Stablecoin Billionaires - A Descriptive Analysis of the Ethereum-based Stablecoin Ecosystem

While Ethereum, the second largest public blockchain after Bitcoin, was experiencing a story of success over the past few years, Stablecoins grew to an integral component within the Ethereum-based decentralized finance (DeFi) ecosystem. Various issuers of cryptographically-secured digital money emerged and in particular Tether, the most widespread dollar-pegged cryptocurrency, received much attention in recent time. Using empirical data broadcasted by seven different Stablecoins and their Smart Contracts, I statistically examine both Ethereum's Stablecoin landscape as a whole and the individual token ecosystems. By incorporating qualitative information obtained by analyzing the contracts' Solidity code, I discovered extreme concentrations of power within the token contracts of centralized Stablecoins, while decentralized tokens such as DAI lacked structural power imbalances. Analyzing the costs, empirical data revealed that transfers of Stablecoins that are primarily used in decentralized environments, were on average up to eight times more expensive than those carried out with `centralized' tokens, which could be traced back to an efficiency trade-off in favor of decentralization. Furthermore, I found that emission- and destruction policies have a decisive impact on basic statistical measures such as the mean transfer amount or its standard deviation, both of which depend heavily on the applied distribution mechanism of a token. In an attempt to identify `High Rollers' within the analyzed ecosystem, the exchanges Huobi and Binance were exposed as Stablecoin billionaires, who together control 20 % of Ethereum's Stablecoin supply. Finally a very unequal distribution of wealth in favour of crypto-exchanges and DeFi applications was revealed.


Introduction
Since the pseudonym Satoshi Nakamoto (2008) first described Bitcoin, more than a decade went by. As a response to the financial crisis in 2008, Bitcoin represented the first decentralized digital cryptocurrency and has experienced a story of success since its creation. Later, with the proliferation of Ethereum from 2014 onwards, the range of possibilities grew enormously with the introduction of the Ethereum Virtual Machine (EVM) and Solidity, a programming language for creating Smart Contracts. The decentralized platform allows anyone to build applications on top of it, ranging from file-storage solutions or decentralized autonomous organizations (DAOs) to tokenized assets and financial derivatives. (Buterin, 2013) Especially, decentralized finance (DeFi) received much attention in recent years and many new projects came up with approaches integrating already existing financial applications into the decentralized environment Ethereum provides. Stablecoins, defined by the Financial Stability Board (2020) as "a crypto-asset that aims to maintain a stable value..." (p. 4), have been part of the DeFi-universe from the beginning and already in 2012, Willett (2012) published the first description of a protocol enabling the creation of new cryptocurrencies on top of Bitcoin and named it Mastercoin. Two years later, the first Stablecoin named Realcoin, later renamed to Tether, was issued. (Rizzo, 2014) Since then, the market has experienced enormous growth and as of July 1 st , 2020, the five largest Stablecoins by market capitalisation on coinmarketcap.com are Tether, USD Coin, Paxos Standard, Binance USD and TrueUSD, with Tether representing the third most capitalised cryptocurrency after Bitcoin and Ether. All these tokens, together with the 6 th and 7 th largest Stablecoins HUSD and Dai, which both have a market capitalisation of more than 100 million dollars, make up the Stablecoin ecosystem examined in the following study.
Research in related fields mostly investigates Ethereum as a network via graph analysis or deals with price developments of Ether and other tokens. (Chen and Bellavitis, 2020;Baumöhl and Vyrost, 2020;Lyons and Viswanath-Natraj, 2020) This study aims at providing quantitative insights into specific activities that occur inside the individual token ecosystems and further attempts to link multi-ple accounts to real-life entities in order to identify several 'High Rollers' within Ethereum's Stablecoin landscape. The ultimate purpose of this work is to provide a descriptive snapshot of the analyzed environment as of July 1 st , 2020. The result is a broad quantitative overview of both the whole ecosystem and the individual tokens, realised through descriptive statistics and carried out without neglecting more in-depth investigations regarding the circulating supply of each token and the wealthiest entities within the analyzed market. This was accomplished by relying on visualizations and further interpretations of tables, charts and the contracts' source code. Besides the identification of Stablecoin billionaires, results show an incredible dominance of Tether, the effects of decentralized finance on certain Stablecoins, the impact of divergent issuing and burning policies on particular statistical measures and extreme wealth inequality within the analyzed ecosystem. To the best of my knowledge, no descriptive analysis of Stablecoins, which incorporates the Solidity code of specific Ethereum-based Smart Contracts to explain descriptive measures, has been carried out yet. A Github Repository 1 with the Python scripts, created to retrieve the relevant data, verify its correctness and completeness and prepare it for further processing, was published together with the complete data set 2 and all charts used in this thesis. Everything is accessible under an open-source licence to ensure that fellows can build upon it. As a consequence, the results of this thesis, from the data collection to the mining process up to the construction of the charts, can easily be reproduced.
The following work is structured as follows: The first section deals with background information on the analyzed Stablecoins, covering the organisations behind the respective cryptocurrencies and their token-emission procedures. The second part contains an in-depth elaboration of Ethereum as the fundamental base-layer for other cryptocurrencies, with emphasis on the ERC-20 Standard and the implementation of Solidity's Events. Afterwards, related works and the applied methodology are outlined in detail, followed by a description of the data set.
The main part of this work first analyzes token events, continues with descriptive statistics on accounts that own Stablecoins and finishes with the identification of Stablecoin billionaires. Finally, a conclusion is drawn. The following section provides a detailed description of the seven Stablecoins incorporated in the this paper and outlines some noteworthy insights for the main analysis. In general, there is no common definition of a Stablecoin. Bullmann et al. (2019) in an EZB Occasional Paper Series, for example, abstractly defined Stablecoins as "digital units of value that are not a form of any specific currency (or basket thereof ) but rely on a set of stabilisation tools which are supposed to minimise fluctuations of their price in such currency(ies)" (p. 9). The Ethereumbased tokens that are analyzed in this work can roughly be distinguished by their asset of collateral, their issuers and their contracts' features. Moreover, internal contract restrictions and governance mechanisms are exciting differentiation criteria that are directly reflected in the figures.

