IIP-29 Activate DPI Intrinsic Productivity

[Kiba]

iip: 29
title: Activate Intrinsic Productivity
status: Proposed
author: Kiba Gateaux (@kiba)
created: 2021-04-01

Summary

This is a concrete and actionable plan to start earning yield on 2 assets within DPI - SUSHI and YFI.

The yield strategies proposed avoid all the issues discussed previously - namely liquidity issues from lending and potential loss of funds by using them as backstops.

Utilizing ~12% of our assets under management in two very simple, low risk strategies we can provide $1.7M in additional income to DPI holders.

Motivation

A brief history on intrinsic productivity discussions for Coop newcomers:

• Origins
• Initial explorations 1 + 2 + 3
• INDEX as staking backstop
• Community survey on intrinsic productivity
• First ‘real’ productivity proposal

The Coop hasn’t really made any progress in implementing intrinsic productivity since the first proposal by OA six months ago. I think the two main reasons are 1. We focus on putting the whole portfolio to work at once instead of picking the low hanging fruit 2. We only explored two yield options previously - lending and protocol level staking. Essentially we were trying to overfit acceptable yield strategies that worked for all assets at once instead of finding the best yield strategies possible for each individual asset.

The goal of this new intrinsic productivity strategy is to focus on as few assets as possible and earn as high yield as possible on them while living by the “boring” values of the Coop.

Also the main point of differentiation for all DPI competitors is that they earn yield on assets. By activating intrinsic productivity for DPI we solidify it as the premier index product in every possible aspect.

Description

Yield strategies:

1. SUSHI → xSUSHI → ~5% APY
2. YFI → yvYFI → ~20% APY (actual vault says 30% but used number from official stats dashboard)

Both of these strategies have all the following properties:

1. Not used as a backstop with risk of slashing
2. Not lent out and 100% backed at all times for redemptions so no liquidity issues
3. No loss strategy (e.g. LPing), always get back more tokens than notional amount in DPI
4. Passive set-and-forget with auto-compounding rewards accrued to yield bearing asset directly
5. No vesting schedule on rewards
6. Long term farms that don’t require crop rotations
7. Low smart contract risk with >$120M in yvYFI and >$1B in xSUSHI representing ~10% and ~45% of each tokens total supply respectively with no hacks or exploits to date.
8. Retain full governance rights of tokens while earning yield on them

If you go back through all the intrinsic productivity posts, these two strategies completely avoid all of the issues we had with utilizing underlying assets. Even better than avoiding all the issues with old strategies, they earn higher yield than any proposed strategy so far.

90% of all assets in DPI will be staked. Funds will be pulled from staking strategies to replenish reserves if less than 5% of notional amount is liquid leaving 5-10% of all funds available to cover rebalances and redemptions. Brian from Set said redemptions aren’t an issue because users would just receive SUSHI/xSUSHI/YFI/yvYFI instead of just SUSHI/YFI. Most importantly, we can retrieve all 90% back at any time without any risk to our positions or profitability so we should maximize yield on these two strategies. As far as I can tell from this chart, there hasn’t been more than a 6% drop in a token over a 3 day time period so this leaves us a very safe margin for rebalances.

DPI Productivity Revenue Projections

Asset	DPI TVL	% of DPI	% Staked	% APY	DPI Productivity Revenue	% Earned on DPI TVL
YFI	134,000,000	0.076	0.75	0.20	1,527,600	1.14
SUSHI	134,000,000	0.047	0.75	0.04	188940.00000000003	0.141

DPI productivity revenue = $134M TVL * 1.3% yield = $1,716,540~~

You can play around with these numbers here.

Yield Distribution

An important consideration for distributing revenue from DPI yield farming is making sure we do it in a way that doesn’t put DPI at risk of becoming a security. The original proposal avoided this risk by diverting excess profits to the Coop but after discussing, the community is mostly aligned on giving all value back to DPI holders and not overplaying the securities concern.

100% of farming proceeds will be returned back to DPI holders. Tokens earned from yield go into DPI’s vault, thereby increasing the value of each individual unit of DPI and increasing the allocation of SUSHI and YFI which then gets rebalanced out every month. This is how Set Protocol natively operates today.

