bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 2쪽 중 1쪽
Morpho Blue Whitepaper
Mathis Gontier Delaunay
mathis@morpho.org
Paul Frambot
paul@morpho.org
Quentin Garchery
quentin@morpho.org
Matthieu Lesbre
matthieu@morpho.org
October 2023
Abstract
Morpho Blue is a noncustodial lending protocol implemented for the
Ethereum Virtual Machine. Morpho Blue offers a new trustless primi-
tive with increased efficiency and flexibility compared to existing lending
platforms. It provides permissionless risk management and permissionless
market creation with oracle agnostic pricing. It also enables higher collat-
eralization factors, improved interest rates, and lower gas consumption.
The protocol is designed to be a simple, immutable, and governance-
minimized base layer that allows for a wide variety of other layers to be
built on top. Morpho Blue also offers a convenient developer experience
with a singleton implementation, callbacks, free flash loans, and account
management features.
1 Introduction
1.1 Motivation
Traditionally, financial services rely on trusted institutions to process transac-
tions and secure value. By allowing trustless code execution, blockchains and
smart contracts challenge this paradigm. Decentralized finance (DeFi) aims to
make financial services more open, efficient, and resilient than its traditional
alternatives [9].
Two paths have emerged for building DeFi platforms: decentralized funds
and protocols [8]. On the one hand, decentralized funds such as Compound
[14] or Aave [18] are platforms similar to traditional funds, but they leverage
onchain infrastructure and decentralized governance to function. While this
approach offers a passive and liquid user experience, it has trust and scalability
issues. For instance, decentralized funds rely on Decentralized Autonomous
Organizations (DAOs) and trusted contractors to manage the platform’s assets.
They accomplish this by monitoring numerous risk parameters [7] and upgrading
large smart contracts. DAOs are also not best suited for operational scaling,
1
and they can often become a bottleneck as the protocol grows. Additionally,
due to their maintenance costs, these platforms necessitate fees from their users.
On the other hand, protocols such as Uniswap [2] are trustless and unopin-
ionated smart contracts performing financial operations. This path has demon-
strated remarkable resilience over the years. However, this primitive approach
comes with usability tradeoffs at the protocol level. For instance, profitable liq-
uidity provision on UniswapV3 [3] may require active and sophisticated position
management.
What we will present is in line with the vision that DeFi should be orga-
nized in layers around trustless and open protocols, like the Internet. Onchain
and offchain applications can be built on minimalist DeFi primitives to manage
positions, handle compliance, or enhance usability for the user, for example. De-
pending on their needs, users can delegate some of their work to different layers.
This model allows them to benefit from the efficiency created by the network
effect around the primitive while remaining fully usable. A good example of
this approach is UniswapX [4], which simplifies and optimizes the experience
for UniswapV4 [1] traders who may have difficulty identifying the appropriate
swap routing and managing MEV protection.
In 2023, Morpho Optimizer [10] has become the third-largest lending plat-
form on Ethereum, with $1 billion in deposited assets. This initial platform
was developed on top of Aave and Compound to enhance the efficiency of their
interest rate model. However, developing the most significant platform built on
top of decentralized lending platform helped us realize that they would never be
sufficiently scalable, efficient, and resilient to reach the next level of adoption.
As stated in the initial whitepaper [10], Morpho must metamorphose to become
fully autonomous and enhance the current state of DeFi lending.
1.2 Introducing Morpho Blue
This paper introduces Morpho Blue, a new trustless and efficient lending proto-
col with permissionless market creation. Morpho Blue implements elementary,
immutable, and isolated lending markets. On a market, suppliers deposit the
loan token into the smart contract. On the other side, borrowers provide the
collateral token to secure their loan whose Loan-To-Value ratio (LTV) can go
up to the market’s Liquidation Loan-To-Value ratio (LLTV). Above this limit,
the account will be eligible for liquidation. Assets are priced with the mar-
ket’s oracle. Borrowers pay lenders interest given by the market’s Interest Rate
Model (IRM). To create a market, one must specify: a loan token, a collateral
token, an oracle, a LLTV and an IRM. Both LLTV and IRM are chosen from
governance-defined collections.
