (2004);
Presumption only, can challenge
NW > $2M;
AITL > $139K (2008);
Conclusive test, cannot challenge
For 10 years: taxed on U.S.-source
income; estate and gifts subject to
U.S. taxation
180-day limit on U.S. visits
Same as above
Annual filings with $10K
penalty for non-filing; 30-day
limit on U.S. visits
NW > $2M;
AITL > $168K (2019)
Exit tax: mark-to-market capital
Annual filings until exit tax
gains tax (deemed realization) with
obligations are met
$725K exemption (2019)
Notes: The column “Test for tax-motivation” indicates the tests which are applied to an individual who
expatriates during the given time period; if an individual is deemed to be a “covered expatriate” under the
tests, then the corresponding consequences (tax and other) apply.
Academic research on expatriation has mainly appeared in law journals, and generally focuses
on detailed components of related legislation or proposed changes to the expatriation tax system
6
Arsenault (2009) provides further information on the first two changes. For the Form 8854 filing requirement, see
the amendment history of IRC §7701(n); the 2004 AJCA added §7701(n), stating that an expatriating individual is
still treated as a citizen or resident of the U.S. until that individual “provides a statement in accordance with Section
6039G.”
7
Exceptions include deferred compensation items, specified tax deferred accounts, and interest in non-grantor trusts.
8
For expatriations during 2019 the first $725K of gains are exempt. Figure 26 Figure 26in the Appendix shows how
the exempted amount has changed over time.
9
Expatriating individuals are still required to file Form 8854 under IRC §6039G, but after the 2008 HEART Act’s
removal of IRC §7701(n), failure to file Form 8854 no longer carries the consequence that an individual is treated as
a U.S. citizen or resident for tax purposes until the form is filed. This change lowered the cost of non-filing and may
help explain the large share of expatriating individuals in recent years without Form 8854 filings.
7
(Arsenault 2009, Kwong 2009, Manolakas and Dentino 2012, Craig 2012). Westin (2000)
provides a comprehensive overview of the expatriation tax system prior to the reforms of the
2000s. More recently, Ahn (2015) studies the HEART Act and notes an increase in expatriations
following the introduction of the deemed realization tax that can be seen in public data from the
Federal Register.
Mason (2016) provides a thorough evaluation of various arguments for and against
citizenship-based taxation. In response to Mason, Kim (2017) argues in favor of citizenship
taxation and discusses how citizenship renunciation rates for the U.S. compare to other highincome countries. Noting the difficulty of defining a denominator when calculating the
renunciation rates, Kim provides several plausible estimates based on 2010 and 2013 foreign
diaspora data and relying on aggregate counts of renunciations, and concludes that the U.S. is not
a serious outlier.10 Kim also notes that “we lack empirical studies on the specific motivation of
renunciation,” a concern also raised by Kudrle (2015). This is precisely the gap that this paper
aims to fill. More recently, De Simone, Lester, and Markle (2020) study how U.S. individuals
responded to FATCA. Although their paper focuses on portfolio investments based in foreign tax
havens, the authors also make use of the public Federal Register data to plot the annual counts
and suggest that the recent rise in U.S. expatriations could be related to FATCA.
This paper is the first to study in detail and quantitatively the connection between citizenship
renunciation and citizenship-based taxation. There is a related literature in economics which
studies the connection between taxes and migration, generally studying residence-based taxation
(Mirrlees 1982, Kleven, Landais and Saez 2013, Akcigit, Baslandze and Stantcheva 2016,
Kleven, Landais and Muñoz, et al. 2020). The distinction between residence-based and
citizenship-based taxation is important because changing one’s residence is more reversible than
changing one’s citizenship (and may carry different costs as well). By using IRS data including
Form 8854 filings, which allow for a more detailed study of the population of those renouncing
10
Kim’s estimates of renunciation rates show that the highest rates were in jurisdictions with military draft systems,
with the top three rates observed for South Korea, Singapore, and Taiwan. While the relative comparison of rates
across jurisdictions is certainly of interest, the many factors influencing citizenship decisions make it difficult to
draw conclusions from these cross-jurisdiction comparisons. By focusing on the decisions of individuals specifically
with respect to U.S. citizenship, observing trends over time, and using individual microdata, much can be learned
about the motivation for citizenship renunciation and its connection to the tax system.
8
citizenship, this paper makes an important contribution to measuring and understanding the
incentives to maintain or renounce citizenship under a citizen-based taxation system.
2.3.
Tax enforcement and foreign financial activity
In the last decade, major changes have been made in the enforcement environment affecting
financial activity by U.S. citizens living or holding financial accounts abroad. Johannesen et al.
(2020) describe the introduction since 2008 of “a range of enforcement initiatives targeting
owners of offshore accounts”: ad hoc legal action and information exchanges; bilateral treaties;
and FATCA.
Ad hoc legal action against Swiss banks included so-called “John Doe summonses”, which
allowed the IRS to request information from foreign banks about their U.S. citizen customers
without identifying the specific customers in advance. 11 The IRS was authorized to use these
summonses beginning in July 2008 against UBS, and subsequently against other large banks
including HSBC and Credit Suisse. In addition to the ad hoc legal steps, the U.S. government
signed bilateral information exchange agreements with several countries deemed to be tax
havens.12 These agreements allowed the IRS to request foreign bank account information for
specific taxpayers in tax evasion cases. As Johannesen et al. note, citing Sheppard (2009), these
agreements are relatively restrictive, requiring specification of taxpayer identities in advance and
evidence to justify the request, and thus may not be effective deterrents of offshore tax evasion.
Finally, a new reporting regime requiring systematic information exchange on U.S. citizen
account holders between foreign financial institutions (FFIs) or foreign tax authorities and the
IRS was introduced in 2010, as part of FATCA. This may have affected U.S. citizens living
abroad in two main ways. First, the IRS would now have better access to third-party reporting on
income and assets for these individuals. Second, these individuals now faced increased costs
(either financial costs or compliance costs) in their dealings with FFIs, as those FFIs themselves
faced increased costs in complying with FATCA. Dharmapala (2016) studies how a unilateral
11
If required to specify customers in advance, the IRS would not have been able to meaningfully pursue the relevant
information. U.S. taxpayers hiding assets did not notify the IRS of their holdings, and thus could not be identified ex
ante and specified in requests for information.
12
Between 2008 and 2010, the U.S. signed such agreements with six jurisdictions: Liechtenstein, Luxembourg,
Malta, Monaco, Panama, and Switzerland.
9
reporting regime (like FATCA) affects the cost to FFIs of providing financial services and how
this in turn affects incentives for tax-compliant behavior by foreign residents. Belnap, Thornock,
and Williams (2019) study foreign countries’ and FFIs’ participation in automatic information
sharing with the IRS and show that FFI participation was near-universal (97% of FFIs
participated in automatic information sharing) and costly.
These enforcement changes are relevant to the study of citizenship renunciation because each
change either made it more difficult, or less attractive, to be a U.S. citizen living and maintaining
financial accounts abroad. This paper is the first to study carefully the potentially unintended
consequence of these changes in tax enforcement – increased U.S. citizenship renunciation by
U.S. citizens living abroad.
3. Data
The main source of data for this study is an IRS database of former U.S. citizens who have
renounced their citizenship since 1998. Individuals who expatriate are required to meet with a
consular official, resulting in a Certificate of Loss of Nationality (CLN), and to file a form with
the IRS (Form 8854, the Initial and Annual Expatriation Statement, intended to be filed along
with the income tax filing for the year of expatriation). The State Department notifies the IRS of
each CLN, which the IRS then matches with the Form 8854 filings they receive from taxpayers.
In practice, some individuals have only one of the two forms, and the IRS database represents the
union of renouncing individuals based on CLNs, Form 8854s, or both. In this paper I study only
those renunciations occurring between 1998 and 2018, to allow for a lag in 8854 filing and
ensure a more complete picture of the renunciations occurring in each year. The database also
includes information about some of the individuals relinquishing long-term residency status
(rather than U.S. citizenship). Because this information is not entirely complete—not all such
individuals are included in the database—I restrict my focus in this paper to former citizens.
For all individuals in the database, I observe the date of renunciation and the destination
country or jurisdiction. For individuals with Form 8854 filings I observe reported net worth as of
the date of expatriation. Other fields of interest on Form 8854 that are not available for study at
this time include more details on how foreign citizenship was acquired, as well as a breakdown
10
of assets by asset category. For those with Social Security Numbers (SSN) or Taxpayer
Identification Numbers (TIN), I link to other relevant tax filings. 13 About 70% of those
renouncing have these identifiers. I include all individuals in each analysis where possible,
although at times this is not feasible (e.g., when studying pre- renunciation income, which
requires linking to income tax filings).
In addition to data on those renouncing citizenship, I also rely on information about the
population of U.S. tax filers who are filing from abroad. 14 This allows me to observe the base of
individuals residing abroad who could potentially renounce their U.S. citizenship.
4. Description of renouncers
This section answers the first of my three research questions: Who is renouncing? Previous
studies have had to rely exclusively on publicly available information, which in practice has
meant only the names of individuals expatriating each quarter as reported in the Federal Register.
I provide more detailed information on these individuals, including their prior U.S. tax filing
behavior (and the resulting inferred location, i.e., in the U.S. or abroad), self-reported net worth
and income, and destination jurisdictions.
4.1.
Overall counts
Figure 1 shows the annual count of all former citizens who have renounced citizenship, as
identified in the IRS database, from 1998 to 2018. There is a gradual increase in annual counts
during the 2000s, followed by a more marked increase since 2011. This is the pattern of
renunciations that was available for study prior to this paper, using only publicly available
information about those expatriating. 15
13
These include Form 1040 (Income Tax), Form 1116 (Foreign Tax Credit), Form 2555 (Foreign Earned Income
Exclusion), and Form 709 (Gift Taxes).
14
I am especially grateful to Tom Hertz at the IRS for developing these data.
15
Figure 27 in Appendix B shows the annual count using publicly available information, with counts for 1962-1994
from the Joint Committee on Taxation (1995) and counts for 1998-2020 from the Federal Register.
11
Figure 1: Annual count of U.S. citizenship renunciations
Notes: This figure plots the count of former citizens who renounced citizenship each year, as identified in
the IRS database for years 1998-2018.
4.2.
Prior presence in the U.S.
Presented with the overall increase, a policy-relevant question is, are these individuals
“leaving the U.S.”, or instead individuals who already were living abroad and chose to drop U.S.
citizenship? To answer this I link individuals to their pre-renunciation income tax filings and
infer their locations from the addresses reported on those filings. Most individuals are required to
file Form 1040 each year, even those living abroad. I categorize each individual into one of a few
buckets: those that filed at least once from a U.S. address before renouncing ( “Movers”); those
that filed income tax returns but never from a U.S. address (“Droppers”); and those for whom we
cannot observe pre-renunciation locations (either because they have no filings or have no TIN). 16
Because this method relies on data for tax filings available in the years prior to renunciation, I
16
This is an imperfect proxy that in general would bias towards classification as a Mover, as some individuals may
maintain addresses in the U.S. even while living abroad, or may use a U.S.-based tax preparer’s address on their
filings. Note that because not all renouncing individuals are primary filers, I search for tax filings associated with
their TIN as either primary or secondary filers, to ensure I gather as much pre-renunciation location information
about each individual as possible.
12
limit this classification to those renouncing in 2005 or later 17; I then use five years of prerenunciation tax returns to classify each individual as Movers or Droppers.
Figure 2 shows the count of renouncing individuals each year, split by this classification. The
gray bars represent the Movers – those who can be thought of as “leaving the U.S.”. The orange
bars represent the Droppers, those who were filing returns but always from a foreign address. In
blue are those with a TIN but no filings, or without TINs or SSNs to match to tax returns (this
latter group is likely comprised mainly of Droppers, i.e., those who were not present in the U.S.
prior to expatriation, which would explain why they have no filings or no TINs). While the
annual count of Movers has increased slightly, most of the of the recent increase is by Droppers.
In later sections I will study further what can explain this increase in Droppers, arguing that it is
primarily an unintended consequence of the increased compliance costs resulting from FATCA
and other offshore financial enforcement.
Figure 2: Annual count of renunciations, split by pre-renunciation tax filing locations
17
The IRS database of income tax returns starts in earnest with returns for tax year 1998.
13
Notes: This figure plots the count of individuals renouncing each year, split by their classification based on
Form 1040 filing behavior in the five years prior to renunciation. Movers are those who filed at least once
from the U.S. during those five years; Droppers are those who filed always from abroad. Renunciations
prior to 2005 are excluded to ensure sufficient pre-renunciation data are available.
4.3.
Income and wealth
It is also interesting to consider how these individuals compare to others in terms of income
and wealth. I begin by comparing renouncers to other foreign filers and the full population of
U.S. tax filers, in terms of total and wage income, and then compare income within renouncers,
between Movers and Droppers. I then do a similar comparison for reported net worth.
Figure 3 reports the mean values of total income and wage income in the year prior to
renunciation, and compares this to two other groups: (i) all other filings from foreign addresses,
and (ii) a sample of the full population of Form 1040 filings. In orange are renouncers who were
the primary filer for a linked 1040 in the year prior to renunciation. 18 In blue are all other Form
1040 filings from foreign addresses for the given year, and in gray are a sample of all Form 1040
filings. The vertical dashed lines represent three key dates related to expatriation tax law: 2004
AJCA (raising the net worth threshold for covered expatriate designation), 2008 HEART Act
(introducing the mark-to-market exit tax), and 2010 FATCA (increasing information reporting of
foreign financial accounts held by U.S. citizens). To illustrate the influence of a few outliers on
the mean value among renouncers, the dashed line removes the top 10 individuals for each year.
18
I use the prior year to ensure a full year’s income is reported. In the year of renunciation itself, those renouncing
citizenship file a Form 1040 representing the portion of the year they are a citizen, and may file a Form 1040 NR for
the remaining portion of the year after they have renounced.
