Understanding the Paxos Gold Stablecoin

  • Analysis
  • November 11, 2019

This report on the Paxos Gold stablecoin examines how the realities of holding gold in vaulted storage (rather than making interest off fiat collateral) upend the economics of existing stablecoins. PAXG’s fee structure undermines its economic attractiveness relative to traditional vehicles for gold exposure and could pose challenges for PAXG operating needs. These observations highlight the importance of analyzing underlying business fundamentals in designing cryptoeconomic and token-based systems.

Key Takeaways

  • PAXG represents the latest, and perhaps most appealing of an extended history of efforts to create digital gold products.
  • The typical economic structures of stablecoins, where interest on deposited funds provide the operating margins for stablecoin promoters, are upended in the context of a physical commodity requiring the payment of storage fees.
  • This modified economic approach forces PAXG to recoup its operating expenses elsewhere, and this arguably undermines PAXG’s appeal as a stablecoin, a reality that recalls the importance of assessing the underlying business models of cryptoeconomic systems.
  • Creation, trading, and redemption fees imposed by PAXG are not insignificant, and materially influence the economics of trading or holding PAXG, particularly at smaller scales.

Understanding the Paxos Gold Stablecoin

Paxos’ new Paxos Gold (PAXG) gold-backed stablecoin represents the latest addition to the expanding realm of cryptocurrency stablecoins. Representing an off-chain collateralized stablecoin that is nominally independent of fiat currencies, PAXG is backed by allocated gold holdings in established gold storage facilities. As such, PAXG offers both a novel iteration on earlier attempts to create a compelling digital gold product, as well as a substantial contribution to the cryptocurrency and specifically stablecoin ecosystems through its novel commodity backed construction. PAXG also, because of the novelty of backing a collateralized stablecoin with non-fiat assets, requires unique measures in terms of construction and operational structures to make PAXG an economically sustainable enterprise for its sponsors. Those measures are worth considering, both for their own sake, and relative to how they are likely to have substantial bearing on PAXG’s prospects and likely paths of adoption. Ultimately the structures Paxos is forced to adopt are likely to represent substantial challenges in its efforts to develop a wide following.

A Unique Contribution to the Stablecoin Ecosystem

Two widely-noted trends combine to shape the context of PAXG’s emergence: a series of digitized gold products that both predate and include the emergence of cryptocurrencies, and the development of a growing number of off-chain collateralized stablecoins. The history of attempts to create digitized gold products includes products like E-Gold, launched in 1995, blockchain-based DigiX Gold, or GoldMoney, which was launched in 2001 and is currently listed on Toronto’s TSX exchange, to cite several.

One approach to digital gold is represented by Goldmoney, for instance, a TSX-listed online precious metals holding company that allows its clients to purchase allocated, custodied, and redeemable precious metals online, and to save or spend their holdings via a prepaid credit card linked directly to their holdings. Goldmoney charges 0.5% on buy and sell transactions, also charging variable storage fees that range from 0.01% to 0.018% for gold to 0.033% to 0.041% for silver, while crediting clients by using transaction fees as credits against storage fees. While an interesting approach to creating digital savings options as well as spending mechanisms, Goldmoney’s customer holdings of $1.8 billion suggest that, while it has a certain amount of traction, its features have not been found especially compelling when compared with vehicles such as traditional ETFs or directly custodied precious metals. (In comparison, GLD, the leading exchange-traded gold ETF, holds around $40 billion in customer assets.) Given the company’s somewhat lengthy history, it would appear that customer demand for gold in this form has simply never materialized.

Another well-known effort to develop a market for digital gold is Digix Gold (DGX), amongst the earliest efforts to represent gold on a blockchain. DGX sources its Singapore-vaulted gold from LBMA-approved refiners, and represents that gold on the blockchain as ERC-20 tokens. The company has relatively low 0.13% transfer fees, and charges an annualized 0.60% fee for storage fees, although those were suspended during 2018. While these fees appear relatively low as well, and in many ways the product appears appealing and accessible, Digix Gold’s current $5.1 million market valuation suggests the token has failed to capture the imagination of the investing public. Whether this is unique to DGX, and equally perhaps Goldmoney, or a reflection of a limited interest in holding gold in a digital format is unclear, but it is easy to wonder if those inclined to allocate assets to physical gold might perceive digital formats for holding gold as somewhat paradoxical?