“Returning all value to DPI holders is an obligation because passing as much value as possible back to the end user is a guiding principle of DeFi” (paraphrased from @BigSky7 and @trx314) and will help grow DPI unincentivized TVL. It also reduces complexity and smart contract risk of implementing IP versus the original proposal’s custom yield splitting scheme which lets us push IP to production much faster.

If we later receive guidance/warnings that DPI is a security because of IP we can always pause IP and evaluate new technical architectures based that are inline with this guidance.

Implementation Details

• Use assets in current DPI vault
• Use WrapAdapter for xSUSHI that already has an open PR.
• Create WrapAdapter for Yearn v2 vaults to access yvYFI
• 90% of all SUSHI + YFI will be put into farms.
• With Yearn v2 vaults, if you try to withdraw more than what is in the unutilized vault reserves you incur a fee. I asked Set team and they said this will be implemented by manager not in the WrapAdapter contracts directly so the Coop will have to decide what our acceptable “slippage” fee is. As far as I can tell there is nearly $100M or 80% of YFI in the vault held in reserve at the moment which is 20x more than we would be staking so it seems unlikely this would happen but we should have a contingency plan anyway.
• Farming program will run indefinitely until Coop votes to end or change strategies

Out Of Scope

• Establishing working group for IC asset management
• Farming strategies for other assets besides xSUSHI and yvYFI
• Smart contracts to manage farming positions

Changelog

• 2021/04/20

Proposals

Approve xSUSHI and yvYFI farming strategies

• yes
• no

Allocate 500 INDEX from Treasury committee for research and development bounty into Farming Manager smart contract

• yes
• no

[verto0912]

We have talked about intrinsic productivity for a long time. I don’t think anyone would argue that we don’t need it meaning that it all comes down to implementation. My understanding here is limited, so I will leave it to others to comment on the technical aspects of this proposal and help guide the conversation.

[Kiba]

There was a small issue what the yvYFI vault that was discovered. No funds were ever at risk, it was a minor accounting mistake in the smart contract. Users have been refunded and the fix patched already.

yearn/yearn-security/blob/master/disclosures/2021-04-02.md

[dylan]

Note: this post was originally made in #product and has been moved to #proposals / converted into an IIP.

[granddaddycool]

This seems like a well thought out plan. It’s more nuanced than prior proposals. I’m not an expert on the technical aspects of implementation but it seems worthy of further analysis.

[Thomas_Hepner]

@Kiba This is an excellent, very well-thought out post. I really like the incremental approach (i.e. one asset at a time) to intrinsic productivity. I think this is a simpler approach that can return yield to DPI and INDEX holders faster than trying to wait for the perfect intrinsic productivity solution.

I am a bit confused by the Yield Distribution section. I do not think it makes sense to reimburse streaming fees to DPI holders. In the traditional fund management world, it’s very common to have an expense ratio (i.e. what is paid to manage a fund) and a yield (i.e. dividend paid to investor).

In my view, all yield farming income should be passed back to DPI holders either in the form of income or through additional DPI purchases. The reason for this is that at this stage in the Coop’s development we should be targeting growth, not income. DPI still only has a market capitalization of ~$150m; the Coop will generate more profits by increasing the demand for DPI by offering yield than by extracting profits through yield farming.

At this point in time, I think it makes sense to pass back yield farming income in terms of additional units of DPI; this also has the impact of “reimbursing” the streaming fee (i.e passing back 1.3% yield with 0.95% streaming fee means I would be paid 0.35% to buy DPI on an annual basis).

[scott_lew_is]

great post kiba.

i would suggest being more aggressive than 75%.

perhaps we could do 90% with a public function to re-establish the ratio at 90% should it get outside a band from 87%-93%?

what do you think about adding aSNX to the list?

[Kiba]

@scott_lew_is

perhaps we could do 90%

I’m down. I thought even 75% was too much for the Coop’s conservative nature but if there is support let’s bump it up to 90%. As an ape I would do 100% and figure out rebalances in the farming contract which would bring in an extra $500k.