14
Figure 3: Comparison of income for renouncers, foreign filers, and all tax filers
Notes: This figure compares the income of renouncers in the year prior to renunciation to two comparison
groups: all other foreign filings, and a sample of the population of Form 1040 filing. For renouncers, only
primary filers with linked filings are included. The three vertical dashed lines represent three key dates
related to expatriation tax law: the 2004 AJCA, the 2008 HEART Act, and 2010 FATCA. The solid line
includes all individuals; the dashed line removes the top 10 in each year.
Figure 3 demonstrates that the average income of those renouncing each year has changed
dramatically over time, and that outlier individuals play an important role in driving the annual
averages. Those renouncing during the window between 2004 (AJCA) and 2010 (FATCA) were
on average much higher income, relative to those expatriating in the 2010s; and this is true even
when removing the top 10 individuals each year. Prior to 2010, those renouncing were higher
income, on average, than both other foreign filers and the broader U.S. filer population. Since
2010, those renouncing have had lower income, on average, than other foreign filers, but still
higher than the U.S. filer population overall. Similar trends appear when considering the median
values instead of the mean (see Appendix, Figure 15).
15
For more detail about the income of renouncers, consider Figure 4, which compares the
income just for renouncers, with averages calculated separately for Movers and Droppers. 19 For
both groups, incomes were higher during the 2005-2010 time period, but the big outliers for total
income are among the Movers, not the Droppers. The groups also differ in terms of their source
of income; Movers have higher average total income, but Droppers have higher average wage
income. The dramatic influence of the top 10 individuals each year on the average total income
among Movers is a stark example of the nature of the renunciation policy problem: although
most individuals have a small revenue impact, a handful can have a significant effect; I discuss
this in further detail in Section 6. As above, similar trends are seen in the median values (see
Appendix, Figure 16).
Figure 4: Comparison of income among renouncers, Movers vs. Droppers
Notes: This figure compares the income in the year prior to renunciation for renouncers with linked Form
1040 filings as primary filers. The mean values are calculated separately among Movers and Droppers.
Renouncers with no filings or no TINs are excluded.
19
Those without filings or TINs are excluded due to lack of income data.
16
Moving from income to wealth, I begin by grouping the renouncers based on their net worth
as reported on Form 8854. I construct buckets using the thresholds for covered expatriate
designation: $622K (the threshold prior to the AJCA, i.e., prior to June 2004) and $2M (the
threshold since the AJCA, i.e., after June 2004). Figure 5 shows the annual count, grouped by
reported net worth.20
Figure 5: Annual count of renunciations, split by reported net worth
Notes: This figure plots the count of individuals renouncing each year, split by reported net worth. Prior to
June 2004, reported net worth data are not available. This figure starts with the first full year of available
data, 2005.
A few patterns are worth noting. First, although there has been a small rise in the number of
renunciations by those reporting net worth of at least $2M (the green bars), these still represent a
relatively small share of the total. Second, there has been more substantial growth in the number
reporting between $622K and $2M in net worth (the blue bars); this group is relevant because it
20
Reported net worth is only completely available since mid-2004, when Form 8854 began to require all filers to list
their reported net worth; prior to this change, only those with net worth above the tax-motivation threshold ($622K
in early 2004, adjusted upward for inflation over 1998-2004) were required to report this information.
17
represents the individuals who prior to the AJCA would have been designated as covered
expatriates, but after the raising of the net worth threshold no longer faced such designation. At
the same time, there was similar growth in those reporting less than $622K (the orange bars).
Finally, an important pattern is the persistent large share of renunciations without Form 8854 or
without reported net worth data, (the gray bars). Although filing Form 8854 is a necessary step to
fully complete one’s citizenship renunciation, a significant number of individuals still have not
done so. Some of this pattern in more recent years could reflect that some file Form 8854 with a
lag (this likely explains the difference between 2017 and 2018 – those who renounced in 2018
and plan to file Form 8854 may still be finalizing their filings). Although this non-filing limits
the ability to draw comprehensive conclusions about the wealth of all renouncers, useful
information can still be gleaned by studying those for whom data are available.
I next consider how the wealth of those renouncing each year has changed over time, and
whether this differs for Movers and Droppers. Figure 6 reports the mean reported net worth of
those renouncing each year since 2005, separately for Movers and Droppers (only including
those with reported net worth data available). The patterns are similar to those above for income:
Movers are wealthier than Droppers; average renouncer wealth during the 2004-2010 period was
notably higher than in more recent years; and removing the top 10 individuals in each group each
year has a dramatic effect on the average values. Similar patterns emerge when considering the
median values (see Appendix, Figure 17).
18
Figure 6: Comparison of reported net worth among renouncers, Movers vs. Droppers
Notes: This figure compares reported net worth among those renouncing each year, separately for Movers
and Droppers. Only those with reported net worth data available are included. The left panel includes all
Movers and Droppers; right panel drops the top 10 Movers and Droppers, by reported net worth, each year.
Finally, I consider how the wealth distribution among renouncers compares to the population.
Table 3 shows the count of renouncers from 2005-2018 by their reported net worth, as well as the
total reported net worth in each group. The share of the population in each net worth group is
included, based on the 2019 Survey of Consumer Finances (these population estimates are for
households, and thus weighted towards higher amounts, relative to the renunciation statistics
which are for individuals). I provide two estimates for the share of renouncers in each net worth
group: the first assumes that all renouncers without reported net worth data are in the <$1M
group; the other excludes those without reported net worth data (i.e., it assumes those without
reported net worth data are distributed the same as those with data).
19
Table 3: Comparison of reported net worth groups
Expatriates
Population
Expatriates
Expatriates
Reported
net worth
Share, assuming Share, excluding
Number missing are <$1M
missing
Share
(housholds)
Total reported Share of
net worth ($B) total
Median
age
<$1M
$1-2M
$2-10M
$10-100M
$100M+
10,700
4,240
1,430
470
50
Has 8854, no RNW
No 8854, no RNW
1,860
17,040
Total
35,790
82.7%
11.8%
4.0%
1.3%
0.1%
63.4%
25.1%
8.5%
2.8%
0.3%
88.1%
5.7%
5.1%
1.0%
0.1%
$3.90
$6.18
$6.20
$13.76
$18.52
8.0%
12.7%
12.8%
28.3%
38.1%
47
56
53
51
45
47
45
100.0%
100.0%
100.0%
$48.56
100.0%
Notes: This table reports statistics for individuals who renounced between 2005 and 2018. Renouncer
counts are rounded to the nearest 10 for disclosure purposes. Population share is based on household shares
in the 2019 Survey of Consumer Finances.
Renouncers are relatively wealthier than the population. Specifically, millionaires are
relatively more common: 17% of renouncers (assuming none of those missing data are
millionaires) versus the estimate of 12% among households in the U.S. population; and note that
estimates for the U.S. population share of millionaires among individuals are lower, around 510%.21 While those renouncing are on average wealthier than the population, the numbers also
reveal the relatively small scale of ultra-wealthy expatriations. Between 2005 and 2018 only
about 50 renouncers reported net worth above $100 million. However, although small in number,
these individuals may have an outsize impact on policy; their decisions to expatriate tend to
show up in the news and spur legislative changes. 22 Interestingly, above $1 million, the median
age at expatriation decreases with reported net worth.
Taken together, the information on income and wealth shows that those who have chosen to
renounce citizenship were on average higher income and higher wealth than the population, but
21
The 2018 Credit Suisse Global Wealth Report estimates that 17.35 million Americans were millionaires, or 7.1%
of the adult population.
22
For example, legislative changes in the 1990s reportedly came about because then President Bill Clinton read
about the tax-motivated expatriation of six wealthy Americans in Forbes magazine (Cooper and Melton 1995). More
recently, Senators Chuck Schumer and Bob Casey proposed a bill to punish Facebook co-founder Eduardo Saverin
for his pre-Facebook IPO expatriation (Romm 2012). The bill, titled the Expatriation Prevention by Abolishing TaxRelated Incentives for Offshore Tenancy, or Ex-PATRIOT Act, failed to make it out of committee.
20
this average obscures significant heterogeneity within the renouncer population: a few outliers in
each year strongly influence the average values. The pattern over time shows that average
income and wealth among renouncers has been trending down.
4.4.
Destination jurisdictions
Finally, I provide information about renouncers’ destination jurisdictions. “Destination” is
perhaps a misnomer given that many of these individuals always lived in the foreign jurisdiction
or moved there many years prior to dropping U.S. citizenship. Nonetheless, destination here
refers to the foreign jurisdiction listed as an individual’s country of tax residency (when reported)
or general residency (when tax residency is not reported or available). 23 Renouncers’ destinations
are of interest generally, and may also provide some information about whether taxes are an
important factor in the expatriation decision.
Figure 7 shows the share (in Panel A) and count (in Panel B) of renouncers in each year going
to the top five destination jurisdictions (by total count from 1998-2018), and all others. Over
time renunciation has become more concentrated in the top five destinations, with the share
going to destinations outside these top five falling from about 50% in the 2000s to 30% in 2013,
although this share ticked back up to 40% by 2018. In recent years, the share renouncing to
Canada has risen dramatically. Also of note is the sharp rise and gentler fall in renunciations to
Switzerland.
23
In almost all cases, tax residency and general residency are the same: more than 99% of the records with both tax
residency and general residency have the same jurisdiction reported for both.
21
Figure 7: Share and count of renunciations to top destination jurisdictions
Notes: Panel A plots the share of individuals in each year renouncing to each of the top five jurisdictions,
or all others. Panel B plots the count of individuals renouncing to each of these jurisdictions, or all others.
I next consider how the pattern of renunciations to certain jurisdictions relates to the base of
U.S. citizens filing from those jurisdictions. If renunciation were equally likely regardless of
where a U.S. citizen living abroad is located, then the number of U.S. citizens filing from a
22
jurisdiction should correlate perfectly with the number of U.S. citizens dropping their citizenship
in that jurisdiction. To test whether the data follow such a pattern, I construct two rankings: first,
I rank foreign jurisdictions by the average number of U.S. tax filings received each year from
each jurisdiction; second, I rank the same foreign jurisdictions by the average number of
renunciations each year listing that jurisdiction as their destination. I then produce a scatterplot
of these rankings. I do this exercise separately for the years 2007-2010, and 2011-2018, in order
to test whether the patterns change before and after FATCA. 24
Figure 8 shows the rank-rank plots described above. A few patterns are worth noting. First,
most jurisdictions fall close to the 45-degree line, suggesting that the correlation between foreign
filings and renunciation is strong, on average. Second, there are clusters of jurisdictions that fall
away from the 45-degree line. Above the line are jurisdictions whose renunciation rank is higher
than their foreign filer rank; U.S. citizens filing from these jurisdictions are more likely to
renounce citizenship, on average, than those filing from other jurisdictions. The prevalence of tax
havens among these clusters suggests that tax considerations do play a role in some citizenship
decisions.25 Those below the line are jurisdictions where renunciation is less common than would
be expected, based solely on the number of foreign filings. The difference between the pre- and
post-FATCA plots also suggests the composition of renouncers may have changed between the
two time periods. I study these jurisdiction patterns further in Section 5.1.3.
24
At present I have comprehensive data on foreign filings by year and jurisdiction only since tax year 2007. If in
future these data are available for earlier years, one could extend this analysis to include those additional years.
25
In this discussion, and later in Section 5.1.3, I rely on the list of tax havens used in Johannesen et al. (2020). As
they note in footnote 1, “This list does not have any official role in IRS enforcement efforts; the IRS does not have
an officially accepted definition of a tax haven.”
23
Figure 8: Comparing jurisdictions by renouncer rank vs. foreign filer rank
Notes: This figure plots each jurisdiction’s rank based on renunciations (the average annual count of
individuals reporting the jurisdiction as their destination when renouncing) and foreign filings (the average
annual count of U.S. tax filings received from the jurisdiction). The ranks are calculated separately for the
years 2007-2010 and 2011-2018 to test whether patterns change before and after FATCA.
Finally, one might ask whether the top destinations of renouncers differ when focusing on
particular sub-groups (e.g., the wealthy). In general, this is not the case; although there are some
small differences, the top jurisdictions are consistent when looking within various subgroups. In
the Appendix, Table 7 shows the top ten destinations within each reported net worth group, and
Table 8 the top ten destinations within each renouncer classification (Mover vs. Dropper).
5. Explaining the increase in renunciations
This section addresses the second of my three research questions: Why are individuals
renouncing? I focus first on explaining the recent increase, which as shown above is mainly
driven by Droppers. I then consider renunciations by Movers and their connection to U.S.
expatriation tax policy.
24
5.1.
Explaining renunciations by those already living abroad
I develop a framework for the decision of those living abroad to maintain or drop citizenship
using a simple option value approach. I then use this framework to motivate empirical tests, first
using individual-level data to test various determinants of renunciation and confirm that age is
positively correlated with renunciation, then using jurisdiction-level data to test what
characteristics correlate with greater renunciation frequency, and finally using a difference-indifferences approach to show that increased compliance costs help explain the recent increase in
renunciations.
5.1.1.
Theoretical framework
For U.S. citizens living abroad, U.S. citizenship can be thought of in an option value
framework. For those abroad, U.S. citizenship represents an American-style call option in which
the foreign resident U.S. citizen retains the right to return to the U.S. to live or work at some
point in the future. Typically, option value can be decomposed into time value and intrinsic
value. Time value for the option on U.S. citizenship corresponds to age: as individuals get older,
the remaining time in which they can exercise the option decreases, leading the value of that
option to decrease as well. All else equal, this suggests that the probability of renunciation should
increase with age.