Tether USD
Tether is the most famous Stablecoin and is issued by the Tether Limited. The asset maintains its price-stability because officially every token on the blockchain is backed by reserves that are held by the issuing company behind the Stablecoin.
According to the company's website, tether.to 3 , the token's collateral does not exclusively consist of cash or traditional currency equivalents, instead, other assets like loans might be used to back the cryptocurrency. In the past, the Tether Limited and Bitfinex, a leading crypto-exchange with very close organisational relationships to Tether, faced criticism after Griffin and Shams (2018) claimed and proofed that the Stablecoin was used to manipulate Bitcoin prices during the 2017 boom. The first Tether Stablecoins were issued using the Bitcoin Omni Layer Protocol and over the years several blockchains were added. Today, Tether can be found on multiple decentralized blockchains and as depicted in Table 1, the largest share of the total Tether supply was issued on Ethereum. Among others, the cryptocurrency was created on Bitcoin, Tron, EOS and Algorand, nevertheless, with a supply of more than six billion dollars, the Ethereum-based 3 Last accessed 20 Jul 2020 Ethereum  6, 037  50  25  0.048 6, 112  Omni  1, 335  0  --1, 335  Tron  2, 882  ---2, 882  EOS  5  ---5  Algorand  4  ---4  Others  22  ---22 10, 043 50 25 0.048 10, 118 Dollar-Token USDT constitutes by far the largest share (some 60 %) of the company's crypto-assets and was first created in November 2017. In fact, Tether was mainly issued pegged to the US-Dollar, however, there are also Euro and Renminbi variants on the Ethereum blockchain and since January 2020 a gold-pegged token called Gold Tether (XAUt) 4 exists. The Tether EUR 5 token contract was created on the 1 st of January 2018, but its tokens have never been distributed, resulting in a total of only 15 token holders. 6 The Renminbi Contract 7 was created in April 2019 and is rather inactive too. 8 The following analysis will show the huge acceptance of the USDT token among the Ethereum community, henceforth focuses on the contract at address 0xdAC... 9 , which refers to the Tether USD token contract, when referring to Tether. Ethereum's Ether uses 18 decimals and many tokens follow this example, however, the USDT token has 'only' six decimals and is therefore still 10, 000 times more divisible than traditional fiat currencies.

