PNAS Nexus, 2023, 2, 1–12

https://doi.org/10.1093/pnasnexus/pgad051
Advance access publication 28 February 2023
Research Report

Complex systems of secrecy: the offshore networks

of oligarchs
a,
Ho-Chun Herbert Chang *, Brooke Harringtonb, Feng Fu c
and Daniel N. Rockmorec,d
a
Annenberg School for Communication and Journalism, University of Southern California, Los Angeles, CA 90007, USA
b
Department of Sociology, Dartmouth College, Hanover, NH 03755, USA

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

c
Department of Mathematics, Dartmouth College, Hanover, NH 03755, USA
d
Department of Quantitative Social Science, Dartmouth College, Hanover NH 03755, USA
*To whom correspondence should be addressed: Email: [email protected]
Edited By: Noshir Contractor

Abstract
Following the invasion of Ukraine, the USA, UK, and EU governments–among others–sanctioned oligarchs close to Putin. This approach
has come under scrutiny, as evidence has emerged of the oligarchs’ successful evasion of these punishments. To address this problem, we
analyze the role of an overlooked but highly influential group: the secretive professional intermediaries who create and administer the
oligarchs’ offshore financial empires. Drawing on the Offshore Leaks Database provided by the International Consortium of Investigative
Journalists (ICIJ), we examine the ties linking offshore expert advisors (lawyers, accountants, and other wealth management
professionals) to ultra-high-net-worth individuals from four countries: Russia, China, the USA, and Hong Kong. We find that resulting
nation-level “oligarch networks” share a scale-free structure characterized by a heterogeneity of heavy-tailed degree distributions of
wealth managers; however, network topologies diverge across clients from democratic versus autocratic regimes. While generally
robust, scale-free networks are fragile when targeted by attacks on highly connected nodes. Our “knock-out” experiments pinpoint
this vulnerability to the small group of wealth managers themselves, suggesting that sanctioning these professional intermediaries
may be more effective and efficient in disrupting dark finance flows than sanctions on their wealthy clients. This vulnerability is
especially pronounced amongst Russian oligarchs, who concentrate their offshore business in a handful of boutique wealth
management firms. The distinctive patterns we identify suggest a new approach to sanctions, focused on expert intermediaries to
disrupt the finances and alliances of their wealthy clients. More generally, our research contributes to the larger body of work on
complexity science and the structures of secrecy.

Keywords: network science, sanctions, secrecy, complex systems, power laws

Significance Statement
We offer a new approach to targeting economic sanctions—a vital nonmilitary tool in disputes among nations. Sanctions have been
particularly important in the international response to Russia’s 2022 invasion of Ukraine. But oligarchs close to President Putin have
too often evaded sanctions on their overseas assets. Our analysis of the largest public database of offshore finance suggests a more
effective strategy: sanctioning the professional wealth managers who administer the oligarchs’ financial networks. Using network sci­
ence enhanced by sociological insights, we pinpoint a crucial vulnerability—the concentration of Russian wealth in the hands of a few
managers—that can inform more successful policy interventions. Our findings lay the groundwork for a new line of research into
complex systems of secrecy.

Introduction the ultra-wealthy (1)—these sanctions yielded dramatic media

coverage of yacht and luxury villa seizures.
Russia’s invasion of Ukraine in February 2022 was swiftly met
But regulatory reviews have exposed major gaps and “short­
with sanctions imposed by nations around the world, including
comings” (2) in these sanctions regimes. Shortly after they were
the USA, the UK, Canada, Australia, Switzerland, Singapore, the
European Union, and its member states. Among the many sanc­ imposed, evidence emerged showing some prominent Russians
tions regimes put in place, those targeted at individual oligarchs evading sanctions through the assistance of elite intermediaries:
close to Putin garnered particular attention. Informed by econom­ trusted experts who secretly transferred the oligarchs’ financial
ic research showing that Russian elites hold a staggering 60% of assets (3) and spirited away treasures like priceless art collections
their wealth offshore—compared to the 10% global average for beyond the reach of the law. The same happened following efforts

Competing Interest: The authors declare no competing interest.

Received: September 6, 2022. Revised: December 16, 2022. Accepted: February 2, 2023
© The Author(s) 2023. Published by Oxford University Press on behalf of National Academy of Sciences. This is an Open Access article
distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits
unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
2 | PNAS Nexus, 2023, Vol. 2, No. 3

to sanction Russian oligarchs after the 2014 invasion of Crimea (4). and nation-level offshore financial networks. This suggests a sur­
Oligarchs’ ability to escape sanctions persists because govern­ prising result: the most effective and efficient way to punish oli­
ments have been “slow to address the enabler problem,” in part garchs may be to sanction their offshore intermediaries, the
due to the exorbitant money and time required to investigate wealth managers.
the secretive mechanisms used by those intermediaries. These networks can be robust yet “super-fragile”, which is to
Any efforts to sanction or curb elites’ accumulation and abuses say that (as shown in our series of “knockout” experiments) delet­
of wealth face the challenge of secrecy. It is an old problem: the ing a very small number of intermediaries can effect significant
19th-century sociologist Georg Simmel observed that secrecy structural damage to some nation-level oligarch networks.
links all groups seeking wealth in order to escape from the rule Furthermore, fine-scale structural differences appear at the na­
of law, whether they be members of the nobility, bandit gangs, tional level, reflected in patterns of relational homophily. These
or other “predatory” associations (5). In the contemporary context fine-grained differences across scale-free networks—which we
of offshore finance, secrecy is the main product sold to elites (6–9). link to divergent political systems and historical path dependencies
The result is what anthropologist Bill Maurer calls “non-locatable —constitute our second discovery. Our finding extends Simmel’s