The intrinsic value of the option on U.S. citizenship comprises many components. First,
consider that for a typical financial option, the value of that option increases with the volatility of
the underlying asset. Similarly, the value of U.S. citizenship should increase as volatility
increases. Volatility in this case could include global economic uncertainty and the political
stability of foreign countries relative to the United States; those living in more stable countries
may consider themselves less likely to want or need to exercise the option to return to or work in
the U.S., and thus be more likely to renounce U.S. citizenship. Other components of the intrinsic
value could include the tax rates of the foreign country relative to the U.S. and the relative value
of the foreign country’s passport. For those living in countries with lower relative rates, the value
of the option on U.S. citizenship would be lower, while for those in countries with a relatively
more valuable passport, the option value of being able to use one’s U.S. passport would be lower.
25
Finally, in addition to the value of the option, consider the cost of maintaining it. This cost has
always included remitting one’s annual tax liability, if any, as well as the compliance costs,
including time and effort, of annual filing of U.S. tax returns. These compliance costs have
increased in recent years, with additional forms required for many taxpayers, both by tax
agencies and financial institutions. In the next sections I test whether the predictions of this
framework are borne out in the data.
5.1.2.
Individual determinants of renunciation
To begin testing the implications of the options model, I focus first on identifying
characteristics associated with the costs and benefits of the decision of those living abroad to
renounce citizenship. Although not all the reasons someone might choose to renounce are
captured in tax filings, administrative microdata still allow me to test how several key
characteristics relate to renunciation.
The base for this study is the set of all U.S. tax filings by those filing from abroad. This
includes Form 1040 filings, and other linked tax form data, for those filing from abroad for tax
years 2007-2017. Among these filings, I identify the individuals who ultimately renounce
citizenship, and flag the tax year prior to the year in which they expatriate, dropping subsequent
filings for these individuals if they appear. 26 As noted above, I consider the tax information in the
year prior to the year of expatriation as the most relevant, because it is represents a complete year
of earnings and other taxpayer decisions. The final dataset contains about 17,000 instances of
citizenship renunciation (I include only primary filers, and am unable to include individuals
without TINs or linked tax filings), out of more than four million tax filings from those living
abroad.
I develop a simple linear probability model, regressing Renounce (the decision to renounce
citizenship in the following year) on a set of individual-year covariates and, in some
specifications, jurisdiction, year, or jurisdiction X year fixed effects:
𝑅𝑒𝑛𝑜𝑢𝑛𝑐𝑒
= 𝛽(𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑠 ) + 𝛼 + [𝛼 ] + [𝛼 ] + 𝜀
26
Some individuals who expatriate continue to file Form 1040 or Form 1040 NR after renunciation, depending on
their income sources and other circumstances.
26
These covariates include: total positive income (TPI) in millions of dollars; wages as a share
of total positive income (0 if no TPI); a dummy indicating the taxpayer had a positive tax
liability; dummies indicating whether a taxpayer had nonzero values reported for Schedule C or
Schedule E income, respectively 27, a dummy indicating that a charitable contribution deduction
was claimed on Schedule A, a dummy indicating Form 709, the U.S. Gift and GenerationSkipping Transfer Tax Return, was filed; and a dummy indicating a taxpayer received any notice
from the IRS. In some specifications I include age (in years), though this slightly lowers the
observation count because of some missing data on dates of birth. Table 9 in the Appendix
presents summary statistics for these variables.
The results of the basic linear probability model are shown in Table 4. 28 The dependent
variable is coded as 100 or 0, so that the coefficient estimates represent the effect in percentage
points for each covariate, holding all others constant. The different columns include various
combinations of fixed effects, culminating in column (6) with year X jurisdiction fixed effects
included (so that the model seeks to explain the decision to renounce within a jurisdiction in a
year). Figure 18 in the Appendix plots the coefficient estimates, scaled by the mean probability
of renunciation, to show the estimated percent change in the probability of renunciation resulting
from a 0 to 1 change in each binary covariate. The figure also compares the coefficient estimates
when including or excluding Movers, showing similar coefficient estimates.
27
Schedule C includes income and loss from a business or profession practiced as a sole proprietor; Schedule E
includes income and loss from rental real estate, royalties, partnerships, S corporations, estates, trusts, and residual
interest in real estate mortgage investment conduits (REMICs).
28
In this main specification, seeking to explain the recent increase in Droppers, I include only the Droppers as
renouncers, excluding Movers from the dataset in any year where they appear. I also run the models including all
renouncers, and the results are nearly identical; see Table 10 in the Appendix.
27
Table 4: Individual linear probability model results
Dependent variable:
[1]
Total Positive Income
($ millions)
Binary: Renounce in following year (100/0)
[2]
[3]
[4]
[5]
[6]
0.0102
0.0102
0.0108
0.0069
0.0074***
0.0071
(0.0064)
(0.0063)
(0.0064)
(0.0056)
(0.0018)
(0.0055)
Wage share
(% of TPI)
-0.0844
-0.055
-0.1119
-0.1203*
-0.1086***
-0.1044**
(0.0783)
(0.0603)
(0.0736)
(0.0549)
(0.0075)
(0.0386)
Positive tax liability
(1/0)
-0.0805
-0.0834
-0.0695
-0.0724
-0.0662***
-0.0645
(0.0641)
(0.0650)
(0.0630)
(0.0466)
(0.0059)
(0.0446)
Any Sch C income
(1/0)
0.0886***
0.0937***
0.0672**
0.0597***
0.0482***
0.0518**
(0.0248)
(0.0278)
(0.0229)
(0.0177)
(0.0088)
(0.0176)
Any Sch E income
(1/0)
0.0066
0.0042
-0.0104
0.0135
-0.0052
0.0033
(0.0372)
(0.0383)
(0.0366)
(0.0269)
(0.0081)
(0.0262)
Schedule A charity
(1/0)
-0.0136
-0.0218
0.0176
-0.1208
-0.0984***
-0.0963
(0.0881)
(0.0856)
(0.0903)
(0.0926)
(0.0112)
(0.0926)
Filed gift tax return
(1/0)
2.2867***
2.2807***
2.2498***
2.2302***
2.1869***
2.1767***
(0.4204)
(0.4172)
(0.4143)
(0.4277)
(0.0551)
(0.4175)
Received any notice
(1/0)
0.1070**
0.1023*
0.0313
0.1163*
0.0398***
0.0447
(0.0544)
(0.0547)
(0.0197)
(0.0529)
(0.0080)
(0.0280)
0.0019
0.0022***
0.0022**
(0.0013)
(0.0002)
(0.0009)
Age
(years)
Constant
Year FE
Jurisdiction FE
YearXJurisdiction FE
Observations
Adjusted R2
Mean dep. var.
0.3954***
0.2890***
(0.1356)
(0.0712)
No
No
No
No
No
No
Yes
No
No
No
Yes
No
Yes
Yes
No
No
No
Yes
4,831,000
0.0005
0.343
4,790,000
0.0005
0.343
4,831,000
0.0013
0.343
4,831,000
0.0045
0.343
4,790,000
0.0053
0.343
4,790,000
0.0066
0.343
Notes: *p<0.1; **p<0.05; ***p<0.01. Standard errors, clustered by year and by jurisdiction, are shown in
parentheses. The dependent variable is coded as 100 or 0 so that the coefficient estimates represent the
effect in percentage points for each covariate, holding all others constant. “Movers” that can be linked to
Form 1040 filings as a primary filer are excluded here; results when including them are shown in Table 10
in the Appendix.
28
The results suggest several individual characteristics connected with the decision to renounce
citizenship. Filing a gift tax form, which is relatively rare in general, is very strongly associated
with renunciation (consistent with a pattern I demonstrate later related to the net worth threshold
for covered expatriate designation). The presence of Schedule C income is positively associated
with renunciation, while having a higher wage share of income is negatively associated with
renunciation (interesting given the pattern shown in Section 4.3 that Droppers had relatively high
wage income, suggesting that those renouncing had both high wage income and non-wage
income). Having a positive tax liability is negatively associated with expatriation, although this is
only statistically significant at standard levels in one specification. However, if the association is
truly negative, this would be consistent with an explanation in which long-term foreign resident
U.S. citizens drop citizenship because of increased compliance costs (filing new and more
complicated forms), not because of tax liability itself.
Most relevant to the option value framework, the results show that age is significantly, and
positively, correlated with the decision to renounce. This is consistent with the prediction that as
individuals age, the time value of their option on U.S. citizenship decreases, leading to lower
values for that option and renunciation becoming more common.
5.1.3.
Jurisdiction characteristics and renunciation frequency
The previous section tested whether certain individual characteristics, observable in tax
filings, correlate with the decision to renounce in a way consistent with the option value
framework. In this section, I similarly test whether characteristics of the jurisdictions from which
foreign-resident U.S. citizens file their taxes correlate with the prevalence of renunciations from
those jurisdictions. The option value framework predicts a higher value of U.S. citizenship (and
thus a lower rate of renunciation) for those living in foreign jurisdictions where they perceive a
higher probability of wanting or needing to exercise the option by returning to live or work in the
United States.
29
For this test, I collapse the individual-level data to a jurisdiction-level dataset and estimate the
following equation:29
𝑅𝑒𝑛𝑢𝑛𝑐𝑖𝑎𝑡𝑖𝑜𝑛 𝑠ℎ𝑎𝑟𝑒 = 𝛽 𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑠 + 𝜀
The renunciation share is defined as the total number of renunciations in a given jurisdiction
from 2008 to 2018, divided by the unique set of U.S. tax filers from that jurisdiction over the
period 2007 to 2017. The denominator approximates the set of “potential renouncers” – those
who filed from abroad and could have chosen to renounce U.S. citizenship. Dividing the total
number of renunciations to a jurisdiction by this set of potential renouncers gives an outcome
value that allows comparison of the relative frequency of renunciation across jurisdictions.
The covariates are motivated by the option value framework. First are three binary variables
indicating whether a jurisdiction is designated as a tax haven, relying on the designations in
Johannesen et al. (2020); offers citizenship-for-sale, based on Christians (2017); and is majority
native English-speaking. Next, I include separately the average percentile rank of the jurisdiction
on two measures from the World Bank’s Governance Indicators: the Rule of Law and Political
Stability indices (higher values indicate better governance). I also include the average percentile
rank of each jurisdiction’s passport value according to the Henley Passport Index, a ranking of
passports based on the number of destinations accessible without a prior visa (higher values
indicate a more valuable foreign passport). Finally, I include the average annual change in real
GDP, according to the IMF. Summary statistics are shown in the Appendix, Table 11.
The results are shown in Table 5 below. Tax haven jurisdictions are associated with higher
renunciation shares, consistent with lower taxes motivating renunciation for at least some
individuals. The Rule of Law index is also positively correlated with renunciations, consistent
29
I take this approach to focus specifically on the jurisdiction characteristics and to capture associations over a
longer time period (collapsing across years), relative to the individual approach above. However, I also test the
relationship between the jurisdiction characteristics and the probability of renunciation by merging the
characteristics into the individual-level data and running similar specifications to those in the prior section, replacing
the jurisdiction fixed effects with the characteristics I discuss in this section. The results are generally consistent
between the two approaches, though the two are not directly comparable: the individual approach studies the
decision to renounce in a given year, while the jurisdiction approach studies the frequency of renunciations over a
longer time period. See Table 12 in the Appendix for the results of the individual-level regression with jurisdictionlevel covariates.
30
with the option value framework’s prediction that individuals living in more stable jurisdictions
anticipate a lower likelihood of exercising the U.S. citizenship option, and thus are more likely to
renounce. Similarly, jurisdictions with more valuable passports are associated with higher
renunciation shares. In these specifications, citizenship-for-sale (CFS) is negatively correlated
with renunciation share; I also run specifications excluding CFS, or including a tax haven X CFS
interaction (see Table 13 in the Appendix). Removing CFS does not materially affect the other
covariate estimates, and the haven interaction suggests that the CFS effect is driven by the few
non-haven CFS jurisdictions, like Bulgaria and Serbia, where renunciation is relatively
uncommon. The lack of an effect for the Political Stability index likely reflects the strong
correlation between the Rule of Law and Political Stability indices. Overall, the results are
generally supportive of the predictions of the option value framework.
31
Table 5: Jurisdiction-level regression results
Dep. var.: Total renunciations/unique foreign filers
[1]
[2]
[3]
[4]
Tax haven
(1/0)
0.0262**
0.0266**
0.0353**
0.0171*
(0.0125)
(0.0129)
(0.0156)
(0.0088)
Citizenship-for-sale
(1/0)
-0.0095
-0.0112*
-0.0155*
-0.0068
(0.0064)
(0.0065)
(0.0082)
(0.0053)
English-speaking
(1/0
-0.0056
-0.0121
-0.0157
-0.0052
(0.0073)
(0.0087)
(0.0097)
(0.0059)
Rule of Law index
(percentile)
0.0323***
0.0182**
0.0201***
(0.0093)
(0.0072)
(0.0062)
Political Stability index
(percentile)
-0.0032
0.0015
-0.0042
(0.0069)
(0.0078)
(0.0051)
Passport ranking
(percentile)
0.0120**
0.0088
(0.0057)
(0.0055)
Change in Real GDP
(percentage points)
Constant
Observations
Adjusted R2
0.0002
(0.0005)
0.0091***
-0.0038
-0.0047
-0.0033
(0.0010)
(0.0033)
(0.0036)
(0.0030)
213
0.0852
205
0.1707
196
0.2134
187
0.2253
Notes: *p<0.1; **p<0.05; ***p<0.01. Heteroskedasticity-robust standard errors are shown in parentheses.
5.1.4.
Testing the compliance cost explanation
I now turn to an analysis which seeks to explain the increase in renunciations seen in the past
decade. Many public press articles about the recent increase include anecdotes attributing the
increase to increasing compliance costs for U.S. citizens living abroad; academic articles have
posited this explanation as well (e.g., Kudrle (2015), De Simone, Lester and Markle (2020)).
