The cryptocurrency moment: Return considerations for a new asset class

Cryptocurrencies, or “digital assets,” (DAs) first appeared on the radar screens of professional asset allocators about five to eight years ago, an event perhaps best marked by the launch of bitcoin futures on traditional financial exchanges in December 2017. Many analysts have spent the subsequent years asking whether these new assets have any legitimate uses and whether they should have any value at all. How can we be sure that bitcoin is not a mere “Ponzi scheme”?

At T. Rowe Price, we began asking similar questions a few years ago. With the help of an outside consultant, we mapped the range of possible future states for DAs into the scenarios shown in Figure 1.

We considered two primary dimensions: regulatory acceptance (the horizontal axis in Figure 1) and market acceptance (the vertical axis). In mid‑2021, our initial assessment was that we were in scenario 2: “Niche Use Cases.” We recognized that DA use cases were emerging but were mindful of their very limited adoption within traditional finance channels and a crypto‑skeptical, heterogeneous regulatory landscape.

During this initial phase, we believed the question of whether DAs had any merit (or value) at all was the right one to focus on.

Moving beyond niche uses

We now believe that the value question has been answered in the affirmative. In our view, DAs do have merit as an asset class, and they should—generally speaking—have a financial value that is “not zero."¹Accordingly, we now believe that we are firmly in the third scenario shown in Figure 1, “Esoteric Allocation,” and on the path to scenario 4, “Accepted Asset Class.” What has changed?

• The highly successful launch of spot bitcoin exchange-traded funds (ETFs) in early 2024 has shown that traditional investor interest in DAs is substantial. Bloomberg reported that the largest two new ETFs, issued by BlackRock and Fidelity, both topped the all‑time leaderboard of first‑month, post‑launch inflows out of 5,535 ETF launches over the past 30 years.
• After the 2024 U.S. elections, the regulatory pushback and confusion over DAs likely has peaked in the world’s largest economy. The incoming Trump administration seems poised to take a constructive stance toward the asset class, a meaningful shift from the skeptical bias that characterized the prior administration.
• Just as importantly, jurisdictions outside the U.S. for years have been more realistic about the staying power of DAs and have focused on regulatory initiatives to bring them within the scope of traditional financial regulation while permitting their continued growth and evolution.
• The DA industry not only survived the collapse of FTX Trading Ltd., but has thrived in its shadow. The criminal prosecutions stemming from FTX’s demise revealed the fundamental distinction between fraud, as such, on one hand, and DAs as a technological and economic innovation on the other. Human corruption can quickly reduce the value of anything to zero, and our existing legal frameworks likely are well equipped to deal with such events even if they involve new technologies.
• Specific to the financial industry, important new use cases are emerging for DAs to support traditional transaction processes. There are numerous ambitious pilot projects seeking to “tokenize” intra‑industry flows and settlement processes, and—not to be outdone—even central banks are evaluating the merits of so‑called central bank digital currencies.

Based on these and other considerations, many investors now believe that DAs have a role to play in our economy and that their fair value indeed is “greater than zero.” This means that different considerations are now rising to the forefront when contemplating what to do with this new asset class.

Toward a return framework

This paper focuses on what we believe could be the most important question: Should passive holders of digital assets expect a level of investment return that would merit their inclusion in diversified portfolios? We will focus here on framing the key issues and will only hint at possible ways of answering this question. We hope to flesh out potential concrete return frameworks in a future paper.

"Should passive holders of digital assets expect a level of investment return that would merit their inclusion in diversified portfolios?"

We start by analogizing DAs to traditional asset classes (stocks, bonds, and gold). Armed with that framework, we examine the return potential for DAs, focusing on bitcoin. We argue that traditional sources of long‑term return generation are not present, creating a need for alternative frameworks specific to the economics (or “tokenomics”) of how DAs operate as a protocol.

Finally, we show how such a localized analysis potentially can establish a rationale for attractive long‑term returns. We conclude the paper with a cursory discussion of other considerations that must go into an investment case for DAs.