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

structures of domination”—a strategic obscurity surrounding who work on the limits of trust among elites linked by secrets, and aligns
owns which assets, making it all but impossible to tax, sanction, or with qualitative research indicating high levels of distrust among
otherwise hold elites accountable (10). This creates considerable high-net-worth individuals (16). This distrust can extend to the point
uncertainty about whom to target with policy measures, and how. of paranoia in some cases, particularly when it comes to sharing in­
In this paper, we introduce a network approach to the study of formation about the sources and locations of their wealth—even
oligarch sanctions, with implications for both scholarship and when dealing with the professional wealth managers who most
policy-making. Our analysis offers an efficient alternative strategy need to know (7, 13, 17). The significance of secrecy and distrust ap­
for achieving policy objectives targeting the offshore wealth of oli­ pears to increase in importance under two conditions characteristic
garchs, particularly those linked to authoritarian regimes. Using of autocracies: corruption and lack of confidence in institutions such
the Offshore Leaks Database provided by the International as the rule of law (18). Hence, qualitative research suggests that oli­
Consortium of Investigative Journalists (ICIJ) we create and analyze garchs from non-democratic regimes like Russia tightly restrict their
the global offshore networks linking high-net-worth individuals to circle of advisors, doling out information only on a “need to know” ba­
the professional intermediaries who manage their offshore wealth. sis and exclusively to a handful of trusted experts—usually chosen
We focus on elite clients from four countries: Russia, China, Hong by word of mouth from family and friends, rather than by shopping
Kong, and the USA. While our primary interest is in the recently around for the most knowledgeable or effective provider (7).
sanctioned Russian elites, this larger group of countries provides The relational patterns we discover in these financial networks
a useful general context for our findings. Recent economic ana­ are effectively networks of secrecy and are shaped in part by his­
lyses identify these four nations as the most economically unequal torical path dependencies—notably the legacy of British colonial­
in the world, with the greatest concentration of wealth in the ism, which created the legal basis for most of the offshore world
0.001% of the population: the realm of oligarchs (11). (19). More specifically, since British Common Law is the “operating
Our analysis of these “oligarch networks” provides the first system” of offshore finance, access to the system is easier for indi­
quantitative test of two key implications from prior ethnographic viduals whose native countries share those legal institutions and
research (12) on the offshore fortunes of high-net-worth individu­ the English language (7, 20). These historical path dependencies
als. This group includes oligarchs, defined as ultra-wealthy people make it easier for wealthy individuals from the USA and Hong
who use their fortunes to exert a disproportionate influence on Kong—themselves former British colonies—to obtain expert off­
politics—domestic and international. By building on in-depth eth­ shore advisory services close to home. In contrast, the system is
nography with network analysis using “big data” from offshore less accessible to elites from places that were never colonized
leaks, our investigation answers recent calls to leverage synergies and do not share the language or legal institutions that are dom­
between qualitative and quantitative research (12). inant offshore (19, 21). This may explain the distinctive patterns of
The first implication we examine from earlier qualitative stud­ Chinese and Russian engagement with the offshore financial sys­
ies is that intermediary wealth managers—rather than the elites tem. For Chinese high-net-worth individuals, Hong Kong is their
they serve—are the linchpins binding the global system of off­ primary access point to expert wealth managers providing off­
shore finance, without whom the system could not function (7). shore services; as a former British colony and one of the world’s
The second implication we study is that, despite their many com­ most established offshore financial centers, Hong Kong is re­
monalities, there remain key distinctions among oligarchs from plete with professionals who speak Chinese, understand
different countries, particularly in the patterns of their relation­ Chinese culture and politics, and therefore can serve Chinese
ships with wealth managers (7, 13). clients better than wealth managers from outside the region.
Our first discovery is that oligarch networks have a “scale-free Since Russian oligarchs have no equivalent to Hong Kong—
structure,” meaning that they are degree-heterogeneous net­ that is, no offshore financial center on the borders of their
works with heavy-tailed degree distributions. Oligarch networks country—these individuals instead access the system through
thus possess the characteristic structural vulnerability of scale- experts based in a diverse range of countries, many of which
free networks: they can be disrupted most efficiently through tar­ are former British colonies. We quantify these differences in re­
geted attacks manifested as the deletion of a few highly connected lationship diversity through a novel application of the
nodes (14). This is the so-called, “robust yet fragile” property of “Herfindahl–Hirschman Index” (HHI), a well-known diversifica­
scale-free networks, a term coined by Albert et al. (14) and charac­ tion metric of market concentration (22).
teristic of many networks, including the Internet (15). We identify More generally, our work fits into the broader body of network
financial intermediaries—wealth management professionals, science research that has been instrumental in unveiling secret­
such as the lawyers, accountants, bankers, and others who spe­ ive, deceptive, and even criminal behavior. For instance, a net­
cialize in serving the ultra-rich—as a class of highly connected no­ work approach has shed light on the ways criminal alliances
des who constitute an overlooked source of fragility in the global can be leveraged for more effective law enforcement (23–25).
 Chang et al. | 3

Other examples include the study of the structure of state- entities, who are the intermediaries included in the network. The
controlled attacks on social media during elections (26–28), on characteristics of the nation-level networks are given in Table 1.2
the communication patterns within terrorist organizations (29),
and on price-fixing conspiracies (30, 31). Research on networks
of illegal activity has identified brokers–the operational coordina­ Country-level comparisons
tors of grand larceny and other covert offenses–as the linchpins of Table 1 summarizes the characteristics of the nation-level oli­
criminal enterprises; their removal devastates the network over­ garch networks. All networks are bipartite, connecting clients
all (32–34). These brokers are characterized by high betweenness with intermediaries; while highly disconnected, they are domi­
centrality, which indicates how crucial they are to facilitating nated by a sizable giant (connected) component. Both the aggre­
shortest paths. However, only a handful of studies (35, 36) have gate network and the nation-level networks have scale-free
applied this analytic technique to the study of offshore wealth degree distributions (see Fig. S2 in the Supplementary Material).
management. None as yet have conducted systematic assess­ Clients are disproportionately represented in the spike at the
ments of the expert networks’ robustness in the face of disrup­
small degree, reflecting a strategy of trusting just a handful of