These articles highlight that over the past decade there has been a general increase in offshore
financial enforcement, including FATCA, as well as ad hoc legal and information actions and
bilateral treaties that, for certain countries, increased the flow of information to the IRS about
32
U.S. citizens’ financial assets and earnings abroad. In general this meant that foreign financial
institutions (FFIs) faced increasing compliance costs when working with U.S. citizen customers,
and they became less willing to do so.30 Thus for those already living abroad, maintaining U.S.
citizenship in the 2010s brought additional costs (e.g., difficulty dealing with local financial
institutions and increased filing requirements). 31 If one wished to remain abroad then these costs
were only avoidable by dropping U.S. citizenship. In what follows, I test empirically whether the
compliance cost narrative is consistent with the patterns visible in the data on renunciations.
In one test, I compare renunciation trends between jurisdictions that signed FATCA-related
Inter-Governmental Agreements (IGAs) with the U.S. and those that did not. The hypothesis is
that jurisdictions which sign IGAs are more cooperative in implementing increased enforcement
measures, which could induce citizenship renunciation because of increased hassle costs of
dealing with FFIs as a U.S. citizen, or by leading U.S. tax avoiders and evaders to drop
citizenship in an attempt to avoid detection. Since FATCA was passed in 2010, about half of all
foreign jurisdictions have signed IGAs specifying the terms under which foreign financial
institutions (FFIs) identify and report information about U.S. accounts. 32 These IGAs were
signed over time starting in 2012, with most signed by 2015 (Figure 19 in the Appendix plots the
annual counts, and Figure 20 shows a map shaded by each jurisdiction’s IGA status). For
purposes of this test, the meaningful comparison is the binary distinction of whether or not a
jurisdiction is an “IGA jurisdiction” (i.e., one that ultimately signs a FATCA IGA) rather than the
timing of the signing itself, because the obligations imposed on FFIs by FATCA are not created
by the IGAs; rather, FATCA imposed those obligations when it was passed, and the IGAs help
30
Press reports describe numerous anecdotes of U.S. citizens abroad facing such difficulties. See, e.g., Williams
(2014), “U.S. expats find their money is no longer welcome at the bank” and Graffy (2015), “The law that makes
U.S. expats toxic.” Some of these difficulties are only now starting to arise, as FATCA implementation was not
necessarily immediate; France, for example, was set to start reporting information in 2020, prompting an August
2019 article warning of pending bank account closures for 40,000 U.S. citizens (Goncalves 2019).
31
One important group of individuals who were particularly affected by the enforcement changes were those hiding
assets abroad. These individuals faced an ever-increasing likelihood of being discovered by the IRS. One response to
this would be to come clean, pay any necessary penalties, and maintain U.S. citizenship. Another response would be
to drop U.S. citizenship in an attempt to “sneak out” before the hidden assets could be discovered. However, because
hidden assets are unobservable it is not possible to test directly whether individuals with such assets were more
likely to expatriate following the increased enforcement actions.
32
There are two models of IGAs. Under Model 1, FFIs report information to a local agency which then
communicates with the IRS on an automatic basis. Under Model 2, FFIs communicate directly with the IRS. The
U.S. Treasury reports the countries with IGAs here.
33
the FFIs reconcile their FATCA obligations with any obligations under their domestic (non-U.S.)
law.33 By signing an IGA, a jurisdiction is identifying itself as one that is proactive about
implementation of the new enforcement system under FATCA, and thus more likely to have led
to increased compliance costs for U.S. citizens living there. 34 The main specification shown
below includes all signed IGAs, and is robust to alternate specifications limiting to only IGAs
signed in earlier years, and to also including jurisdictions that have “Agreements in Substance”
but have not yet signed an IGA.35
In another test, I split jurisdictions based on whether they are considered tax havens, relying
on the designations in Johannesen et al. (2020). Because Switzerland was subject to very focused
enforcement actions prior to FATCA in 2008 and 2009 (see Section 2.3 above), I run the tax
haven test twice, either including or excluding filings from Switzerland. The expectation for the
relative difference between havens and non-havens is ambiguous. FATCA increases the
compliance costs of those who are compliant; if those living in non-haven jurisdictions are more
likely to be compliant (and thus bear the full weight of increased compliance costs), we would
expect to see larger increases in renunciation in those non-haven jurisdictions. Conversely, if
those living in tax havens are mainly there for tax evasion or avoidance purposes, and perceive
FATCA as a signal that they will face increased scrutiny in future years, we may see a larger
increase in renunciations by these haven residents in an attempt to escape the U.S. tax system
before any detection of potential wrongdoing. Thus, it is not clear ex ante whether we should
expect to see relatively more or less renunciation activity in havens relative to non-havens after
FATCA.
These tests use a difference-in-differences approach, relying on the same underlying
individual-year level data as in the previous section, as follows:
𝑅𝑒𝑛𝑜𝑢𝑛𝑐𝑒
= 𝛼 + 𝛽 𝐼𝐺𝐴 ∗ 𝑃𝑜𝑠𝑡 + 𝛽 𝐼𝐺𝐴 + 𝛽 𝑃𝑜𝑠𝑡 + 𝛾(𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒𝑠 ) + 𝜀
33
Dharmapala (2016) uses a simple theoretical model to understand the effects of FATCA and IGAs.
News coverage of IGA negotiations and signings in IGA jurisdictions may also have exposed U.S. citizens living
there to relatively more information about FATCA and its obligations for U.S. citizens and for FFIs.
35
The three jurisdictions with signed IGAs with the highest average annual U.S. tax filers during the sample period
were Canada, the United Kingdom, and Israel; the three most frequent with Agreements in Substance were China,
Peru, and Indonesia; the three most frequent non-IGA jurisdictions were Argentina, Lebanon, and Egypt.
34
34
where 𝐼𝐺𝐴 is an indicator equal to one for jurisdictions that ultimately signed an IGA 36, 𝑃𝑜𝑠𝑡
is an indicator for tax years 2010 or later (i.e., renunciations in 2011 or later), and 𝐼𝐺𝐴 ∗ 𝑃𝑜𝑠𝑡 is
their interaction.37 For the tax haven tests, IGA is replaced by a dummy variable indicating
whether the jurisdiction is designated as a haven or not. As before, Renounce is an indicator
equal to 100 if an individual renounces citizenship the following year, and 0 otherwise.
Figure 9 shows graphically the difference in the trends (the average share of foreign filers
from a jurisdiction who renounce the following year) between IGA and non-IGA jurisdictions
(top panel) and between haven and non-haven jurisdictions (bottom panel). Tax years 2007-2009
are the “pre” period to the left of the red line, indicating the passing of FATCA in December
2010, while tax years 2010-2017 are the “post” period. The IGA jurisdictions clearly experience
a rise in the share of filers renouncing citizenship, relative to the non-IGA jurisdictions. And
among havens, after removing Switzerland, we see that the haven share is relatively constant,
while the non-haven share rises after FATCA goes into place. Comparing the haven patterns
reveals that Switzerland’s renunciation share began to increase prior to FATCA, coincident with
the 2008 and 2009 enforcement actions specifically targeting Switzerland.
36
In the main specification, I include all IGAs through the end of 2019; I also test alternative specifications
including only IGAs signed through 2017 or 2015 and find consistent results (see Table 14 in the Appendix).
37
The results are robust to instead using a specification with year and year X IGA dummy variables, rather than
pre/post-FATCA. Figure 23 in the Appendix shows the coefficients on the year X IGA covariates, in a pattern
consistent with the trend in average renunciation shares shown in Figure 9.
35
Figure 9: Annual share renouncing for specified jurisdiction splits
Notes: This figure plots the average share of individuals renouncing in the following year, based on the
data underlying the individual regressions and splitting the sample into groups based on IGA jurisdictions
or tax haven jurisdictions. Both the IGA and Haven patterns are robust to a series of alternate specifications
(IGA definition, jurisdiction restrictions, inclusion of Movers as well as Droppers); see Figure 21 and Figure
22 in the Appendix.
The results of the corresponding regressions are shown in Table 6 below and confirm that
renunciation became relatively more common in IGA jurisdictions after FATCA, relative to nonIGA jurisdictions. For the tax haven tests, the most relevant results are those excluding
Switzerland, given the special attention paid to Swiss activities for several years prior to FATCA.
We see that consistent with the graph, the relative rate of renunciation from havens vs. nonhavens was smaller after FATCA vs. before FATCA. This is consistent with the compliance costs
explanation, in which those living in non-haven jurisdictions experience newly increased
compliance costs under FATCA, and are more likely to drop citizenship in response, while those
in haven jurisdictions do not experience as strong an increase in compliance costs.
36
Table 6: Individual difference-in-difference results
Dependent variable:
Post
IGA jurisdiction
Binary: Renounce in following year (100/0)
[1]
[2]
[3]
0.0372*
0.3013***
0.2969***
(0.0211)
(0.0974)
(0.0994)
0.4611***
0.4582***
(0.0881)
(0.1168)
0.4343
-0.3151***
(0.5484)
(0.1151)
Yes
Yes
0.0559*
(0.0334)
Post X IGA
0.3144***
(0.1003)
Tax haven jurisdiction
Post X Haven
Individual covariates
Sample
Observations
Adjusted R2
Mean dep. var.
Yes
Excl. Movers
4,831,000
0.0013
0.343
Excl. Movers Excl. Movers and Switzerland
4,831,000
4,686,000
0.0025
0.0012
0.343
0.291
Notes: *p<0.1; **p<0.05; ***p<0.01. Standard errors clustered by year and by jurisdiction are shown in
parentheses. IGA jurisdiction is an indicator for jurisdictions that signed a FATCA IGA in or before 2019;
Haven is an indicator for jurisdictions designated as tax havens in Johannesen et al. (2020). Post is an
indicator for tax years 2010 and later. These results are robust to various alternate variable definitions and
jurisdiction restrictions; see Table 14 (IGA test) and Table 15 (haven test) in the Appendix.
Taken together, these two tests provide empirical evidence consistent with the compliance
cost explanation for the recent increase in renunciations. The option value framework predicts
that if compliance costs increase, renunciations should increase, and the two tests here show that
renunciations became relatively more common after FATCA in jurisdictions where resident
individuals were more likely to experience increased compliance costs, namely, IGA-signing
jurisdictions and non-tax haven jurisdictions.
37
5.2.
Evaluating the effects of expatriation tax law changes
The previous section focused on explaining the recent increase in renunciations by Droppers.
What about Movers? Although in fact a small share of the total, those who at one point filed
from the U.S. and subsequently moved abroad and renounced citizenship represent more of the
stereotypical individual that may come to mind when thinking about expatriation and citizenship
renunciation. Indeed, these are the types of individuals cited by legislators when discussing the
expatriation tax system, and who in the past have apparently prompted changes to that system.
Understanding their behavior is important for evaluating the effects of prior tax law changes, and
considering future policy.
5.2.1.
General trends in renunciation by Movers
As noted above, Movers may have a large set of reasons for moving abroad and renouncing
their citizenship, including family or other ties abroad, but it is also possible that tax
considerations play an important role in their decisions. To better understand the relationship
between renunciations by Movers and the tax system, I begin by showing their annual counts;
this is the solid gray line in Figure 10 below (the same as the gray bar in Figure 2 above). I then
adjust this count in two ways. First, note that my preferred categorization of individuals as
Movers or Droppers relies on five years of pre-renunciation tax filings, to allow for the fact that
some of those moving from the U.S. take several years to settle in before renouncing. One
drawback of this approach is that it limits observations to those renouncing in 2005 or later. To
address this, I produce an alternate categorization based only on tax filings in the two years prior
to renunciation, shown with the blue line. Second, because much of the public press and
legislative focus on this subject has centered on wealthy or high-income Movers, I produce a set
of counts restricting to those Movers who have high net worth (above $622K) or high income
(AGI greater than $200K in the year prior to renunciation), shown with the dashed lines.
38
Figure 10: Annual count of renunciations by Movers
Notes: This figure plots the annual count of renunciations made by those designated as Movers, either based
on having filed from the U.S. at least once during the prior five years (in gray) or two years (in blue). Total
counts are shown with solid lines, and those including only high wealth (net worth > $622K) or income
(prior-year AGI > $200K) are shown with dashed lines. Vertical dashed lines indicate years with legislative
changes: 2004 (AJCA), 2008 (HEART Act), and 2010 (FATCA).
Over time there has been an increase in renunciations by Movers, although the increase in the
past decade is less extreme than that seen above for Droppers. The number is still small, with
annual counts of around 100-200 during the 2000s, and 300-400 in the 2010s. The increase in
renunciations by Movers between 2004 and 2008 could indicate that the tax law changes in those
years had some effect; I explore this further below. The acceleration in renunciations after 2010
suggests that the increase in offshore financial enforcement may also have played a role in the
renunciation decisions for Movers, just as was seen above for Droppers; perhaps some U.S.
citizens who previously would have moved abroad but maintained citizenship chose instead to
renounce that citizenship when facing increased compliance costs during the 2010s.
5.2.2.
Relating tax law changes to renunciations
How did the 2004 and 2008 tax law changes affect individuals’ decisions of whether and
when to renounce? The 2004 AJCA made two important changes to the expatriation tax system:
39
(1) it raised the net worth threshold for designation as a covered expatriate from $622K to $2M;
and (2) it removed the ability to challenge one’s designation as tax-motivated, replacing it with a
strictly objective test based on net worth, past-five-years average tax liability, and certification of
compliance with the last five years of tax filings.
Consider how these two changes would affect the costs and benefits of renunciation. The net
worth threshold change would lower the cost for certain individuals. For individuals with true net
worth between $622K and $2M, renunciation prior to the change would have included
designation as a covered expatriate and the ensuing effort to either challenge that designation or
deal with the next-10-years tax consequences. After the change, these individuals could renounce
and report their true net worth without being designated as covered expatriates. The change may
also have affected some individuals with true net worth above $2M, as the cost of getting under
the threshold was lowered.38 Individuals with true net worth below $622K would be unaffected,
as both before and after the change they were not at risk of covered expatriate designation. In
sum, the change lowered the cost of expatriation for those with net worth above $622K, and
especially for those above $622K and below $2M. All else equal, this predicts more renunciation
by such individuals as a result of the increase in the net worth threshold.