USD Coin
The second largest Stablecoin by market capitalisation is called USD Coin (USDC) and is issued by the crypto-exchange Coinbase. Developed by CENTRE and initiated as an open-source project by the companies Circle and Coinbase, the Stablecoin maintains its price peg to the US-Dollar through fiat money securi-4 Tether Gold: XAUt Token Address: 0x4922a015c4407F87432B179bb209e125432E4a2A 5 Tether EUR Address: 0xAbdf147870235FcFC34153828c769A70B3FAe01F 6 see etherscan.io/token/0xAbdf147870235FcFC34153828c769A70B3FAe01F [Last accessed 1 Jul 2020] 7 Tether CNH Address: 0x6E109E9dD7Fa1a58BC3eff667e8e41fC3cc07AEF 8 see etherscan.io/token/0x6E109E9dD7Fa1a58BC3eff667e8e41fC3cc07AEF [Last accessed 1 Jul 2020] 9 Tether USD Address: 0xdAC17F958D2ee523a2206206994597C13D831ec7 ties stored in bank accounts.  As explained in the Whitepaper, customers must meet Know-Your-Customer (KYC) requirements before they can deposit fiat in exchange for tokens. This can be done through the official web application, which is maintained by Circle itself, or via a licensed token-issuing partner. Such partners are authorised to then interact with the CENTRE network in order to issue tokens in exchange for dollars. This means that every time someone requests USDC, the tokens are either freshly minted by an issuer or taken from its reserves. The same applies for the redeeming process: When tokens are redeemed, they get burned. In theory, this approach could decentralize the emission process to a certain degree, as it distributes the privilege to mint new tokens to multiple entities that are all part of the CENTRE network. (Centre, 2018) To which extend differences between a centralized issuing policy and a more decentralized one can be observed in the figures, is examined in the main part of this thesis. The USDC token contract can be found at address 0xA0b... 10 and was published to the blockchain on the 3 rd of August 2018. Just like the Tether token, USDC has six decimal places and as of July 1 st , 2020, the Stablecoin has a market capitalisation of around 950 million dollars. 11

Paxos Standard, Binance USD & HUSD
Paxos Trust Company is a fintech founded in 2012 that distributes its own fiat- and FDIC-insured banks were selected to hold the fiat funds that are deposited by customers. (Cascarilla, 2018) The tokens represent the 3 rd (PAX), 4 th (BUSD) and 7 th (HUSD) largest Stablecoins by market capitalisation on Ethereum and their contracts can be found at the addresses 0x8E8... 14 for Paxos, 0x4Fa... 15 for Binance USD and 0xdF5... 16 for HUSD. While PAX and BUSD both have 18 decimals, HUSD uses eight decimal places. Finally, the token contract of Paxos Standard was published on 8 September 2018, HUSD on the 18 th of July 2019 and the contract of Binance USD on September 5, 2019.

TrueUSD
TrustToken, the company behind the 5 th largest Ethereum-based Stablecoin on CoinMarketCap, published its token contract already in 2018. TrueUSD is a Stablecoin that is collateralized with US-Dollars and as stated on the company's website, only customers who passed KYC procedures are able to purchase and redeem tokens. Every TUSD is freshly minted when fiat money is deposited into an escrow account and vice versa, KYC-approved customers can redeem their tokens for dollars. (TrustToken, 2020) Besides the dollar version, the company also issued Ethereum-based Stablecoins backed by reserves of British Pound (TrueGBP), Australian Dollar (TrueAUD), Canadian Dollar (TrueCAD) and Honk Kong Dollar (TrueHKD), but similar to Tether, only the US-Dollar-pegged tokens are widespread and frequently used. 17 The token contract of TrueUSD was originally deployed on 5 March 2018 at the address 0x8dd... 18 and through the implementation of a call-forwarding mechanism into the token contract, which allows requests to be forwarded to another contract, the Stablecoin is able to perform upgrades. This feature was activated on the 4 th of January 2019, when the token contract officially began delegating to the address 0x000. then, token transfers are logged by the new contract, which is important to be considered because for this reason a complete picture of the TUSD transfers and its total supply can only be reconstructed by fetching the logs of both contracts. 20 In the following work, the focus is on the US-Dollar version of TrustToken's Stablecoins because the others count significantly low numbers in transfers and token holders. Lastly, the token has 18 decimals and as of July 1 st , 2020, TrueUSD represents the 5 th largest Stablecoin on Ethereum with a market capitalisation of around 140 million dollar. Furthermore, the MKR token holders can adjust the DSR to regulate the demand of DAI tokens in order to stimulate its price and keep it pegged to one dollar. For example, in times of high demand they might decrease the DSR to lower the demand. The above mentioned Stability Fee finances the DSR and is paid by those who unlock their collateral for repaying DAI loans. The amount of the fee is determined by the MKR holders and again provides them a tool to regulate the price of DAI through an adjustment of the fee, which consequently increases/decreases the total demand. Technically, every token transferred into the DSR contract gets afterwards burned by being sent to address(0) 24 , which represents the address that is typically used to send destroyed tokens to. The same happens to DAI tokens that are paid back in loan settlements, constituting 22 see docs.makerdao.com/ 23 see blog.makerdao.com/say-goodbye-to-cdps-and-hello-to-maker-vaults/ 24 Address: 0x0000000000000000000000000000000000000000