Returns from traditional assets

To merit inclusion in a traditional multi‑asset portfolio, an asset not only needs to have a positive fair value but also an expectation of a positive future return. This paper, then, will consider the return requirements for a long‑term, strategic allocation to DAs.² Specifically, we will focus on the excess return above cash, which is the measure of the reward for taking on investment risk rather than allocating to cash as the risk‑free asset.

Further, we focus on the so‑called arithmetic or per‑period expected return. This is distinguished from the geometric return, which represents an asset’s long‑term growth rate. The wedge between the arithmetic and the geometric expected return is created by volatility. Highly volatile assets may have a positive arithmetic expected return but a negative geometric expected return. Such assets still may be worthy of inclusion in diversified portfolios because the volatility risk is diversified away at the portfolio level, making the arithmetic expected return the relevant measure for portfolio construction.

Consider stocks and bonds as the obvious examples of this long‑term focus. We have many decades of historical returns showing that both asset classes have exceeded cash returns over time, as well as a solid theoretical foundation for why their returns should exceed cash over time. Both asset classes carry a risk premium (the equity risk premium for stocks and the duration premium for bonds), whereby the long‑term holder is rewarded with excess returns for taking on a financial risk that the security issuer wishes to offload. As a result, most portfolios today rely overwhelmingly on those two asset classes.

Now, consider gold as another example. Few will argue that the fair value of gold is zero or claim it is a Ponzi scheme. However, it is difficult to support the notion that its long‑term, structural price return should exceed cash (or inflation). Gold is a commodity; like all commodities, its long‑term prospects are a function of fundamental supply and demand. One can put together a positive or negative thesis depending on one’s view of those drivers. There is no apparent “evergreen” reason for gold to earn excess returns over time. And so, unsurprisingly, a 2022 study by the World Gold Council reported that only 7%–18% of allocators surveyed in different regions had institutional allocations to gold.³

The case for “digital gold”

Turning to DAs, we are sympathetic to the “digital gold” thesis for owning bitcoin. This theory views bitcoin and other tokens as hedges against financial collapse. The provider of one recently launched bitcoin ETF has emphasized this rationale,⁴ and it certainly can lead to a return‑based investment thesis: In an inflationary/financial repression environment, one can reasonably expect real yields and returns (i.e., after subtracting inflation) to be negative for most financial assets, including cash. Thus, any asset that merely keeps up with inflation will have a return that exceeds cash, thus satisfying the return requirement. Bitcoin might well fall into that category, thanks to its limited supply.

The challenge with focusing narrowly on this argument is twofold. First, the same thesis works for any real asset—so why not buy gold, art, or unimproved land? Many institutional investors already have these “real assets” in their portfolios, and we believe that DAs need to bring more to the table to compete. Secondly, wouldn’t it be nice to have reasons to invest in DAs that do not require investors to bet on some form of a debt‑infused financial collapse?

DA advocates also point out that the growing adoption of a digital gold thesis (or, relatedly, adopting something like bitcoin as a reserve asset) itself could lead to very attractive returns during that adoption cycle. In fact, it is easy to view the bitcoin bull market that was in progress as of late 2024 as an example of that thesis playing itself out.

We are sidestepping these arguments here solely because they do not provide a truly evergreen thesis. Both may be plausible reasons to consider bitcoin. Still, they strike us as less durable and universal than the investment rationales underpinning the traditional equity and fixed income holdings that dominate most liquid portfolios. The case for including those assets rests on their ability to add durable economic value and on their role as risk transfer vehicles with no expiration date. Returns accrue because these assets have a genuine purpose in our economy that goes beyond the objectives of their owners. This paper aims to lay the groundwork for looking at DAs through that same lens.

