Skip to content

LibraryCredit and stable value2017Design paperCorpus record

The Dai Stablecoin System

Maker. MakerDAO.

MakerDAO's 2017 description of Dai: a liability minted against overcollateralised vaults, kept near a dollar target by fees, a collateral auction, and an emergency shutdown. It is the reference design for crypto-collateralised stable value, as distinct from a fiat-backed token.

Maker's Dai is a liability minted against collateral locked in a vault, kept near a dollar target by a stability fee, a liquidation auction, and an emergency shutdown that settles to the collateral.

The five-minute read

Dai is debt, not a wrapped bank deposit

A user locks collateral, mints Dai up to a ratio, and owes the system. The dollar target is a policy. The asset behind a given Dai is the collateral and the auction, not a named cash account.

The stability fee is monetary policy

It accrues on the debt. Raising it is supposed to reduce minting when Dai is cheap. The paper treats the fee as a tool, which means someone sets it.

Liquidation is an auction

If the collateral value falls through the ratio, the vault is sold to cover the Dai. The auction only works if bidders show up near the oracle price.

Shutdown is part of the design

Emergency shutdown settles the system. Dai holders are supposed to be able to claim collateral. A stablecoin paper that cannot describe its stop condition is less honest than this one.

Oracles and keepers are inside the security model

The paper does not work without a price, and the price is not produced by the vault math. Keepers who bid in auctions are part of solvency.

One action, walked through

  1. A user opens a vault, locks an accepted collateral, and draws Dai up to the liquidation ratio.
  2. A stability fee accrues on the drawn amount. The user must repay Dai plus the fee to unlock collateral.
  3. Oracles update the collateral price. If the ratio breaks, the vault is sent to auction.
  4. Keepers bid Dai or the collateral, under the auction rules of the version being cited, to cover the debt.
  5. If governance triggers shutdown, further minting stops and Dai is settled against the remaining collateral at the settlement price.

The argument, unpacked

Overcollateralisation is only as good as the sale

A 150 percent ratio does nothing if the collateral gaps through the oracle and the auction clears far below. The paper's protection is a process: ratio, oracle, auction, penalty. Quote the process, not the adjective.

Governance can change the deal

Fees, ratios, debt ceilings and accepted collateral are parameters. A user who opened a vault under one set can be liquidated under a later set. That power is not an embarrassment added by cynics. It is how the 2017 system steers the peg.

Single-collateral and multi-collateral are different papers in practice

The 2017 story is one collateral and a simpler auction. Multi-collateral Dai adds more oracles, more auction types and more ways to be wrong. Say which one you mean when you say Dai.

What has to be true

  • The oracle price is close enough to a market where keepers can actually sell.
  • Auctions complete. A stalled auction during a crash is how a surplus becomes a shortfall.
  • Debt ceilings and governance delay are set so one collateral cannot swallow the system unnoticed.
  • Shutdown settlement is implementable. A settlement price that cannot be redeemed is a paragraph, not a backstop.

What happened after the paper

Maker moved from single-collateral Sai to multi-collateral Dai, added new auction mechanics after the March 2020 event, and later added real-world collateral that behaves nothing like ether in a crash. The 2017 paper is still the right anatomy: vault, fee, oracle, auction, shutdown. It is not a description of every later collateral type, and it is not a promise that the peg holds.

What to check before you use the idea

  • Which collateral, which ratio, and which oracle?
  • What do keepers bid in a liquidation, and did that mechanism change after 2020?
  • Who sets the stability fee, and how fast can it change?
  • What does a Dai holder receive in shutdown, in the version being cited?

Terms

Vault
The position that locks collateral and mints Dai. It is a debt, not a deposit account.
Stability fee
Interest on the minted Dai, used as a peg tool.
Liquidation ratio
The collateral multiple below which the vault is auctioned.
Emergency shutdown
The stop that settles Dai against collateral and ends further minting.

The problem the paper names

A token that simply promises a dollar is a claim on a bank account. A token that is minted when someone locks a volatile asset is a claim on that collateral and on the liquidation process. Maker's paper specifies the second object.

What the design proposes

  • A vault locks collateral and mints Dai up to a limit set by a collateralisation ratio.
  • A stability fee accrues on the debt. That fee is a monetary tool in the paper, not a service charge hidden in a terms document.
  • If collateral value falls through the ratio, the position is liquidated through an auction.

How the mechanism is specified

  • Keepers and price oracles are part of the safety system. The paper does not work without a price.
  • Emergency shutdown settles the system to collateral. It is the 'stop' the design is willing to name.
  • Multi-collateral Dai and later modules extend the 2017 single-collateral story. Name the version.

What this page does not treat as proven

  • Overcollateralisation only protects the peg while the collateral can be sold for something near the oracle price.
  • Governance parameters can change the risk after a user opened a vault. That power is part of the system, not an embarrassment to be omitted.
  • This is not a claim that Dai has held or will hold a peg. Pegs are empirical.

Why a venture studio still reads it

The studio test for a 'decentralised dollar' is the Maker paper's inventory of moving parts: collateral, ratio, fee, oracle, auction, shutdown. A pitch that has a token and none of those parts is a different, weaker object.

This is Blockchain Lab's reading of a public design paper. It is not the paper, not a copy of it, and not an offer of tokens, equity, custody or a partnership. Later network behaviour can diverge from the text. Nothing here is investment, legal or technical advice.

Research status: Design paper. Last reviewed: 1 October 2026. This is a reading of a public paper, not investment, legal or security advice.