For the security of its consensus algorithm, Tezos relies on bakers that place security deposits, equal to about 7.6% of the supply of tez, and attract delegations. Those bakers are compensated for the work they are doing and the bond they lock up by receiving block and endorsement rewards.
But how does the protocol know-how just much to pay bakers? The answer is, it doesn’t. In order to deal with this, a reward is paid, about 4.6% of the supply is distributed as a reward to bakers every year. Note that the security deposit used to be about 10% of the supply and the reward 5.5% but the absolute numbers have not been updated to grow with the supply since launch.
A 4.6% reward paid to people putting 7.6% of stake is effectively a 60% a year return on the stake, before node operating costs. This is a high reward. However, because bakers compete to attract delegations that do determine who gets rewards, much of this return ends up being paid to delegators. Assuming bakers pay about 85% of the rewards to the delegators, and with 75% of the network delegating, that’s about 5.2% reward per delegate against 4.6% inflation. Net-net, this is equivalent to about 0.6% deflation.
The way the network determines how much security costs is via this emergent mechanism of payment to delegators. Look at it as a refund for excess inflation.
There are two ways though in which this isn’t exactly equivalent to deflation:
- There are nominal effects at play that make people value rewards even if they are inflationary. This is a true free lunch in the sense that it helps with SoV status.
- Depending on jurisdiction, the tax treatment can be unfavorable.
The first is an argument in favor of this model, the second against it.
A few more arguments against it are:
- It could potentially create a compliance burden to operator bakeries
- It means there’s an opportunity cost to not delegating a tez, which becomes an issue for applications where tez is pooled. ctez solves that problem, but it adds a layer of indirection.
Is it possible to design a system where the cost of security is determined directly, without this mechanism? Yes, and it wouldn’t require a huge change to the codebase, although of course it would have huge consequences on the baking ecosystem and on the economics of tez.
The way this could be achieved is as follow:
- Decree that tez at risk (i.e. held as a frozen bond by the baker) carries twice as much block creation right as a delegated stake.
- Set the global inflation rate to be (2/(100 x))^2 where x is the global fraction of tez staked (at risk). So if bakers altogether place 20% of the whole supply at stake, the global inflation rate would be 0.5% per year. If they only place 10%, it would be 2% per year to incentivize more stake to be placed, if it’s 30% it would be 0.22% because that’s plenty of economic security and there’s no need to inflate all that much to attract stake.
You can pick any curve you want here, the general idea is to find an equilibrium between amount staked and inflation. If we have a rough idea of the target amount staked, this will generally pick the lowest possible inflation to be around that value.
Of course, this means that delegation and payments to delegates would become a much smaller part of the staking experience. Out-of-protocol custodial arrangements to delegate would become more common as opposed to simple delegation.
There are pros and cons to doing this, but I’m writing it so that there’s a concrete proposal out there that can be debated. I’m often asked about the topic and whether inflation could be smaller. If that’s the goal, this is how to do it. There are some benefits:
- tax efficiency
- simpler narrative, no need to explain non-dilutionary inflation
- better composability of tez for defi
- can simplify baking
There are some costs:
- loss of reward narrative
- changes the role and landscape of delegation
- no transfer from inactive delegators to active delegators
Personally, I lean towards this being positive on the balance and have for a couple years, but this is a big change to the economics of the chain and it’s worth a thorough think about.
In the meantime, it would still be wise to increase the bond requirement back to over 10% or more, and consider whether to increase back the issuance in conjunction or not.