The Morpho Governance cannot halt the operation of a market or modify
its LLTV, IRM, and oracle. However, it has the ability to expand the range
of LLTV and IRM options available for market creation. Additionally, in each
market, the governance can enable a fee ranging from 0% to 25% of the total
amount of interests paid by borrowers.
2
In the remainder of this whitepaper, we first describe how the protocol ex-
ternalizes risk management and how lenders can delegate this work to a permis-
sionless risk management layer. The latter enables anyone to rebuild any user
experience with any risk profile on top of a primitive, which features permis-
sionless market creation, oracle-agnostic pricing, and bad debt accounting.
Then, we explain specifics of the design. Notably, the minimalist approach
allows for full auditability of the protocol and unlocks new levels of efficiency
for interest rates, collateralization factors, and gas consumption.
Finally, we detail how the codebase of Morpho Blue improves the expres-
siveness of integrations and power users by utilizing notable code patterns such
as a singleton contract, callbacks, free flash loans, and account management.
2 Externalized Risk Management
Removing the DAO risk management bottleneck is critical for DeFi lending to
become more resilient and scalable. The current approach severely restricts the
number of supported assets and confines users to a single risk-return profile[7].
A primitive protocol removes these downsides, but it may imply a more com-
plex user experience and fragmented liquidity. In this section, we first explain
how management layers can be rebuilt on top of Morpho Blue to re-aggregate
liquidity and provide a passive user experience with diversified exposure for
those who prefer it. Then, we describe how Morpho Blue externalizes oracle
risk while still retaining the benefits provided by oracles. Finally, we outline
how Morpho Blue eliminates the risk of bank runs and can continue running
indefinitely without requiring any built-in management, regardless of market
conditions.
2.1 Permissionless Risk Management
Morpho Blue is designed to be the foundational layer of decentralized lending.
It is immutable, and the MORPHO governance cannot manage users’ funds (see
section 1.2).
The protocol is designed to leave the choices up to the users, notably by let-
ting anyone create markets with any loan asset, collateral asset, risk parameters,
or oracle. For example, one could create the market DAI backed by WETH as
collateral with a LLTV of 90%, and using Chainlink[6] as the oracle.
While this flexibility offers a broader diversity of use cases, lenders must
actively select the market in which they want to provide liquidity. This can
lead to liquidity fragmentation and introduce complexities in risk management,
which may deter lenders with less knowledge. To solve this problem, Morpho
Blue is designed to serve as a basic building block that allows for the addition
of more layers of logic. These layers can enhance the core functionality by
handling risk management, or even compliance, and simplify the user experience
for passive lenders.
3
Morpho Blue lending markets can be assembled to reproduce a multi-asset
lending pool. For example, a vault could accept WETH from passive lenders
seeking a risk-adjusted and liquid experience, then have a risk expert allocate
deposits across various Morpho Blue markets where WETH is the loan asset.
Here, the risk expert abstracts the complexities by selecting markets with ap-
propriate collateral, LLTV, and oracles on behalf of users. In this case, WETH
in the vault could be evenly allocated between stETH/WETH (97%, Chainlink,
IRM) and cbETH/WETH (95%, Chainlink, IRM).
The example above illustrates how a Compound III ETH [13] experience can
be replicated on top of Morpho Blue. It is possible to recreate any lending pool
with any asset and any risk management method on this unique, trustless, and
efficient primitive. As a result, it is also possible to build a vault similar to the
AaveV3 WETH lending market.
With decentralized funds, when two platforms offer similar risks, their liquid-
ity is completely segregated. However, with Morpho Blue, users with different
risk profiles can share their liquidity on some common primitive markets. As
a result, markets will benefit from deeper liquidity, creating network effects for
all participants. Additionally, risk management solutions can easily compete,
leading to better and cheaper products for end users.
It is important to note that if the risk management of a platform built on top
of Morpho Blue loses its users’ funds, for either bad management or a technical
issue, it will not affect the security of the primitive.
4
The following figure illustrates how Morpho Blue separates and modularizes
risk management by separating it from the lending protocol.
2.2 Oracle-Agnostic Pricing
For a lending protocol to operate efficiently, it must have an accurate notion of
price. This price information is obtained by querying trading markets, whether
they are internal or external to the protocol.
External price feeds, often called oracles, have various designs with different
accuracy and security properties. For example, some oracles can provide fre-
quent and accurate price updates, but they may be considered too centralized
or manipulable to be at the core of a lending protocol.