Ethereum
In order to conduct a deeper analysis of the discussed Stablecoins, it is essential to be familiar with the basics of their common underlying platform, namely Ethereum, the largest decentralized blockchain, which serves decentralized applications (DApps). Ethereum is an open-source project and was first described by Buterin (2013 (0), representing "0x0" in hexadecimal format. A message call transaction, is used to simply transfer Ether to another account or to interact with a contract. The latter is done by passing additional information in binary format to the contract, alongside the transaction. As defined in Ethereum's Yellowpaper, a transaction contains a data field, which can hold an unlimited size byte array and is used to pass further commands to the recipient of the transaction. Two out of many possible instructions could be a token transfer and the creation of tokens. As only EOAs are capable to carry out transactions, contracts interact with each other through message calls, which enable them to send Ether to an EOA or to execute another contract's function. (Wood, 2014) The discussed categorisation and definition of a transaction is crucial to be aware of when it comes to analyzing ERC-20 tokens. All the analyzed Stablecoins represent Smart Contracts on Ethereum and therefore require to understand the differences between EOAs and Contracts, but also the relationship between message calls and transactions.

ERC-20 Standard
Contracts on Ethereum may be used for various applications, but especially Initial Coin Offerings (ICOs) received much attention in recent years. ICOs allow 27 see also medium.com/authereum/bytecode-and-init-code-and-runtime-code-oh-my-7bcd89065904 for further details about contract creations and the EVM Code [Last accessed 15 Jul 2020] companies to finance themselves through the sale of (pre-mined) tokens in exchange for fiat money or other cryptocurrencies such as Bitcoin or Ether. In order to standardize the interface for the rapidly increasing number of tokens that were created on the Ethereum blockchain, Vogelsteller and Buterin (2015) proposed the ERC-20

Methods
As described in Table 3, each function has its own Method ID, which consists of the first four bytes of the Keccak-256 hash of the Signature, which in turn is defined as "the canonical expression of the basic prototype without data location specifier." (Read the Docs, 2020) This can be clarified with the following example: The transfer function takes two parameters, an address for the recipient, and an uint256, which is an unsigned 256-bit integer, for the amount to send. Consequently, the Method ID of the transfer function equals the first four bytes of: Keccak-256( transfer(address,uint256) ) = a9059cbb2ab09eb219583f4a59a5d0623ade346d962bcd4e46b11da047c9049b The Method ID specifies the function a contract should execute. This is done by sending a transaction to the contract with the Method ID included in the first four bytes of the data field. The address and the transfer amount are integrated in the data field too. An example of a data field in a transaction that triggers a token transfer might look like the following: The first four bytes represent the Method ID, followed by 32 bytes including the recipient-address and finally, 32 bytes containing the amount of tokens in hexadecimal format end the data payload. For a token with zero decimals the above data field would trigger a transfer of 16 tokens to the address 0xabc...01 . 28

Events
According to the ERC-20 Standard by Vogelsteller and Buterin (2015) to represent the address that sends and the one that receives tokens. The data field is not indexed and for transfer events it usually includes the amount of tokens to transfer in hexadecimal format and padded to 32 bytes too. 30 Summarizing, a typical transfer event might take the following form:

Transactions vs. Transfers
Analyzing ERC-20 tokens, it is important to carefully distinguish between transactions and transfers.  Figure 1, on the left side, the token balance of address 0xA... decreased by ten while the balance of address 0xB... increased by the same amount in state 1 (s1), compared to the address' balance in s0.
Notably, not all transfer events are triggered by transactions directly to a token contract: On the right side, Figure 1 illustrates a scenario in which a Multisig-  They elaborated why TrueUSD, Paxos Standard, and Gemini Dollar were the best 'safe haven' candidates, resulting from a negative dependence on non-stable coins.

