Burn unstaked tokens instead of rewarding staked tokens

When a delegator receives staking rewards, this will probably be seen by tax authorities as income. In reality though, the delegator is simply maintaining their financial position in the system, as their tokens are being inflated along with everyone else. Unless this is dealt with somehow, it will result in the system leaking value out to taxes at a compounding rate. There’s a good thread about this possibility here: https://twitter.com/ceterispar1bus/status/1113116321925877760

Here’s a good summary of the issue:

Yes. Said the same to someone last week. If everyone stakes, no one gets richer, but the taxing authorities still demand their cut of the income. The higher the “yield,” the higher the portion of the market cap that gets eaten by gov each year.

— Ben Davenport (@bendavenport)

A simple solution suggested in the thread:

The better tax answer is to forfeit (burn) unstaked coins. But again, no one ever asks me before designing these things, so this is purely hypothetical.

— Alice Tax (@towneslaw)

What this could look like is that all unstaked Atom accounts are reduced by the “inflation rate” every block and there would be no inflation rewards. This would be weird but Atoms are not meant as a general medium of exchange so I think Atom holders can deal with it.

Maybe there’s an accounting method that can properly deal with this, but if there isn’t, then the system will slowly be drained of value by taxes on income that isn’t really income.

There’s also the possibility that there is an accounting method to deal with this, but it is complicated. For example, if you had to sell and rebuy all your tokens every year to realize a loss to offset your staking “income” (I’m not an accountant, this is purely hypothetical). This would push people towards using custodial validators who could handle this operation for them, making the system centralized.


I am imagining a buyback&burn of stock by its corporation. It reduces the total supply of the stock which usually results in risen stock price.

And risen stock price also results in tax, so I see the same result in both way.

Is the taxation of dividend and trade profit have significant difference in US tax law?

Profits from stocks held for less than a year are taxed at your ordinary income tax rate. Ordinary dividends earned on your stock holdings are taxed at regular income tax rates, not at capital gains rates.

It seems like both have same rate of taxation.

But, if the reward is dealt as interest rate, the tax rate will be lower than the two described above.

There are a lot of complexities here.

How do you account for sub-uatom burns. For inflation and fees we track sub uatom rewards and then only truncate on withdrawl.

Iterating over the entire account set might be acceptable if we amortize for a large number of blocks and do this infrequently.

we probably need to devlope some sort of amortization strategy so the cost of iterating over every account is spread over many blocks.

When a delegator receives staking rewards, this will probably be seen by tax authorities as income.

It will be seen as capital gains and what is the problem with that ?
I don’t really get the purpose of this question. Are you writing a proposal whose sole goal is tax avoidance ?

I am not a tax expert but would like to share my dummy’s thoughts. Let’s use an example. If the total intrinsic value of 200M atoms is USD1,000M, this means each atom worths USD5.

Assuming the intrinsic value of the tokens have not changed after 5% inflation. So the total number of atoms is now 210M while the total intrinsic value is still USD1,000M. Each atom worths an intrinsic value of USD4.76.

Assuming there is only one guy holding all atoms and he has staked all atoms and earn all the 5% inflation. Assuming the market price of atom fully reflects the intrinsic value. So the 5% of atoms, which is 10M atom is subject to income tax, which is now worths USD47.6M.

At the same time the price per atom has dropped from USD5 to USD4.76. The difference of USD0.24 is now an impairment loss (foreign exchange loss?) and should be tax deductible depending on jurisdiction. So this means the originally 200M atom has suffered a foreign exchange loss of 200M * USD0.24 = USD48M.

In this case, the guy has actually experienced an operating loss of USD0.4M and should not be subject to income tax.

1 Like

You rounded, there would be no loss. It’s 47.6M in gains and 47.6M capital loss, not 48M.

1 Like

No, rewards given by the system to stakers will be seen as income. I have no problem getting taxed on capital gains for tokens that have appreciated between the time I bought them and the time I sold them.

The purpose of inflationary rewards to stakers is to take money from the people who are not staking, and give it to people who are staking. This is to incentivize people to stake. I think we all agree on that, right?

Taxes aside, this mechanism is economically equivalent to one where you simply burn the Atoms of those who don’t stake. But it is implemented in such a way, that without clever accounting, delegators will be liable for income tax every year, rather than capital gains when they sell.

You bring up “tax avoidance”. What most people don’t like about tax avoidance is not the fact that less taxes are collected. If you want the government to collect more taxes, just vote for a tax increase. What people don’t like about tax avoidance is that it involves clever accounting which is out of reach of ordinary taxpayers. Because of this, rich people often end up paying less in taxes.

The current inflationary rewards create this exact situation. As you can see, people have gotten started on the complicated accounting strategies already in this thread. It’s not going to be possible for the average person staking Atoms to pull this stuff off, especially if they only have a few hundred $ worth of Atoms anyway.

I predict that large stakers who can afford sophisticated legal advice will be able to offset the income taxes using various mechanisms. But small stakers will be stuck paying full tax.

Worse, it could incentivize centralization. Imagine if you are a small Atom holder, and you could make 7% staking your atoms. This would be ~5% after income taxes. And you’d have to sell some of your Atoms every year to pay these taxes. Alternatively, you could transfer your Atoms to a fund which would take custody of them, get the staking rewards, and offset the income taxes through one of the many complicated but completely legal methods that are available. Then your shares in this fund would appreciate every year by the full 7%, and you’d only have to pay capital gains tax when you sold them.

