Token Supply Regimes & Market Microstructure

Fees, MEV, Issuance… and the Risk Premium of a PoS Token

Sep 09, 2025

In Proof-of-Stake (PoS), the mix of cash-flows—burned base fees, validator tips & MEV, and net issuance—determines who gets paid (stakers vs. non-stakers), how volatile those payments are, and how much risk compensation the token demands. We:

write an accounting identity that maps fee/MEV/issuance policies into per-token cash-flows for stakers and for passive holders;
show the microstructure channels (EIP-1559, PBS/MEV auctions, MEV-burn) that make those cash-flows smoother or spikier;
propose testable predictions and a simple empirical design connecting supply mix to excess returns—in line with recent evidence that staking ratios and “crypto carry” predict returns.

Setup: who earns what under PoS?

Let \(S_t\) be circulating supply and \(P_t\) the token price. Over a small interval \(\Delta t\), the per-token cash-flow decomposition is:

\( \underbrace{\tfrac{\Delta S_t}{S_t}}_{\text{net supply change}} = \underbrace{i_t}_{\text{gross issuance}} - \underbrace{b_t}_{\text{burn (EIP-1559 base fee)}} - \underbrace{s_t}_{\text{slashing/burn-like}} \)

Validator revenue per token splits into (i) block issuance; (ii) tips; (iii) MEV rebates; while base fees are burned by design in EIP-1559. Burning removes supply (a buyback-like benefit to all holders) and—crucially—hardens the fee mechanism against collusion between block producers and users. Tips & MEV instead accrue to proposers/builders and are volatile.

Define per-token flows (normalized by supply \(S_t\)):

\(Y^{\text{stk}}_t\): staking carry (issuance + tips + MEV share paid to validators) per staked token;

\(Y^{\text{burn}}_t \equiv b_t\): buyback yield to all holders via burn;

\(Y^{\text{dil}}_t \equiv i_t\): dilution (hurts passive holders; neutral for stakers if exactly offset by rewards).

For a non-staker, the expected excess return (risk premium) over \(\Delta t\) can be sketched as:

\(E\!\left[R^{NS}_{t \to t+\Delta t}\right] \approx \underbrace{E\!\left[\Delta \ln P_t\right]}_{\text{price}} + \underbrace{Y_t^{\text{burn}}}_{\text{buyback}} - \underbrace{Y_t^{\text{dil}}}_{\text{dilution}} \, \)

For a staker,

\(E\!\left[R^{ST}_{t \to t+\Delta t}\right] \approx E\!\left[\Delta \ln P_t\right] + Y_t^{\text{burn}} - Y_t^{\text{dil}} + \underbrace{Y_t^{\text{stk}}}_{\text{staking carry}} \,\)

Key idea: changing the mix between burn, MEV/tips, and issuance re-allocates cash-flows and changes their volatility. Both effects move the required risk premium.

Microstructure channels that (de)smooth cash-flows

EIP-1559 base-fee burn (TFM): burns the history-dependent reserve price; reduces collusion attack-surface; turns utilization into a buyback-like stream that scales with demand. Smoother than MEV, less cyclical than pure issuance.
MEV (tips/auctions): highly spiky and crisis-sensitive; revenues cluster in stress (liquidations, de-pegs) and concentrate among few builders, increasing cash-flow risk to stakers and price uncertainty to all holders.
PBS / MEV-burn proposals: enshrined PBS and MEV-burn smooth and/or redistribute MEV (e.g., burn or socialize spikes), shifting weight from volatile validator income to system-level burn. Prediction: lower required return for non-stakers (more buyback, less dilution; lower variance), ambiguous for stakers (lower mean-variance of carry).

Theory meets data: staking, carry, and returns

Recent evidence (NBER 2025) links staking ratios, reward rates, and excess returns: (i) staking ratio correlates positively with reward cross-sectionally but negatively in time series; (ii) staking ratios positively predict excess returns; (iii) crypto carry premia arise from convenience yields and platform usage. Our decomposition gives a mechanism: higher staking share changes who internalizes fees/MEV/issuance and how risky that income is.

