Paramater Change Proposal: Inflation Rate Change

Cosmos Hub Inflation variable modification: Inflation Rate Change

Quick Summary of issue

There are 6 main variables that control the maximum, minimum, & change of the atom inflation rate for the cosmos hub. These variables can be found inside the mint module via the attached hyperlink: CosmosParametersWiki/Mint.md at master · gavinly/CosmosParametersWiki · GitHub

In this proposal we will be looking at adjusting the inflation rate change variable.

  • When the cosmos hub inflation dynamics were originally designed, the goal was for the hub to go from the minimum rate (7%) to the maximum rate (20%) in roughly one year after a shock unbond had occurred. Thus, for the variable “inflation rate change” a value of .13 was chosen. Unfortunately, in practice the variable didn’t work as intended due to the fact the change in the inflation rate for the hub is proportional to the Target Bonded / current bonded ratio.

How to select “inflation rate change” AKA Maximum Slope of the inflation curve

The purpose of changing the network’s inflation rate is to protect it from unbonding shocks that threaten to compromise the security of the network. When the bonded ratio gets below the goal bonded ratio (currently at 66% on the hub) the inflation rate & effective yield of staked atoms goes up in order to incentivize holders to bond new ATOM’s, providing additional security to the network. Inversely, if we regain a desired amount of staked tokens, the yield will drop, tdecreasing effective yields for all delegators. Ideally the inflation rate starts changing fast, optimizing network security over monetary hardness. To figure out what an appropriate selection for the cosmos hub would be, I created excel sheets to run through all the different scenarios to find what value made the inflation rate react in an optimal manner. I came to the conclusion that 1 (AKA 100% per year is the maximum slope of the inflation curve) gave the most ideal characteristics, with the added benefit of simplifying the equation. I’ll give some examples on how it would react. Since unbonding is what we are protecting against, I will look at flash unbonds while the inflation rate is at the floor due to that being the #1 time of vulnerability.

Scenarios of Shock Unbonds #1. Cosmos hub Bonded ratio flash crashes to 60%, which is not bad, but still 6% below the target. With our current variable of .13, it would take approx. 9.6 years to make it to the ceiling rate of 20%. On the other hand, if this variable was 1, the hub would reach it ceiling 17.2 months after the unbond shock (assuming bonded % stays at 60% the entire time for simplicity).

#2 Cosmos hub Bonded ratio flash crashes to 50%, starting to get a little scary, but only 16% below the 66% target. With our current variable of .13, it would take aprox. 4 years to make it to the ceiling rate of 20%. On the other hand, if this variable was 1, the hub would reach it ceiling 6.4 months after the unbond.

#3 Cosmos hub Bonded ratio flash crashes to 35%, NOT GOOD!!! We need to get more atoms staked ASAP! With our current variable of .13, it would take aprox 25 months to reach the ceiling rate of 20%. On the other hand, if this variable was 1, the hub would reach its ceiling 3.3 months after the unbond and the inflation rate would be increasing at a rate of 3.91% per month.

All of the data above, graphs, and much more can be found at Cosmos Inflation - Google Sheets

Possible Risks / Benefits

I will split this up into two sections, doing nothing & doing the proposed changes.

Inflation rate Change (Max Slope of inflation curve)

1a) Doing nothing Risks / Benefits: In this case, I do believe there are risks to do nothing. Currently the inflation rate of the hub is not very reactive to unbond shocks (as seen above). This can lead to network security risks if an economic actor with bad intent wanted to try and destabilize the hub. On the other hand, changing any variable within a system can cause unwanted consequences. There is no way to get rid of this risk entirely, luckily the inflation mechanics are fairly simple and easy to model with excel spreadsheets. Additionally, none of these changes will affect the consensus mechanism, they will just impact the minting of new atoms.

1b) Changing rate to 1 Risks / Benefits: As stated above, there are always risks to changing variables within complex systems. Changing this variable will undoubtedly greatly increase the reactivity of the hub to unbonding shocks. Overall, this seems very beneficial due to the extremely slow reaction time of the hub currently. Also, it brings the hub closer to the original goal of 1 year from the min to the max in a unbonding shock scenario. Due to the fact the inflation curve slope is proportional to the goal vs current bonded ratio, it isn’t possible to hit the original goal directly, but this brings it much closer to reality.

Conclusion

I know there isn’t a right answer for maximum slope of the inflation curve. I fully expect this to be fine-tuned over the coming years / decades. Future proposals may adjust this number by a factor of (.5 – 2+) but this gets us in the right ball park.

Many thanks to Ethan Buchman for the insights provided during drafting.

11 Likes

I support this proposal and think it is long overdue for these variables to be checked over and updated. I agree that the current Inflation Rate Change is too low and support raising it to 1.

7 Likes

I also support this proposal. Seems to be a no-brainer to protect network and allow quicker adjustments to staking rewards to incentivize new/current stakers.

1 Like

I also support this and vote yes. From what I read it will definitely incentivize current stakers to keep their ATOMs bonded and also encourage new investors to stake their Atoms. I plan on buying more ATOMs each month and of course staking them. Thank you for this proposal.

1 Like

Supporting this one. But, I think to help on this subject, there is a need to add more validators to join the chain. Currently, it’s not possible for newcomers to take place.

1 Like

I mean, I guess I’m not against it, but it feels like that level of shock might not be fixed even by slowly increasing the rate of minting new atoms over 3 months. I would imagine just the “shock” even at minimum rate (7%) would be enough to get more atom minted quickly (because at that point, a 7% inflation rate of new atoms would be split between only about a third of all atom, which would make the APR for the ones still holding on over 20% which would be extremely attractive alone. While of course at max (20%) split between a third of the atom, giving the stakes a 60% APR is obviously even more attractive, I’m not sure that I buy that it would naturally get to that point… people would start jumping on naturally far before it reaches that, no matter the speed, unless there is some more nefarious manipulation at play, and I’m not sure this proposal would be able to defend against whatever that particular issue might be, either.

I’d vote for it, if only because it means I’d get a big ATOM windfall if some rich guy decides to buy two-thirds of all ATOM and not stake it… but I just think I don’t foresee a viable example of what this specifically would counter.

With both Gravity DeX and Osmosis coming soon, I definitely think this needs to be addressed. I am in support.

2 Likes

I have updated the github draft with @JoeDirtay 's latest and will go on chain tomorrow.

Nice to see this finally on chain! Definitely important to address.

I would imagine just the “shock” even at minimum rate (7%) would be enough to get more atom minted quickly (because at that point, a 7% inflation rate of new atoms would be split between only about a third of all atom, which would make the APR for the ones still holding on over 20% which would be extremely attractive alone. While of course at max (20%) split between a third of the atom, giving the stakes a 60% APR is obviously even more attractive, I’m not sure that I buy that it would naturally get to that point

This is a good point, and we’d benefit from more analysis of the dynamics here - if the stake drops to 35% and APR is rising fast, stake is unlikely to be at 35% for long. But given the possibility for other opportunities to yield ~20% for some time, we probably do want the hub to be more responsive regardless. I can also imagine in the future we may want something more sophisticated here like a PID controller, but this adjustment is a good start.

Slightly related, Delphi team recently proposed a PID controller for money markets (Dynamic Interest Rate Model Based On Control Theory - Delphi Digital)

1 Like