Inflation versus dilution, and what real staking yield means

Inflation in a token supply is the rate at which new units are created against the existing base. Dilution is what happens to one holder's proportional share when new units go to someone else. The two are often used interchangeably and are not the same quantity: a holder who receives new units at exactly the average rate experiences the inflation and none of the dilution.

The arithmetic

Let the supply grow by a rate i over a period, and let a holder's unit count grow by r over the same period through staking rewards. The holder's share of the network changes by approximately r - i, and exactly by (1 + r) / (1 + i) - 1. The subtraction is the whole idea behind real yield in its supply sense.

Three cases follow. Where r exceeds i, the staker's proportional share rises, funded by holders who did not stake. Where r equals i, the share is unchanged and the reward is compensation for participating rather than a gain in claim. Where r falls below i, the staker's share falls more slowly than a non-staker's but still falls.

The base is what makes this precise. If rewards go only to bonded units and only part of the supply is bonded, then the reward rate paid to stakers is higher than the overall inflation rate by roughly the reciprocal of the staking ratio. A chain issuing a small percentage of supply annually can advertise a much larger nominal rate to stakers when a minority of the supply participates, and the same chain's rate falls as participation rises with no change in the issuance rule. The advertised rate is therefore partly a statement about how many other holders are staking.

A hypothetical case makes the relationship concrete. Take a chain that issues new units equal to two percent of supply over a year and pays all of it to bonded units, with half the supply bonded. Stakers collectively receive two percent of supply spread over half the supply, so their nominal rate is about four percent. A staker's proportional share rises by roughly two percent over the year, and a non-staker's falls by roughly two percent. If participation later rises to four fifths of supply with the issuance rule unchanged, the nominal rate falls to about two and a half percent while the real change in a staker's share is still about two percent, because the same issuance is now spread across more bonded units. The headline moved by a third; the substance did not move at all.

Burning changes the sign of the correction. Where fees are destroyed, the relevant figure is net issuance, and a chain whose burn exceeds issuance over a period had negative supply growth for that period. Substituting net for gross issuance in the subtraction is what makes the resulting real rate meaningful, and it also makes the real rate depend on network usage rather than only on the schedule. It also makes the real rate a backward-looking measurement over a chosen window, since usage in the next window is not known, and the choice of window is a methodology decision that changes the published number.

Where the reward actually comes from

The distinction between the first row and the rest is the substantive one. Issuance-funded rewards move claim from non-stakers to stakers and create nothing. Fee-funded rewards represent users paying for a service, which is genuinely external to the holder base and is closer to the intuition most readers bring from equity. MEV sits awkwardly in between: it is real income to block producers and it is extracted from other participants' transactions, so it is a transfer within the market rather than a payment for a service.

This is why chains describe the fee-funded proportion of validator income as the more durable component. A protocol whose rewards are entirely issuance-funded is redistributing among holders; one whose rewards are substantially fee-funded is passing along what users pay. Both can be perfectly functional designs. They are not the same statement about where value comes from.

Two incompatible uses of "real yield"

The term carries two meanings in circulation, and they are not reconcilable. In the sense used above, it means a nominal rate less supply growth, by analogy with a real interest rate less price inflation. In the other, common in application-layer protocols, it means rewards paid out of protocol fees in an established asset rather than in newly minted governance tokens, and there it is a claim about the funding source, not about a subtraction.

Both usages are established, so the term alone is ambiguous and the underlying construction has to be checked. On this site, real staking yield is the first sense: nominal reward rate adjusted for net supply growth, computed in units, with the method stated in methodology.

What the adjusted number still does not capture

A real staking rate computed in units is not a return. It says nothing about the asset's price in any currency, and a positive real rate in units is fully compatible with a loss measured in a currency, as the price term dominates the reward term over most horizons. It also ignores several deductions that are real: operator commission, the tax treatment of rewards in the holder's jurisdiction, the cost of the unbonding period, the possibility of slashing, and, for staking through a protocol, smart contract risk.

Compounding conventions add another gap between quoted figures. Rewards on some chains accrue continuously and are automatically restaked, on others they accumulate until claimed and earn nothing until they are. A rate quoted as an annual percentage yield assumes reinvestment that may not happen by default, and a rate quoted as a simple annual rate on the same protocol will be lower for no economic reason. Comparing two published rates without checking the convention compares two different calculations.

Finally, the rate itself is not fixed. It varies with participation, as the arithmetic above shows, with usage where fees are part of the reward, and with governance decisions about the issuance rule. A published figure is a recent measurement, not a contracted rate, and treating it as the latter imports an assumption from fixed income that does not hold. The staking ratio is therefore worth reading beside any rate, because it explains a large part of why the rate is where it is.

Annualized inflation, the burn-adjusted rate, and the nominal and real staking rates appear together on the staking pages so the subtraction is visible rather than implied. The next lesson turns to the decisions that set these parameters in the first place.