@scott_lew_is

what do you think about adding aSNX to the list?

aSNX is probably the best yield we’ll get for SNX that fits majority of the criteria. I don’t think Set Protocol has Aave v2 integrated yet so that would have to be added to R&D and increase time to deployment (cc @dylan plz fact check me)

@Thomas_Hepner Great point. Giving all returns to DPI holders is actually way easier and wouldn’t require new smart contracts to be written and would definitely help with growth.

It looks like I removed it before posting but the main consideration for only reimbursing fees and not giving all yield to DPI holders is that providing profits could make DPI a security which would really fuck up our adoption and business development. I took this from comments by @overanalyser and @reganbozman across a couple channels so they can explain more or correct me if I’m wrong. I personally don’t care either way. As a dev and author of proposal I’d prefer to give all yield to DPI since thats way less complex, can be deployed faster, and would grow DPI faster but that would be sacrificing long term success for short term gain (assuming my understanding of securities issue is correct)

[overanalyser]

I’ve voted no to this proposal for a number of reasons.

I like many features of the proposal including allocation of the income to reducing the fees of zero without becoming yield generating (I suspect that yield generation for holders would look more “security like”) and the fact that we can capture good yields on a couple of components so we don’t need to do everything in the first iteration.

So, why am I against it? I think it’s a combination of:

1. The majority (over $95 m) of DPI holders are currently paying 0.95% and unincentivised. So we have a pretty good product market fit. I suspect that many of these customers want something simple and easy to understand. Adding intrinsic productivity to DPI, may cause some of these customers to move elsewhere.
2. We are currently working on a number of external integrations that we believe would be good for the coop and our customers (exchange listing, MakerDAO, AAVE, sDPI). Changing the structure of DPI could a) make calculation of the fund NAV more challenging, b) delay these integrations.

I do still believe that Intrinsic productivity is a desirable goal of the coop. However, I think that the proposed SDI fund is a much better product to lead these efforts for a number of reasons:

1. The target customer is very DeFi aware and I would expect to be not be concerned about the added complications that IP brings.
2. The target customer is likely to be price sensitive and more likely to reject a streaming fee.
3. The methodologist is happy to include intrinsic productivity in the fund structure from launch.
4. The panned route to market is via a single AMM SDI:sUSD pool, and we will no be pushing for CEX listing, use as collateral
5. The fund will contain three core components at over 20% each (wBTC, wETH and DAI) which will have multiple options for yield generation.

As such, I would consider SDI to be a much better candidate for the initial implementation of intrinsic productivity.

One final comment; Recent product launches by PieDAO and BasketDAO have shown that there is appetite for yield generating products (which I think are likely securities). However, how much of this is driven by the generous staking rewards remains to be seen.

[trx314]

My take on this very important proposal is that capturing the productive value of these tokens is not an option, it is an obligation. We are talking about DeFi: over time, all of these tokens will have utility and be able to generate value to their active holders/stakers in one way or another.

By not capturing the productivity value of these tokens, the DPI product would basically track only a subset of the whole value that these tokens have. DPI will become a lot less competitive compared to other products capturing the full value.

I am therefore totally for activating the Intrinsic Productivity.

However, I agree with @Thomas_Hepner regarding the Yield Distribution, and this is why I voted “No” to the proposal. In my opinion, the DPI token should capture all the value generated by the assets composing it. The yields should not be in any way distributed to the Coop.

Make the DPI token capture these yields will make it more attractive and competitive because of the performance impact, but also have a positive effect on the Index Coop image (I, wearing the DPI holder hat, would really find it abusive if the Index Coop would generate additional revenue by using “my” assets to generate yields).

Regarding the concerns expressed by @Kiba and @overanalyser:

• risk for DPI to be considered as a security: I think this would apply only if the yields were distributed as cash flows (like a dividend). The other issue with distributing yield directly to holders is that it can become a taxable event. This is why I would rather reinvest the yields in the DPI assets. It could work this way: each month the yields are collected and used in the rebalancing process. It will increase the DPI performance over time by accruing the yields in the assets, without changing the index methodology. Would it be complex technically?

• impact on external integrations: if the yields are reinvested into the DPI assets as described above, my understanding it that it would not change the DPI token structure and not interfere with these very important integrations (exchanges, Maker, Aave, etc.).