Diversification arguments (e.g., a possible lack of correlation between DAs and stocks and bonds) also are helpful but are insufficient to motivate a structural allocation to any asset, including DAs. A low correlation to traditional assets lowers the bar for the excess return required to justify an allocation. However, the expected return above cash still has to be positive unless the correlation is markedly negative and the asset acts as an outright hedge of those traditional assets.⁵

Consider that by randomly selecting offsetting long and short positions in individual stocks, one can easily create a portfolio that is uncorrelated to equity markets and that will slightly trail cash due to implementation costs. Who would want to buy that?⁶

Taking the next step

In sum, DAs like bitcoin and Ethereum have arrived at the “gold” stage. We believe they have legitimate value, and the next step in their evolution as an investable asset class will be the formation of a long‑term return thesis that can justify a structural allocation in diversified, multi‑asset portfolios. The remainder of this paper focuses on that need.

We begin by drawing on the equity analogy again to unpack how returns are generated. Imagine a company that makes and sells chairs. The key constituents in this economic process are represented in Figure 2.

• The company is the physical entity that combines labor, materials, and managerial skill to make and sell chairs. The shareholders are the passive owners of this joint‑stock company. They are not involved in its operations, but they absorb all of the economic risk (including the possibility of bankruptcy and losing their entire investment) in return for receiving any generated profits.
• The customers buy and use chairs. In this example, we assume that the chairs sell for USD 100 (revenue), comprising USD 85 in costs (labor and materials) and USD 15 in profit. The typical customer would be willing to pay up to USD 130 per chair, so the real value added is USD 130, even though the price is only USD 100. This gain—the ability to obtain something for the price of USD 100 that the customer individually values at USD 130—is the added value that flows to the customer.

Now, let us investigate how the USD 15 profit accrues to the passive shareholders. There are two ways:

1. The company distributes the profit in the form of dividends.
2. The company retains the profit. All else being equal (i.e., assuming no change in how the stock market values the business), the share price should increase by that same amount.⁷

For most companies, the second avenue—sustained earnings growth—is the main mechanism by which long‑term shareholders are rewarded for taking equity risk. However, note that the market only values undistributed profit growth because, at some point, there will have to be a distribution to shareholders (even if it’s just in the “terminal value” of the company). The profits cannot legally go anywhere but to shareholders, and the only question is when.

Stocks and bonds are both bundles of legal (contractual) rights that entitle the owners to specific future cash flows from the issuers. These rights are the conduits through which the equity risk premium is harvested at the individual company level. The managers or founders of the company want to offload the economic risk of profit fluctuations and are willing to part with the profits in return. The shareholders are willing to take the deal because they believe in the business’s future profitability and because there is a legal mechanism that secures their rights to those profits.

Given the economic risk, the stock should trade at a price just low enough to attract enough stockholders to absorb all outstanding shares.

Why digital assets are different

Now, let us apply this same framework to a generic DA. Here, too, there are three key constituencies, as shown in Figure 3.

The consensus workers that operate the blockchain’s consensus mechanism play the same role as the company in the equity example above. This is where the “work” happens of adding validated new transaction blocks to the blockchain in a decentralized and secure manner. The ability to organize value‑additive work in such a manner is at the heart of blockchain technology as an economic innovation. Software code and technology replace the traditional company for the limited set of services that can be “produced” using blockchain technology.

In the case of bitcoin, the consensus workers are the miners who use expensive, customized computer equipment (plus lots of electrical power) to find new blocks. Bitcoin is a “proof of work” (PoW) protocol, as physical (computational) work is required to add new blocks.

Other protocols, like Ethereum, run on a “proof of stake” (PoS) mechanism where certain participants have the right to create new blocks (and obtain associated rewards) in return for pledging large amounts of the underlying DA and exposing themselves to certain loss risks. Consensus workers are willing to do this work because the blockchain protocol rewards them with units of the associated DA for doing so.

Next, there are the users. In the DA model, these are the network participants who submit new transactions to the blockchain. In some cases, as with bitcoin, this transaction might be akin to a money or value transfer. In other cases, like Ethereum, it might be the exercise of computational work supporting an application rather than a direct value transfer. Either way, users generally need to pay fees for these transactions. These fees are accrued in the associated DA and run through the blockchain, being included as part of the transaction.

Users can specify higher fees to increase the likelihood that their transactions will make it into the next block, as consensus workers are incentivized to give higher‑fee transactions preference in assembling potential new blocks. We can think of these fees as the equivalent of the USD 100 in company revenue (costs plus profits) in the chair company example shown in Figure 2.

