Vote-Escrow After Curve: A Decade of veToken Models, Compared

Curve shipped vote-escrow tokenomics in May 2020. Five years later, every third governance token claims to have a “ve” mechanic. Most of them don’t. Here’s what actually works, what doesn’t, and why.

The original: veCRV

Curve was facing a hard tokenomics problem: it needed to issue CRV emissions to bootstrap LPs, but emission farmers immediately dumped, depressing the token. Michael Egorov’s answer was vote-escrow:

• Lock CRV for any duration up to 4 years. Receive veCRV proportional to locked_amount × time_to_unlock / 4_years.
• veCRV holders direct gauge weights — i.e., decide which Curve pools receive the next epoch’s CRV emissions.
• veCRV holders also receive 50% of trading fees (post-2020 governance vote).
• Locked CRV cannot be sold. veCRV cannot be transferred. The lock decays linearly until expiry, at which point CRV is returned to the holder.

Three things happened that the designers may or may not have anticipated:

• Locking pressure became real. By 2021, ~50% of CRV supply was locked in veCRV. Circulating supply collapsed. Buy demand for CRV grew purely to lock and direct emissions.
• The bribe market emerged. Stablecoin protocols realised that buying veCRV votes was cheaper than paying their own emissions. Convex (CVX) emerged to aggregate veCRV at scale. Bribe markets like Votium and Hidden Hand professionalised the meta-game.
• Governance became plutocratic and entrenched. Whoever locked early and never unlocked dominated decisions for years. Newcomers couldn’t catch up without buying tokens at progressively worse prices.

veCRV solved emission dumping. It also created a permanent governance oligarchy. Both are features, depending on who you are.

The successors

Velodrome / Aerodrome (V2 ve(3,3))

Velodrome on Optimism (and later Aerodrome on Base) introduced liquid vote-escrow: veVELO is an NFT, not a frozen position. Holders can sell their lock, transferring voting power without unwinding it. The rest of the design adopts Curve’s veCRV almost wholesale: 4-year max lock, gauge votes for emissions, 100% of fees to voters of the pool you voted for.

Aerodrome currently runs the most active emissions market on Base — gauge weight votes on Aerodrome are arguably the most economically valuable governance votes in DeFi. The liquid-NFT design lets institutional holders custody and rebalance positions without unstaking penalties.

Pendle (vePENDLE)

Pendle adopted vePENDLE in 2022. Mechanics are similar to veCRV — 2-year max lock, voting on which pools receive emissions, fee accrual to lockers. The interesting twist: actual protocol fees from yield-trading volume make vePENDLE one of the few ve-tokens with cash-flow backed yield rather than emission-funded yield.

Pendle is the canonical “ve done well” example because the protocol generates real fees (yield trading on $5B+ TVL of yield assets). Most ve copycats lack this — their lockers earn yield denominated in their own emission, which is a self-referential loop.

Frax (veFXS)

Frax adopted vote-escrow in 2021. veFXS holders direct emissions across Frax’s growing collection of products (Frax Stablecoin, frxETH, Fraxlend, etc.). The interesting feature: veFXS receives a 50% share of all protocol revenue across products, paid in stables — so the “real yield” story is structurally similar to Pendle.

The wrappers: Convex and friends

Three months after veCRV launched, Convex Finance introduced cvxCRV: deposit CRV, receive cvxCRV which represents a perpetually-locked veCRV position. cvxCRV is liquid (tradable on-DEX), receives boosted CRV rewards, and CVX holders direct the underlying veCRV votes.

Convex won. By 2022 it controlled > 50% of veCRV. The strategic outcome: Curve’s governance was abstracted to Convex’s governance. Whoever held CVX effectively controlled CRV gauge weights. A bribe paid in $X to CVX voters could redirect $5–10X of CRV emissions.

This pattern repeated for Frax (cvxFXS), Velodrome (lots of variants), and a dozen smaller ve-tokens. The lesson is structural: any vote-escrow design will eventually be wrapped, and the wrapper will accrue the real governance. You cannot prevent this with tokenomics — only by writing the wrapper yourself, the way Aerodrome did.

Why most ve copycats fail

A ve mechanism only creates real lock pressure if locking is incentive-rational for someone — i.e., the rewards for locking exceed the opportunity cost of liquidity. Three failure modes:

1. Locking yields the protocol’s own emission

If your “ve yield” is just more of the same emission you’re locking, you’ve built a snake eating its tail. Locked holders receive a share of inflation, but inflation is what dilutes them. The math nets out to roughly zero real return. Without external value flowing in (fees, bribes, external rewards), nobody locks beyond the marginal whale who needs the votes.

2. No fee accrual + no governance value

If voting on emissions doesn’t direct meaningful capital, the votes aren’t worth anything. Curve’s gauges direct $100M+ of CRV per year; those votes have value. A 5M-FDV fork’s gauge votes direct $200K/year. Nobody locks 4 years for that.

3. The ve unlock cliff

Curve’s veCRV decays linearly to zero over the lock duration. Many copycats use cliff-unlock schedules where lockers receive constant voting power until the unlock date. This breaks the Schelling point — the day before unlock you have full voting power; the day after you can dump. Holders just wait out the lock. Use linear decay.

Modern derivatives

Liquid restaking + ve-style governance

EigenLayer, Renzo, Kelp and the other restaking protocols of 2024 borrow ve mechanics for governance over operator allocations. The lock is in restaked ETH (liquid via LRT) rather than the protocol’s own token, which de-couples lock pressure from token economics. Whether this is “ve” depends on your definition.

Staked governance with auto-compounding

Some 2023+ launches abandoned ve in favour of simpler “stake to govern”: stake X tokens, receive auto-compounding share of fees, un-stake any time but with a cooldown. Less lock pressure than ve, but more liquid and less governance-game-able. Most modern launches choose this over ve for simpler reasons.

When to use ve in your design

Three filters. If you can’t answer yes to all three, don’t use ve:

• Do you generate meaningful protocol fees? Without external value flowing in, ve becomes a snake eating its tail. Curve, Pendle, Aerodrome, Frax all pass. Most fork ecosystems don’t.
• Do votes direct meaningful capital? Gauge weights need to direct enough emissions or fees to be worth locking 4 years to influence. If your treasury is $5M, your gauge votes aren’t worth meaningful lock pressure.
• Are you OK with a wrapper emerging? Convex-equivalents will appear. Plan for it: either build the wrapper yourself (Aerodrome) or accept that governance abstracts a layer up (Curve’s relationship with Convex).

For most launches, ve is overkill. The simpler “stake to receive fees + cooldown to unstake” design captures most of the alignment benefit without the governance theatre. Reserve ve for protocols where directing emissions is itself the primary product.

Further reading

• Curve docs: understanding veCRV ↗ — the canonical reference.
• Pendle’s vePENDLE design ↗ — the cleanest modern implementation.
• Aerodrome’s liquid ve-NFT model ↗ — the post-Convex version.