Chapter 17 · Where Yield Actually Comes From
Who pays the money behind this APY?
- Capabilities practiced
- Financial LiteracyResearch
- Hands on
- Pick three high-APY pools and break down each yield source: fees, interest, and subsidies.
- AI Lab
- Have AI break down a yield product's cash flows, then use on-chain data yourself to verify when the subsidy component ends.
A Real-World Problem
Three interfaces on the same day.
A banking app shows a tiny demand-deposit rate. An on-chain lending protocol offers 12% on stablecoins. A newly launched pool displays 340%.
All say annualized. Your instinct is to compare their size.
But they answer different questions. You broadly understand bank interest: deposits finance loans, the bank retains the spread and pays you a little. Chapter 16 explained the second: borrowers pay depositors, with a protocol share deducted.
Can you identify who pays the third?
This chapter teaches one action that may save more money than any other in Part A: before comparing APY, identify the payer.
Because money never appears from nowhere. Every extra cent in your account comes from somewhere else. Not finding its source does not mean none exists; it means you have not found it.
Thought Experiment
Three village shops accept deposits and promise interest.
Shop A lends to people buying farm machinery at 8%, pays you 6%, and retains 2%.
Shop B uses deposits to operate a market stall and distributes exchange fees proportionally.
Shop C prints its own vouchers and distributes them daily according to deposits. A voucher currently trades for one dollar, producing a quoted 200% annualized return.
All three tell the truth. Ask:
If no new customers enter the village tomorrow, what happens to each shop?
- A continues. Farmers repay using income generated by machinery, independently of new depositors.
- B depends on market activity. Trading villagers continue paying fees; a quiet market generates almost none. The amount fluctuates, but the source is clear.
- C keeps printing vouchers, but with no new buyers their price declines. Today's price produces 200%; halving it tomorrow produces 100%, and vouchers already received lose half their value too.
The difference is where the money comes from:
- A: borrowers' real income.
- B: traders' fees.
- C: the next voucher buyer.
The first two sources are outside the deposit scheme; the third is inside: you and those entering after you.
Your Decision
A pool displays 180% APY. What do you do?
What Happens
Break the 180% into parts. Suppose you find:
| Observation | Value |
|---|---|
| Pool TVL | $1 million |
| Genuine 24-hour fees | $137 |
| Daily subsidy | 240,000 protocol tokens |
| Current token price | $0.02 |
| Remaining subsidy | 30 million tokens |
Calculate:
Annualized fees = 137 × 365 ÷ 1,000,000 = 5.0%
Annualized subsidy = 240,000 × 0.02 × 365 ÷ 1,000,000 = 175.2%
Subsidy lifetime = 30,000,000 ÷ 240,000 = 125 daysThe headline becomes a different statement:
"Traders fund a genuine 5% annualized return. The protocol supplies another 175% in its own token, with 125 days remaining."
Same facts and numbers; now you understand them.
Examine the properties of each source:
| Source | Payer | Stops when | Denomination | |
|---|---|---|---|---|
| Fees | Trading, swapping, minting, other activity | Product users | Usage stops | Dollars or major assets |
| Interest | Borrowing cost | Borrowers | Borrowing stops | Dollars |
| Subsidies | Issued or released protocol tokens | Other holders and future buyers | Allocation ends or governance stops it | Protocol token |
The last column is crucial. The first two arrive as dollars; the third arrives as a fluctuating token continually sold by you and every other recipient.
The 175% uses today's token price. Reward issuance creates selling pressure, lowering price and yield, while real-time interface updates can keep the headline looking high.
Four other things are often labeled yield:
| Displayed as yield | Actual meaning |
|---|---|
| Newly issued tokens' paper value | An unrealized claim whose sale depresses its price |
| Part of your principal | Principal funding early payments |
| Later participants' principal | New funds paying old users: the definition of a Ponzi structure |
| Unrealized gains | Collateral or LP appreciation that may differ on exit |
If the yield source is unclear, it may be you. This is literal: in three of these rows, holders ultimately pay themselves.