The other trend relevant to Paxos Gold is that of off-chain collateralized stablecoins. While the stablecoin genre is described as creating a cryptocurrency exhibiting less volatility than prominent cryptocurrencies such as Bitcoin, in practice, the off-chain collateralized stablecoin genre—where assets are held on a 1-to-1 basis by some centralized custodian who promises to maintain this accurate backing and frictionlessly honor any redemption requests from the stablecoin in question back to fiat—has come to be seen as pegging the stablecoin to a fiat currency as the method of avoiding the volatility associated with cryptocurrencies. Commonly, the fiat in question has been the US Dollar, although most recently the range of asset has begun to expand to include a wide range of leading national fiat currencies, and even baskets of currencies, often using methods drawing inspiration from the IMF’s Special Drawing Rights (SDR) that is used to backstop international finance.

Insofar as PAXG lies at the intersection of these trends of digital gold and stablecoins, it can properly be understood as advancing each trend in significant ways. By creating a blockchain-tradeable representation of allocated physical gold sponsored by a highly legitimate group and with recognized legal and regulatory permissions and licenses, PAXG may be properly understood as having developed an approach whereby gold-as-money can easily interact with other financial assets. From the cryptocurrency perspective, it meaningfully expands the range of stablecoins, and creates a richer ecosystem where a sophisticated portfolio of stablecoins could relatively easily be assembled, composed, for instance, of USD, Euro, GBP, and Gold linked stablecoins and allowing a genuinely stable asset base to back a portfolio’s stablecoin element.

The Significance of Physical Assets as Stablecoin Backing

While a regulated, gold-back stablecoin is a meaningful contribution to the genre by substantially diversifying the range of assets backing individual stablecoins, backing a stablecoin with physical gold impacts the stablecoin’s construction and operations in significant respects. Most significantly, PAXG’s reserve assets do not earn interest, but rather, require fees for vaulting services to hold the assets underlying the stablecoin’s collateralization. This represents a fundamental distinction between PAXG and fiat-backed stablecoins, one that ultimately has material impacts upon the stablecoin’s construction. Such differences in construction and operation are worth understanding, as their impacts are likely to be enduring. More significantly, this illustrate a challenge stablecoin promoters will be forced to confront as a more diverse range of asset-backed stablecoins is brought to market with backing other than fiat currencies.

As opposed to a commodity-backed stablecoin, there are two points of significance in terms of backing a stablecoin with fiat assets held in custodial bank accounts. One is that the process of creating fiat-collateralized stablecoins is a straightforward, low-cost one, for which fees are generally not required. This is because the sponsors of fiat-backed stablecoin are able to absorb much of the administrative overhead associated with operating the stablecoin in question, by virtue of being able to earn interest on the fiat holdings backing the stablecoins.

The case of gold is different, and introduces substantial problems to this model. Backing a stablecoin with verifiable, allocated gold holdings, as opposed to merely tracking gold’s price, completely transforms a stablecoin’s operating economics. A critical point is that (allocated, unlent) gold holdings maintained in a vault do not earn interest in the way fiat held in a bank account does. On the contrary, gold held in a vault actually requires storage fees. Thus, the economics of an off-chain collateralized stablecoin are transformed when gold, not fiat, becomes the backing for the asset.

Gold vaulting fees at recognized holding facilities are non-trivial. Gold ETFs generally see substantial portions of their operating expenses consumed by storage fees. GLD, the largest of the gold ETFs, has a 0.40% expense ratio, including storage fees. Contrasted with a fiat backed stablecoin, which may earn as much as 2% on its fiat holdings, this almost 2.5% difference in the economics of the stablecoin demonstrates how the difference in the backing-asset is significant. This ultimately has considerable implications for how PAXG is structured and operates, and the observations here are likely to apply to other commodity-backed stablecoins that might be created in the future.