Related Works
Recently, Lyons and Viswanath-Natraj (2020) analyzed the stability mechanisms behind the largest Stablecoins as of March 2020 and integrated the tokens USDT, USDC, PAX, BUSD, TUSD and DAI into their study. They pointed out how arbitrage possibilities contribute to the price-stabilization of USDT tokens, while the emission-policy only played a secondary role. In addition, they were able to confirm the safe haven hypothesis by analyzing the Covid-19 crypto-downturn.
In contrast to existing work, the following analysis will provide basic statistical measures to describe the Stablecoin landscape and integrates additional information gained by examining the contracts' Solidity code to explain the empirical data conclusively. This interdisciplinary approach allows the study of quantitative data through the interpretation of code to obtain a holistic picture of the whole ecosystem. While price data is completely ignored, qualitative information is included in an attempt to identify specific powerful accounts that have been granted certain rights within a token ecosystem. Additionally, the applied methodology pushes the boundaries of existing research because it enables not only a comparison of the largest Stablecoins by numerical measures, but also by qualitative ones and enables to characterise token contracts based on how they implement and restrict certain functions and events within their code.

Descriptive Analysis
As outlined in the chapter Related Works, the most common approach to analyse blockchain data is taking a network-theory perspective to examine an ecosystem consisting of nodes and edges. In addition, a quantitative-descriptive approach can be applied by analyzing the data with statistical measures to provide a static snapshot of the token landscape. Since the data of many famous blockchains like Bitcoin or Ethereum is publicly available and accessible without any restrictions,

Data Collection
The data used for the quantitative part of this work was retrieved using the public API of Etherscan.io 37 , which is an alternative to setting up a full-node or using an Infura node 38 and allows to filter a defined block range for addresses and indexed topics. Hence, as a first step, the Solidity code of each contract was reviewed to determine the functions that lead to event emissions. following attributes were collected: Block timestamp , block number , transaction index , transaction hash , transf er f rom , transf er to , transf er amount , gas price and gas used .
The same fields apply to all the events that influence the supply of tokens, except that transf er f rom and transf er to turn to transf er address , representing the account where tokens were burned from or issued to. In total, the analysis is based  The x-axis represents time and respectively ranges from the first day a Transfer Event occurred until June 30 th , 2020. The y-axis shows the transfers/day.
According to an analysis of TokenAnalyst (2019) T r a n s f e r s USDT Others Focusing on Paxos Standard and its representative chart, an increase in transfers has been recorded from January 2020 onwards. According to recent research by Le Calvez (2018) which sent and received about five to seven times more DAI transfers on these two days than usual. As the latter discussed contracts are decentralized exchanges, the peak in DAI transfers that occurred during the Covid-19 crash in the first half of March may have caused an increased demand for Stablecoins, which supports the assumption of Allaire (2020) and the Safe Haven hypothesis by Baumöhl and Vyrost (2020). Nevertheless, as not only the withdrawals from the DSR contract increased, but even more the deposits, it cannot be concluded that there was an increased demand for liquidity.
Analyzing the amount of transferred tokens (see Table 5