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

tions such as sanctions. This is our distinctive contribution,
intermediaries, with an average degree of 1.036 across the clients
realized as the deletion of certain expert nodes in the offshore fi­
of all countries. Fig. 2a isolates the degree distributions of inter­
nancial networks of oligarchs.
mediaries for clients based in the USA, Hong Kong, China, and
Our investigation and the research cited above can be viewed
Russia, respectively. Each are well-fit by a power-law distribu­
as a part of a growing body of work on what we call the “complex
tion (see test statistics in Supplementary Material). This is con­
systems of secrecy.” The scale-free structure we uncover in oli­
sistent with—although not required for—a growth process in
garchs’ offshore networks, particularly their overlooked vulner­
which one person’s choice of wealth manager (which is a signal
ability to attack via removal of central wealth managers with
of trust) inspires others to trust in the same intermediary. This
dense ties, exemplify a phenomenon identified in new research
preferential attachment process creates a “Matthew Effect” for
on “complex secrets” (37). This novel theory of secrecy builds on
the intermediaries.
Simmel’s work to posit that the accumulation of wealth and
While the slopes of their fits are close, there is a marked distinc­
power by elites can be most effectively achieved when key actors
tion between the distribution patterns for Russia and China versus
and data necessary to maintaining secrets are distributed across
those for the USA and Hong Kong (full robustness checks of the
multiple—often unlikely—locations in a network. We extend the
powerlaw fit can be found in Table S3 of the Supplementary
implications of this theory by showing that wealth managers re­
Material). Indeed, when we consider the kernel density estimates
present these overlooked repositories of secrets within the com­
below, we find the curves for China (black) and Russia (red) over­
plex networks of oligarchs’ offshore fortunes. This network
lapping and much flatter than those of the USA (green) and Hong
framing, characterization of roles in terms of network position,
Kong (blue). Together, this suggests divergent macroscale behav­
and distributed structure suggest that our findings and methods
ior between autocratic and democratic nations. This may be be­
can establish a basis for more effective policy interventions, and
cause oligarchs perceive wealth managers with fewer clients as
for more fruitful research on secrecy as a complex system.
being at lower risk of breaching secrecy, simply because there
are not as many people interacting with the professionals—hence,
there are fewer points of potential data leakage. Previous research
Results on Mafia networks suggests that this is a particular concern
Our analysis focuses on the network of relationships connecting among clients linked to corruption or criminality (18).
clients (also known as beneficiaries), wealth managers (also However, Fig. 2b shows how the degree distribution may not
known as intermediaries and offshore entities (such as corporate necessarily correspond to the ratio between beneficiaries and
service providers).1 In total, the data comprise 1,970,448 such en­ intermediaries. China has by far the largest number of clients
tities (the nodes in the network), along with 3,273,524 relation­ per intermediary, followed by the USA, then Russia and Hong
ships: edges connecting intermediaries, clients, and entities. Kong. We attribute this variation to supply and demand for wealth
From that superset, we extract a subset of interest, consisting of management services. For example, while China now has more
ultra-high-net worth individuals from countries known to pro­ billionaires than any other country, and is producing new billion­
duce oligarchs—Russia, China, Hong Kong and the USA—for a to­ aires at a rate three times that of its nearest competitor (the USA)
tal of 79,458 clients and 143,795 edges. Fig. 1 visualizes the (38), the supply of Chinese-speaking wealth managers has not
network of intermediaries and 90,155 of their clients from caught up to this new demand; therefore, each wealth manager
Russia (RUS), China (CHN), Hong Kong (HKG), and the USA (USA) serving Chinese high-net-worth individuals carries a dispropor­
(see data and methods). Centered around these clusters are inter­ tionately large number of clients. Hong Kong’s low number of cli­
mediaries, marked in yellow. ents per intermediary may reflect the high concentration of
We further partition this aggregate network into four nation-level wealth managers working there, creating intense competition
oligarch networks, one each for Russia, China, Hong Kong, and the for clients. In any case, these figures show two things: first, bene­
USA. Each nation-level network is constructed by first extracting ficiaries greatly outnumber intermediaries in all cases; second,
all beneficiaries from a specific country. We then identify entities sanctions on intermediaries would affect many oligarchs at once.
they are associated with, and the intermediaries who created the
2
We use the terms “intermediaries” and “wealth managers” interchange­
1
We use the terms “beneficiary,” “intermediary,” and “entities” in keeping ably, and the terms “clients” and “beneficiaries” interchangeably. Both the
with the norms of the financial services industry (8, 56), as reflected in the ter­ terms “intermediaries” and “wealth managers” signify experts whose formal
minology found in the ICIJ dataset. We emphasize that none of the terms are training is in the law, accountancy, banking, tax advisory services and related
technical social network measures; for example, “intermediary” does not con­ fields. By the same token, the word “beneficiaries” defines “clients” in relation to
note centrality in an arbitrary network. That said, the oligarch networks are bi­ the legal structures that wealth managers create—as in the phrase “beneficiar­
partite (connecting intermediaries, clients, and offshore entities) and in this ies of offshore trusts.” In our case-study, we further define “oligarchs” as a sub­
case, intermediaries do in fact lie between clients who are not connected to set of ultra-high-net-worth individuals who devote some of their time and
each other. wealth to influence affairs of state, whether domestic or international.
4 | PNAS Nexus, 2023, Vol. 2, No. 3

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

Fig. 1. Offshore bipartite financial networks constructed using the ICIJ offshore leaks database. For clarity, shown here is the partial network of 79,458
intermediaries and their clients from Russia (RUS/red), China (CHN/purple), Hong Kong (HKG/green), and the USA (USA/blue). Nodes are either
beneficiaries or intermediaries, visualized through physics-based verlet integration. This is a subset of a greater network of 1,970,448 nodes and 3,273,524
edges.

Table 1. Oligarch network statistics. that various connectivity markers of the networks collapse
completely when a very small number of highly connected
Country Clients Intermed Edges LGC size % GC
nodes—ties to wealth managers—are removed. Fig. 3 shows the
CHN 32,045 1,601 48,239 22,235 66.1 results of our iterative removal of the top three wealth managers
RUS 6,311 510 8,512 4,267 62.6
(by degree) from the offshore networks of high-net-worth individ­
USA 15,450 1,632 32,253 8,647 50.6
HKG 25,661 3,665 54,791 15,785 53.8 uals from each of the four countries in our study (for the full list of
All 4 79,458 5,711 143,795 69,916 60.1 managers, see Table S2 in the Supplementary Material). We then
assessed global network metrics, normalized on the values from
The rightmost column, “% GC” counts the proportion of the largest connected the original network. Specifically, we measured the impact of
component (LGC) in each network. Note that they are each dominated by a
large “giant component” (GC). node removal on (a) network size, (b) the number of triangles, (c)
redundancy, and (d) the clustering coefficient. Size refers to the
cardinality of the network, which diminishes when nodes are re­
Previous work by Albert et al. (14) shows that “targeted attacks” moved; however, if a beneficiary is connected to multiple inter­
on the most highly connected nodes can destroy the connectivity mediaries, then removal of one intermediary will not remove
of a scale-free network. This property obtains for each of the four the beneficiary. The term triangles refers to the number of triplets
nation-level oligarch networks we construct. However, we also that share clients. This tells us how diversified clients are: for in­
find that Russian and Chinese oligarch networks are character­ stance, if client a and client b both employ wealth managers x and
ized by “super fragility” in the face of such attacks. This means y, then even with the removal of x they will remain part of a
 Chang et al. | 5

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

Fig. 2. A Matthew Effect of trust in accumulated advantage for offshore
wealth managers. A small fraction of these experts serve
disproportionately large numbers of clients. Shown are degree properties
by clients of certain countries. a) Shows the degree distribution of these
intermediaries. b) Shows the ratio of beneficiaries to wealth managers/
intermediaries.

triangle (a–y–b). Redundancy measures the total number of pair-

wise paths between beneficiaries. Finally, the clustering coeffi­
cient represents the fraction of completed triangles out of all
possible “triangular” interactions. This is yet another measure of
connectivity (see Methods for detailed descriptions).
The divergent patterns here are quite striking: while
the offshore networks of clients from the USA and Hong Kong
are relatively unperturbed by the removal of intermediaries,
the networks of clients from Russia and mainland China
collapse after removal of just one or two intermediaries. All
four measures—size, triangles, redundancy, and clustering co­
efficient—diminish significantly after the top intermediary no­
des are removed.
Our findings are consistent with qualitative research on the
central role of expert wealth managers in the functioning of the
offshore system. In particular, our analyses indicate that these
professional intermediaries are the key nodes in the network of
global private wealth: they are the point of contact uniting mul­ Fig. 3. Wealth managers as the points of vulnerability in oligarchs’
tiple oligarchs and offshore structures across national boundaries offshore financial networks, measuring a) size, b) number of triangles, c)
the clustering coefficient, and d) redundancy. Noticeably, the offshore
(8, 9, 39). Together, these results suggest why previous sanctions networks of high-net-worth individuals from China and Russia are much
regimes targeting individual oligarchs or offshore jurisdictions more vulnerable to such targeted knock-outs of wealth managers than
have proved easier to bypass than policy-makers expected: they the networks of high-net-worth individuals from the USA and Hong Kong.
6 | PNAS Nexus, 2023, Vol. 2, No. 3

have mistakenly targeted the spokes of a wheel rather than the

hub around which the whole system turns.
Fig. 3 further suggests that power is not only a matter of con­
centration and quantity of clients, but of (dis)trust and quality
of the client relationship. This is another point suggested by pre­
vious qualitative research that we have been able to test and val­
idate through analysis of big data. This aligns with Fig. 2b, which
shows Russian and Chinese elites share the secrets of their wealth
with a much smaller set of professional intermediaries than do
their peers from relatively more free and democratic societies,
such as the USA and Hong Kong.
A natural follow-up question is whether degree centrality is the
best criterion for knock-out from a disruption viewpoint. Degree