The removal of the ability to challenge covered expatriate designation should work in the
opposite direction, raising the cost of renunciation for individuals who previously could have
successfully challenged their covered expatriate designation. Consider two wealthy individuals
(above the $2M threshold), identical in every respect except that one has no ties abroad, while
the other does have strong ties in the country to which they plan to renounce. Prior to this change
the former individual would be designated as a covered expatriate and may have some difficulty
challenging that designation; the latter would also be designated as covered but would have an
easier time challenging that designation. After the 2004 removal of the ability to challenge, both
individuals would be designated as covered and remain so. The effect of this removal is thus a
change in the relative cost of renunciation: for the individual with strong ties abroad, the relative
cost of renunciation has increased when compared to the cost for an individual without strong
38
For example, consider someone with $2.1M in true net worth; prior to the change, they would need to somehow
lower their reported net worth by nearly $1.5M to fall below the $622K threshold, but only by $100K to fall below
the new $2M threshold.
40
ties abroad. All else equal, this predicts relatively fewer renunciations by those with ties abroad,
and thus relatively more by those without such ties.
The 2008 HEART Act’s introduction of the mark-to-market exit tax was a more fundamental
change to the expatriation tax system. It changed the consequences of covered expatriate
designation from an uncertain future liability based on an income over the next 10 years with an
immediate, up-front tax liability based on unrealized capital gains above an exemption threshold
(although this liability could be temporarily deferred). This change could in principle push in
different directions. On the one hand, an up-front liability could be perceived as more costly than
the uncertain future liability, and thus make renunciation seem more costly than under the prior
system. On the other hand, the ability to pay the one-time exit tax and cleanly walk away may
have been more desirable to some individuals, relative to the lingering connection to the U.S.
that would persist under the next-10-years system. Whether the mark-to-market tax would be
more or less desirable than the earlier system would also depend crucially on the extent of an
individual’s unrealized capital gains; someone with significant wealth but relatively low amounts
of unrealized capital gains would face little or no liability under the mark-to-market tax, which
exempts the first several hundred thousand dollars of gains. In sum, the change from the earlier
system to the mark-to-market exit tax was certainly a significant change, but its effects would
likely not push unambiguously in the same direction for all individuals.
5.2.3.
Trends around the tax law changes
To further understand how the AJCA and HEART Act affected the number of renunciations,
and the types of individuals renouncing, we can look for evidence in the patterns of renunciation
around the tax law changes. Consider Figure 11, which shows the count of renouncing
individuals, grouped by reported net worth, in each half-year time period from 2003-2010.
Beginning with the effect of the AJCA, and focusing on the net worth threshold change, we
would ideally compare the number with net worth between $622K and $2M, before and after the
threshold change, to see whether under the post-AJCA regime in which they are no longer
designated as covered expatriates, their numbers rise. Unfortunately, data on net worth is not
available for the pre-AJCA renunciations, and even post-AJCA, many individuals either do not
have a filed Form 8854 or do not have reported net worth data available.
41
Nevertheless, we can observe that the number of renunciations increased after the AJCA
change; if the expatriate provisions in the AJCA were intended to discourage renunciations, a
simple assessment of the trend suggests they may not have achieved that goal. We can also
observe that after the net worth threshold was increased, there were a handful of individuals with
reported net worth in the $622K to $2M range (the green bars); it is possible they were induced
to renounce by no longer facing the cost of covered expatriate designation. At the same time,
however, there were a similar number of renunciations by those with reported net worth above
$2M (the top, light blue bars), confirming that renunciation was still desirable for some
individuals even when facing the costs of covered expatriate designation. Without further detail
on the net worth of all renouncers, both before and after the 2004 law change, it is difficult to
draw firm conclusions about the effect of the change.
Figure 11: Annual count of renunciations, before and after AJCA and HEART Act changes
42
Notes: This figure plots the count of individuals renouncing in each half-year period grouped by reported
net worth. For 2004, the periods are split around June 4, 2004, when the net worth threshold for designation
as a covered expatriate increased from $622 thousand to $2 million, as part of the AJCA. For 2008, the
periods are split around June 18, 2008, when the HEART Act’s mark-to-market tax provisions went into
place. For pre-AJCA expatriations, we can only observe whether an individual was above the net worth
threshold (in gray) or not (in orange). Post-AJCA, most individuals filing Form 8854 report net worth, and
thus can be grouped into three buckets based on the threshold changes, though there are still many
individuals (in darker blue) without reported net worth data.
Focusing on the patterns around the 2008 HEART Act, some interesting patterns are visible.
In the few periods prior to the HEART Act change, the number of renouncers reporting net worth
of $622K-$2M and above $2M increased noticeably. This could reflect individuals accelerating
their renunciations to avoid the mark-to-market tax, which was in discussion for at least several
months prior to being passed and signed into law on, and affecting expatriations on or after, June
17, 2008.39 The number of high-wealth renunciations fell in the second half of 2008 and first half
of 2009, which again would be consistent with individuals moving renunciation forward to avoid
the mark-to-market tax. However, the confounding effects of the financial crisis may also have
affected the ability of U.S. citizens to move abroad or affected their decisions about whether to
incur the costs of renunciation, and could also help to explain the drop in renunciations. Again,
more complete information about these individuals would help to say more with greater
certainty.
5.2.4.
Responses to the net worth threshold
The data patterns discussed above suggest that some individuals responded to the changes in
expatriation tax law. Further evidence of taxpayers responding to the expatriation tax rules can
be seen by examining the pattern of filings with reported net worth above and below the $2M net
worth threshold for designation as a covered expatriate.
As described above in Section 2.2, expatriating individuals are subjected to a test that
determines whether they are a covered expatriate. The test has three components, any one of
which results in designation as a covered expatriate: (1) net worth above a threshold; (2) pastfive-years average income tax liability above a threshold; and (3) failing to certify compliance on
39
Reichenberg Sherr (2008) notes that the expat provision ultimately passed as part of the HEART Act is “similar
to…the expat provision in a prior bill, H.R. 3997, which was passed by both the House and Senate in December
2007 but did not get enacted due to other differences between the House and Senate bills.”
43
U.S. taxes for the five years prior to expatriation. Covered expatriate status results in additional
filing requirements, as well as potential additional tax liability. Prior to the HEART Act in 2008,
this tax liability was based on income earned during the 10 years following expatriation, which
could be liable for U.S. income taxation. Since the HEART Act, this tax liability is a mark-tomarket exit tax based on the value of all assets owned on the day prior to expatriation, with taxes
applied to gains above a statutory exemption.
For most covered expatriates, the net worth threshold is the crucial component. 40 Since mid2004, the net worth threshold has been constant at $2M. A histogram of renouncers’ reported net
worth around this threshold reveals a strong response, as shown in Figure 12: a sharp drop-off in
the number of renouncers reporting net worth just above the threshold. This figure shows the
aggregate histogram for all renunciations from mid-2004, when the AJCA took effect and net
worth data become widely available, through 2018. Although not presented here for disclosure
reasons, the pattern is also visible within each year. 41 There are several plausible explanations for
this drop-off: some potential renouncers with net worth above $2M may have been discouraged
from renouncing; some may have taken actions to reduce net worth below the $2M threshold (for
example, by making gifts or charitable contributions); and some may have reported net worth
lower than their actual net worth, in order to appear below the threshold. In addition, recall that
only about half of renouncing individuals have a filed Form 8854 with reported net worth data
available; it is possible that some individuals with net worth above the threshold chose not to file
Form 8854.
40
Among all covered expatriates, nearly 90% are over the net worth threshold, while only about 25% are over the
average income tax liability threshold. The evidence suggests there is little direct response to the average income tax
liability threshold, in that there is no bunching below the threshold (see Appendix, Figure 25). One explanation is
that it is harder for taxpayers to adjust an average based on past-5-years income tax liabilities than it is to adjust
reported net worth at the point of expatriation.
41
That the pattern is visible both before and after the HEART Act suggests that covered expatriate designation was
viewed as costly even without the mark-to-market exit tax consequences introduced under the HEART Act.
44
Figure 12: Histogram of reported net worth around $2 million
Notes: This figure plots the count of renouncers in each $100K bucket around the $2M threshold for
designation as a covered expatriate. Renunciations with a filed Form 8854 and available reported net worth
data, after the AJCA (mid-June 2004) through 2018, are included. The drop-off in filings with reported net
worth occurs exactly at the cutoff for covered expatriate designation, suggesting it is this cutoff that is
driving the observed behavior; there is no drop-off at either $1M or $3M, suggesting that round-number
bunching can be ruled out as an explanation for the observed pattern (see Figure 24 in the Appendix).
There is evidence that for some taxpayers, gifts may have been used to get below the
threshold. 8% of the individuals who report net worth of $1-2M would have had net worth above
$2M if gifts they reported making in the 0-2 years prior to renunciation were added to their
reported net worth. A handful of individuals similarly would move from below the threshold to
above it if their pre-renunciation charitable contributions were added to their reported net
worth.42 Still, even after adjusting the reported net worth amounts to include recent gifts and
charitable contributions, a large “hole” to the right of the threshold remains. One feature of Form
8854 (the expatriation tax form) is that it requires individuals to provide a balance sheet with
42
For this analysis, I rely on gift amounts as reported on Form 709 and charitable contributions reported on
Schedule A.
45
assets listed by asset type; although not presently available, these data could in future be used to
further explore the patterns shown here.
In sum, there is evidence consistent with the hypothesis that renunciations were responsive to
these tax law changes, although it is not possible to draw conclusions with certainty. When
pairing the patterns shown here with the income and wealth trends discussed earlier in Section
4.3, the strongest trends seem to be that a group of high-wealth and high-income individuals
chose to renounce after the AJCA and before the HEART Act. If these renunciations were taxmotivated and made in anticipation of the mark-to-market exit tax, this suggests that the exit tax
was perceived by many taxpayers as costly and worth avoiding (a view further supported by the
strong and observable behavioral response to the net worth threshold).
Although it is not possible to give a single answer to the question “Why are they
renouncing?”, the preceding analyses help to provide some resolution. The results suggest that
the recent increase in renunciations was caused by increased compliance costs for those already
living abroad, and that some individuals’ renunciation decisions during the mid-2000s were at
least in part a response to changes, or expectations of changes, to the expatriation tax system.
6. Policy consequences
Building on the findings above about who is renouncing and why, this section answers my
third and final research question: What are the policy consequences? I first consider the revenue
impacts of recent renunciations. I then discuss what lessons can be learned from the policy
changes over the last two decades and conclude by discussing what these findings suggest about
the value of U.S. citizenship.
6.1.
Revenue impacts
Considering the revenue impacts of recent renunciations, the evidence suggests that, although
most do not have any effect, or at most a small one, a handful of renunciations by very highwealth and high-income individuals could have substantial revenue impacts.
A simple way to think about the direct revenue impacts of renunciations is to consider the tax
liabilities renouncers had in the years leading up to their renunciation and assume that these
46
liabilities would have continued had they not renounced. Focusing on the year just prior to
renunciation, Figure 13 shows that the share of renouncers with no tax liability in the year prior
to renunciation has increased markedly; since 2013, about two-thirds of renouncers linked to a
Form 1040 filing as the primary filer in the year prior to renunciation had no liability on that
return. These linked returns represent between half and two-thirds of all renouncers; the
remainder are mostly those without TINs, or with TINs but no linked filings, who also likely had
no U.S. tax liability. Including them in the proportions would further increase the share of
renunciations with no revenue impact.
Figure 13: Pre-renunciation tax liability
Notes: This figure reports the share of individuals renouncing each year with a pre-renunciation tax liability
of zero, <$1,000, or >$1,000, among those who are linked as a primary filer on a Form 1040.
Although many, indeed most, renunciations probably have a negligible revenue impact, this is
not universally true, nor is it necessarily the case that this pattern will hold indefinitely. As shown
earlier in Section 4.3, a handful of wealthy and high-income renouncers can have an outsize
impact on the average net worth and income of those renouncing, and thus on the estimated
revenue impacts. If policymakers are concerned about renunciation purely from a revenue
perspective, the wealthy and high-income are where their focus should continue to be. The
experience of the past two decades does provide some evidence that policy can help discourage
renunciation by these individuals. The prevalence of especially wealthy and high-income
47
individuals among those renouncing between 2004 and 2008 suggests that the introduction of the
mark-to-market tax was perceived as costly, and thus may have had some success in
discouraging subsequent high wealth and income taxpayers from renouncing (although unable to
stop those who could renounce before its enactment, an issue I discuss below). In addition, for
those still choosing to renounce, the mark-to-market tax helps to mitigate the revenue impact.
One high-profile example of this is the renunciation of U.S. citizenship by Facebook co-founder
Eduardo Saverin.43 Although his renunciation meant the U.S. lost out on future income and estate
tax revenue, this was at least partly offset by his exit tax liability (which, according to reports in
the public press, likely was in the hundreds of millions of dollars (Benoit 2012).
Table 16 in the Appendix provides an additional set of summary statistics that support these
conclusions about the revenue impacts of recent renunciations. More than half of Droppers had
no liability during all five years prior to their renunciation. Movers are more likely to have had
non-zero liabilities, but for most individuals, these are still relatively small. The median non-zero
liability for Movers in the year prior to renunciation was about $12K, or $8K when considering
the average over the five years prior to renunciation. These median values are about 10 times
smaller than the mean values, again illustrating that a few outliers have a large impact while most
individuals do not.
Of course, liabilities can change from year to year and assuming that they would stay constant
may not always be correct. A more refined estimate of the revenue impacts of renunciation could
take several routes. To get a more precise estimate of the direct revenue impacts, one could more
carefully forecast what the path of tax liabilities would have been, absent renunciation. This
could consider the path of liabilities prior to renunciation, as well as the renouncer’s age and
assumptions about retirement age and life expectancy. In addition, the revenue impacts should
include estimated effects on future estate tax liabilities, and the revenue raised from expatriation
tax liabilities of covered expatriates. Still, even taking account of these refinements the
conclusion is unlikely to change: most renunciations have had minimal revenue impact, but a
handful probably had a significant impact.