Build a Model
Four yield-source questions
Ask in order. If the first cannot be answered, the rest have no foundation.
- Who pays?
- Why do they pay?
- How long can they keep paying?
- What risk am I taking in exchange?
| Question | Investigation | What missing answers mean |
|---|---|---|
| Who pays? | Revenue pages, fee flows, reward contracts | You may be the payer |
| Why? | Service received and alternatives | No clear business reason; payments may stop |
| How long? | Remaining rewards ÷ daily issuance, or demand trends | Unknown end date |
| What risk? | Contracts, prices, liquidation, liquidity | A priced return exchanged for unpriced risk |
Split APY into three components
Take away this formula:
Displayed APY = fee APY + interest APY + subsidy APY
Fee APY = period fees × annualization factor ÷ TVL (dollars)
Interest APY = borrow rate × utilization × (1 − protocol share)
Subsidy APY = daily units × current token price × 365 ÷ TVL
(token-denominated; its price changes)
Remaining subsidy days = remaining reward units ÷ daily unitsThe first two form Real Yield: dollar-denominated returns from real payers. They are often much smaller than the headline and are what remains after subsidies end.
APR and APY differ
Four-digit annualized figures usually originate here:
| Daily return | APR, simple annualization | APY, daily reinvestment |
|---|---|---|
| 0.1% | 36.5% | 44.0% |
| 0.5% | 182.5% | 517% |
| 1.0% | 365% | 3679% |
APY assumes daily reinvestment for an entire year, and hides a stronger assumption: the reward token's price remains unchanged.
If that price falls 90%, constant token issuance does not deliver the displayed dollar return. A four-digit APY should prompt investigation of compounding assumptions.
Yield can stack across layers
Some products stake an asset, collateralize its receipt, borrow to provide liquidity, then farm with the LP claim.
Each layer can generate genuine yield, but two things happen:
- Returns add: four layers at 3% produce about 12%.
- Risks multiply: four independent 95% annual survival probabilities combine to 81%.
These risks are often correlated: a base-asset failure affects every layer. Returns grow linearly while risk compounds, an essential feature of these structures.
What It Is Called
APR annualizes simply; APY assumes reinvestment at a specified frequency.
Check which the interface displays and its compounding frequency. Daily compounding turns 365% into 3679%, describing the same daily rate.
Locking tokens to support chain consensus or protocol security in exchange for rewards.
Distinguish chain-level staking, paid through issuance dilution and transaction fees, from application-protocol staking, paid through subsidies or revenue sharing. Inspect the specific design.
Chapter 14's payments from pool traders, distributed by share.
This is identifiable Real Yield, but must be assessed alongside divergence loss. Ignoring that cost while citing fee APY is a common mistake.
Chapter 16's flow from borrowers to depositors, minus a protocol share.
Ask further: what do borrowers use the funds for? If they pursue a higher-subsidy farm, the ultimate interest source is still a subsidy.
Protocol-token rewards attracting liquidity or users.
Inspect daily issuance, remaining allocation, and adjustment authority. Together they turn a black box into a contract with an identifiable end date.
Subsidies are not automatically scams. Buying early growth with tokens is a legitimate business choice; users must understand the payment and its expiry.
Pledging already-staked assets or their receipts to other protocols for additional rewards.
One collateral pool secures several systems. Rewards add, while correlated risks affect multiple layers together. List each layer's slashing conditions separately.
Returns denominated in dollars or major assets and funded by genuine protocol revenue.
Its virtue is survival after subsidies end, rather than high rates. This is the number relevant to surviving the next cycle.
Hands-on
Choose a triple-digit pool, a moderate-yield pool, and a pure lending pool. No capital is required.
Record interface figures. Note the displayed rate, APR or APY label, TVL, and observation time.
