Inflation

Token inflation and price inflation are different measurements

Central banks define inflation as a rise in the general price level of goods and services, measured against a basket and driven by demand conditions, supply conditions and expectations.¹ Token inflation measures one thing: how much the supply of one asset grew over a period. The two share a word and very little else.

The confusion is not harmless. Describing 8 percent issuance as inflation invites the reader to reach for macro intuitions, that it erodes purchasing power at a predictable rate, that it is a policy lever, that a rate rise somewhere can tame it. None of that transfers. Token issuance dilutes a claim on whatever value the network captures. Whether it shows up in price depends on demand, which the schedule does not control and cannot promise.

So use the precise words in your own documentation. Supply growth is what you are measuring. Dilution is what it does to a holder. Inflation is the macroeconomic word and it belongs to the macroeconomic measurement.

Gross issuance, then net of burns

Gross issuance is what the schedule mints. Net supply change is that number minus whatever the protocol destroys. EIP-1559 made the distinction concrete for Ethereum by introducing a per-block base fee that is burned rather than paid to a block producer, and the specification carries a section headed ETH Burn Precludes Fixed Supply for precisely this reason.² The burn scales with usage, so total supply becomes a function of activity rather than of a schedule.

The design consequence is direct. Any chain with a usage-linked burn has a supply number nobody can forecast from the emission schedule alone, so publish both. Gross issuance is what the protocol committed to. Net change is what happened. A project quoting only the flattering one of the two is telling the reader which one it is.

The reverse move deserves more scepticism than it gets. Buyback and burn programmes funded from a treasury are often presented as offsetting inflation. They offset the number, not the mechanism, and they run for exactly as long as the treasury does.

The denominator is where the number gets massaged

The same issuance produces three different headline rates. Measured against max supply the base is largest, so the rate is smallest. Against total supply it lands somewhere in the middle. Against circulating supply, the only base reflecting the tokens a holder is actually diluted against today, it is largest.

We treat presenting inflation against max supply as a disclosure failure rather than a presentation choice. Anyone reading the table professionally recomputes it against circulating supply in about a minute, and discovering that the published figure needed recomputing costs more credibility than the higher figure ever would have.

State the base every time. Eight percent of circulating supply is a sentence. Eight percent is not.

Real yield is the reward minus the dilution funding it

A staking reward paid in the same token being issued is not income until the issuance is netted out of it. If supply grows 10 percent over a year and a staker earns 8 percent, the staker's share of supply has fallen. The arithmetic is 1.08 divided by 1.10, roughly negative 1.8 percent, and no framing turns that into a yield.

That is why a headline staking APR can be negative in real terms, and why the comparison a reader needs is reward rate against supply growth rate rather than reward rate alone. Where the reward is funded from fees rather than from issuance, the arithmetic changes and the number survives the netting. Almost nobody says which of the two they are doing. Say it.

In our view the second-order effect matters more than the first. Once a reward is understood as dilution defence rather than income, staking becomes something a holder does to avoid losing ground rather than to gain any. Participation climbs, the reward per staker falls, and the design has bought a high staking ratio without buying anything else.

On some chains inflation is a dial, not a schedule

The Cosmos SDK mint module publishes no curve. It recomputes an annual inflation rate every block from the ratio of bonded to total supply: below the goal bonded ratio the rate climbs toward a maximum, above it the rate falls toward a minimum, and at the goal, 67 percent in Cosmos Hub, it holds constant.³

Those parameters are governable and they move. A Cosmos Hub governance forum post from January 2024 published the values then operating on the chain: an inflation ceiling of 10 percent, a floor of 7 percent, a goal bonded ratio of 67 percent and 6,311,520 blocks per year.⁴ The module's own shipped defaults carry a 20 percent ceiling, so the live band was roughly half the default.³ Read a chain's operative parameters, not its documentation's defaults.

If your design puts inflation under governance, say so plainly and name the band. A governed rate is a legitimate choice. A governed rate described as fixed is a disclosure problem waiting for its first vote.

How we want an inflation number disclosed

Four lines, and they fit in a small table. Gross issuance for the period. Burns for the period. Net supply change. And the base the percentage is computed against, named rather than implied. Add a fifth line if the rate is governable: the floor, the ceiling, and who votes.

That table does something a paragraph cannot. It makes the number checkable against chain data by anyone who cares, which is the only thing that makes publishing it worthwhile. It also forces the real question into the open, which is whether the value the network captures is growing faster than the supply is.