43
According to the quarterly publication of expatriating individuals in the Federal Register, Saverin renounced his
U.S. citizenship in the first quarter of 2012, prior to Facebook, Inc’s IPO on May 18, 2012 (77 FR 25538).
48
A secondary impact on revenue could come from “brain drain”. Academic literature on this
topic has focused mainly on high-skilled migration from less-developed to more-developed
countries (Gibson and McKenzie 2011), but in principle it could also matter for the U.S. It is
possible that some Movers’ renunciations could generate negative spillover effects in the U.S.,
for example if Movers close or relocate U.S. businesses when they move abroad, or postpone
entrepreneurial activity and innovation until after renunciation. Given the small number of
Movers to date, this is unlikely to have had a significant impact, but in principle could become
relevant if future policies led to increasing numbers of Movers. This is less of a concern when
considering the impacts of renunciation by Droppers, as their economic activity is probably
concentrated abroad and thus their renunciations are unlikely to have spillover effects in the U.S.
Finally, consider that the incentives affecting the outflow of citizens (renunciation) should
also affect the inflow of citizens (immigration and naturalization). The academic literature on
immigration points to economic incentives as one factor determining whether, when, and where
individuals migrate (Freeman 2006). As shown above, expatriation tax rules did affect
renunciation decisions by some U.S. citizens on the margin, particularly the wealthy and highincome. It is plausible that these rules, and the tax costs and benefits of U.S. citizenship, would
similarly affect the decisions of those considering in-migration to the U.S. Mason (2016) raises
the concern that citizenship taxation could discourage marginal wealthy or high-income
migrants; Kim (2017) disagrees, arguing that it is U.S. immigration law, not tax law, that is the
real obstacle for highly skilled and educated immigrants. The key determinant of the importance
of the tax law effect is the existence of at least some individuals considering in-migration who
are on the margin. If the distribution of those considering in-migration is comparable to those
considering renunciation, then U.S. tax law could discourage some individuals on the margin
from migrating to the U.S. or naturalizing once in the U.S. Given the relative magnitudes (for the
U.S., naturalizations are two orders of magnitude higher than renunciations, as I discuss below),
this could have significant implications for U.S. tax revenue and economic activity.
6.2.
Policy lessons
Studying the renunciation responses to recent tax policy changes reveals two additional
lessons. First, unintended side effects matter: FATCA appears to have induced some U.S. citizens
49
abroad to renounce citizenship, and the resulting social cost should be considered when
evaluating FATCA. Second, timing matters: the timing of the AJCA and HEART Act legislation
may have allowed some high-wealth individuals to renounce in advance of the exit tax taking
effect.
The analysis in Section 5 showed that the increase in renunciations in the last decade was in
part an unintended side effect of FATCA and other related policies that, while having some
positive revenue impacts, imposed additional compliance costs on those maintaining financial
accounts abroad. Does the U.S. value those foreign-resident U.S. citizens? I argue that the
answer is yes. It may at first seem that these individuals’ welfare should be discounted; because
such individuals are often called “accidental Americans” one might think their renunciations do
not have a social cost. The treatment of citizenship under U.S. nationality law, however, suggests
this is not the case. In principle, the U.S. could further restrict citizenship but so far has not. This
reveals that the U.S. indeed places some value or social benefit on maintaining citizenship for
these individuals.44 Thus, the U.S. loses value, or experiences a social cost, when those abroad
renounce citizenship, and this cost should be included when evaluating the overall effects of
FATCA.
The experience of renunciations during the 2000s also illustrates the importance of policy
timing, and how anticipatory action can partially negate some of the intended effects of
legislation. As shown above, the years between the AJCA and HEART Act saw a handful of
wealthy individuals renouncing citizenship, perhaps influenced in part by a desire to renounce
before the imposition of the mark-to-market exit tax which was being discussed but not yet
implemented. A resulting lesson is thus that the speed of debate and implementation becomes
more important when considering a policy that is intended to target a small group of people who
are sophisticated and well-informed about potential policy changes.
6.3.
The value of U.S. citizenship
Finally, it is important to put recent renunciations in context. This paper focuses on those
dropping citizenship, motivated by the recent increase in renunciations. However, to evaluate the
44
My thanks to Dhammika Dharmapala for a helpful discussion about this topic.
50
effect of the tax system on citizenship decisions overall, consider the rest of the picture: most
individuals do not choose to renounce citizenship, and there is also a large number each year
gaining U.S. citizenship.
Consider first the naturalizations: although the relative increase in renunciations over the last
decade is remarkable, the net flow (naturalizations less renunciations) is still vastly tilted towards
in-migration. Figure 14 plots the annual count of naturalizations (those receiving U.S.
citizenship) in gray, and renunciations (plotted with negative values) in orange. The
renunciations are just barely distinguishable at the bottom of the graph, two orders of magnitude
smaller than the naturalizations. In every year between 1998 and 2018, the number of
naturalizations was above 400,000. This compares to a total of roughly 40,000 citizenship
renunciations between 1998 and 2018.45
45
As noted above, due to data accessibility I focus in this paper on citizenship and not long-term residency status,
but similar arguments can be made for the long-term resident population, with similar conclusions about the effect of
the tax system on individuals’ decisions. In each year, the number of individuals relinquishing long-term residency
status is far lower than the number applying for it.
51
Figure 14: Naturalizations vs. citizenship renunciations
Notes: This figure plots in gray the total number of U.S. naturalizations each year from 1998-2018, from
DHS, 2019 Yearbook of Immigration Statistics, Table 20; in orange with negative values are the annual
counts of those renouncing citizenship.
What about those who already have citizenship, and choose not to renounce it? This describes
almost all U.S. citizens. There are more than 300 million such individuals, and typically fewer
than 5,000 renouncing each year. The number of renunciations is still tiny even when compared
to the stock of U.S. citizens abroad, who could more readily renounce. Although the exact
number of U.S. citizens living abroad is not known, some estimates put it at perhaps nine
million, and the number filing taxes from foreign addresses is more than one million per year. A
few thousand renunciations per year thus represents, as a conservative upper bound, less than
half of 1 percent of those living abroad.46 This suggests another lesson from the fact that the
increased compliance costs under FATCA induced some individuals abroad to drop their U.S.
46
I am not the first to draw this comparison; a similar point was made by Elise Bean in her testimony before the
House Subcommittee on Government Operations in a hearing titled “Reviewing the Unintended Consequences of
the Foreign Account Tax Compliance Act,” held on April 26, 2017. In some respects, the discussion of renunciations
is similar to that of corporate inversions: although the absolute number occurring is relatively small, there is still
significant public press and legislative focus on the issue.
52
citizenship: those costs did not induce vastly many more foreign-resident U.S. citizens to drop
citizenship, implying that for those individuals the maintenance of U.S. citizenship was worth
incurring the resulting financial and hassle costs of complying with new regulations, and thus
that they place a relatively high value on U.S. citizenship.
7. Conclusions
Because the U.S. tax system applies to its citizens’ worldwide income and estates, citizenship
and taxes are more closely connected for the U.S. than for nearly any other country. Using
administrative tax microdata on the population of individuals who have dropped U.S. citizenship
over the past twenty years, this paper demonstrates that this connection can have substantial
impacts on taxpayer behavior, including the decision to maintain or renounce citizenship.
The preceding analyses provide a detailed understanding of who is renouncing and why. The
recent increase in renunciations has come mainly from those who have long filed U.S. taxes from
abroad – that is, mainly from Droppers, not Movers. These Droppers’ renunciations were
primarily an unintended side effect of the increased compliance costs brought on by FATCA and
other offshore financial enforcement during the 2010s. And although renouncers on average are
wealthier and higher-income than the U.S. population, most recent renouncers had low or zero
pre-renunciation U.S. tax liability, suggesting that their renunciations may not have a significant
revenue impact.
The evidence reveals that citizenship decisions are connected to U.S. tax policy, most notably
that the compliance costs of increased offshore enforcement may have led thousands of U.S.
citizens abroad to drop their citizenship; for these individuals the costs of renunciation, both
financial and emotional, surely were quite large. Still, the total number of renunciations remains
relatively small, whether compared to estimates of the remaining population of U.S. citizens
living abroad, filing taxes from foreign addresses, or newly gaining U.S. citizenship, and in
purely financial terms, the revenue impact of their renunciations is likely to be small. All this
together suggests that U.S. citizenship has historically been perceived as valuable by most who
hold it, and remains so today.
53
That citizenship decisions and the tax system are connected should be accounted for when
considering changes to the tax system. The attractiveness of citizenship renunciation depends
crucially on the current tax system as well as expectations about its future, relative to alternative
foreign tax systems. Individuals determine the expected costs and benefits of retaining or
dropping citizenship, factoring in the potential for future tax increases (or decreases) or even
entirely new taxes, such as an annual wealth tax. This determination may be particularly relevant
for younger individuals facing a future stream of annual tax liabilities, for entrepreneurs
considering the potential future net-of-tax gains to their innovation, and for the wealthy
considering potential future estate tax liabilities. Those considering moving to the U.S., or
naturalizing as U.S. citizens, may also be influenced by the tax system. Policymakers should not
ignore citizenship renunciation and naturalization as potentially important margins of response.
54
References
Ahn, Julie Jee Eun. 2015. "The HEART Act of 2008: To Deter Expatriation, But Why More
Expatriation." North Carolina Journal of International Law and Commercial Regulation
4: 1021-84.
Akcigit, Ufuk, Salomé Baslandze, and Stefanie Stantcheva. 2016. "Taxation and the International
Mobility of Inventors." American Economic Review 106 (10): 2930-81.
Arsenault, Steven J. 2009. "Surviving a Heart Attack: Expatriation and the Tax Policy
Implications of the New Exit Tax." Akron Tax Journal 24 (2).
Belnap, Andrew, Jacob Thornock, and Braden Williams. 2019. "The Long Arm of the U.S. Tax
Law: Participation Rates and Costs related to Automatic Information Sharing." Working
paper.
Benoit, David. 2012. "Facebook Co-Founder Saverin Fires Back At 'Misinformation'." The Wall
Street Journal, May 17.
Christians, Allison. 2017. "Buying In: Residence and Citizenship by Investment." Saint Louis
University Law Journal 62: 51-72.
Cooper, Kenneth J., and R.H. Melton. 1995. "Tax breaks for wealthy expatriates sparks class
warfare charges." The Washington Post, March 31.
Craig, Bradford. 2012. "Congress, Have a Heart: Practical Solutions to Punitive Measures
Plaguing the HEART Act’s Expatriate Inheritance Tax." Temple International &
Comparative Law Journal 1: 69-102.
De Simone, Lisa, Rebecca Lester, and Kevin Markle. 2020. "Transparency and Tax Evasion:
Evidence from the Foreign Account Tax Compliance Act (FATCA)." Journal of
Accounting Research 58 (1): 105-53.
Dharmapala, Dhammika. 2016. "Cross-border tax evasion under a unilateral FATCA regime."
Journal of Public Economics 141: 29-37.
55
Freeman, Richard B. 2006. "People Flows in Globalization." Journal of Economic Perspectives
20 (2): 145-70.
Gibson, John, and David McKenzie. 2011. "Eight Questions about Brain Drain." Journal of
Economic Perspectives 25 (3): 107-28.
Goncalves, Pedro. 2019. "FATCA could push French banks to close up to 40,000 accounts."
International Investment, August 14.
Graffy, Colleen. 2015. "The law that makes U.S. expats toxic." The Wall Street Journal, October
10.
Johannesen, Niels, Patrick Langetieg, Daniel Reck, Max Risch, and Joel Slemrod. 2020. "Taxing
Hidden Wealth: The Consequences of U.S. Enforcement Initiatives on Evasive Foreign
Accounts." American Economic Journal: Economic Policy 12 (3): 312-46.
Joint Committee on Taxation. 1995. "Issues presented by proposals to modify the tax treatment
of expatriation: A report." JCS-17-95.
Kim, Young Ran (Christine). 2017. "Considering 'Citizenship Taxation': In Defense of FATCA."
Florida Tax Review 20 (5): 335-70.
Kleven, Henrik, Camille Landais, and Emmanuel Saez. 2013. "Taxation and International
Migration of Superstars: Evidence from the European Football Market." American
Economic Review 103 (5): 1892-1924.
Kleven, Henrik, Camille Landais, Mathilde Muñoz, and Stefanie Stantcheva. 2020. "Taxation
and Migration: Evidence and Policy Implications." Journal of Economic Perspectives 34
(2): 119-42.
Kudrle, Robert T. 2015. "Expatriation: A Last Refuge for the Wealthy?" Global Policy 6 (4):
408-17.
Kwong, Yu Hang Sonny. 2009. "Catch Me If You Can: Relinquishing Citizenship for Taxation
Purposes After the 2008 HEART Act." Houston Business and Tax Journal 9: 411-44.
56
Manolakas, Christine, and William L. Dentino. 2012. "The Exit Tax: A Move in the Right
Direction." William & Mary Business Law Review 3: 341-418.
Mason, Ruth. 2016. "Citizenship Taxation." Southern California Law Review 89: 169-240.
Mirrlees, James A. 1982. "Migration and Optimal Income Taxes." Journal of Public Economics
18 (3): 319-41.
Reichenberg Sherr, Eileen. 2008. "New Expatriation Tax Rules Enacted." The Tax Adviser,
August.
Richards, Paul J. 2016. "Delays, costs mount for Canadians renouncing U.S. citizenship." The
Globe and Mail, February 9.
Romm, Tony. 2012. "Pols plan would tax Facebook expat." Politico, May 17.
Sheppard, Lee. 2009. "Don't Ask, Don't Tell, Part 4: Ineffectual Information Sharing." Tax Notes,
March 23.