• complexity for DPI holders: if the yields are reinvested into the DPI assets as described above, DPI holders do not even need to know it, they will just benefit from a better product performance

In conclusion, I think the idea would not be to make DPI a “Yield Generating” product, but to make it a product capturing the full value of what it tracks, which is DeFi tokens having productive value.

[Kiba]

@overanalyser

Adding intrinsic productivity to DPI, may cause some of these customers to move elsewhere.

There’s no extra work for DPI customers and they get free money, I don’t see why any customers would be against that for a tiny amount of additional smart contract risk. In fact, investors should be more interested in buying DPI when they see our approach to researching and implementing farming so we can mitigate risks and maximize rewards for them.

@overanalyser

Changing the structure of DPI could a) make calculation of the fund NAV more challenging, b) delay these integrations.

I thought NAV price wouldn’t be a big deal since they all use offchain price feeds but after talking with Joseph from DFP I can see why intrinsic productivity makes those integrations harder. I think a delay for them to do extra DD on DPI is fine. The entire industry is moving to greater capital efficiency (uni v3, BentoBox, aToken Curve pool, uncollateralized loans on IronBank, etc.). Short term we might have to wait a bit longer but the benefits from increased capital efficiency for DPI holders, IC, and the entire ecosystem is worth it.

One way to get around the NAV issue is to have a staking contract for DPI which gives back a stkDPI token or whatever and yield is distributed only to stkDPI holders. We’ve discussed this in the past in relation to managing DPI solvency when depositing in Aave/Compound but it could also be used as an abstraction layer to keep DPI price ‘constant’ and let stkDPI price fluctuate with yield. I’m not sure where this falls on the securities spectrum. This staking contract can be added to the R&D for this proposal in addition to the Farming Manager contract.

SDI is also fantastic for intrinsic productivity. Our solution should be applicable across all IC products otherwise it’s more work for IC devs to add productivity, for protocols to securely integrate, and for investors to properly evaluate risk/reward for each product we release.

While not simple, it’s not incredibly difficult for someone like BasketDao to take DPI, redeem it for underlying, farm during the month, and mint new DPI before rebalances. We would lose income from lower DPI TVL and from not taking a cut of yield.

[DarkForestCapital]

Looks like the og proposal plus the extras captured in this response would form a pretty strong base to start from.

It’s mentioned in the thread but we are seeing bDPI steal some of our lunch by offering the yielding version of DPI (perhaps also something to do with the 300% APR rewards they are lobbing at it) so it’s good to see it lighting a bit of a fire within the Coop.

I’d be interested to see @reganbozman’s thoughts on the security aspect given how hard he’s worked with the Singapore law firm on getting a non-security letter for DPI as-is. Might be worth leaning into that relationship and asking them what would change if we recycled the yield back to holders.

Great stuff as always Kiba!

[trx314]

@Kiba

One way to get around the NAV issue is to have a staking contract for DPI which gives back a stkDPI token or whatever and yield is distributed only to stkDPI holders.

This solution would work, but involves several actions by the DPI holders. Also the yields would obviously be considered as income for tax purpose.

What about simply reallocating the yields into the DPI positions? Would it be technically possible to do the following: the IC farming manager smart contract collects the yields periodically in the form of tokens (sushi, YFI, etc.) then send these tokens to the DPI smart contract, to add them to the positions of the index? Everything else stays equal. No additional actions required to the DPI holder, no staking contract, etc.

The NAV would be 100% accurate at the time of the yield allocation, then slightly diverge progressively until the next allocation. The more frequent the allocation, the more accurate the NAV will stay, but the slight divergence between each allocation is probably acceptable until a more complex solution is implemented (NAV computation taking into account in real time the accruing yields).

[reganbozman]

I fully support this proposal and find it quite frustrating that we have made close to zero progress on intrinsic productivity since launch.

There will always be reasons to delay things - exchange listings, integrations, etc. These things are ongoing and will not be ‘done’ anytime soon.