Transaction fees are important because they provide evidence of the added economic value created by DA networks. The fact that users are willing to pay fees for adding new transactions demonstrates that the network is fulfilling a real‑world purpose and generating real‑world utility.

Consider that over the 12 months ended in August 2024, slightly more than USD 1.3 billion in fees was expended on the bitcoin network to support a total transaction volume of USD 2.4 trillion.⁸ The International Monetary Fund (IMF) has reported that bitcoin used for cross‑border flows can amount to a sizable fraction of gross domestic product for certain developing countries with limited official capital flows.⁹

The role of passive DA holders

Passive DA holders have been notably absent so far in our discussion of a generic digital asset. Despite the potential for DA protocols to create economically meaningful value, there is no general reason (one that applies to all DAs) why passive DA holders should play any role in the system at all, let alone be compensated for holding the asset.

Recall that a stock or a bond is a bundle of rights to certain cash flows for the holder. However, while blockchain participants certainly are subject to the law of the land, the economic relationship between the different parties is defined entirely by software code, not by contractual rights.

Importantly, this does not mean that there are no rules. Instead, the rules are provided by a different socioeconomic convention, one that quite possibly is more reliable for certain applications. Still, this does mean that in many cases there is no built‑in mechanism for passive DA holders to directly participate in any “cash flows” from the value added by the protocol. And with that, many traditional forms of security analysis are off the table.

To be sure, we are not saying a rationale to expect positive returns cannot be found. We are just saying that it is not obvious that one exists simply because a digital asset has a legitimate use and a positive value. There is a real possibility that individual DAs could be highly successful in terms of adoption and value added for users without any rational expectation that passive owners of the DA will earn a positive return.

"...we believe that the digital asset space may end up looking a lot more fragmented as an asset class than the stock market."

To proceed from here, then, we need to wrestle with the tokenomics of how each individual DA is structured and then determine, on a case‑by‑case basis, whether passive holders should expect to participate in the flourishing of the network. The answer most likely will look different for different groups of DAs. For this reason, we believe that the digital asset space may end up looking a lot more fragmented as an asset class than the stock market.

Where digital asset returns might come from

What are some DA‑specific frameworks that could provide an expected return thesis? We start with bitcoin since it is the largest DA by value and has a comparatively simple protocol in which, at least facially, there is no clear role for mere passive holders. Nothing in the protocol gives any role to passive holders beyond the ability to transfer bitcoin to someone else.

Still, a possible investment thesis comes into view when we recognize that this transaction use constitutes a critical difference with gold. We showed earlier that, unlike gold, bitcoin is used for real‑world transactions that are economically significant. Bitcoin is sometimes called digital gold, but “digital money” seems more appropriate to us because monetary frameworks that are unavailable in gold analysis may lead the way toward an investment framework.

The throughput on the bitcoin blockchain is too small (and validations take too long) for small, day‑to‑day transactions. However, in return, the bitcoin blockchain is very secure—a lot of network participants work hard to get relatively few transactions done right. This makes it attractive for important but infrequent transactions. So perhaps a “digital armored truck” is the better analogy.

How would one articulate a positive expected return for such a vehicle? As a thought experiment, imagine that we froze the U.S. money supply as measured in USD while the real economy continued to grow. Assuming that aggregate transaction activity (the transfer of goods and services) remained proportional to the economy’s size, the same nominal money stock would then have to support a growing value of real‑world transactions.

The only way that could work would be if the aggregate price level trended downward—deflation, in other words. This, in turn, would give holding money a positive expected return in terms of what it could buy. USD 100 might buy 20 hamburgers today, but 21 hamburgers five years from now.

Growing with the economy: A potential source of bitcoin return

Now, consider bitcoin as having the same dynamics concerning its on‑chain transaction volumes. The protocol famously fixes the final supply of bitcoin, and it also puts governors on how many transactions can be put through the system. Block space is fixed, and new blocks arrive no faster than 10 minutes on average.¹⁰ This means that if the underlying demand for USD‑valued transactions on the bitcoin network continues to grow, the price of the good being transacted (i.e., U.S. dollars) will need to deflate relative to bitcoin as the base currency.