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

centrality is a simple indication of the number of clients an inter­
mediary has. Betweenness centrality, reflecting the number of
shortest paths in which a node lies on, is another natural choice
and might also be used to target nodes which exhibit high broker­
age leverage (in the network scientific term).3
We test this alternative centrality measure by running the
same knock-out experiments as before but based on betweenness.
For direct comparison purpose, we show the ratio, r of the result­
ing knock-out impacts by choosing betweenness versus degree
centrality, across the four robustness measures. Here, r > 1 means
greater retention of the disrupted network, hence less impact on
the network, and indicates knockout based on betweenness cen­
trality is less effective. On the other hand, r < 1 indicates the op­
posite is true. Fig. 4 shows these ratios for the four regions.
For Hong Kong and the USA, knockout based on degree central­
ity performs monotonically better. For China, the first choice in
ranking betweenness and degree centrality coincides, which pro­
duces a ratio of r = 1, but the performance of degree centrality-
based knockout increases significantly onwards. For Russia how­
ever, choosing the intermediary by betweenness centrality to
knock out may actually generate a better first response, as the ra­
tio dips below 1, before increasing above 1.
This extended result affirms that finding the most crucial no­
des in knockout experiments is not a simple matter: the optimal
choice is context-dependent. For instance, a common scheme in
computer science is modularity maximization (40), which would
ensure the maximal knock out of client nodes. However, this
does not necessarily correlate with redundancy across clients,
which is based on the Cartesian product on other possible paths.
Thus, in the context of offshore finance, the optimal strategy is
likely a weighted combination of two-step betweenness, degree
centrality, and mutual nodes across the set of knock-out manag­
ers. This is because redundancy is fundamentally defined on both
size (how many knocked out) and connectedness (reduction in the
product space), and with each country having its own unique con­
figuration. The findings illustrated in Figs. 3 and 4 represents the
ramifications of the scale-free structure of the oligarch networks,
however; as indicated by Fig. 2b, there are still important distinc­
tions to be found in the nature of the edge’s spatial relations.
In Fig. 5a and b, we drill down into the discrepancy between the Fig. 4. Ratio of knock-out results, based on degree centrality or betweenness
power-law and edge quantity findings in the offshore networks of centrality, for measures of a) size, b) number of triangles, c) the clustering
coefficient, and d) redundancy. r > 1 indicates knocknout based on degree
Russian versus Chinese oligarchs. As these figures show, Russian
centrality is more effective; r < 1 indicates knockout based on betweenness
wealthy elites choose from a far more geographically distributed centrality is more effective. For both USA and HKG, degree-based removal
group of advisors than do the Chinese, who draw their wealth man­ leads to better knockout performance. For CHN and RUS, betweenness-based
agers primarily from Hong Kong. This is likely due to the aforemen­ removal can yield equal or better results, if choosing one.
tioned path dependencies, which—for reasons of British colonial
history—put one of the world’s most important offshore financial who can speak Chinese and understand the cultural, political and
centers right on China’s doorstep, complete with wealth managers legal concerns of Chinese clients. Russians lack that convenient ac­
cess portal to the offshore world, and thus by necessity must seek
3
We thank one of the reviewers for this suggestion. intermediaries elsewhere. This spatial heterogeneity should be
 Chang et al. | 7

most powerful in Russia, several of whom have been subject to

international sanctions numerous times over the past decade
(42). The relational pattern shown in Fig. 6a, which is the induced
network on the 26 oligarchs (see Methods), underscores the ex­
treme concentration of these oligarchs’ offshore networks, which
can be encapsulated in just 124 nodes. Note, here we expand our
set of nodes to include three additional types. The nodes labeled
“Business entity” represent the low-level offshore service pro­
viders wealth managers hire to do the paperwork of creating
trusts and corporations. The nodes labeled “Nominee Admin”
are people who rent their names for use in public documentation,
to preserve the privacy of oligarch clients using offshore struc­
tures. Finally, nodes labeled “Offshore Address” represent the lo­

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

cations for which trusts and corporations are set up. In these
figures, cyan denotes intermediaries and yellow represents the
oligarchs; in addition, blue nodes represent entities, green denotes
nominees, and gray represents offshore addresses.
Fig. 6c shows that intermediary firms such as Markom
Management—a boutique wealth management agency based in
London—not only link oligarchs to numerous offshore entities,
but to other oligarchs. For example, Markom works with three
members of the Rotenberg family, who are among the wealthiest
people in Russia as well as being close personal friends of
President Vladimir Putin. In Fig. 6d, we see that Boris Rotenberg
is linked via an entity—a company in the British Virgin Islands
that does trust services paperwork, presumably hired for
Rotenberg by Markom—to Alisher Usmanov, the oligarch “en­
trusted with servicing financial flows” for Putin’s own offshore
wealth. The same pathway links Rotenberg to Vladimir
Kiriyenko, son of Putin’s current Deputy Chief of Staff (43). One
might conclude from this that the entities should be sanctioned;
but as we know from qualitative research, entities are down­
stream in the system from wealth managers, and much more nu­
merous (7). Entities providing paperwork are easily replaced if
sanctioned or eliminated, because the skill set they call upon—
compliance with local laws, making sure signatures are in the
right places and fees paid to corporate registrars—is readily repli­
cated. Wealth management skills, however, are far more com­
plex, cross-national and not easily substitutable. That is why
oligarchs hire and work with wealth managers, not corporate ser­
vice providers or nominees.
Fig. 5. Location of intermediaries for high net-worth clients from a)
Russia versus b) China. Whereas Russians employ a geographically Finally, Fig. 6e shows the extreme concentration of Alisher
diverse set of intermediaries, Chinese clients employ wealth managers Usmanov’s offshore holdings under the auspices of a single Isle of
principally located in Hong Kong. This is equivalent to a diversity index of Man-based wealth management firm, Bridgewaters Limited. This il­
0.151 and 0.068 for Russia and China, respectively. lustrates what a “highly connected node” in a scale-free offshore net­
work looks like in practice. Taken together, the images in Fig. 6a–e
acknowledged both by scholars and in future sanctions regimes. illustrate that wealth management intermediaries are central not
Quantitatively, we compute a diversity index (DI) based on the in­ only to Russian oligarchs’ offshore financial networks, but also to
verse Herfindahl–Hirschman Index (HHI), where higher DI indi­ their social and professional networks. The intermediaries should
cates greater diversity (see Supplementary Material). This yields therefore be a prime target for sanctions and other efforts aimed
0.151 and 0.068 for Russia and China, respectively, indicating at disrupting the activities of those oligarchs. Quantification of these
that Russian oligarchs’ ties to wealth managers are on the order motifs, at a larger scale, is ripe for future research.
of four times as diverse as those of Chinese oligarchs. Table 2 provides more detail on these concentrated local net­
works (see Methods). Sorted by the number of sanctioned oli­
garchs affiliated with each management firm, Table 2 shows an
Russian oligarchs sanctioned following the 2022 inverse relationship between the number of sanctioned oligarchs
Ukraine invasion and the firms’ total number of clients. In other words, the more a
In Fig. 6, we isolate the offshore networks of the 26 Russian oli­ wealth management firm serves sanctioned Russian oligarchs,
garchs subject to sanctions on their offshore wealth following the fewer clients that firm has in total. This supports earlier quali­
the February 2022 invasion of Ukraine; these are a subset of tative research which suggested that oligarch–intermediary rela­
Russians sanctioned by the UK, USA, and EU who also appear in tionships were characterized by an unusual, almost familial
the ICIJ’s offshore leaks database (41). This small group includes level of intimacy quite unusual for professional work, but com­
some of the wealthiest people in the world, and some of the mon in criminal networks such as those of Mafia families (7).
8 | PNAS Nexus, 2023, Vol. 2, No. 3