Westin, Richard A. 2000. "Expatriation and Return: An Examination of Tax-Driven Expatriation
by United States Citizens, and Reform Proposals." Virginia Tax Review 20 (1): 75-190.
Williams, Geoff. 2014. "U.S. expats find their money is no longer welcome at the bank."
Reuters, June 11.
57
Appendices
Appendix A. Additional figures and tables
Figure 15: Comparison of income for renouncers, foreign filers, and all tax filers, mean and median
Notes: This figure compares the income of renouncers in the year prior to renunciation to two comparison
groups: all other foreign filings, and a sample of the population of Form 1040 filing. For renouncers, only
primary filers with linked filings are included. The three vertical dashed lines represent three key dates
related to expatriation tax law: the 2004 AJCA, the 2008 HEART Act, and 2010 FATCA. The solid line
includes all individuals; the dashed line removes the top 10 in each year.
58
Figure 16: Comparison of income among renouncers, Movers vs. Droppers, mean and median
Notes: This figure comparison the income in the year prior to renunciation for renouncers with linked Form
1040 filings as primary filers. The mean values are calculated separately among Movers and Droppers.
Renouncers with no filings or no TINs are excluded.
59
Figure 17: Comparison of reported net worth among renouncers, Movers vs. Droppers, mean and
median
Notes: This figure compares reported net worth among those renouncing each year, separately for Movers
and Droppers. Only those with reported net worth data available are included. The left panel includes all
Movers and Droppers; right panel drops the top 10 Movers and Droppers, by reported net worth, each year.
60
Table 7: Top destinations, split by renouncer reported net worth
Renouncers, split by reported net worth
Rank
All renouncers
$2M+
$622K-2M
$0-622K
No RNW
No 8854
1
Canada
Canada
Canada
Canada
Canada
Canada
2
Switzerland
United Kingdom Switzerland
Switzerland
Switzerland
Switzerland
3
United Kingdom
Switzerland
United Kingdom United Kingdom United Kingdom United Kingdom
4
Germany
Hong Kong
Hong Kong
Hong Kong
Germany
Germany
5
Hong Kong
Australia
Australia
Germany
South Korea
South Korea
6
Australia
France
Germany
Australia
China
Singapore
7
South Korea
Germany
France
Netherlands
Norway
Hong Kong
8
Singapore
Singapore
Singapore
Taiwan
Hong Kong
Australia
9
Taiwan
Taiwan
Taiwan
France
France
Taiwan
10
France
China
Belgium
Singapore
Japan
Belgium
Notes: This table presents the top 10 destinations among all renouncers, and then within each reported net
worth group. All renunciations between 2005 and 2018 are included.
Table 8: Top destinations, split by renouncer classification
Renouncers, split by classification
Rank
All renouncers
Mover
Dropper
No Filings
No TIN
1
Canada
Canada
Canada
Switzerland
Switzerland
2
Switzerland
United Kingdom Switzerland
Canada
Canada
3
United Kingdom
Switzerland
United Kingdom United Kingdom Germany
4
Germany
Hong Kong
Hong Kong
Germany
United Kingdom
5
Hong Kong
Taiwan
Australia
South Korea
South Korea
6
Australia
Germany
Germany
Hong Kong
Singapore
7
South Korea
South Korea
France
Australia
Hong Kong
8
Singapore
China
Netherlands
France
Taiwan
9
Taiwan
Australia
Singapore
Singapore
Belgium
10
France
Singapore
New Zealand
Taiwan
Australia
Notes: This table presents the top 10 destinations among all renouncers, and then within each reported
classification. All renunciations between 2005 and 2018 are included.
61
Table 9: Summary statistics for individual-level regression data
Variable
N
Mean
St. Dev.
Min
Pctl(25)
Median
Pctl(75)
Max
Binary outcome (100/0)
Renounce in following year
4,831,000
0.343
5.849
0
0
0
0
100
Covariates
Total Positive Income ($M)
Wage share of TPI (%)
Age (years)
4,831,000
4,831,000
4,790,000
0.14
0.64
48
1.48
0.44
17
0
0.00
16
0.02
0.00
34
0.05
0.95
46
0.11
1.00
60
1,200
1.00
100
Binary covariates (1/0)
Positive tax liability
Had any Sch C income
Had any Sch E income
Received an IRS notice
Made Sch A charity deduction
Filed a gift tax form
4,831,000
4,831,000
4,831,000
4,831,000
4,831,000
4,831,000
0.38
0.12
0.14
0.15
0.07
0.00
0.49
0.32
0.34
0.35
0.25
0.05
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
1
1
1
1
1
1
Diff-in-diff covariates (1/0)
IGA jurisdiction
as of 2019
as of 2017
as of 2015
Tax haven jurisdiction
Post (tax years >= 2010)
4,831,000
4,831,000
4,831,000
4,831,000
4,831,000
0.932
0.930
0.873
0.082
0.766
0.25
0.26
0.33
0.27
0.42
0
0
0
0
0
1
1
1
0
1
1
1
1
0
1
1
1
1
0
1
1
1
1
1
1
Notes: This table reports the summary statistics for the individual-year level data used in regression analysis
with results shown in Table 4. The population is all Form 1040 filings from foreign addresses for tax years
2007-2017, excluding any Mover renouncers. Values are rounded for disclosure purposes.
62
Table 10: Individual regression results, including all linked renouncers
Dependent variable:
[1]
Total Positive Income
($ millions)
Binary: Renounce in following year (100/0)
[2]
[3]
[4]
[5]
[6]
0.0171
0.017
0.0176
0.0137
0.0140***
0.0137
(0.0105)
(0.0104)
(0.0105)
(0.0098)
(0.0018)
(0.0097)
Wage share
(% of TPI)
-0.0934
-0.071
-0.1216
-0.1345**
(0.0804)
(0.0614)
(0.0748)
(0.0558)
(0.0077)
(0.0385)
Positive tax liability
(1/0)
-0.0692
-0.0717
-0.0579
-0.0632
-0.0562***
-0.0546
(0.0655)
(0.0665)
(0.0644)
(0.0472)
(0.0061)
(0.0453)
Any Sch C income
(1/0)
0.0843***
0.0874***
0.0624**
0.0570**
0.0433***
0.0469**
(0.0278)
(0.0309)
(0.0254)
(0.0196)
(0.0091)
(0.0195)
Any Sch E income
(1/0)
0.0127
0.0105
-0.0046
0.018
-0.0011
0.0077
(0.0381)
(0.0392)
(0.0376)
(0.0270)
(0.0083)
(0.0264)
Schedule A charity
(1/0)
-0.0052
-0.0131
0.0264
-0.1158
-0.0928***
-0.0907
(0.0915)
(0.0894)
(0.0938)
(0.0972)
(0.0115)
(0.0975)
Filed gift tax return
(1/0)
2.5150***
2.5109***
2.4779***
2.4555***
2.4139***
2.4046***
(0.4208)
(0.4180)
(0.4150)
(0.4284)
(0.0568)
(0.4188)
Received any notice
(1/0)
0.1163**
0.1123**
0.0376*
0.1241**
0.0454***
0.0506*
(0.0560)
(0.0563)
(0.0197)
(0.0543)
(0.0082)
(0.0279)
0.0016
0.0020***
0.0019
(0.0013)
(0.0002)
(0.0012)
Age
(years)
Constant
Year FE
Jurisdiction FE
YearXJurisdiction FE
Observations
Adjusted R2
Mean dep. var.
-0.1293*** -0.1250***
0.4168***
0.3304***
(0.1370)
(0.0720)
No
No
No
No
No
No
Yes
No
No
No
Yes
No
Yes
Yes
No
No
No
Yes
4,835,000
0.0006
0.343
4,793,000
0.0006
0.343
4,835,000
0.0014
0.343
4,835,000
0.0046
0.343
4,793,000
0.0054
0.343
4,793,000
0.0067
0.343
Notes: *p<0.1; **p<0.05; ***p<0.01. Standard errors, clustered by year and by jurisdiction, are shown in
parentheses. The dependent variable is coded as 100 or 0 so that the coefficient estimates represent the
effect in percentage points for each covariate, holding all others constant. All expatriates that can be linked
to Form 1040 filings as a primary filer are included.
63
Figure 18: Scaled coefficient estimates for selected covariates, individual LPM
Notes: This figure plots the coefficient estimates for selected covariates of the fully saturated linear
probability model reported in column 6 of Table 4 (excluding Movers) and Table 10 (including Movers).
The estimates are scaled by the mean dependent variable to show the estimated effect on the probability of
a given type of expatriation, in percent. 95% confidence intervals using standard errors clustered by year
and jurisdiction are shown around each point estimate. The covariate for filing a gift tax form is excluded
from this figure because it dominates the others, with a scaled point estimate suggesting it is associated
with a more than 500% increase in the probability of renunciation.
64
Table 11: Summary statistics for jurisdiction-level regression data
Variable
N
Mean
St. Dev.
Min
Pctl(25)
Median
Pctl(75)
Max
Outcome
Renunciation share
213
0.013
0.030
0
0
0.005
0.0128
0.3415
Covariates
Tax haven
Citizenship-for-sale
English-speaking
WGI Rule of Law
WGI Political Stability
Passport value
Average change in RGDP
213
213
213
205
205
197
188
0.174
0.047
0.094
0.490
0.497
0.469
3.218
0.380
0.212
0.292
0.284
0.286
0.297
2.370
0
0
0
0
0.012
0.015
-4.609
0
0
0
0.251
0.255
0.214
1.600
0
0
0
0.474
0.493
0.415
3.177
0
0
0
0.728
0.769
0.733
4.766
1
1
1
0.996
0.985
0.979
11.509
Notes: This table reports the summary statistics for the jurisdiction-level data used in the regression analysis
with results shown in Table 5. The population is all jurisdictions with any U.S. tax filings for tax years
2007-2017. Certain covariates are only available for a subset of jurisdictions. I define citizenship-for-sale
jurisdictions as those that began such a program prior to 2017: Antigua and Barbuda, Bulgaria, Comoros,
Dominica, Grenada, Malta, St. Kitts and Nevis, Serbia, St. Lucia, and Vanuatu (Christians 2017).
65
Table 12: Individual-level regression results with jurisdiction-level covariates
Dependent variable:
[1]
Binary: Renounce in following year (100/0)
[2]
[3]
[4]
[5]
[6]
Total Positive Income
($ millions)
0.0102
0.0068
0.0073
0.0108
0.0074
0.0079
(0.0063)
(0.0060)
(0.0063)
(0.0063)
(0.0073)
(0.0072)
Wage share
(% of TPI)
-0.055
-0.0922**
-0.0961**
-0.073
-0.1093**
-0.1122**
(0.0603)
(0.0422)
(0.0418)
(0.0567)
(0.0389)
(0.0389)
Positive tax liability
(1/0)
-0.0834
-0.0869*
-0.0883*
-0.0727
-0.0772
-0.0784
(0.0650)
(0.0465)
(0.0462)
(0.0638)
(0.0442)
(0.0445)
Any Sch C income
(1/0)
0.0937***
0.0864***
0.0887***
0.0749**
0.0691***
0.0710***
(0.0278)
(0.0236)
(0.0221)
(0.0274)
(0.0199)
(0.0198)
0.0042
-0.0044
-0.0044
-0.0129
-0.0208
-0.0207
(0.0383)
(0.0354)
(0.0348)
(0.0377)
(0.0342)
(0.0338)
-0.0834
Any Sch E income
(1/0)
Schedule A charity
(1/0)
-0.0218
-0.114
-0.1148
0.0084
-0.0828
(0.0856)
(0.0912)
(0.0928)
(0.0874)
(0.0930)
(0.0937)
Filed gift tax return
(1/0)
2.2807***
2.2270***
2.2497***
2.2419***
2.1897***
2.2117***
(0.4172)
(0.4222)
(0.4266)
(0.4108)
(0.4141)
(0.4183)
Received any notice
(1/0)
0.1023*
0.1066*
0.1055*
0.025
0.032
0.0311
(0.0547)
(0.0569)
(0.0569)
(0.0191)
(0.0283)
(0.0286)
0.0019
0.0020*
0.0019*
0.0023*
0.0024**
0.0023**
(0.0013)
(0.0011)
(0.0011)
(0.0012)
(0.0010)
(0.0010)
Tax haven
(1/0)
0.7891**
0.8146**
0.7949*
0.8255**
(0.3659)
(0.3698)
(0.3657)
(0.3698)
Citizenship-for-sale
(1/0)
-0.2167
-0.2056
-0.2018
-0.1946
(0.1338)
(0.1445)
(0.1373)
(0.1484)
English-speaking
(1/0
0.2075***
0.2124***
0.2043***
0.2109***
(0.0435)
(0.0453)
(0.0568)
(0.0572)
Rule of Law index
(percentile)
0.2542***
0.4921***
0.2257***
0.4409**
(0.0672)
(0.1799)
(0.0655)
(0.1752)
Political Stability index
(percentile)
0.2708***
0.2558***
0.2842***
0.2603**
(0.0775)
(0.0755)
(0.0854)
(0.0877)
Age
(years)
Passport ranking
(percentile)
Change in Real GDP
(percentage points)
Constant
Year FE
Observations
Adjusted R2
Mean dep. var.
-0.3995*
-0.3806
(0.2280)
(0.2226)
-0.035
-0.0374
(0.0255)
(0.0253)
0.2890***
-0.2045**
0.0171
(0.0712)
(0.0810)
(0.1428)
No
No
No
Yes
Yes
Yes
4,790,000
0.0005
0.343
4,786,000
0.0030
0.343
4,752,000
0.0030
0.343
4,790,000
0.0014
0.343
4,786,000
0.0038
0.343
4,752,000
0.0039
0.343
Notes: *p<0.1; **p<0.05; ***p<0.01. Standard errors clustered by year are shown in parentheses.