Put differently, the bitcoin price in USD will need to increase so that the same amount of bitcoin can cover a higher USD transaction volume. Thus, we can back into a structural expected return thesis for bitcoin if we can identify structural or secular reasons for continued, long‑term growth in the use of bitcoin for secure value transfer across the globe.¹¹

This type of return thesis is truly long term. It shows how passive bitcoin holders, like equity owners, could participate in economic growth without having to rely on supply/demand cycles or on bitcoin taking share from other transaction systems.

Potential Ethereum return sources

As to Ethereum, it is notable that its current incarnation as a proof‑of‑stake asset has two features that could provide durable sources of return.

• First, there is a yield feature stemming from regular token distributions to holders who “stake” their Ethereum to participate in the creation of new blocks. Averaging around 3.5% (as of May 2024), this yield is programmatically distributed to staking participants in return for the risk they take of losing their stake if they should, intentionally or unintentionally, violate any of the built‑in requirements for disseminating new blocks. In terms of the diagram in Figure 3, this mechanism represents an easily implementable way for passive holders of Ethereum to become consensus workers and also participate in the value added by the protocol directly.
• Secondly, the base fees associated with Ethereum transactions are “burned” (i.e., removed from circulation) rather than passed on to validators. All else being equal, this deflationary mechanism should support Ethereum prices (and returns) measured in USD. Note, however, that unlike bitcoin, the supply of Ethereum is not capped programmatically, so the price support available depends on the level of net issuance.

Both of these mechanisms are programmatic features of the protocol itself, representing a durable and structural linkage between holding the DA and either obtaining a cash flow (akin to a dividend) or benefiting from the DA equivalent of a buyback. Both are software code equivalents of the “bundle of rights” that enables investors in traditional securities to earn long‑term returns.

There are other examples of durable linkages between DA ownership and participation in the value added by the underlying protocols for different DAs. These tend to be specific to the tokenomics of the protocols involved. Accordingly, unearthing a long‑term investment thesis for owning DAs will require a form of security analysis that is inherently flexible and highly localized. We hope to lay out concrete return frameworks for various DAs in our subsequent analytical work.

Other investment considerations

Digital assets have a deserved reputation for raising complex investability and implementation considerations. While it would be out of scope here to provide a full discussion of these issues, we can briefly acknowledge the main ones and seek to put them in perspective. However, investors will need to do their own work to make themselves comfortable with this unique new asset class.

DA investability questions are captured by the letters “E” (for environmental) and “S” (for social impact) in the acronym ESG (environmental, social, and governance).

The energy demands of bitcoin mining

Starting with the environment, many potential investors are taken aback by the perceived large energy usage of the mining process that sustains bitcoin as a PoW protocol. Indeed, bitcoin miners currently consume about 100–200 terawatt hours of energy, annualized,¹² representing about 0.1%–0.2% of global energy consumption. However, it is important to realize that this issue is specific to PoW protocols and does not apply to the growing range of DAs, such as Ethereum, that are governed by alternative consensus mechanisms like PoS.

0.1%–0.2%
Share of global energy consumption used for bitcoin mining.

Staying with bitcoin, one can rationalize its energy consumption on two levels. At the highest level, we can ask if the power usage (and its likely environmental impact) can be justified given the economic value that the protocol provides. To many skeptics, the energy consumption in the mining process seems entirely wasteful since mining computers use that energy to solve what essentially are artificial  problems in order to create new blocks that, to the skeptics, don’t seem to produce anything of real‑world value.

However, this artificially “wasteful” process also underpins the key value proposition of bitcoin as a transaction ledger—its immutability, censorship resistance, and (to a degree) privacy, while at the same time being transparent and not requiring a central entity to run the system.

If it were easy (i.e., cheap in terms of energy consumption) to add new blocks, then bitcoin would not be secure. By analogy, we can justify the energy consumption of our refrigerator through its economic benefits, such as the ability to enjoy fresh food and the mitigation of food waste and food‑borne diseases.