While we’re at it, we should also look at adding simple solutions like aTokens and xTokens (see here: A Simple Solution to Make DPI Productive)

Yes, we should do a legal analysis of this but I don’t see a strong reason to not implement this barring that

[Kiba]

@reganbozman The simplest and fastest way to market is to pass on all yield to DPI holders which also seems to be the preferred solution for our users and DPI growth. How fast can get that legal analysis to determine if that would make DPI a security?

@trx314

What about simply reallocating the yields into the DPI positions?

That’s how it would work today on Set Protocol if we started yield farming without trying to take profit to the Coop. But then we would be giving excess profits to DPI holders which would put us in securities realm (an issue I dont care about but others do) Actual mechanics of how yield gets distributed was part of my original draft proposal but I took it out so that it could be decided in R&D what the best distro model is for what we setup.

@trx314

The NAV would be 100% accurate at the time of the yield allocation, then slightly diverge progressively until the next allocation.

Yea that’s exactly the problem that OA is talking about, difference between NAV and market price will affect different systems e.g. MakerDAO only cares about NAV because that is what’s liquidatable. sDPI cares about market price but people will price in yield differently - some might price it in once it’s farmed, some when yield is actually put back into DPI, some might price DPI with forward looking yields so there isn’t a single price for sDPI to value DPI at. Even if you want to rely on Chainlink’s price aggregation to get an average, that price can fluctuate a lot if nodes use different valuation strategies, nodes’ data sources change valuation strategy behind the scenes, or not all nodes respond on each job.

I think Maker is more concerned with the risk of farming and DPI solvency for liquidations where as sDPI is concerned about pricing. @overanalyser if no one at Set has I can talk to both Maker and Synthetix teams about what technical issues they think would arise from intrinsic productivity

[BigSky7]

@trx314

capturing the productive value of these tokens is not an option, it is an obligation

@trx314

the DPI token should capture all the value generated by the assets composing it

Passing as much value as possible back to the end user is a guiding principle of DeFi. We compete by making the most profitable tokens. People hold our tokens to make money. If some other token gets them the same risk profile but makes them more money, they can be expected to switch.

@DarkForestCapital

Looks like the og proposal plus the extras captured in this response would form a pretty strong base to start from.

Agreed - @trx314 and @Thomas_Hepner nail some key points and concerns. It seems like we have broad alignment on the outline here and potential updates.

@overanalyser

We are currently working on a number of external integrations that we believe would be good for the coop and our customers (exchange listing, MakerDAO, AAVE, sDPI). Changing the structure of DPI could a) make calculation of the fund NAV more challenging, b) delay these integrations.

@reganbozman

There will always be reasons to delay things - exchange listings, integrations, etc. These things are ongoing and will not be ‘done’ anytime soon.

@overanalyser I hear the point here - but I strongly agree with @reganbozman. We should not let regulatory uncertainty or ambiguity around future plans keep us from innovating.

Regarding the NAV issue - this seems like this is solved naturally through arbitrage. Say we have a $50 million dollar sDPI <> ETH pool on Uniswaps. sDPI will never be at 100% NAV except during that time of yield allocation, however price arbitrage from the minting contract should keep divergence from NAV acceptably low for consumers.

Another thought - is keeping $DPI as is ($DPI Classic) and also launching this $sDPI possible? This would allow for a natural migration and potentially even increase our AUM as some users choose to gain exposure to both assets. This would also greatly expand the arbitrage possibilities to help maintain sDPI nav. And finally this will build a ton of flexibility - if an exchange or Maker doesnt want to list sDPI we can let them use DPI. We make money either way.

Great work @Kiba and thanks everyone for this awesome conversation. I learned a ton from reading this discussion - these dialogues are where true unforkable value is born!

[Thomas_Hepner]

I took the time to read through most of the Intrinsic Productivity (IP) governance posts that have taken place since December, and while I am still confused about many aspects, I have come to a few beliefs and a conclusion.

Beliefs:

1. The Index Coop is Slow and Bureaucratic: IP discussions have been ongoing for ~7 months now. My sense is that organizationally the Index Coop is in a logjam and @reganbozman and @Kiba are trying to break the logjam and make incremental steps that bring immediate value to DPI holders while maintaining a long-term strategic view of the product. I echo @reganbozman’s comment that it simply unacceptable that DPI’s KNC position is being diluted at ~10% annually.