Understanding PAXG’s Fee Structure

PAXG has developed reasonable measures to address these hurdles, specifically, by creating a tiered range of creation and redemption fees, as well as assessing modest fees when PAXG tokens are moved outside of Paxos wallets and transferred on-chain. While the fees are reasonable when examined closely, as is seen below, they also introduce unique dynamics into how traders and investors might choose to use PAXG. How these structures, and the ways users might develop strategies around them, might influence the viability of operating PAXG remains to be seen.

A potential indication that Paxos itself is not entirely certain that its planned fee structures will be sufficient to fund PAXG’s operations can perhaps be identified in the PAXG whitepaper. In mentioning the storage fees that PAXG will incur for its gold holdings, Paxos has indicated that it will not charge storage fees to clients holding PAXG tokens. However, its specific language on the Paxos Gold user-guide webpage, that ‘Paxos does not currently charge gold storage fees,’ does not exactly inspire confidence that, over the longer-term, these fees will not begin to be passed through to investors, rather than absorbed by Paxos and recovered in other ways. This hedged language may also be a form of acknowledgement from the Paxos team that the fee structure it is proposing remains, in effect, untested. Should the projected income from generation, destruction, and trading fees not prove capable of supporting the underlying organizational overhead these storage fees will need to be passed through to customers, a step that would likely have significant impacts on the economics of the PAXG tokens.

In terms of specific fees, PAXG charges a tiered creation and redemption fee for each PAXG token created, as well as transaction fees for on-chain transaction of PAXG tokens that are moved outside of PAXG wallets.

PAXG Creation and Destruction Fee

Table of PAXG Creation and Destruction Fees for 0.01, 75, and 20000 PAXG

Translating these fees into dollar values and percentages of the original invested amount the fees represent, the table below illustrates the fees different investment amounts, ranging from the minimum of 0.01 PAXG token up to 15 million dollars, would pay for the creation of PAXG tokens and the subsequent transfer of those tokens to a 3rd party wallet.

Fees for PAXG Creation and Transfer to a 3rd Party Wallet

table for fees for PAXG creation and transference

While the minimal creation fee of 0.02 PAXG (translating to $30.20 when the gold spot price is around $1500, as it was in early September,) is not excessive, it nevertheless represents a substantial amount for an investor looking to purchase $250 or even $1,000 of PAG tokens. The declining scale of fees as a percentage of the original investment leaves the creation fees, and eventually destruction fees which operate on the same sliding scale as creation fees, appearing increasingly less significant on a relative basis the more the invested amount increases. Across the spectrum of the investment sums modeled in the table above, the fees merely to generate then move into one’s wallet are not trivial.

In either case, however, while these fees might not be excessive in the world of commodities trading and investments, where storage costs are understood to be a portion of the economic calculation behind a trade or investment, whether these fees will be understood or considered quite substantial In the cryptocurrency space remains to be seen. This contrast is particularly true given that when compared to a stablecoin such as USDC which has virtually no creation, redemption, or storage fees, the contrast is even more glaring. Even Tether, the most prominent, widely used, and criticized of stablecoins has only minimal fees associated with it.

A further appreciation of the impact of the fees associated with the structure of PAXG come through in the table below, which illustrates the impact of paying the PAXG creation fees, transferring the tokens to a 3rd party wallet, eventually making an on-chain transfer back to a Paxos wallet, and paying the redemption fees. The associated fees assessment is developed assuming an unchanged gold price for the period of the trade.

Purchase and Transfer to a Wallet, Return transfer and Cash-out (gold price unchanged)

table for purchase, transfer, return, and cashout fees

As is clear in the right hand column which illustrates the fees as a percentage of the original capital investment, the impacts of the PAXG fee structure is not insignificant. These amounts represent a substantial initial overhead an investor must clear before a trade turns positive.

The impact of the PAXG fee structure is even more clear when considered in the context of a rising gold price. The table below models the same fees for creation, on-chain transfer of the PAXG tokens, return transfer, then redemption. The only difference is the 10% higher gold price at the time of redemption. Note that this results in a higher set of return transfer and redemption fees.

Purchase and Transfer to a Wallet, Return transfer and Cash-out (gold price + 10%)


The divergent returns at different investment levels within a 10% rising gold price scenario are seen in the below chart. Note also that the below chart considers pretax scenarios.