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

Fig. 6. The offshore networks of the 26 sanctioned Russian oligarchs who appear in the ICIJ’s offshore leaks database. These oligarchs are connected
through just 124 nodes (a), forming a rather small social circle of wealth, power and influence. Panels (b), (c), (d), and (e) show different clusters of
oligarchs connected via wealth managers, business entities, or offshore locations. Blue denotes entities, green denotes officers, cyan denotes
intermediaries, and yellow the oligarchs.

Table 2. Intermediary firms ordered by the total number of of clients—in Appleby’s case including the Queen of England—
sanctioned Russian oligarchs they are affiliated with.
are careful about protecting their own reputations through exclu­
Intermediary Sanctioned Entities Clients sion of clients who might be tainted by involvement in corrupt or
oligarchs criminal activities.
In contrast, smaller boutique firms such as Markom and
Markom Management Ltd. 6 140 369
American Corporate Services, Inc. 4 311 540 American Corporate Services Inc—which represent a vanishingly
Appleby Trust (Isle of Man) Limited 4 458 3,566 small number of clients compared to Appleby’s—likely service a
I&T Consulting Ltd. 3 138 483 disproportionately large segment of the Russian oligarch market
G.S.L. Law & Consulting 3 2,097 3,487
while taking on a much smaller client base overall. From the inter­
Dr. K. Chrysostomides & Co. 2 69 207
Ryon Ltd. 2 305 620 mediaries’ perspective, this suggests a corporate strategy focused
Consulco International Ltd. 2 3,168 5,499 on charging higher fees to legally and reputationally risky clients.
Appleby Services (Bermuda) Ltd. 2 3,645 72,316 From the perspective of Russian oligarchs, this illustrates an ex­
Dietrich, Baumgartner & Partner 1 41 162 treme concentration or localization of trust in the hands of a
Bridgewater Limited 1 200 666
tiny number of professionals who serve clusters of closely tied
Christodoulos Vassiliades 1 287 562
Andersen Business Services, Inc. 1 521 1,350 family and friendship groups. This pattern helps explain why
Lotus Holding Company Limited 1 1,218 2,797 Russian elites’ offshore networks would be so vulnerable to tar­
geted attacks on their intermediaries (Fig. 3).
From this limited sample, there is a negative correlation between the number of
sanctioned oligarchs and each firm’s affiliated total number of clients overall,
suggesting the special vulnerability of Russian oligarchs’ offshore networks to
interventions focused on their wealth managers.
Discussion
Our findings suggest that the offshore financial system operates
Table 2 shows that the Bermuda-based multinational offshore as a scale-free network, both globally and on the level of nations,
law firm Appleby represents the largest number of clients overall and thus shares the same type of structural vulnerabilities as oth­
in the offshore leaks database; however, they only represent two er scale-free networks, such as the Internet. They are robust in
sanctioned Russian clients, Kerimov, and Arkady Rotenberg. some respects, enduring through random deletion, but some are
This may be because large multinationals serving a wide range “super fragile” (in terms of the hallmarks of connectivity) to
 Chang et al. | 9

targeted node removal. Most connectivity is concentrated in a few secretive phenomena like financial corruption have proved robust
nodes: in the case of Russian oligarchs’ offshore networks, the and resistant to change due to certain patterns in evolved struc­
high-connectivity nodes consist of a small group of wealth man­ tures that enable the distribution and flow of information (37).
agers. The network structure we have uncovered can explain Our analysis of ties between oligarchs and their wealth managers
why earlier rounds of sanctions directed at those oligarchs—the not only represents an empirical case of such a complex system of
sparsely connected nodes in our analysis—could largely be secrecy, reified in a network structure, but suggests that such sys­
evaded by some prominent individuals, despite direct seizure of tems may be characterized by the specific network topology we
some of their assets. Our findings suggest that future sanctions identify. Our results diverge from prior findings on criminal net­
should be directed at the professional intermediaries who con­ works (33) showing that the most sensitive nodes are character­
struct and maintain the offshore system for the oligarchs’ benefit. ized by high betweenness and low degree. Specifically, we find
We identify these expert advisors as overlooked chokepoints in that if restricted to just one intermediary, choosing by between­
the global financial network. ness centrality may yield greater disruption to the underlying fi­
To obtain these results, we used data from the offshore leaks nancial networks of oligarchs from autocratic countries.