I don’t want to “avoid taxes”, I want to make it so that large and small Atom holders pay the same taxes.

1 Like

Yeah I agree with your goals here.

What needs to be done is

  1. Figure out a coin destruction algorithm that doesn’t require iterating over every account.

  2. Figure out how this will work with exchanges. Do they need to run a version of the coin destruction algorithm themselves to distribute the destroyed coins to each customer account?

If what I suspect to be the future where we hit 95%+ staked coins, I would recommend stopping inflation through governance entirely at some point and this become moot.

1 Like

Yea, the ideal scenario would be to have tx fees be healthy enough to incentivize stakers on their own.

I’ll think about the coin destruction. It probably only needs to be done before coins are spent. But I agree that it could wreak havoc on custodial services by breaking a core assumption. Maybe it’s best that we just get to a post-inflation system ASAP by having lots of apps and tx’s on the hub.

I went on about this problem for a while back in 2017 when I first started learning about taxes for real. I also always strongly favoured demurage to inflation because even though “in theory” they might have the same effect, in practice the psychology of losing coins vs getting new coins is profoundly different, and obviously the market doesn’t correct the prices as quickly as the inflation is issued. So you have my heart here :stuck_out_tongue:

In any case, making this work for custodial services was always cited as the main challenge. Though I think it would just mean that such services would require one address per client, and wouldn’t be able to pool funds. Is there more to it then that? Of course users of the service would need to know they are subject to the demurage rates still …

Custodians aside, if we had a UTXO model it would be a bit simpler - coins could be deducted whenever you go try to spend the output according to how old it is. Of course, since the effective demurrage rate changes according to the amount stake, you’d have to partition the age of the UTXO into intervals according to the rate in each interval, and then get a final answer.

For non-UTXO system like ours, perhaps we could burn coins every time the account balance changes (whether it’s debited or credited). Otherwise the averaging work might be too complex.

I’m actually hopeful something like this would be doable without too much complexity. Let’s try to get a real spec written!

1 Like

Okay I think I have design for a system that avoid the tax problems but avoids the structural problems with naive demiurge but allows treating staking rewards as capital gain.

High level overview of the proposed system.

  1. Staked coins are converted to a discrete entity called a bond. The Initial voting power of a bond is proportional to the number of coins burned to create it. A bond is permanently attached to a validator. The voting power of a bond increases on every block by some percentage. Slashing reduces the voting power of every validators bond.

  2. Bonds can be directly transferred between users via a bond transfer transaction.

  3. Bonds can be liquidated into staking tokens. When a bond is liquidated, the voting power disappears and a proportional number of staking tokens are minted.

This emulates demurrage because voting power become more expensive in terms of available staking tokens permanently.

It solves the liquidity problem with earlier versions of the demurrage design where voting power becomes too expensive to purchase.

The accrual of voting power to bonds should not be correctly treated as income because it is non-fungibly bound to the slashing risks of the validator and continued operation of the validator.

But I believe the system would still more less work with voting power being constant and only the liquidation value accruing block by block.


How is this different than if inflation was auto-bonded?

1 Like

It’s mostly about how the UX is presented which unfortunately is relevant to tax authorities.

it would also be effectively opt-out autobonding which opts into the income tax regime.

1 Like

And what’s the relevance of

Bonds can be directly transferred between users via a bond transfer transaction.

It makes the bonds an appreciating capital asset where gains can be realized without destruction. I think it strengthens that case that native staking rewards are not income.

So if I understand correctly, the differences from the current system basically boil down to:

  1. inflation is auto-bonded
  2. delegations can be transferred to another delegator

And the goal of (2) is to convert the bond from a non-transferable and income generating account to a transferable and appreciating capital account.

It’s a cool idea, but I’m not sure the transferability really makes the case for it not being income. For instance the closest analogy I can think of are shares where the dividends are paid in the shares themselves. I believe those dividends are still treated as income. So it seems that so long as there are increases in the native denomination of an asset, it may be deemed as income. Unless I’m missing something?

Also, I’d be a bit concerned that transferable bonds might have unintended consequences on the economic security since it means they can change hands without going through an unbonding period, and what was once a commitment to wait before you could liquidate becomes no longer so.

As for

It solves the liquidity problem with earlier versions of the demurrage design where voting power becomes too expensive to purchase.

Are these designs written up anywhere?

But there is no increase in the native denomination of the bonds. 1 bond always equals 1 bond. It’s just the denomination of unexercised liquidation option that is increasing.

Transferable bonds do not weaken economic security because the bond is still attached to the validator and still slashable. Liquidation is the only way to recover a fungible asset with no potential slashing risks.

Isn’t the voting power increasing?

And maybe not the economic security per-se, but something about it’s time preference. If 100% of atoms were bonded, you’d have to wait 3 weeks (at least in protocol) before you could acquire any stake. If bonds were liquid, you could get them right away. Not sure exactly how this should be interpreted w.r.t some definition of security. Possibly it’s irrelevant, but just pointing it out.

you can’t change the validator that bond is attached to in my proposed construction. So selling bonds doesn’t change voting power.

I’d probably suggest removing redelegation in this design.

Some reasoning that this might not help.

You have to pay income taxes on imputed interest if nothing actually pays out.

https://www.sec.gov/fast-answers/answerszerohtm.html H/t Arthur Breitman