Paxos Gold vs. Gold Return, Scenario of 10% Gold Price Increase

Chart comparing paxos gold versus gold returns

While PAXG may offer additional appeal for holders, and potentially have more complex economics if one considers that PAXG tokens will easily integrate into crypto lending projects, this is only simplistically considered an uncomplicated positive. As this blog post discussing divergent lending rates between different financial products usefully reminds, higher interest rates, such as those received via crypto lending projects, are rarely merely a reflection of market inefficiencies. Rather, they are probably more accurately considered the market’s collective analysis of liquidity and capital return risk inherent in any lending context. Thus, the idea that PAXG tokens can be employed as collateral for crypto lending schemes, as Paxos was quick to point out in its public unveiling of PAXG, is a more complex proposition than it might appear at a casual glance.

Undoubtedly, there will be traders who purchase PAXG as it is poised to rise, use it as collateral for some other trade that will prove successful, and will achieve outstanding compound returns. Fewer tales of triumph are likely to be heard from those who purchase PAXG prior to a decline in the gold price and proceed to leverage it as collateral for a trade that also proves unsuccessful. Compound losses of 50% will undoubtedly be as easily achievable as total returns in this context. But the larger point, beyond the ability to use PAXG as a source of leverage, is that its own fee structures and investment prospects are not especially compelling, as was seen in the bar chart above. This reality should not be downplayed with suggestions that PAXG has unique capabilities of being easily leveraged.

But regardless of how successful one may prove with prove to be with their own margin trades, it is also worth noting that, in a broader context, PAXG is unremarkable in offering exposure to allocated gold that can subsequently be leveraged. This suggests a comparison of PAXG to one of the major gold exchange traded funds can provide a number of useful observations.


In terms of easy exposure to gold, major gold ETFs such as GLD already offer straightforward, low-commission exposure to allocated gold in accessible formats. While traditional financial instruments such as ETFs cannot compete with the uninterrupted, round the clock trading of blockchain-based financial instruments, there are a number of other points that make clear the complexity of simple comparisons between these instruments. (Note also that PAXG, in a nod to its real-world foundation in physical assets, is a blockchain-based asset that can only be purchased or redeemed 23 hours a day during weekdays, with additional downtime during the weekends, in accordance with account reconciliation trading breaks and weekend closures of the London-based price setting operations of the London Bullion Market Association, gold’s dominant trading body.)

That State Street’s GLD ETF, for instance, has more than $40 billion in assets under management attests to the appeal of this product. Moreover, the simplicity and low costs of trading GLD through brokerage accounts, where even a retail investor can trade unlimited volumes for $4.95, further illustrate the appeal of this product. The low expense ratio of GLD, merely 0.40%, means that a $1 million holding would only lose $4,000 annually to expense and the low costs of this product. Perhaps most importantly for institutional investors, GLD’s daily liquidity of more than $1.5 billion reduces concerns of being unable to exit substantial positions without moving the market, a critical concern for institutional investors and a serious challenge for PAXG if it intends to build a following amongst institutional investors.

Equally significant, given that a major appeal of PAXG is presented as its ability to be leverage within cryptocurrency lending protocols, is that assets such as GLD, when held in brokerage accounts, can already be leveraged and employed as collateral for additional trades. In this regard, a comparison of lending practices in crypto and traditional finance can help to better understand this distinction. Mainstream equities accounts generally allow up to 50% of an asset’s value to be borrowed against, and professional investors often have higher limits. Margin loans too are frequently low-cost, especially for professional investors where prime brokers compete for their business. Compared to crypto lending vehicles such as SALT, these opportunities look highly competitive. Given SALT’s published restrictions on loan-to-value lending ratios, for instance, which appear to constrain lending ratios to between 30% and 70% of an asset’s value, this does not meaningfully increase the appeal of PAXG as an institutional vehicle. This is because, for instance, at an equally compelling base fee structure, and without transfer fees of the sort PAXG is proposing, margin loans of a highly competitive sort can be found using traditional investment vehicles. This would appear to represent a challenge to Paxos hopes of developing a broad and deep institutional following for PAXG.