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

databases created by ICIJ from the Panama, Paradise, and However, in general, knock-out based on intermediaries’ degree
Pandora Papers leaks to conduct a network attack analysis. We centrality is more effective.
show that for Russian oligarchs—as well as their peers in China Both sets of results suggest that the information structures
—removal of just one or two intermediaries collapsed their off­ underlying complex systems of secrecy are diverse and context-
shore networks entirely. The fragility of these networks is not a dependent in patterned ways; future research should explore
surprise: offshore finance is by definition an elite phenomenon in­ these patterns in greater depth. Even in the settings in which
volving a few hundred thousand ultra-high-net-worth individuals scale-free structure is observed our experiments show that their
—including just under 2,700 billionaires (44)—served by an even points of fragility may require new measures for identification.
smaller group of intermediaries (7, 11, 21). But the distinctive As we find the set of most sensitive nodes depends on a country’s
“super fragility” of some offshore networks is a novel finding. unique network configuration, this suggests exciting new direc­
Our analysis pinpoints the greatest vulnerabilities in those net­ tions of inquiry focused on pinpointing optimal knock-out targets,
works, and identifies important national variations in their struc­ contingent on characteristics of the clients’ country of origin. We
tures, which we expect may be influenced by geography, history, look forward to extending these analyses, adding texture to our
language, and governance. understanding of the adaptations and variations from which se­
Like any study of elite phenomena, our research is limited by crecy networks emerge.
data that can only provide fragmentary snapshots of a system As a practical matter, the results of our attack analysis indicate
that is observably far larger than anyone has measured. This is that future sanctions strategies against rogue states should
a perennial problem with research on elites, and particularly on consider targeting the intermediaries serving the country’s elite,
elite wealth: information is extremely scarce because it is inten­ instead of or in addition to sanctions on the individual beneficiar­
tionally shrouded in multiple levels of secrecy (7, 8). Thus, data ies themselves. We may soon have data to test this claim. In early
points are few and representativeness difficult to assess. Our ana­ June, the US banned the provision of some offshore expertise to
lysis drew from what is now the best data available: the 6.94 ter­ Russian oligarchs—including accountancy and corporate forma­
abytes of data and 46.8 million records comprising the Panama, tion, which are crucial for creating and maintaining the structures
Paradise, and Pandora Papers leaks. We were fortunate to have that shroud oligarchs’ assets in secrecy (47). Shortly thereafter,
such information, since so much previous research on offshore the EU and the UK implemented similar measures, banning inter­
wealth has had to rely on indirect or imputed measures (45). national law and tax experts from those jurisdictions from serving
Still, there is much more to learn and our findings should be put sanctioned Russian oligarchs (48–50). These bans are backed by
to the test against the offshore data likely to emerge in future. A penalties including imprisonment and huge fines.4
potential limitation of our study is that some in network science Given our analysis showing Russian elites’ reliance on inter­
have questioned the generalizability of scale-free networks (46). mediaries based in the UK (and its dependent territories) and in
Fortunately, our results are robust regardless of whether the off­ Cyprus (an EU country), these bans could represent a meaningful
shore network is scale-free; as long as the network structure is blow to the oligarchs’ ability to access their troves of offshore
characterized by extreme heterogeneity in the density of inter­ wealth or move them beyond the reach of sanctioning countries.
mediary node connectivity, our findings hold. Our goodness-of-fit Crucially, the ban on the provision of intermediary services is sep­
appears sufficiently robust, but the broader take-away is that two arate from and more extensive than freezing or seizing any par­
characteristics uncovered by our analysis—heavy-tail behavior ticular asset; rather, the ban means cutting off the expertise
and preferential attachment—align closely with the findings of pipeline linking sanctioned individuals to their offshore wealth.
previous qualitative research, forming a coherent, consistent pic­ It is a more encompassing punishment than losing access to a spe­
ture of oligarchs’ offshore networks. cific bank account or yacht or private jet.
For scholars of inequality, globalization and finance, our work This approach to sanctions via bans on the provision of expert
offers several new insights that may be fruitful for future re­ professional services has been around for decades, and derives its
search. First, our findings substantiate the implications of previ­ legal basis from defense ministry regulations. Such bans were typ­
ous qualitative research (12) suggesting that professional ically imposed on the transfer of data deemed important to na­
intermediaries are pivotal in linking and maintaining the offshore tional security and technology related to nuclear, chemical, and
financial system, and thus in producing some of its most harmful
outcomes, such as exploding wealth inequality and elite corrup­ 4
In the USA, current law imposes civil penalties on individual intermediar­
tion (7–9). There should therefore be more social scientific atten­ ies up to $250,000; for criminal convictions, the penalty is up to $1 million in
tion to such actors. fines and 20 years in prison (see 50 USC 1705 at https://uscode.house.gov/). In
the UK and EU, such bans on the provision of expert intermediary services
Second, our study could be particularly useful in building out are characterized by the same basic purpose and mechanisms, as well as simi­
an emerging theory of “complex secrets,” which posits that lar penalties.
10 | PNAS Nexus, 2023, Vol. 2, No. 3

biological weaponry (51). That such rules are now being applied to Country-level comparisons
financial, legal, and accounting expertise indicates growing recog­ To assess the country-level behavior of beneficiaries, we first filter
nition by policy practitioners that offshore wealth management on beneficiary officer nodes, then extract their connected entities.
can threaten international security and stability. This is consist­ From the edge list, we then extract the intermediaries connected
ent with our analysis illustrating the significance of these profes­ to these entities.
sionals as chokepoints in global financial networks. Our work thus
also motivates a closer look at country-specific sanctions for more Bc = {b ∈ B | CNTRY(b) = c}
policy-oriented outcomes, and inspires subsequent work to take Ec = {e ∈ E | e ∈ nei(b)∀b ∈ Bc } (1)
up more nation-level policy comparisons (52, 53). Ic = {i ∈ I | I ∈ nei(e)∀e ∈ Ec }
Most importantly, sanctioning professional intermediaries has
Note, it is possible to subset Bc , Ec , and Ic directly from the log-level
a solid track record of effectiveness in achieving policy goals. For
data. Upon acquiring the set of intermediaries tied to beneficiaries
example, the current US policy vis-a-vis intermediaries and
Russian oligarchs was preceded by a similar ban on intermediar­ from country c, we first consider the ratio of these two sets: BIcc . We

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

ies working with government officials in Iranv—which the US then consider the degree distribution of Ic for each country, which
sanctions as “State Sponsor of Terrorism.” This approach has is then plotted as a power law.
been credited with “contributing to Iran’s decision to enter into
a 2015 agreement that put limits on its nuclear program,” in
part because the country’s leadership “could not access its foreign Node-removal experiment
exchange assets held abroad”(54). The purpose of this analysis is to assess the robustness of financial
Finally, the highly tailored impact of this strategy makes it far networks in general. We choose the USA, China, and Russia for
less morally fraught than broad-based sanctions which can de­ comparison, adding in Hong Kong due to its role in facilitating
prive whole nations of resources like grain, medical supplies, Chinese investments and known role as a hub for intermediaries.
and fuel. While there is no known data on the frequency with For each country c, we identify the top 3 frequently occurring
which bans on elite intermediary services are used as part of sanc­ intermediaries in Ic . We then iteratively remove i1 , i2 , and i3 .
tions packages, this strategy is certainly less well-known publicly At each iteration, we assess the size, clustering coefficient,
than resource sanctions (despite the occasional breathless media number of triangles, and redundancy. Size denotes the cardinality
coverage of luxury asset seizures). We hope that our study will of the set of nodes, which serves as one view of redundancy in
contribute to more widespread knowledge, discussion, and appli­ these networks, since if a client only invests through one
cation of this more precisely targeted tool of international policy. intermediary, they will be removed. The clustering coefficient is
given as:

Num. of closed triangles

C= (2)
Methods Num. of possible triangles
Dataset As such, the number of triangles provides a more raw and less
In its natural form, each entry in the IJIP database can be one of scaled measure of such redundancy. Intuitively, if clients tend
five node classes: to diversify their intermediaries, more triangles will be present.
Lastly, our own measure of redundancy counts the possible paths
• Officers—Individual people who are related to entities. We between any two beneficiaries for a given country. We approxi­
further split this into three classes. Beneficiaries: The direct mate this by taking the square sum of the cardinality of all con­
recipients and benefits of offshore accounts; Nominees: nected components. Formally, let CC(k) denote the set of
Individuals who are instated to manage these accounts; connected components when k intermediaries are removed.
(Officer) Intermediaries: Those employed to set-up these Then the normalized redundancy metric R is given by:
accounts.
Σa∈CC(i) |a|(|a| − 1)
• Entities—The businesses set-up for beneficiaries. These can R= (3)
Σb∈CC(0) |b|(|b| − 1)
be corporations, foundations, or trusts.
• Intermediaries—Companies that help set-up these offshore For size, clustering coefficient, and triangles we use the built-in
entities. NetworkX package for direct computation.
• Addresses—Addresses registered to the three other types of
nodes named above.
Sanctioned oligarch-level network
These four node classes are then structured through an edge Using fuzzy-word matching, we extracted 26 of 41 sanctioned oli­
list. Relational in nature, the general structure is as followed: di­ garchs. We then constructed an undirected subgraph between all
rected edges from officers to entities, intermediaries to entities, the 26 oligarchs, restricted to one degree away (up to one direct
and from officers, entities, and intermediaries to addresses. In intermediary). This yielded connected components, for which
this study, we focus especially on the tripartite structure of offi­ we computed the shortest paths between these connected compo­
cers, entities, and intermediaries, and use this tripartite network nents via oligarchs.
to induce a bipartite network between intermediaries and officers Fig. 6 shows how the sanctioned oligarchs are directly connected
(for a visualization of the tripartite structure, see Fig. S1 in the and allow us to observe the minimal paths among these clusters of
Supplementary Material). We further filter on the officer class, oligarchs. Given the set of intermediaries linked to the Russian
classifying manually the top 50 most common classes (i.e. share­ Oligarchs (which we denote as IRO ), we further expanded the net­
holder, ultimate beneficiary, nominee executive). The top 50 work by one degree to tabulate the number of entities and clients
classes account for 99.2% of all entries in the relational edge list each of these intermediaries serves, which gives the power of inter­
(see Table S1 in the Supplementary Material). mediaries for their sanctioned clients as shown in Table 2.
 Chang et al. | 11

Although our analysis is based on the first wave of sanctions— References

as of March 1st—we do not expect the network structures and pat­
terns we discovered to change with subsequent waves of sanc­ 1 Alstadsæter A, Johannesen N, Zucman G. 2018. Who owns the
tions (55). We leave this an empirical question for future research. wealth in tax havens? Macro evidence and implications for glo­
bal inequality. J Public Econ. 162:89–100.
2 Wolcott R, Wolff B, Ehret T. 2022. The fog of sanctions. Thomson
Reuters [accessed 2022 Dec 12]. https://www.thomsonreuters.
Acknowledgement com/en-us/posts/wp-content/uploads/sites/20/2022/07/Russia-
The authors would like to thank Neils Johannesen, Finn Tarp, Bob Sanctions-White-Paper-2022.pdf.
Rijker, and UNU-WIDER for feedback. H.C. thanks the ANN group 3 Gregory J. 2022 Oct 11. UK businessman charged with “helping
for feedback. For publication funding, the authors thank the Bill Russian oligarch evade sanctions”. BBC News [accessed 2022
and Melinda Gates Foundation (Grant No. OPP1217336). Dec 12]. https://www.bbc.com/news/uk-63218643.
4 Bertrand N. 2020 Jul 29. Congressional probe: Russian oligarchs

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

using art to evade sanctions. Politico [accessed 2022 Dec 12].
https://www.politico.com/news/2020/07/29/probe-russian-oligarchs-
Supplementary material
evade-art-sanctions-386154.
Supplementary material is available at PNAS Nexus online. 5 Simmel G. 1906. The sociology of secrecy and of secret societies.
Am J Sociol. 11(4):441–498.
6 Harrington B. 2012. Trust and estate planning: the emergence of a
Authors’ contributions profession and its contribution to socioeconomic inequality 1. In:
Sociological forum. Vol. 27. Wiley Online Library. p. 825–846.
D.R., F.F., B.H., and H.C. conceived the analysis and experiments;
7 Harrington B. 2016. Capital without borders. Cambridge, MA: Harvard
H.C. analysed the results. All authors wrote and reviewed the
University Press.
manuscript.
8 Hoang KK. 2022. Spiderweb capitalism: how global elites exploit fron­
tier markets. Princeton, NJ: Princeton University Press.
9 Winters JA. 2011. Oligarchy. Cambridge, UK: Cambridge University
Data availability Press.
The data underlying this article are publically available on the ICIJ 10 Maurer WM. 1995. Complex subjects: offshore finance, complex­
website. All code used to analyze the dataset is available here: ity theory, and the dispersion of the modern. Social Rev. 25(3&4):
https://github.com/herbertfreeze/RussianFinance. The data used 117.
can be acquired from https://offshoreleaks.icij.org/pages/database. 11 Piketty T, Yang L. 2021. Income and wealth inequality in Hong
Kong, 1981–2020: the rise of pluto-communism? Available at
SSRN 3888118.
12 Grigoropoulou N, Small ML. 2022. The data revolution in social
Notes
science needs qualitative research. Nat Hum Behav. 6:904–906.
1. We use the terms “beneficiary,” “intermediary,” and “entities” in 13 Harrington B, Strike VM. 2018. Between kinship and commerce:
keeping with the norms of the financial services industry (8, 56), fiduciaries and the institutional logics of family firms. Fam Bus
as reflected in the terminology found in the ICIJ dataset. We em­ Rev. 31(4):417–440.
phasize that none of the terms are technical social network meas­ 14 Albert R, Jeong H, Barabási AL. 2000. Error and attack tolerance of
ures; for example, “intermediary” does not connote centrality in complex networks. Nature. 406(6794):378–382.
an arbitrary network. That said, the oligarch networks are bipart­ 15 Doyle JC, et al. 2005. The “robust yet fragile” nature of the
ite (connecting intermediaries, clients, and offshore entities) and Internet. Proc Natl Acad Sci USA. 102(41):14497–14502.
in this case, intermediaries do in fact lie between clients who are 16 Sherman R. 2019. Uneasy street. Princeton, NJ: Princeton
not connected to each other. University Press.
2. We use the terms “intermediaries” and “wealth managers” inter­ 17 Marcus GE, Hall PD. 1992. Lives in trust: the fortunes of dynastic fam­
changeably, and the terms “clients” and “beneficiaries” inter­ ilies in late twentieth-century America. Boulder, CO: Westview Press.
changeably. Both the terms “intermediaries” and “wealth 18 Gambetta D. 2000. Mafia: the price of distrust. Trust: Making and
managers” signify experts whose formal training is in the law, ac­ Breaking Cooperative Relations. 10:158–175.
countancy, banking, tax advisory services and related fields. By 19 Ogle V. 2017. Archipelago capitalism: Tax havens, offshore mon­
the same token, the word “beneficiaries” defines “clients” in rela­ ey, and the state, 1950s–1970s. Am Hist Rev. 122(5):1431–1458.
tion to the legal structures that wealth managers create—as in 20 Palan R. 2002. Tax havens and the commercialization of state
the phrase “beneficiaries of offshore trusts.” In our case-study, sovereignty. Int Organ. 56(1):151–176.
we further define “oligarchs” as a subset of ultra-high-net-worth 21 Chavagneux C, Murphy R, Palan R. 2013. Tax havens: how global­
individuals who devote some of their time and wealth to influ­ ization really works. Ithaca, NY: Cornell University Press.
ence affairs of state, whether domestic or international. 22 Rhoades SA. 1993. The Herfindahl-Hirschman index. Fed Res Bull.
3. We thank one of the reviewers for this suggestion. 79:188.
4. In the USA, current law imposes civil penalties on individual 23 Duijn PAC, Klerks PPHM. 2014. Social network analysis applied to
intermediaries up to $250,000; for criminal convictions, the pen­ criminal networks: recent developments in Dutch law enforce­
alty is up to $1 million in fines and 20 years in prison (see 50 ment. In: Networks and network analysis for defence and secur­
USC 1705 at https://uscode.house.gov/). In the UK and EU, such ity. p. 121–159. New York, Cham: Springer International
bans on the provision of expert intermediary services are charac­ Publishing.
terized by the same basic purpose and mechanisms, as well as 24 Klerks P. 2004. The network paradigm applied to criminal organ­
similar penalties. isations: theoretical nitpicking or a relevant doctrine for
12 | PNAS Nexus, 2023, Vol. 2, No. 3

