A new working paper from the European Central Bank argues that governance in decentralized finance is far less decentralized than the industry narrative often suggests. Examining token ownership and voting behavior across Aave, MakerDAO (now Sky), Ampleforth, and Uniswap, the paper finds that control over key protocol decisions is concentrated among a narrow set of token holders, delegates, and exchange-linked addresses. The findings raise difficult questions for both market participants and regulators about accountability, transparency, and what “decentralized” should mean in practice.
The ECB study, published as Working Paper No. 3208, analyzed two snapshots in time—November 2022 and May 2023. The four protocols were selected because they represented different segments of DeFi activity and, at the time of data collection, collectively accounted for roughly 32% of Ethereum’s total value locked. That gives the paper meaningful reach beyond a narrow case study, especially as these protocols have long been among the most influential governance systems in the sector.
Ownership Is Concentrated at the Top
The paper’s headline conclusion is stark: the top 100 holders controlled more than 80% of governance token supply across the four protocols studied. Concentration was especially pronounced in Aave and Uniswap, where the top five holders controlled nearly half of all governance tokens. Ampleforth was even more concentrated, with its top five addresses holding close to 60%.
Researchers then attempted to determine who stood behind those addresses. Their findings suggest that a significant share of governance power does not sit with a broad community of independent users. Instead, in most of the protocols examined, roughly half or more of total holdings could be traced either to the protocols themselves—through treasuries, founders, and developer allocations—or to centralized and decentralized crypto exchanges.
Among centralized platforms, Binance was identified as the largest holder across all four protocols, with holdings ranging from 2% to 15% depending on the specific token. That matters because exchange-held tokens may represent assets custodied on behalf of users, but they may also create ambiguity around who actually controls or influences governance power at critical moments.
Delegation Does Not Eliminate Concentration
The ECB paper also looked beyond raw token balances to examine who actually votes. In theory, delegation can help governance function more efficiently by allowing token holders to assign voting power to more engaged or informed participants. In practice, the study suggests that delegation can reinforce concentration rather than disperse it.
Top voters across the sample were mostly delegates—individuals or organizations that received voting power from smaller token holders. But identifying those delegates proved difficult. Researchers used public web searches, Github activity, social media, governance forums, and blockchain analytics tools from Crystal Intelligence in an effort to map addresses to real-world actors. Even so, about one-third of top voters remained unidentifiable.
Among those that could be identified, individuals accounted for about 21% of top voters, while Web3 companies made up around 19%. Venture capital firms and university blockchain societies also appeared in the data. In Uniswap’s case, the most prominent voter in both observation periods was Andreessen Horowitz (A16z). By May 2023, the firm had received delegated voting power from 125 separate addresses, making it the protocol’s top voter in the ECB’s dataset.
Power Structures Appeared Stable Over Time
One of the more important implications of the study is that governance concentration did not meaningfully improve between the two sample dates. The ECB researchers found that the basic distribution of ownership and voting influence remained largely stable from late 2022 to mid-2023. That stability cuts in two directions. On one hand, it suggests governance systems are not rapidly shifting or fragmenting. On the other, it indicates that entrenched power structures may be durable and difficult to dilute through normal market processes alone.
In other words, simply waiting for tokens to circulate more broadly may not be enough to produce genuinely decentralized governance. If insiders, treasuries, early investors, and heavily delegated entities continue to dominate voting, then the formal presence of onchain governance does not necessarily translate into broad-based control.
What Protocols Actually Vote On
The ECB paper categorized 248 governance proposals across the four protocols to understand what types of issues were actually being decided. The largest category involved risk parameters, which accounted for 28% of proposals. These included decisions on loan-to-value ratios, debt ceilings, stability fees, and emergency shutdown mechanisms—topics with direct financial consequences for protocol users and system resilience.
Asset listing proposals represented another 23% of the total. By contrast, proposals focused on governance structure itself made up only 1% of the sample. That imbalance suggests that protocol communities spend far more time adjusting economic and operational settings than revisiting the distribution of governance power or redesigning governance architecture.
For critics of DeFi governance, that finding reinforces a common concern: governance systems may be active, but they are not necessarily self-correcting. If structural questions rarely reach the proposal stage, then concentration can persist even in protocols that hold frequent votes.
Regulatory Accountability Remains Murky
From a regulatory perspective, the paper concludes that governance token holders, developers, and centralized exchanges cannot currently serve as reliable entry points for supervision. The core problem is traceability. Blockchain addresses are pseudonymous, delegation chains can be opaque, and the actual parties influencing governance decisions may not be publicly visible.
That makes it difficult for authorities to determine who should be held responsible when governance decisions affect users, counterparties, or market stability. Even in cases where wallets can be linked to organizations or known participants, the underlying legal relationships are often unclear. A governance token holder may vote directly, abstain, or delegate. An exchange may custody tokens without exercising their associated rights, or it may play a more active role. In such an environment, assigning responsibility becomes far more complex than in traditional corporate structures.
The study is especially relevant in the context of the European Union’s Markets in Crypto-Assets Regulation (MiCA), which exempts services provided in a “fully decentralized” manner. The ECB paper argues that applying this standard in real-world cases is extremely difficult. Based on the protocols examined, none came close to a robust or genuine decentralization threshold. Most retained meaningful control in the hands of insiders or other concentrated actors.
Possible Paths Forward
The authors outline several possible responses. These include mandatory disclosure of token holder affiliations, legal frameworks designed specifically for DAOs, and hybrid governance models that combine blockchain-based decision-making with more traditional legal accountability. The paper also points to the framework used by the Danish Financial Supervisory Authority as one possible starting point for evaluating whether an offering or service is genuinely decentralized.
The broader implication is not necessarily that DeFi governance is invalid, but that it may need clearer standards. If protocols seek recognition as decentralized systems, regulators may increasingly ask who holds the tokens, who receives delegated power, who can move proposals, and who benefits from entrenched influence. Without better visibility into those questions, claims of decentralization may face growing skepticism.
DeFi Governance Versus Traditional Corporate Governance
The ECB paper draws a direct comparison between DeFi governance and traditional corporate governance. In both systems, voter participation is often low and a relatively small group of active participants shapes outcomes. But traditional finance has legal duties, proxy-voting rules, stewardship codes, and established channels for liability. DeFi generally does not.
That contrast is central to the paper’s argument. Onchain voting can be transparent at the transaction level, but it does not automatically create institutional accountability. If the people behind key wallets are unknown, and if delegates can accumulate influence without corresponding legal obligations, then governance may be visible onchain while still opaque in practical terms.
For the DeFi sector, the study is likely to intensify debate over whether current governance models are a stepping stone toward decentralization or evidence that many protocols remain governed by a small inner circle. For policymakers, it offers support for a more skeptical reading of “fully decentralized” claims. And for token holders, it is a reminder that governance rights may exist formally, yet still be shaped disproportionately by a handful of actors with the scale, organization, or delegated trust to dominate outcomes.