On issuance, optimal policy trades security vs. dilution (Jermann 2025). Higher issuance raises \(Y^{\text{stk}}\) (security budget) but also \(Y^{\text{dil}}\) (non-staker tax). A regime that meets security at minimum variance of holder flows should price a lower premium than one that funds security primarily with spiky MEV/tips.

Testable predictions (signs & intuition)

Let \(r_{t\to t+h}^{\text{ex}}\) be future h-period excess returns. Define observables (rolling k-day averages, normalized by market cap):

- \( b_t \) = burn yield (base-fee burned / cap),

- \( m_t \) = MEV+tips to validators / cap,

- \( i_t \) = net issuance / cap,

- \( \sigma_t^{\text{MEV}} \) = volatility of \( m_t \).

Prediction P1 (Buyback effect): higher \(b_t\) → lower required premium for non-stakers (buyback-like, smoother); sign in regressions: \( \beta_b < 0 \). (Microstructure: EIP-1559.)

P2 (Dilution tax): higher \(i_t\) → higher required premium for non-stakers; \( \beta_i > 0 \). (Optimal issuance must justify its security benefit.)

P3 (Carry vs. risk): higher \(m_t\) raises staker carry \(Y^{\text{stk}}\) but if \(\sigma^{\text{MEV}}_t\) is large, risk compensation rises; net sign depends on variance/price-of-risk. Expect \( \beta_m > 0 \) when \(\sigma^{\text{MEV}}_t\) is high.

P4 (Policy mix): MEV-burn-like redesign (shift mass from \(m_t\) to \(b_t\)) should lower the non-staker premium and smooth returns.

Empirical recipe (ETH or cross-chain panel)

Data.

Burn & issuance: post-EIP-1559 and post-Merge series (burn rate, net issuance) from public ledgers/dashboards.
MEV & builder markets: relay/builder data; spikes around stress events.
Returns & controls: token returns, market factors, usage (tx volume), staking ratio (from validator sets / StakingRewards).

Baseline panel regression (monthly):

\(r^{\text{ex}}_{i,t \to t+h} = \alpha_i + \gamma_t + \beta_b b_{i,t} + \beta_i i_{i,t} + \beta_m m_{i,t} + \beta_\sigma \sigma^{\text{MEV}}_{i,t} + \beta_\theta \,\text{stake_ratio}_{i,t} + X'_{i,t} \delta + \varepsilon_{i,t}\)

Signs expected: \( \beta_b < 0, \; \beta_i > 0, \; \beta_\sigma > 0 \) (spiky income \(\rightarrow\) higher premia), \( \beta_\theta > 0 \)

Event windows. Around fee spikes, liquidations, de-pegs: test whether \( \sigma^{\text{MEV}} \) jumps and premia widen (higher option-implied beta)

What the microstructure suggests for design

Favor burn over tips for base fees (keep EIP-1559 spirit): combats collusion and creates buyback yield for all holders.
Tame MEV variance (PBS++ / MEV-burn): smooths validator income and reduces risk premia borne by non-stakers.
Target issuance via a security budget (not price memes): calibrate iti_tit to security; excess issuance is a pure dilution tax that raises required return.

ETH as a live case (what to measure)

Post-EIP-1559 and post-Merge, ETH’s net supply often reflects burn minus issuance, while validator income combines issuance, tips, and MEV auctions. Track rolling burn yield and MEV variance versus future excess returns to evaluate P1–P4 above. (For raw series, see the canonical burn/issuance dashboard; for MEV, use builder/relay datasets and regulator overviews on spikes.)

References (selected, 2024–2025)

Staking & pricing: Cong, He, Tang (2025) The Tokenomics of Staking (NBER w33640). NBER
Fee mechanism & burn: Roughgarden et al., Transaction Fee Mechanism Design (JACM, 2024); “Post-MEV World” notes on why base fees are burned. ACM Digital Library timroughgarden.org
MEV-burn / PBS: Buterin et al., MEV-burn—simple design (ethresear.ch, 2023); SoK/updates on enshrined PBS (2025). Ethereum Research arXiv
MEV spikes & concentration: ESMA (2025) report on MEV revenue spikes in stress and concentration; builder market data 2024–2025. ESMA arXiv
Optimal issuance: Jermann (2025) Optimal Issuance for PoS Blockchains. Finance Department
Burn/issuance series: canonical ETH supply/burn tracker. ultrasound.money