Copy any fee/reward breakdown for later comparison, treating it as evidence to check.
Calculate fee APY. Find genuine 24-hour or seven-day pool fee revenue.
Fee APY = period revenue × annualization factor ÷ TVLIf no dashboard exists, multiply daily volume by the fee rate and label it an approximation.
Calculate subsidy APY. Inspect the reward contract for:
- Tokens issued per block or day
- Current token price
- Remaining tokens at the reward address
Subsidy APY = daily units × current price × 365 ÷ TVL
Subsidy lifetime = remaining units ÷ daily unitsFor weekly governance allocations, inspect the latest vote and note that next week may differ.
Complete the table.
| Pool A | Pool B | Pool C | |
|---|---|---|---|
| Displayed APY | |||
| Fee APY | |||
| Interest APY | |||
| Subsidy APY | |||
| Component sum versus display | |||
| Remaining subsidy days | |||
| Subsidy adjustment authority |
The fifth row checks consistency: a mismatch means a missing source or different interface assumptions. Investigate either.
Write one sentence per pool. Use this format:
X% is funded by [payer] paying [fee]; Y% is protocol-token subsidy, with roughly Z days remaining.
Three sentences complete the Lab. An unwriteable sentence identifies an incomplete investigation, itself useful information.
AI Lab
Avoid asking only whether yield is sustainable. Request a structure you can verify line by line:
Product: [name / pool address / chain]
Produce a cash-flow table, one independent income source per row:
- Source name
- Specific payer: trader, borrower, treasury, token issuance, etc.
- Denomination: dollars, major asset, or protocol token
- Annualized percentage-point contribution
- Condition under which payment stops
- Source link and observation time
Then answer separately:
1. Reward contract address, daily issuance, and remaining token units
2. Remaining days at current issuance
3. Authority and procedure for changing issuance
4. Remaining annualized return if subsidies become zero tomorrow
Write "not found" for missing data. Do not estimate or use historical memory.Then open the reward address and inspect its balance yourself.
The core property is that remaining subsidy allocation is directly observable on-chain. The model either gets that number right or does not.
Question 4 replaces marketing with a decision-relevant number: what remains when subsidies stop.
Verify these points yourself after using AI
- Can you identify a payer for every revenue row? Probe any row naming only the protocol.
- Did it separate subsidies from total APY?
- Can daily issuance and remaining allocation be verified at the reward address?
- Which date supplies the token price, and is that recorded?
- Does it distinguish APR and APY and state compounding frequency?
- For stacked products, are risks listed by layer rather than only adding returns?
- Do sustainability and Real Yield claims identify specific revenue?
Real-World Case
Protocols subsidized liquidity with their own tokens, displaying triple- and quadruple-digit yields. Funds poured in and TVL looked impressive.
The subsequent process was structural: recipients sold, prices fell, APY fell, funds left, and TVL fell. Lower TVL, the denominator, then lifted APY again and attracted new entrants.
The loop could repeat several times, producing winners and final holders. Without knowing your place in the loop, you are guessing.
An ecosystem offered high fixed stablecoin deposit interest and grew to tens of billions.
Its documented sources were borrowing interest, staking rewards, and a reserve fund covering the gap.
Borrowing demand lagged deposits, enlarging the gap and accelerating reserve depletion. Doubts about remaining reserves triggered withdrawals and collapse within days.
The four questions exposed it: the reserve paid part of yield; reserve balance divided by daily depletion gave a public, shrinking lifetime. No inside information was required.
Restaking uses one base asset to secure multiple systems, each paying rewards.
Returns add and risks compound, with strong correlation from the base asset or bottom-layer slashing.
List each layer's slashing conditions and resulting loss, rather than merely summing APYs.
Following the subsidy era, some protocols emphasized distributing genuine fees to holders or stakers and disclosed revenue.
Rates were generally single digits to the teens, but a real customer paid for a service, rather than an investor buying a token.