Potential Trading Revenue to Paxos

While important aspects of a comparison of a PAXG and an existing gold ETF such as GLD appear to raise some questions about the actual prospects for PAXG to develop a large institutional following, given the apparent advantages and proven utility of existing, traditional products, it is nevertheless worth considering the scenario where PAXG develops a large trading community. The trade fee revenue that would flow to Paxos in such a case are substantial, and would be very likely to ensure the continued existence of PAXG.

One way of considering realistic potential targets for daily trading of PAXG is to examine existing volume in the leading gold ETF. GLD, for instance, trades about $1.6 billion USD per day. If the actual daily trading volume of PAXG establishes itself at a level of $16 million, representing 1% of GLD’s daily traded volume, annual revenue to PAXG would be $1.6 million. This number is unlikely to fund the operating fees of PAXG, which include whatever storage fees they are absorbing. Should the daily trading volume for PAXG establish itself closer to one third of GLD’s volume, an ambitious but potentially attainable goal should PAXG develop a meaningful following, the $528 million in daily trading volume this would represent would bring in around $38 million in trading fees. This sum would likely be sufficient to fund PAXG’s operations. The question, therefore, would appear to revolve around how quickly PAXG can develop a meaningful following and establish significant trading volume.

Annual Revenue to Paxos from Different Potential PAXG Trade Volumes

chart for the annual revenue to paxos from different potential PAXG trade volumes at $1600000000, 1056000000, 28000000, 11680000, and 16000000

A challenge, however, will be that institutional investors will also be able to analyze these structures, and may develop their own tactics around that seek to circumvent these trading fees as a way of lowering the cost basis on their own gold exposure. Will funds and institutions develop strategies where PAXG is never traded outside of Paxos, and what impact might that have? What if a fund were to move PAXG to a Stellar wallet, establish PAXG as an anchor within the Stellar ecosystem, and trade digital representations of allocated gold in Stellar’s low-cost environment, bypassing Paxos’ fees but leaving them to absorb the bullion storage costs? Other strategies fund may develop to circumvent Paxos own fee structures may be even more novel, but what is clear is that there would appear to exist a non-negligible risk of PAXG’s structures being manipulated by an element of their audience, with impacts that would be difficult to assess at this point.

In this regard, it is also worth noting that PAXG fees will not be assessed for trades on Paxos Itbit exchange. This can be interpreted in two ways, both related to Itbit’s institutionally-focused trading clientele. One objective may be an attempt to attract institutional traders to migrate at least parts of their trading to Itbit by promising reduced fees on the trading of PAXG on the exchange. The second may simply be to help ensure the loyalty of existing institutional clients of Itbit by adding additional incentives to trade on Itbit.

Prospects, Challenges, and Larger Implications of the Appearance of PAXG

Observing Paxos development and evolution promises to be telling in a variety of ways. There can be little question of the product appearing to be yet another highly legitimate and confidence inspiring contribution to the growing ecosystem of cryptocurrency fintech products, in addition to being amongst the more credible contributions to the lengthy history of attempts to create digital gold products. That PAXG’s construction, and the thinking behind it, are both top caliber speaks well of the product and gives grounds for optimism regarding its prospects. That said, it will be interesting to observe if the product moves behind niche status and manages to attract a more substantial institutional following. In this regard, while the fee structure appears reasonable given the constraints Paxos is facing, whether PAXG can develop liquid trading markets capable of luring professional investors already using GLD for highly liquid and low-cost exposure to the gold market will be challenging. If it fails to do so, whether PAXG’s fee structure will prove sufficient to support the products operation may come into question. One equally wonders if institutions may develop strategies to game the system. This could be done in a variety of ways, and may alter the product’s development, and perhaps even its survival. Given these uncertainties, and the range of implications for both PAXG itself and the cryptocurrency markets more broadly, it will be interesting to observe the fortunes of PAXG and the variety of institutional and retail dialogues that are likely to develop as observers consider its various points of significance. The outcome of these questions could potentially be far-reaching relative to many of the conversations currently underway related to efforts to tokenize a wide range of physical, real world assets as part of an effort to contribute to the growth and expansion of the cryptocurrency ecosystem.