EIP-7: Redesigning coupons - spreading out the promise of expansion

[alexl]

I’ve already touched on several issues I think are manifesting with the coupon system in the server. I’ll repeat the key points.

Fundamental issues with coupons as a deflation mechanism

1. If odds of an expansionary growth in the near term don’t look very high, which is when burning is needed the most, coupons are mathematically a bad deal. This is evident when you compare it to options trading or use any other bet pricing model.

2. It’s a very unstable burn mechanism. Burning to earn a premium means that there has to be a strong expectation that the required contraction will be followed by an even bigger expansion of the supply. (if we have to do -X, then buying a coupon is betting on +X*premium). This assumes every bearish period is quickly followed by not only a bullish period, but a record breaking one. If it’s not record breaking, then on average, for the -X contraction, less than +X in coupons has been paid out. The requirement for it to be record breaking can disappear when a lot of coupons expire, but this does not help their math. Furthermore, this effect is multiplicative, and even if it works out in one instance, it gets increasingly less feasible after every cycle.

3. One obvious idea might be to try to improve the lucrativeness of coupons by steepening the premium curve, i.e. promising a bigger payout. This does not improve the math at all either. Again, let’s say we have to buy out X in coupon value. Since we agreed premium1 is a bad deal, we’re doing it at a bigger premium2. Now, with the same math as the above, we still need an expansion worth at least Xpremium1. If that’s the amount of expansion we get, premium2 coupons will have ended up worth exactly as much as premium1 ones, except, instead of a fair distribution of the return, whoever uses the most tricks to redeem first will get the full payout. If we reach an expansion of Xpremium2, only then did the higher premium accomplish its goal of being more lucrative. However, as previously noted, the higher amplitude of the quick recovery assumption, the increasingly improbable a full reimbursement becomes. Each linear unit of increase in promised premium is exponentially less likely to realize. At some point increases in premium become nigh-worthless, if not an anti-feature, cause it introduces more random chance as to who manages to cash in on the limited expansion pie and who doesn’t.

4. Rolling coupons might be a good way to buy time, but selling a bad bet at an additional premium, to buy the same bad bet makes it even more negatively skewed.

Strategy Proposal

Draw inspiration from the traditional financial system. Ever since hard backing has been depracticed, careful peg balancing is what currency is. Governments have figured out a multitude of instruments of nudging their currency in the right direction. Of course, there are several fundamental differences between the two systems, but I believe a careful analysis into which strategies could be reused is necessary to achieve a more mature system.

The near term, practical design change

Replace coupons with a version of “treasury bonds” and systematic debt towards ESD burners. Instead of assuming a brief and explosive recovery in the near term, spread it out, and assume a net recovery over a longer period instead. Do this by creating debt that is paid out at a premium. Make it a stable reward for removing ESD out of the supply. A lucrative offer for both existing and, more importantly, new capital that only gets better if the market is in trouble. A guaranteed, if necessarily gradual return with interest is much more desirable in a bear market than a near-term bet. In traditional markets, bonds are valued more highly during recessions.

This still has an assumption of eventual inflationary growth, but a much safer one, and one that governments make and enforce with traditional currencies.

To be determined

How premium is calculated, and the full duration of the bond. How frequently yield is paid out. Whether it’s better to have a debt balance sheet, or tradeable debt tokens. Whether to have one kind of treasury bond or several. Whether to leave room for interest rates to also be adjustable according to market conditions, as they traditionally are, or not. My guess is it we’d go with the simplest option for most of these for now, but refining some of these parameters over the long term may be a good idea.

Also TBD is what happens with paying out yield during contractions. I would say keep them lucrative and guarantee a payout. If paying out during contraction is necessary, we’ll have to compensate by buying more bonds. If this is too much of a burden, I think we’ll have time to realize it and react. It’s a much less explosive system.

What this doesn’t solve - a fundamental issue for a different EIP

The system is way too focused on total money supply. The common misconception about currency is that this is the only driver of inflation/deflation. In reality, it’s a multi-parameter formula, and the only way to manage and adjust the peg is to monitor and react to every single parameter in the equation.

Even if deflation is fixed, if exclusively bonded money ends up being deflated, the accomplished deflation is 0 as the V in the formula below, for bonded money, is 0. There’s still a dependency on both positive influx of new capital and low selling pressure to keep the system stable long term. At some point, we need to focus a lot more on movement patterns of the ESD supply. I believe we could make it more resilient to stagnant markets, but this needs a much longer, separate discussion.

[sabretooth]

I’m trying to understand the concept of a “bond” in the current system of coupons.

So would a bond be a coupon with no expiry where the rewards are vested over a period of time to simulate the paying of interest?

i.e. You lock up 1000 ESD in a bond for 100 Epochs, The reward is 10 ESD, which unlocks linearly over those 100 Epochs.

Is that the way this would work in practise?

[alexl]

That’s mostly how I’m imagining it. They would have to yield back more than the initial $1000. Potentially how much more could be calculated like coupon premium - they get more lucrative as the buyout is more desperately needed, although maybe just skipping straight to a fixed rate would be a good way to avoid price manipulation. It might be good to use market rates of similar “saving” options in defi: 12-ish% APY?

The period would have to be longer than 100 Epochs. It should be roughly as long as the longest contraction we want to be tolerable for the system.

[Beepidibop]

Came here after reading TIP-6.

My one criticism of this approach is as outlined in the proposal already – “This still has an assumption of eventual inflationary growth”

I’m going to throw around some arguably bad ideas out here in hopes that it could inspire an actual method that would convince people to voluntarily reduce ESD supply, even if eventual deflation is guaranteed.

1. We could argue that gambling is a big enough industry that proves people would take any bet, even when it’s proven that the house will always win. With this type of philosophy, we could continue the couponing system in hopes that there will always be irrational actors in a big enough market.
   • Of course, this would ultimately mean that wealth will become concentrated towards people who don’t participate in the couponing mechanism. The system might eventually collapse anyway when rational actors outweigh irrational actors.
2. Maybe DAO lockups can be lengthed (before retraction, say maybe 300 epochs as mentioned elsewhere. But then we’d need to solve the long epoch’s impact during expansion periods) and a haircut is automatically applied (burned) if DAO holders exit when ESD is below peg. This would reduce DAO exit rate, effectively reducing the circulating supply. Entering DAO would then be a decision on whether you think ESD would expand or contract in the next 100 days.
   • This would potentially limit participation in DAO. But the less people in DAO, the more rewards there are pro rata during expansion periods. However, if a contraction period is forseen more than the lock-up period ahead, there would be an exodus from the DAO.
3. We can have multiple methods in place during a recession. Keeping both the couponing and the newer systems.
   • This might be too complicated for new comers and be unpredictable.

[Vamoose]

Longer coupons might help. Nobody is going to buy coupons that they think will expire, and a longer coupon period will cover more estimates of when they might be redeemable. Maybe we could even make it dynamic somehow, where the coupon period will lengthen and shorten automatically depending on whether people are buying enough to make the system happy.

Reducing the DAO exit rate I think is a great idea; it ought to be dynamic. The DAO and its expansion generates a big pool of value, but there might not be enough faith in ESD among people with USDC to support that value. Encumbering the ESD in the DAO with longer lockup times as ESD falls below the peg could stop a bank run from the DAO to Uniswap, without causing problems for people using ESD outside the DAO. If USDC holders are willing to buy ~$5000 ESD per day without the price going below 0.97, say, then that’s the maximum off-rate we should allow from the DAO when below the peg by that amount.

The question there is how you design the system to determine the right lockup time from the Uniswap statistics.