Chapter 26 explores revenue and its recipients. First ask whether Real Yield exists, then how much.
Change One Variable
Subsidy APY halves from roughly 175% to 87.6%, reducing total yield from 180% to 92.6%.
Unspent rewards already received halve in value too. And a dollar-budgeted subsidy issues twice as many tokens to maintain its amount, accelerating dilution.
Check whether issuance is fixed in tokens or dollars. Their behavior during declines differs fundamentally.
Subsidy APY falls to one tenth, from 175% to 17.5%, since issuance remains fixed.
Fee APY depends on trading-volume growth, which usually fails to keep pace, so it often falls too.
High APY erodes itself by attracting capital. When it remains high, ask: why has capital not already competed it down? Risk often explains the answer.
The token gains a clear cash-flow claim and value-capture path.
The treasury cannot accumulate funds for development, growth, incidents, or bad debt. Payout ratios balance current rewards with future resilience.
Governance usually controls distributions, so the ratio can change. Its reliability depends on that voting process.
Compounded nominal yield is about 12.5%.
If each layer has a 97% annual survival probability, independent survival is 91.3%. But shared base-asset failures make the risks correlated and the real outcome worse.
An extra eight percentage points can multiply the chance of total loss. Understand the exchange before judging it. Chapter 24 provides a risk-listing template.
Takeaway Questions
Make it executable: find daily genuine protocol revenue, divide by TVL, and annualize. That is revenue-based return; the interface adds subsidies. Separate them before researching further.
Payers are product users, borrowers, or other token holders. The first two are customers; the third represents dilution. If the first two cannot be found, the third may be the answer, and you are among them.
Real Yield directly captures protocol revenue for holders. Ask how much is distributed, how, and who can change the ratio. The third determines the reliability of the first two.
Calculate remaining subsidy allocation ÷ daily issuance. Then ask what remains at expiry. That remainder represents the protocol's underlying scale.
Chapter Check
First: who pays?
Comparing rates requires a valid reference. Interest and dilution are not interchangeable returns.
Source determines duration and whether a rate can be treated as long-term. If later entrants fund payments, sustainability depends on continued entry rather than product quality.
Use three publicly verifiable formulas:
Fee APY = period fees × annualization factor ÷ TVL
Interest APY = borrow rate × utilization × (1 − protocol share)
Subsidy APY = daily units × current token price × 365 ÷ TVLThe first two are dollar-based Real Yield; the third changes with protocol-token price.
Then calculate remaining reward units ÷ daily issuance = remaining days. These four numbers reveal far more than the headline.
Buying early growth with tokens resembles startup spending on acquisition. Healthy protocols can use subsidies for launch and later replace them with genuine revenue.
Two problems remain:
- Combining subsidy and real revenue in one displayed rate, making token issuance look like interest.
- Omitting the subsidy end date, despite its observability on-chain.
The useful stance is to know the payment type, its end, and what remains afterward.
Usually by compounding daily returns. A 1% daily rate gives 365% APR and 3679% daily-compounded APY.
Three difficult assumptions:
- Daily reinvestment despite Gas costs.
- An unchanged daily rate for a full year despite short subsidy lifetimes.
- An unchanged reward-token price, despite issuance selling pressure.
Inspect assumptions instead of projecting earnings. The number describes a hypothetical world unlikely to occur.
Useful columns include:
- Source, one independent payment per row.
- Specific payer, such as traders, borrowers, or issuance; "protocol" is insufficient.
- Denomination, dollars, major assets, or protocol tokens.
- Percentage-point contribution and valuation date.
- Stop condition, allocation exhaustion, governance, or vanishing demand.
- Remaining duration, calculated whenever possible.
- Modification authority, process, and timelocks.
- Additional risk accepted for that income.
Test the design: can a newcomer identify what remains after subsidies stop? If not, a necessary column is missing.
Key idea
When you see APY, ask for the yield source first; if no one can explain it, the yield is probably coming from you.