Stepping down a level, it is important to note how energy‑efficient bitcoin mining has become. The cost of electricity is a critical consideration for mining profitability. And, unlike most other uses of electricity (like our refrigerators), bitcoin mining does not require electricity to be available in a specific location. Mining operations can locate (and/or relocate) to sinks of very cheap or even excess electricity where they do not compete with or displace other users.

One analyst has cited examples of tapping into stranded hydroelectric power or using energy produced from natural gas that is “flared off” in the extraction of other energy commodities.¹³ Bitcoin mining installations also could help balance demand on electric grids that are increasingly fed by renewable, but less stable, energy sources like wind or solar. Utilities should value having highly price‑sensitive customers who will quickly turn off operations when spot electricity prices exceed a certain level, coming back online just as quickly when prices drop.

For similar reasons, a 2023 study found that global bitcoin mining overall had a power mix that was 59.9% sustainable compared with just 48.5% for a country like Germany, which is leading the energy transition.¹⁴

In summary, to the degree that DAs have legitimate economic use value, the associated energy consumption can be defended as long as it is efficient and commensurate with the value generated. Energy usage is thus a question of degree and should not, in our minds at least, be raised as a binary pass/fail consideration for DAs.

Illicit activities

Moving on to social implications, there are legitimate concerns that the privacy and censorship‑resistant characteristics of bitcoin make it an ideal tool for illicit financial activities. However, as of 2023, blockchain analysis company Chainalysis estimated that only 0.34% of transaction activity at all covered DAs was associated with receiving addresses known to be affiliated with such activities.¹⁵ Obviously, this is only a tiny fraction of overall usage—even allowing for a generous degree of underestimation.

Chainalysis also reported that the share of bitcoin within the illicit activities pool has been declining, with so‑called stablecoins (designed to maintain a stable USD value) gaining share. This should come as no surprise: Criminals like liquidity risk no more than legitimate investors. Additionally, the publicly transparent nature of the bitcoin blockchain, a key feature of the protocol, permits sophisticated forensic analysis to uncover linkages between individual addresses and known real‑world illicit activities. Indeed, this was how Chainalysis was able to estimate the bitcoin share of illicit transactions to begin with. Many illicit users have found out the hard way that the privacy‑preserving aspects of bitcoin only go so far once law enforcement acquires these kinds of forensic tools.

It would also be unfair to DAs to not draw a relative comparison to other transaction vehicles, with cash being the most obvious.¹⁶ As with energy usage, not much would be left in our portfolios if we required all investments to be 100% unaffiliated with possible nefarious uses.

On the other hand, important social benefits from bitcoin and other DAs come into view when we consider countries or societies with a lower level of financial intermediation or less free and reliable institutions. Here, DAs have the potential to unleash legitimate economic activity that official public or private institutions are unwilling or unable to support.

We consider democratization of access to financial services in the developing world to be a noble goal, and it is no surprise that countries like Argentina, Venezuela, and Nigeria are among those where some of the most vibrant adoptions of DAs for everyday transactions have been reported.¹⁷

Portfolio implementation

Turning briefly to implementation issues, it historically has been the case that securely accessing bitcoin or Ethereum required mastery of the sophisticated cryptographic tools used to interact directly with protocol blockchains. DAs are bearer assets: Ownership amounts to knowing the private key for an address; it is not a legal claim. This has made it very challenging to integrate DAs into the traditional trading, settlement, and custody infrastructures that are relied upon, often as legal requirements, in professionally managed portfolios.

Digital asset ETFs represent one solution, by allowing DA holdings to be integrated into the traditional institutional investment workflow just like any other ETF. We have focused here on bitcoin and Ethereum in part for this reason.

Finally, DAs have unusual risk characteristics that make it challenging to integrate them into traditional portfolio construction frameworks. Due to their extreme volatility and pronounced higher moments, considerations that are normally of secondary importance (like deciding between a monthly or a quarterly rebalancing schedule) can have a primary impact on investment outcomes even for a small DA allocation.¹⁸