investigators? Recent developments in the Netherlands. In: 40 Good BH, De Montjoye YA, Clauset A. 2010. Performance of
Transnational organised crime. p. 111–127. London, modularity maximization in practical contexts. Phys Rev E.
UK: Routledge. 81(4):046106.
25 McIllwain JS. 1999. Organized crime: a social network approach. 41 International Consortium of Investigative Journalists. 2022 Apr.
Crime Law Soc Change. 32(4):301–323. List of oligarchs and Russian elites featured in ICIJ investigations.
26 Chang HCH, Chen E, Zhang M, Muric G, Ferrara E. 2021. Social 42 Tognini G, Hyatt J. 2022 Feb. The Forbes ultimate guide to
bots and social media manipulation in 2020: the year in review. Russian oligarchs. Forbes.
In: Handbook of computational social science. Vol. 1. 43 Partridge J. 2022 Mar. Alisher Usmanov: Ex-fencer who – solves
Routledge. p. 304–323. Putin’s business problems’. The Guardian.
27 Chang HCH, Haider S, Ferrara E. 2021. Digital civic participation 44 Hernandez J. 2022 Apr. The number of billionaires has fallen (but
and misinformation during the 2020 Taiwanese presidential they’re worth a collective $12.7 trillion).
election. Media Commun. 9(1):144–157. 45 Farrell H, Newman AL. 2022 May. Weak links in finance and sup­
28 Ferrara E, Chang H, Chen E, Muric G, Patel J. 2020. Characterizing ply chains are easily weaponized. Nature. 605:219–222.

Downloaded from https://academic.oup.com/pnasnexus/article/2/3/pgad051/7059318 by guest on 28 March 2023

social media manipulation in the 2020 U.S. presidential election. 46 Holme P. 2019. Rare and everywhere: perspectives on scale-free
First Monday. 25(11). networks. Nat Commun. 10(1):1–3.
29 Matusitz J. 2008. Similarities between terrorist networks in an­ 47 Amato N. 2022. US sanctions ban provision of accounting, con­
tiquity and present-day cyberterrorist networks. Trends Organ sulting services to Russia. J Account Res. https://www.journalo­
Crime. 11(2):183–199. faccountancy.com/news/2022/may/us-sanctions-ban-provi­
30 Baker WE, Faulkner RR. 1993. The social organization of conspir­ sion-accounting-consulting-services-russia.html.
acy: illegal networks in the heavy electrical equipment industry. 48 European Commission. 2022 Sep. Press statement by president von
Am Sociol Rev. 58:837–860. der Leyen on a new package of restrictive measures against Russia.
31 Enders W, Su X. 2007. Rational terrorists and optimal network 49 Commonwealth & Development Office Foregin. 2022 Dec.
structure. J Confl Resolut. 51(1):33–57. Sanctions in response to Putin’s illegal annexation of ukrainian
32 Bright D, Greenhill C, Britz T, Ritter A, Morselli C. 2017. Criminal regions.
network vulnerabilities and adaptations. Global Crime. 18(4): 50 Thomas A, Histed B, Costelloe G, Kang T. 2022 Oct. Incoming uk
424–441. sanctions on professional services.
33 Morselli C. 2010. Assessing vulnerable and strategic positions in a 51 McGowan M. 2007. Between a rock and a hard place: the export of
criminal network. J Contemp Crim Justice. 26(4):382–392. technical data under the international traffic in arms regula­
34 Morselli C, Roy J. 2008. Brokerage qualifications in ringing opera­ tions. George Wash Law Rev. 76:1327.
tions. Criminology. 46(1):71–98. 52 Gould-Davies N. 2018. Economic effects and political impacts:
35 Dominguez D, et al. 2020. Panama papers’ offshoring network be­ assessing western sanctions on Russia. Technical Report,
havior. Heliyon. 6(6):e04293. BOFIT Policy Brief.
36 Poon JPH, Tan GKS, Hamilton T. 2019. Social power, offshore fi­ 53 Siegel D. 2022. From oligarchs to oligarchy: the failure of us sanc­
nancial intermediaries and a network regulatory imaginary. tions on Russia and its implications for theories of informal pol­
Polit Geogr. 68:55–65. itics. World Affairs. 185(2):249–284.
37 Rilinger G. 2019. Corporate conspiracies and complex secrets: 54 Katzman K. 2022 Feb 2. Iran sanctions. Congressional Research
structure and perception of the Insull scheme in 1930s Service [accessed 2022 Dec 12]. https://sgp.fas.org/crs/mideast/
Chicago. Am J Sociol. 124(4):1043–1089. RS20871.pdf.
38 King H. 2022 Mar. China is making new billionaires much faster 55 Butler S, Neate R, Boffey D, Rushe D. 2022 Mar. Russia: Who is on
than the U.S. the sanctions lists – at a glance.
39 Harrington B. 2015. Going global: professionals and the 56 Boyd JH, Prescott EC. 1986. Financial intermediary-coalitions.
micro-foundations of institutional change. J Prof Organ. 2(2):103–121. J Econ Theory. 38(2):211–232.