To create more resilient lending primitives, some oracle-less protocols di-
rectly incorporated the pricing mechanics at their core. However, by doing so,
the primitive has to handle both a trading and a lending mechanism simulta-
neously. This brings additional complexity, increasing gas costs and limiting
auditability and security.
Morpho is dedicated exclusively to lending rather than trading. Its goal is
to become the most open and efficient lending primitive while still operating in
a trustless manner. This is why the protocol has an oracle-agnostic approach
rather than an oracle-less one.
Morpho Blue has no oracle or trading mechanism built into the primitive.
Instead, anyone can create a market by specifying an address that returns a
price for the loan and collateral assets. Some markets may feature Chainlink
or Uniswap oracles, while others could hardcode the price or use an innovative
mechanism similar to Ajna’s [16]. It is up to the users to decide which oracle
they agree to depend on by choosing the corresponding markets to interact with.
Lenders and borrowers could delegate this work to the permissionless risk man-
agement layer (see section 2.1), should this choice be deemed too complicated.
This way, Morpho Blue remains simple, trustless, and unopinionated while
ensuring optimal efficiency for those who desire it.
5
2.3 Bad Debt Accounting
In this section, we note LT Vthe current Loan-To-Value ratio of the borrower.
This is the ratio of the value of the debt over the value of the borrower’s col-
lateral. The Liquidation Incentive Factor denoted LIF is the factor defining
the bonus percentage given to the liquidator during a liquidation process. The
health of a borrower’s position can be divided into three main categories.
• If LT V < LLT V, the position is healthy, and the borrower cannot be
liquidated.
• If LLT V < LT V≤ 1/LIF , a liquidator can repay part or all of the user’s
debt and seize part of its collateral.
• If 1/LIF < LT V, a liquidator can seize all the collateral by repaying only
a share of the debt. There is no incentive for the liquidator or borrower
to repay the remaining debt. The latter is commonly referred to as bad
debt.
For example, in a market with LLT V= 0.8 and LIF = 1.1 and a position
with $1000 as collateral and a $750 of debt cannot be liquidated. If its debt
grows to $850, then liquidators can repay all the debt and get back part of
the collateral. And if its debt grows to $950, then liquidators can take all the
collateral and repay only part of the debt, leaving bad debt on the market.
The impact of bad debt on lending platforms and the approach to handling
it vary depending on the mechanism. For example, Aave and Compound do
not account for bad debt, meaning that the last lenders to withdraw from the
pool will bear the entire loss alone. To prevent losses, a lender would withdraw
from the pool if they believe the bad debt is significant compared to the overall
size of the lending pool. As more lenders begin to withdraw their capital,
the proportion of bad debt increases, creating a stronger incentive for further
withdrawals. This results in a bank run on lending pools, which could lead to
the collapse of the entire platform.
Lending pools’ DAOs have implemented several measures to mitigate this
existential risk. These measures include compensating for bad debt through
their fees or treasury, paying for insurance funds [19], or simply building a
trusted brand for the platform.
While these mitigations may be suitable for a decentralized funds, managing
profits and losses on behalf of users is not an option for a protocol aiming to be
trustless and scalable. Morpho Blue takes a different approach by accounting
for the bad debt. When a liquidation occurs on Morpho Blue, if the borrower
has outstanding debt but no collateral, the losses are socialized proportionately
among lenders, resulting in an instant loss for lenders. If for some reason liq-
uidators don’t liquidate the position fully and thus do not account for bad debt,
lenders can, if there is enough liquidity, temporarily withdraw their funds to ac-
count for the default without incurring any losses. As for future users, if there
is unaccounted bad debt in the pool, they can account it before safely entering
6
the market. As a result, Morpho Blue markets can continue running indefinitely
in a trustless manner, regardless of market conditions.
2.4 Uncapped Markets
In current lending pools, the governance can set supply caps, constraining the
total amount of a given collateral that can be supplied. This is notably used to
control the exposure of lenders’ funds to each collateral in pools where there are
multiple ones. In addition, by ensuring that the size of the market stays below a
certain threshold, supply caps can, for example, be expected to guarantee there
is enough liquidity on exchange markets for liquidations. The latter is far from
being ideal, though. For example, an external market composed of the same
assets can freely grow, thus challenging these liquidity assumptions and making
the guarantees of caps irrelevant.