Circulating Supply
The following section is divided into two parts, beginning with an analysis of the circulating supply of both the entire Stablecoin ecosystem and the individual tokens and continuing with an in-depth study of the respective token contracts and their supply-regulating functions and policies, which were elaborated and interpreted on the basis of the collected data and the contracts' Solidity code.
To begin with, the Ethereum-based Stablecoin ecosystem, represented by the tokens analyzed, has a market capitalisation of 7, 800, 726, 550 dollars. The circulating supplies of the individual tokens were plotted against time in Figure 5 and clearly demonstrate the leading position of Tether. Focusing on the leftside chart, which compares Tether to the other Stablecoins analyzed, it can be seen that Tether's circulating supply is more than twice as large as the one of the other tokens. By July 1 st , the total supply of USDT amounts to 6, 037, 847, 551 to-   Figure 5, the creation of these tokens happened earlier in April '19.
Since the beginning of July '19, Tether's dominance has never dropped below 60 % again and from July '19 until July '20, the market share of Tether grew by almost 20 %, as depicted in Figure 6.  tion of tokens. Every contract is free to choose the functions that influence its supply and as a consequence, the data set analyzed comprises seven different Stablecoins with six distinct Mint and Burn methods. Only BUSD and PAX follow the same approach, as both broadcast SupplyIncreased and SupplyDecreased events. In the following part, all these supply-controlling functions and their events were treated equally, regardless of how they were named and therefore each time the supply increases/decreases, the terms mint and burn were used as synonyms for issue and destroy. This terminology was even applied for DAI Stablecoin, although the contract does not emit any specific event to track its supply.
However, by tracking the transfers from and to address(0), the circulating supply of DAI tokens could be fully reconstructed.
Analysing the entire data set, a total of 352, 065 mint and burn events were identified and in contrast to the transfer events, Tether is far behind regarding the number of emitted supply-controlling events, which is demonstrated in   Figure 7 illustrates the cumulated token mints per day and shows that most of the Tether tokens were issued around April '20. In figures, 61.9 % of the total Tether supply was created in the first half of 2020 and as depicted in Figure 7, the three largest emissions each covered 300 million tokens, two of 60 Bitfinex: MultiSig 2 Address: 0xC6CDE7C39eB2f0F0095F41570af89eFC2C1Ea828 61 Bitfinex: Deployer 5 Address: 0x36928500Bc1dCd7af6a2B4008875CC336b927D57 which occurred in the first half this year. Shifting the focus to the burned tokens, further analysis reveals that a total of 2, 270, 455 USDT tokens were burned from 32 different accounts. This means that only 0.038 % of the issued tokens got destroyed afterwards, which constitutes an outlier compared to the other tokens. It also emphasises the low usage of supply-controlling functions such as redeem and destroyBlackFunds and underlines the lack of automated procedures that could allow users to redeem tokens for dollars independently. Figure 9 visualizes the few burn events of the Tether USD contract and shows that the largest token-burn took place on 26 September 2019, when a total of 884, 664 USDT were destroyed from address(0). This is possible because the contract code of Tether enables the owner of the contract, which is the Bitfinex Multisig account, to burn tokens from every address, as long as it is blacklisted, which is again a decision determined by the contract's owner. Blacklisted accounts are then no longer able to execute the contract's functions. Users of the USDT token are unable to mint, burn or redeem tokens on their own and thus cannot influence the circulating supply of the asset. By code, only the owner of the Tether USD contract is allowed to execute supply-controlling functions such as mint, burn and redeem, which results in a highly centralized distribution of responsibility and power within the Tether USD ecosystem. Apart the above functions, the owner is exclusively authorised to introduce fees to the Tether contract. The Bitfinex Multisig account could do so by executing the setParams function and providing two unsigned integers as input to determine both the fees as a percentage of the transfer amount and the maximum fee, while the total charge cannot be increased above 0.002 % of the transfer value and also the maximum transfer fee is hard-coded to never exceed 50 USDT. Furthermore, it is incumbent upon the owner to pause and unpause the contract in order to initiate emergency shutdowns in which every balance is immediately frozen and transfers cannot be further processed. Finally, if the owner calls the deprecate function, he will do so by providing an address as input, whereupon the original contract at address 0xdAC... is replaced by forwarding each incoming call to the new address specified by the owner. This shows that although the Tether contract was created at a very early stage, it does not lack of opportunities to introduce fees or even upgrade itself.  Very different figures can be observed for DAI Stablecoin, which is the youngest contract analysed and at the same time represents the one with the most fluctuations in its circulating supply. In total, the DAI token contract counts 318, 087 transfers from or to address (0) to work fully decentralized, which means that everyone who locks an approved collateral in the contract will receive DAI tokens in exchange. Consequently, the circulating supply of DAI is entirely in the hands of its users and no central entity determines whether or not the supply shall be increased. As a corollary, the DAI ecosystem cannot be split into a primary and secondary market, resulting in much lower average mints and burns as shown in Table 6. Finally, adding a temporal dimension as done in Figure 7 and 9, a peak during the already dis-   and consequently represents the most powerful account within the USDC ecosystem. Finally, the token contract is upgradable as it features a proxy contract solution, which means that the contract at address 0xA0b... points to another account that is located at address 0x088... 75 and provides the actual functions the token contract needs. By changing the target address within the proxy contract, upgrades could be implemented, but until July 1 st , the contract has never been upgraded and basically only the account privileged with the Admin role inside the proxy contract would be able to do so. By July '20, the EOA acting as the Admin can be found at address 0x807... 76 Continuing with the token contract of TrueUSD, the figures shown in Table 6 are quite similar to the ones seen at USDC. Nevertheless, the emission-and distribution policy appears to be more decentralized compared to USD Coin: burned from address 0x519..., which has also already been used for emissions.
The examination of the Solidity code of both tokens leads to the conclusion that their basic features are exactly the same: Both contracts do not provide any possibility for users to mine and burn tokens themselves independently, however, each of them has a SupplyController, which is a role capable of creating new tokens and destroying its own. Additionally, the SupplyController is responsible for the further distribution of the minted tokens since every created token is directly issued to the account behind this role, dividing the ecosystem into a primary and secondary market. As of July 1 st , this role is granted to the account at address 0x519... for both Paxos and Binance USD, which implies that a single account controls the entire supply of two individual Stablecoins. In addition, within both contracts the AssetProtectionRole can be assigned to an account, empowering it to freeze and burn third party funds. For Paxos, the account behind the As-setProtectionRole can be found at address 0x87f... 82 and for BUSD, it is the account at address 0xb87... 83 The latter account simultaneously acts as the owner of both contracts and is thus responsible for the allocation of the discussed roles. This indicates that both Stablecoins, Paxos Standard and Binance USD, follow a highly centralized approach, firstly because users cannot directly interact with the contract to mint and burn tokens and secondly, the possibility for specific accounts to burn third party tokens implies a massive power imbalance between participants inside the ecosystem. In addition, both Stablecoins share a very close organisational relationship, as often the same accounts have access to fundamental functions within both contracts. Finally to complete, both tokens were implemented via proxies similar to USD Coin and TrueUSD and are therefore upgradeable.
The last Stablecoin analyzed, HUSD, counts a total of 1, 540, 511, 785 tokens that were issued to 129 different addresses. On the burn-side, more than 92.5 % of the minted tokens have already been burned again, representing the highest burn/mint ratio among all token contracts analyzed. In total, 1, 425, 973, 065 tokens were burned from 75 accounts. Furthermore, the HUSD contract counts   Table 7: Supply-Controlling Roles Itemization of the supply-controlling roles within the token ecosystems and each contract's upgradability; each role was implemented through Solidity's function modifiers 87

