‹ Understanding DeFi Lesson 13 of 16
Contents Lesson 13 of 16

5 min read · professional

Where does a DeFi yield actually come from?

Every yield is somebody's cost. That sentence is not a moral position; it is an accounting identity. When you see an advertised APY, exactly one question separates understanding from guessing: who is paying it, and why would they keep paying? There are only a handful of possible answers.

Source 1 — trading fees (real, and volume-dependent)

Traders pay a fee on every swap; liquidity providers receive it. This is genuine external revenue: the payer is a third party receiving a service, and nothing is being printed.

It is also volatile, because volume is volatile, and — as Unit 2 established — it must be netted against impermanent loss before it means anything. Real income, uncertain quantity, and never the whole picture.

Source 2 — borrowing demand (real, and mechanical)

Borrowers pay interest; suppliers receive most of it. Rates are set by a published utilisation curve rather than by negotiation:

utilisation U = total borrowed ÷ total supplied

supply rate ≈ borrow rate × U × (1 − reserve factor)

Worked. $100 million supplied, $70 million borrowed → U = 70%. Borrow rate 8%, reserve factor 10%:

supply rate = 0.08 × 0.70 × 0.90 = 5.04%

This formula explains a fact that confuses newcomers: the supply rate is always well below the borrow rate, and the gap is not a spread the protocol keeps — most of it is the idle 30% of capital earning nothing so that withdrawals can be honoured. It also explains why supply rates spike when utilisation approaches 100%: the curve steepens deliberately, to bid in more supply and price out more borrowing. High utilisation means high rates and the possibility that you cannot withdraw until someone repays.

Source 3 — token emissions (not income at all)

The protocol prints its own governance token and distributes it to users. Denominated in dollars, this looks identical to the first two sources on a dashboard. Economically it is entirely different: no external party is paying. Value is transferred from existing token holders to users through dilution.

This is the single most important distinction in this unit. Fees and interest are revenue. Emissions are a transfer, funded by dilution, priced at whatever the market pays for the token. The next lesson does the arithmetic of why they decay.

Source 4 — points, airdrop expectations and incentives

A promise of future tokens that do not yet exist, with no defined rate. This is not a yield at all; it is an option on a future distribution, with unknown terms, unknown timing and no obligation to deliver. Displaying it as an APY is a category error.

Source 5 — leverage on the above

The recursive loop from Unit 1 does not create yield. It applies the same underlying return to a larger balance while adding liquidation risk and multiplying smart-contract exposure. The arithmetic does work while nothing goes wrong: $40,000 earning 5% is $2,000 on $10,000 of your own money, which is 20% before borrow costs. What has not happened is the discovery of a 20% yield. It is still a 5% source, and leverage has multiplied the losses by the same four as the gains, added a liquidation price, and put four times as much through the same smart contracts. Magnified return on equity is not a new return.

Source 6 — staking (two sources wearing one number)

The base chain pays validators from new issuance plus the priority fees and ordering value in the blocks they propose. A liquid staking token passes that through minus the operator's cut (Lido's documentation states 10% of rewards) and then trades and lends like any other token. Classify it in two parts: the fee and ordering part is paid by users and is revenue; the issuance part is dilution borne by whoever does not stake, a printing press with a demand side. Restaking lends the same stake to additional services and accepts their slashing conditions too, so the yield rises with the number of ways the collateral can be destroyed. Liquid staking has been the largest category in DeFi by value since 2023, so a yield decomposition without it misreads the biggest number in the system.

The checklist

For any advertised yield, in order:

  1. Who pays? A trader, a borrower, or a printing press?
  2. What share is emissions? Most dashboards separate "base" from "reward" APY. If they do not, treat the whole figure as unexplained.
  3. What is being shorted? Fee income is short volatility; borrowing yield is short liquidity; leverage is short a price move.
  4. How many contracts is it built on? Apply the composability arithmetic from Unit 1.
  5. What would the return be if the incentive token went to zero? That number is the durable part.

The plain statement

A high advertised APY almost always encodes one of two things: risk, or token emissions. Occasionally it encodes a genuine, temporary imbalance — which competition removes quickly, because capital is permissionless and moves fast. And every yield in DeFi, however sourced, sits on top of smart contracts that have repeatedly been exploited for total, unrecoverable losses.

This course does not suggest pursuing yield anywhere, and takes no view on any protocol or rate. It gives you the decomposition.

Try it now

  1. On defillama.com or a similar public analytics site, open the yields section and sort by APY. Look at the top twenty and check how many separate base yield from reward yield.
  2. Pick three listings and answer question 1 of the checklist for each: who is actually paying?
  3. Take any lending market's published supply rate, borrow rate and utilisation, and check them against the formula above. Where the numbers do not reconcile, find the reserve factor — the residual is usually there.