2. Investors want trust and abstraction, not understanding: I disagree with @overanalyser 's comment above that DPI investors would prefer something “easy to understand”. As a DPI holder, you are implicitly making a very speculative investment on frontier crypto technology; you are looking for abstraction, not simplicity of understanding. As a DPI holder, I am trusting (yes, trusting) the Index Coop community to make good technical and financial risk and return decisions within the DPI tokenset with the guarantee that I can always exit the asset if I disagree with those decisions.

3. The Community wants Intrinsic Productivity (IP): The vote in favor of @Kiba’s proposal is 92% yes to 8% no, and the at least half of the no votes are from the two people (@overanalyser @trx314) who are huge supporters of IP products. This means at least 96% of highly engaged community members, as evidenced by voting on this forum post, want IP. Simply put, the Coop has to have a public plan to make this happen.

Conclusion (Proposal): Form a short-lived Intrinsic Productivity team responsible for bringing a plan/ proposal to the community to either (1) update DPI with IP or (2) create a new DPI product with IP. My suggestions for the proposal are:

1. Unanimous Agreement: has support from all project team members as signatories.
2. Ownership: It is clear who is responsible for implementing the plan with clear timelines and commitments.
3. Short-term and Long-term Viewpoints: Takes long-term Coop product development and strategic goals into consideration without letting the “perfect be the enemy of the good” to create immediate value.

Here are my suggestions for the IP team members:

• Team Lead: @anon10525910
• Set Protocol: @setoshi
• DeFi Pulse Index: @scott_lew_is
• IP Research Leads: @overanalyser / @richard / @Kiba
• Treasury Representative: @LemonadeAlpha
• Business Development Representative: @reganbozman

[Kiba]

nice summary @Thomas_Hepner.
I’m meeting with Jo from DFP about IP and starting a working group on Friday already, I’ll bring up your suggestions.

After that talk and over the weekend I’ll go over all the dicsussions in this thread and parts of the proposal I didn’t bring over from my original draft that have come up and update the IIP.

[oneski22]

Hi, big believer that we need to activate IP, however concerned that we need to approach this in a responsible way that:

1. doesn’t add unnecessary regulatory/securities risks to DPI/other indexes.
2. is repeatable
3. doesn’t fragment the market.

NOT AN ENGINEER SO UNSURE OF DIFFICULTY IN BUILDING THIS

High-level sketch of a proposal that I will be fleshing out more:

DPI Intrinsic Productivity Compromise Keeping DPI institution/regulator friendly while offering the more adventurous a yielding vehicle. Index Coop creates a vault contract that allows users to stake DPI for iDPI. This iDPI can be redeemed at any time for the original DPI + any yield generated (in the form of DPI). The contact has the ability to borrow the underlying tokens from the DPI contract, and use strategies approved by INDEX. The returns on assets are converted into DPI keeping things composable.

Benefits:

• Yield product offered to those willing to bear risks without segmenting the market (multiple indices). still one product. DPI.
• Main DPI product is not subject to extra regulatory/securities risks - Loss of tokens due to bad yield strategy only harms iDPI holders, normal DPI users are not affected.
• Repeatable strategy for all other indexes (MVI, future products)
• Clear separation of responsibilities, methodologists focuses on creating index strategies. INDEX holders decide the yield strategies and can balance meta-governance / yield via DAO Votes.

[Kiba]

Change Log:

1. Update Yield Distribution to send 100% to DPI holders
2. Increase % of assets earning yield from 75% to 90%
3. Move description of how yield is returned to Description
4. Add “Out Of Scope” section
5. Reduce INDEX reward to 100 to reflect change in scope. Contract to manage farming no longer needed, only need YFI v2 adapter.
6. Happy 420

@scott_lew_is I ended up not adding aSNX to this proposal to keep it lean. Aave v1 only has 0.07% APY and Aave v2 only has 2% APY plus doesn’t have a Set adapter yet so it would take extra work/funding. I think xTokens are probably a better option as well and Regan + OA like it but I would have to do more research to feel comfortable at a technical level before proposing it.