Morpho Blue has no cap mechanism because it fully externalizes risk man-
agement. Instead, lenders are expected to achieve granular control over their
exposure to different collateral by distributing their supplied volume across the
different Morpho Blue markets. As explained in section 2.1, this work can be
delegated to risk experts.
3 Minimal Lending Markets
In the nascent stages of DeFi, liquidity was an indispensable component. With
few sophisticated actors, DeFi had to reinvent traditional finance mechanisms
to make its markets usable. On the DEX side, constant function market makers
(CFMMs) [2] were able to gather enough liquidity for emerging cryptocurrency
markets by simplifying the market-making process to a “set and forget” design.
On the lending side, Compound and Aave gathered multiple markets in the
same pool, choosing common lending terms through the DAO and reinvesting
collateral, meaning that the collateral itself can be borrowed. Although effective
in bootstrapping DeFi liquidity, this design was forced to make a trade-off:
prioritizing ease of use over efficiency and scalability.
DeFi is evolving. As the number and sophistication of actors increase, pro-
tocols have evolved to allow greater expressiveness. This not only enhances
efficiency [15] for the user but, somewhat paradoxically, can also result in more
streamlined logic and code.
3.1 No Embedded Re-Hypothecation
The first constraint of multi-asset lending pools is the fact that the markets
must remain liquid at all times. Since collateral assets lent out may need to be
liquidated, a consistent level of liquidity is required. Without it, it puts other
markets at risk of bad debt. This means that the markets should not stay highly
utilized for long periods of time. Concretely, it induces quite large spreads and
interest rates to spike at high utilization.
7
In Morpho Blue, by default borrowers’ collateral is not lent out to other
borrowers but rather stays fully liquid on Morpho Blue’s contract. As a result,
the market’s utilization can average higher, lowering the rate spread. Also, rates
do not need to spike at high utilizations, providing more stability to borrowers.
Although, one can create markets with tokenized supply assets or a supply
vault as collateral to do some re-hypothecation.
3.2 Efficient LLTVs
In current multi-asset lending pools, there is typically one LLTV per collateral.
In other words, the LLTV doesn’t depend on the loan asset for a given collat-
eral. However, reasonable models [5][12] for assessing the LLTV associated with
a collateral asset also depend on the loan asset because one cares about the
price of one of them quoted in the other. Consequently, risk managers of such
lending pools must compromise between risk on uncorrelated loan assets and
inefficiency on correlated assets. Aave introduced in its v3 [11] efficiency and
isolation modes, which offer partial solutions to this problem. Similarly, Euler
[21] introduced the concept of borrow factor.
Morpho Blue doesn’t face this issue because LLTVs are defined for a market
with one loan asset and one collateral asset. As a result, collateral ratios are
improved for borrowers, usually leading to more interest generated for lenders.
3.3 Primitive Liquidations
When a borrower’s LTV on a market exceeds the market’s LLTV, the borrower
can be liquidated. Anyone can perform a liquidation by repaying the account’s
debt and receiving the equivalent amount in collateral, along with an incentive
factor, defined by the Liquidation Incentive Factor (LIF).
When establishing a liquidation incentive, there exists no definitive opti-
mal solution. Rather, it is a fundamental trade-off between borrowers’ costs,
liquidators’ incentives, and lenders’ safety. Current protocols have different ap-
proaches to setting it. Aave and Compound have a fixed incentive per collateral
set by governance, which tends to favor lenders. Euler [21] and Angle [20]
have a health-dependent Dutch auction mechanism, meaning that the liquida-
tion incentive increases when the health of the position decreases. This favors
borrowers and disfavors lenders by reducing the liquidatable amount and the
absolute incentive to liquidate above a certain health factor.
Regarding the proportion of the debt that can be repaid by the liquidator, or
“close factor”, there is a similar trade-off. Small close factors favor borrowers,
and large ones favor lenders. Dynamic close factors, such as the ones allowing to
liquidate only what is necessary to put the account back above water, reduce the
immediate incentive and thus favor borrowers. In addition to that, as discussed
in [17], fixed close factors can be bypassed by performing successive liquid프로젝트로 돌아가기