Accounts and Balances
The following section focuses on specific identities within both the Ethereum-

Stablecoin Ecosystem
To begin with, the whole data set includes 8, 372, 918 different accounts and as of July    Table 8, which states that the wealthiest 1 % posses almost 96 % of the total Stablecoin supply, while the richest 0.01 %, represented by 184 accounts, own more Stablecoins than the rest, 1, 841, 983 different accounts in total. In summary, the contrast between accounts with balances even smaller than the average transaction fee and those that are millionaires contributed to a strongly positive-skewed distribution of wealth, while the large number of possible decimal places further reinforced the skewness. Finally, the largest balance in the data set (some 400 Mio.) is about 4 × 10 26 times larger than the smallest one (1/10 18 ).
Remarkably, the whole data contains only eleven accounts that own all seven Stablecoins analyzed. Their addresses are listed in Appendix 4 and the labeled ones are all decentralized exchanges, namely EtherDelta, Bitcratic and finally dForce, which represents a DEX that offers Stablecoin-asset swaps. 88 Merely 1.33 % (19, 703) of the accounts with a positive USDT balance own at least one other Stablecoin. This figure is very different to those observable for USDC, PAX and DAI where 8.85 %, 9.44 % and 9.89 % of the accounts posses at least two different Stablecoins. For TUSD, this percentage is even higher at 22.73 % and only surpassed by HUSD and BUSD, whose tokens are held together with other Stablecoins in 33.66 % and 54.92 % of the cases.  Table 9: Accounts Analysis Descriptive Statistics of the largest Ethereum-based Stablecoins and their token holders as of July 1 st . The columns on the left side represent the number of unique addresses while on the right-side basic statistics are calculated to describe the distribution of the balances. The column '> 10 $ Addr.' shows the number of accounts with a balance larger than 10 dollar. *in thousands **in millions Focusing on the individual tokens that have been statistically examined in Table 9, Tether is by far the most widespread Stablecoin. Four out of five positive accounts in the data set own at least 0.000001 USDT, which stresses the already highlighted broad acceptance of Tether within the Ethereum-based Stablecoin market. In total, 721, 859 addresses posses more than 10 dollars in USDT, and the wealthiest account, which is deeply analyzed in the following subsection, owns 400 million USDT tokens. In contrast, the DAI token shows unusually small numbers comparing the mean balances and the standard deviations. This is due to the distinctive issuing policy of the DAI Stablecoin, which does not require a centralized distribution of tokens to specific accounts such as centralized exchanges, who then spread them.