Jurisdiction-level covariates are averages over the full time period, and not time-varying. Observations are
individual-year.
66
Table 13: Jurisdiction-level regression results, CFS robustness checks
Dep. var.: Total renunciations/unique foreign filers
Robustness 1: Remove CFS covariate
Tax haven
(1/0)
Robustness 2: Include Haven X CFS interaction
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
0.0253**
0.0254**
0.0332**
0.0160*
0.0267**
0.0273**
0.0367**
0.0172*
(0.0121)
(0.0123)
(0.0147)
(0.0083)
(0.0131)
(0.0137)
(0.0168)
(0.0096)
Citizenship-for-sale
(1/0)
-0.0054*
(0.0028)
(0.0024)
(0.0031)
(0.0025)
Tax haven X CFS
(1/0)
-0.0077
-0.0081
-0.0139
-0.0006
(0.0118)
(0.0117)
(0.0145)
(0.0103)
English-speaking
(1/0
-0.0070*** -0.0085*** -0.0065***
-0.0074
-0.0144
-0.0189*
-0.0064
-0.0049
-0.0111
-0.014
-0.0051
(0.0080)
(0.0097)
(0.0111)
(0.0065)
(0.0067)
(0.0079)
(0.0088)
(0.0055)
Rule of Law index
(percentile)
0.0328***
0.0194***
0.0206***
0.0323***
0.0186**
0.0201***
(0.0093)
(0.0072)
(0.0063)
(0.0093)
(0.0074)
(0.0063)
Political Stability index
(percentile)
-0.0036
0.0002
-0.0049
-0.0034
0.0014
-0.0042
(0.0067)
(0.0074)
(0.0050)
(0.0070)
(0.0078)
(0.0051)
Passport ranking
(percentile)
0.0120**
0.0089
0.0112**
0.0087
(0.0057)
(0.0055)
(0.0055)
(0.0055)
Change in Real GDP
(percentage points)
Constant
Observations
Adjusted R2
0.0002
0.0001
(0.0005)
(0.0005)
0.0090***
-0.004
-0.0049
-0.0035
0.0090***
-0.0038
-0.0047
-0.0033
(0.0010)
(0.0034)
(0.0037)
(0.0031)
(0.0010)
(0.0034)
(0.0036)
(0.0030)
213
0.0855
205
0.1693
196
0.2069
187
0.2231
213
0.0814
205
0.1672
196
0.2113
187
0.2210
Notes: This table reports the results of two alternative sets of jurisdiction-level regression specifications,
one without the citizenship-for-sale covariate, and one adding a Tax haven X CFS interaction covariate.
Removing CFS as a covariate does not materially affect the estimates on other covariates. Adding the haven
interaction reveals that the CFS effect is driven by the non-haven CFS jurisdictions, like Bulgaria and
Serbia.
67
Figure 19: Count of jurisdictions signing FATCA IGAs, by year and model type
Notes: This figure plots the count of unique jurisdictions that signed a FATCA-related IGA with the U.S.
in each year, separately for the two types of IGAs.
68
Figure 20: Jurisdictions by FATCA IGA status
Notes: This figure shows world jurisdictions, shaded to indicate their FATCA IGA status, according to the
U.S. Treasury. Jurisdictions with a signed IGA are shaded in orange; those listed as having an “Agreement
in Substance” but not yet signed are shaded in blue. Those with neither a signed IGA nor an Agreement in
Substance are shaded in green.
69
Figure 21: Annual share renouncing between IGA and non-IGA jurisdictions, alternate specifications
Notes: This figure plots the number of individual renouncing each year as a share of prior-year U.S. tax
filers, from either IGA or non-IGA jurisdictions. The four rows correspond to four different definitions of
“IGA jurisdictions”: including those signed through 2019, 2017, or 2015, or adding in those with
Agreements in Substance. The four columns correspond to sample definitions; the main specification
includes only Dropper renouncers; the second column also includes Movers; the third column excludes
Switzerland; the fourth column restricts to jurisdictions with at least 100 U.S. tax filers in one year.
70
Figure 22: Annual share renouncing between haven and non-haven jurisdictions, alternate
specifications
Notes: This figure plots the number of individual renouncing each year as a share of prior-year U.S. tax
filers, from either tax haven or non-tax-haven jurisdictions. The three columns correspond to sample
definitions; the main specification includes only Dropper renouncers; the second column also includes
Movers; the third column restricts to jurisdictions with at least 100 U.S. tax filers in one year.
71
Table 14: Difference-in-difference, IGA test robustness checks
Dependent variable:
Sample:
Binary: Renounce in following year (100/0)
Droppers &
Exclude
Large
Droppers
Movers
Switzerland
Jurisdictions
[1]
[2]
[3]
[4]
IGA definition
Signed through 2019
Signed through 2017
Signed through 2015
Signed or Agreed in
Substance through 2019
Individual covariates
Observations
Adjusted R2
0.3144***
0.3168***
0.2580***
0.3149***
(0.1003)
(0.1023)
(0.0996)
(0.1006)
0.3155***
0.3182***
0.2589***
0.3156***
(0.1000)
(0.1019)
(0.0991)
(0.1003)
0.2912***
0.2869***
0.2305**
0.2897***
(0.0998)
(0.1017)
(0.0991)
(0.1001)
0.2738***
0.2724***
0.2182**
0.2716***
(0.0971)
(0.0989)
(0.0958)
(0.0972)
Yes
Yes
Yes
Yes
4,831,000
0.0013
4,835,000
0.0013
4,686,000
0.0012
4,805,000
0.0013
Notes: *p<0.1; **p<0.05; ***p<0.01. The coefficient estimates on the Post X IGA covariate are shown
under alternate IGA definitions and sample definitions. Standard errors clustered by year and by jurisdiction
are shown in parentheses. The four rows correspond to four different definitions of “IGA jurisdictions”:
including those signed through 2019, 2017, or 2015, or adding in those with Agreements in Substance. The
four columns correspond to sample definitions; the main specification includes only Dropper renouncers;
the second column also includes Movers; the third column excludes Switzerland; the fourth column restricts
to jurisdictions with at least 100 U.S. tax filers in one year.
72
Table 15: Difference-in-difference, haven test robustness checks
Dependent variable:
Sample:
Binary: Renounce in following year (100/0)
Droppers &
Large
Droppers
Movers
Jurisdictions
[1]
[2]
[4]
Switzerland treatment:
Included
Excluded
Individual covariates
Observations
Adjusted R2
0.4343
0.4274
0.4384
(0.5484)
(0.5820)
(0.5591)
-0.3151***
-0.3641***
-0.3301***
(0.1151)
(0.1234)
(0.1145)
Yes
Yes
Yes
4,831,000
0.0025
4,835,000
0.0027
4,805,000
0.0025
Notes: *p<0.1; **p<0.05; ***p<0.01. The coefficient estimates on the Post X Haven covariate are shown
when including or excluding Switzerland, and under alternat sample definitions Standard errors clustered
by year and jurisdiction are shown in parentheses. The three columns correspond to sample definitions; the
main specification includes only Dropper renouncers; the second column also includes Movers; the third
column restricts to jurisdictions with at least 100 U.S. tax filers in one year.
73
Figure 23: Individual IGA analysis, year-by-year specification
Notes: This figure plots the coefficient estimates on the year X IGA dummy variables in a specification
which also includes year dummy variables and individual covariates (i.e., the same as the main IGA
difference-in-difference specification but replacing the Post, IGA, and Post X IGA variables with year-byyear covariates). This alternate specification shows results consistent with the main difference-in-difference
IGA analysis.
74
Table 16: Revenue-relevant summary statistics
Split by prior U.S.-based filing
Split by time period
All
Mover
Dropper
1998-2010
2011-2018
39,300
27,900
20,400
4,600
4,600
3,200
20,000
20,000
14,500
9,100
6,500
5,000
30,200
21,400
15,400
Count with non-zero liability
Among those with TIN: share with non-zero liability
8,700
31%
2,200
48%
6,100
31%
2,400
37%
6,200
29%
Among those with non-zero liability:
Median liability ($)
Mean liability ($)
Total liability ($M)
Median age at expatriation
$4,200
$81,000
$704
52
$11,800
$183,500
$404
51
$3,000
$46,900
$286
52
$9,300
$137,500
$330
50
$3,200
$60,400
$374
52
Total count of renouncers
Count with TIN/SSN
Count with Form 8854
Tax liability one year prior to renunciation
Average annual tax liability over five years prior to expatriation
Count with at least one year of non-zero liability
Among those with TIN: share with non-zero liability
14,800
53%
3,600
78%
10,600
53%
3,800
58%
11,000
51%
Among those with non-zero liability:
Median liability ($)
Mean liability ($)
Total liability ($M)
Median age at expatriation
$2,200
$41,500
$614
52
$8,000
$94,000
$338
50
$1,400
$24,900
$264
52
$6,200
$71,700
$272
49
$1,500
$31,000
$341
53
Notes: pre-FATCA includes expatriations from 1998-2010; post-FATCA from 2011-2018. Values are
rounded to the nearest 100 for disclosure purposes. Tax liability is net of the Foreign Tax Credit.
75
Figure 24: Histogram of reported net worth around $2 million (extended range)
Notes: This figure plots the count of expatriates with past-5-years average income tax liability in each $10K
bucket of liability relative to the covered expatriate threshold in their year of renunciation. The thresholds
are shown above in Figure 26.
76
Figure 25: Histogram of average income tax liability relative to threshold
Notes: This figure plots the count of expatriates with past-5-years average income tax liability in each $10K
bucket of liability relative to the covered expatriate threshold in their year of renunciation. The thresholds
are shown in Figure 26.
77
Appendix B. Additional notes on U.S. citizenship renunciation
Appendix B.1. The process of U.S. citizenship renunciation
Section 349(a) of the Immigration and Nationality Act outlines the seven acts by which a U.S.
national can voluntarily relinquish U.S. nationality: (1) obtaining naturalization in a foreign state
once 18 years or older; (2) declaring allegiance to a foreign state once 18 years or older; (3)
serving in the armed forces of a foreign state, either as an officer or engaged in hostilities against
the U.S.; (4) serving a foreign government if that service requires foreign nationality or
allegiance; (5) making a formal renunciation of nationality before a diplomatic or consular
officer of the U.S. in a foreign state; (6) making a formal written renunciation of nationality in
the U.S. (when the U.S. is at war and the renunciation is approved by the Attorney General); and
(7) committing treason against or attempting to overthrow the U.S. government (8 U.S.C.
§1481).
The fifth option is the main approach taken by those choosing to lose their U.S. citizenship.
The required steps include preparing the necessary forms, meeting with a diplomatic or consular
officer in a foreign state, swearing an oath of renunciation, and paying the renunciation fee. In
most cases, renunciation requires two separate appointments at a foreign embassy or consulate.
In the first appointment, the U.S. citizen is interviewed to confirm that renunciation is being done
out of free will and not under duress. At the second appointment, an oath of renunciation is
sworn. The current fee for citizenship renunciation is $2,350 (until mid-2014 the fee was $450,
and there was no fee before 2010). Because of the recent increase in renunciations, some
embassies and consulates have experienced backlogs of renunciation appointments, leading to
delays or prompting some individuals to travel to other cities and countries to seek earlier
appointments (Richards 2016). Various third-party firms offer services to U.S. citizens
considering citizenship renunciation, promising to assist with the process and often targeting
their marketing at high-wealth individuals. 47
After completing these steps, the State Department processes the renunciation and sends the
individual a Certificate of Loss of Nationality confirming the renunciation of U.S. citizenship.
47
See, e.g., the Nomad Capitalist (https://nomadcapitalist.com/) or 1040Abroad (https://1040abroad.com/about/).
78
Those renouncing citizenship must also file a U.S. income tax form for the year in which they
renounced citizenship and include Form 8854 to complete renunciation for tax purposes (and
remit or make arrangements to remit any associated tax liability). Because expatriation is an
individual process, each individual must file a separate Form 8854, even if filing Form 1040 with
married filing jointly status.
Figure 26 below plots the changes over time in the income tax and net worth thresholds for
covered expatriate designation, as well as the changes in the capital gains exemption available to
those who are deemed covered expatriates and subject to the mark-to-market exit tax (since
2008).
79
Figure 26: Statutory covered expatriate thresholds and gains exemptions over time
Appendix B.2. Prior U.S. history with citizenship renunciation
In the main text of this paper I focus on renunciations from 1998-2018, the years for which
the IRS database with information on those renouncing U.S. citizenship is available and
complete. Using outside sources, it is also possible to provide some context on renunciations
covering a longer time period.
Annual counts of U.S. citizen renunciations are available for the years 1962-1994 from the
State Department, as listed in a report discussing proposals for changes to the tax treatment of
80
expatriation (Joint Committee on Taxation 1995). These can be paired with annual counts of
individuals reported in the Federal Register as having relinquished citizenship, which are
available for the years 1998-2020. Note that because of slight differences in the way numbers
were tracked from year to year and the precise criteria for inclusion, these sources may not be
exactly comparable with each other, nor with the counts I present above. 48 Still, all three capture
a similar idea and allow for consideration of trends over time.
Figure 27 below shows the JCT and Federal Register series, with the JCT numbers in blue and
the Federal Register numbers in green. The longer-term trend shows that renunciations were
actually somewhat more common in the 1960s and 1970s, and had fallen to a relative low by the
2000s, before increasing in the past decade, as discussed above.
Figure 27: Annual count of renunciations (JCT and Federal Register)
Notes: This figure plots the count of U.S. citizenship renunciations from two sources. For the years 19621994, the values are as reported in Joint Committee on Taxation (1995). For the years 1998-2020, the values
are the count of names published in the Federal Register as the “Quarterly Publication of Individuals Who
Have Chosen to Expatriate”, required under IRC §6039G.
48
For instance, the JCT report notes at p. 7 that there may be discrepancies between the definitions used for the yeas
1962-1979 and 1980-1994.
81
This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.