Entity Identification
As highlighted above, the Stablecoin market contains several outlier accounts that are responsible for heavily right-skewed distributions of wealth within both the whole Stablecoin landscape and the individual token ecosystems. Figure 10 shows an extract of 26, 332 accounts and displays information about the activity of each account and its balance. This was accomplished by drawing the number of transfers an address received against the number of transfers initiated by the same address. Finally the dot-size is related to each account's balance. It can be seen that the wealthiest accounts, which are the labeled ones, are spread over both axis, which indicates that high-balance accounts might be used for distinct purposes and as a consequence, either a very high to moderate number of in-and outgoing transfers is observable, or the opposite occurs. Furthermore, without conducting a correlation analysis, a linear relationship between the number of inand outgoing transfers can be assumed.
The account with the largest share of Stablecoins can be found at address 0x7b8... 89 and has a balance of around 400 million dollars. Although this account is not 89 Huobi Address: 0x7b8c69a0f660cD43ef67948976daaE77BC6A019B 1 10 100 1,000 10,000 100,000 1,000,000 T r a n sf e r s out publicly labeled on Etherscan, the transfer history suggests that it belongs to the cryptocurrency exchange Huobi and acts as a Tether treasury for the company: Between the 15 th and 20 th of May 2020, a total of four 'tranches' with respectively 100 million USDT tokens were transmitted from the address 0x674... 90 , which publicly belongs to Huobi, to 0x7b8... and each tranche consisted of five transfers: First, a test transfer of one USDT, followed by 24, 999, 999 tokens and finally 75 million tokens split into three transfers. This whole procedure is further described in Appendix 5, in which the transaction hashes of the transfers in discussion are listed. As of July 1 st , address 0x7b8... received a total of 26 transfers and once sent one USDT token to itself, which represents the only token transfer from this account. In addition to the above investigation, the Tether Rich List 91 , which is officially maintained by Tether Limited, confirms that the exchange Huobi is the entity behind the account in discussion, which holds the largest share (some 6.6 %) of the entire Tether supply on Ethereum, as illus-90 Huobi 2 Address: 0x6748F50f686bfbcA6Fe8ad62b22228b87F31ff2b 91 see wallet.tether.to/richlist [Last accessed 19 Jul 2020] trated in Figure 11. Moreover, the addresses 0x674... and 0xfdb... 92 , which are labeled with Huobi 2 and Huobi 3, represent the 4 th and 5 th wealthiest identities within the Stablecoin ecosystem and the 3 rd and 4 th largest token holders amidst the USDT community. Finally, the account with the 6 th largest Stablecoin balance, located at address 0x106... 93 and labeled as Huobi 9, controls 110, 481, 722 HUSD and 1, 845, 258 USDT tokens, which equals around 96.5 % of the entire HUSD supply. The latter mentioned accounts at the addresses 0x674..., 0xfdb..., 0x106... count about one million in-and outgoing transfers and are in 2 nd , 3 rd and 4 th position in this category, which encourages the fact that Huobi plays a major role in the Ethereum-based Stablecoin ecosystem.
The cumulative balance of all accounts that were identified belonging to Huobi (see Appendix 6) accounts for 11.5 % of the total Stablecoin supply in the ana-  This account owns all analyzed Stablecoins except HUSD and DAI and as illustrated in Table 10, USDT represents the largest fraction of the address' balance, followed by more than 22.5 million BUSD tokens, constituting 12.9 % of the total BUSD supply. Additionally, as depicted in Figure 10,  The 3 rd wealthiest account with the address 0x39A... 102 refers to the label of USDC tokens and holds the largest share of these tokens within the entire ecosystem. In total, 0x39A... controls 20.7 % of the total USDC supply, which is depicted in Figure 11. Notably, Compound represents the only entity among the richest accounts identified in Table 10, which is not an exchange. Additionally, the number of transfers distinguishes the address in discussion from the largest two accounts at the addresses 0x7b8... and 0xBE0...: Compound's account at address 0x39A... counts a total of 88, 470 transfers, which is illustrated in Figure 10, and hence has been much more active than the wealthiest accounts of

Conclusion
The aim of this thesis was to provide a statistical description of the largest Ultimately, blockchain data is always complete and has no measurement errors, providing excellent conditions to fruitfully apply data science.

TUSD Upgrade Transactions
The following table lists the transaction hashes of the transactions that included the command to upgrade the TrueUSD proxy contract, however, the upgrade from the T rueU SD legacy contract to the proxy contract was carried out with a delegated call and does not represent a upgrade within the proxy contract. As a result, the code of the TrueUSD token is spread over 3 accounts. The first upgrade of the proxy contract took place on 3 January 2019 and the last one on 17 September 2020.