System Reserves

Do you have any thoughts on what this might look like?

P.S: thanks for putting out this poll. Ever since i have joined the forums the topic of system buffer has been a sort of a sticking point for me. Glad people are talking about it more.

1 Like

Reposting this here because this has popped up in relation to @LongForWisdom and the MIPX Proposal regarding DAI Protocol Transfer and discussion with @Andy_McCall

Given that Maker governance seems to have relaxed regarding centralization risks on various coins (wBTC, USDC, and who knows what we add next).

Wouldn’t it make more sense to take DAI surplus which one would use as system reserves and general operating funds and convert them into other stablecoin assets: USDC, PAX, USDT, whatever.

This idea originated to buy assets generally (ideally anticorrelated to collateral) and use them to deal directly with injecting and removing DAI liquidity but seems to be applicable to system reserves and general funding uses of capital.

Example: PEG is above $1. Sell DAI and buy USDC, PAX, USDT. Bank these in the System Reserves. In effect one is providing DAI liquidity via direct injection to the markets. Reverse this when one wants to remove liquidity when PEG is below $1. Use these system reserves to pay out on Declarations of Intent add as a system buffer, operating funds, etc.

This same mechanism is easily applied to system reserves generally and allows for use of system reserves for DAI liquidity management via direct injection/removal as well as a source of operating funds. In this sense then we can grow system reserves indefinitely. Personally I’d rather use this secondary system reserve to manage buying MKR via flop type auctions.

While some of what I am suggesting here may require some contract changes other parts of it can be dealt with manually. In principle instead of having automatic MKR purchases when surplus buffer is hit we could just send that DAI into the system reserve. Then we can have governance choose whether and how much system reserve to use to purchase and burn MKR. This is a definite change to how the system works and something MKR holders might not like. But if you look at this from how publicaly traded corporations deal cash management and stock buybacks it is very much the same mechanically. When it comes to building system value, building operating capital is a key component to that, buying back stock is another when one doesn’t want to pay dividends.

I’m not keen on the buy other assets plan for MKR reserves.

Can somebody comment on my simpler proposal, 3 posts up from here?

1 Like

Paying ourself the DSR on our surplus buffer wouldn’t make sense. It’s a net zero transaction.

1 Like

As I stated before my personal preference would probably be to buy tokenized gold (ala PAXG) vs usd stable tokens as it seems to me be less of a regulatory risk in the medium term. With things like GLD already in existence it seems less likely to me that PAXG is going to run into future legal trouble by bringing that on chain. All these USD equivalents on the other hand seem to have gotten a lot of regulatory interest in recent months so it may be wise to be wary there.

Assets aside what you are saying is more or less how i think the reserve might work. It gets dai that it sells for SOMETHING else. We store value in SOMETHING. Then later when we need more DAI we sell SOMETHING for DAI.

Fine, the DSR part of my question is not important. The question is, can we move DAI from the surplus buffer (owned by DAI holders) to another account that is owned by MKR holders. This would reduce the effect of MKR’s DAI savings on the peg because this account would be owned by MKR and not contribute to DAI holders in an emergency shutdown.


I think gold by itself has a number of market risk hazards honestly and it makes more sense to spread out system reserves over multiple assets since one never knows what market liquidity will look like and the more reserve pairs to look at to buy/sell the better.

Remember I first proposed this whole system reserve from a DAI liquidity direct injection/removal mechanic.

PEG > 1 (usually times are tight markets dying for DAI) BUY assets with DAI to provide DAI liquidity
PEG < 1 (markets are good) sell assets for DAI to remove liquidity.

The reason why stablecoins are probably the best here is that we really want the system reserve to try to maintain a constant USD value and not end up buying PAXG when price is high, and selling PAXG when prices is low. It is not clear this will be true generally but it is why one doesn’t want to have (1) one system reserve asset pony which can have a fluctuating value and (2) why one would prefer assets at least if one is managing from a cash perspective that don’t fluctuate at all in value.

In the loosest sense one could have two pieces to a system reserve, assets completely uncorrelated and designed to hold USD value (it is a real debate whether USD is really the value metric vs. a basket of currencies or other values) and another system reserve that is designed to be anti-correlated with collateral price changes to act as a system buffer which should grow when system overall is being hammered with collateral price dropping.

Personally I’d like to spread out centralization risk. so when it came to USDC I would favor MORE using compound cUSDC as it would earn some interest. Not so thrilled with synthetix honestly but one could diversify this with sUSD. Even uniswap v2 stuff might be an interesting model do DAI:PAXG pair and or DAI:USDC etc. Literally spread this whole reserve thing out over a lot of very different types of assets but with a clear mind to constant value ones, and anti-correlated value to collateral classes on Maker as well as spreading out centralization risk as far and wide as possible as well as paying attention to how this affects DAI liquidity when the system needs to call on it.

BTW: One very clear way to spread DAI into markets is to buy cDAI but note the way compound is structured we might not be able to get at that liquidity whereas uniswap DAI:USDC straight up is guaranteed to have capital access even if it drains market DAI liquidity to some extent.

Multiple assets also makes plenty of sense to me big ++. My only concern there would be that it introduces complexity which is something that i would argue this idea probably doesn’t need any more of at this point.

Just as an add on i could definitely be convinced that keeping other dollar pegged assets is a good idea if that were the one that was met with the least resistance on this front. I have my concerns, but my alternative is not without its own drawbacks so who am i to say which is better.

My only position is that we probably need a way for the protocol to keep a balance sheet and if usd peg’d tokens are what people like when it comes to storing value that seems reasonable to me as a first step.

I think so, yeah. I don’t see why that wouldn’t be possible.

Hey I am all for KISS but when it comes to portfolio management usually the wider the throw the better. Pretty much would need a system reserve(s) asset manager :wink: The risk team could signal whether markets need to add or remove X DAI liquidity and then asset manager could select assets to buy/sell and governance signals whether to actually act.

But yeah generally looking at such a beast from ‘minimal’ requirements would be a prudent approach to start.

Great, well I think this would be an easy first step. We can keep reserves in other assets eventually, but segregating DAI that is owned by MKR should be easy.

My only concern with this is that it doesn’t really address this problem of the protocol actually hoarding liquid DAI that the risk team brought up.

Sure, that’s true. It’s not perfect, but it’s an easy step and a substantial improvement on what we have now.

I don’t like the idea of baking offchain assets into the core makerdao protocol. This gives these entities who control usdc, paxg etc a huge amount of influence over whether makerdao is able to function and so kind of removes “censorship resistance”.

I like the idea of increasing our dai buffers. Purchasing dai below peg, potentially selling dai and reducing the reserves if dai is significantly above peg. It seems like the risks people are concerned about with large reserves are vague hypotheticals rather than concrete risks.

Keep in mind if we begin to see any issues that have been brought up we can always decrease the reserves again in the future with minimal overhead. The liquidity issue should only be a problem when dai is significantly above peg, which is when we would be incentivized to reduce reserves anyway.


I think going down the road of purchasing alternative assets (gold, stablecoins, other) provides far too much complexity and risk.

Cyrus has stated he thinks that creating a DAI reserve that is owned by MKR holders is possible, so this alleviates Rune’s concerns. I think then we have all the pieces available to make this work.

I envision the system working as @LongForWisdom proposed.

Once SFs increase, accumulate a DAI reserve of approximately 5% of DAI supply.

Then at a later point, if we get to a point of 0% base rate and DAI is still above peg, the DAI reserve is sold for MKR to try and keep the drop the DAI price.

1 Like

Personally I think this all depends on what the protocol’s VaR is. For instance if the protocol’s VaR is <= say $10MM sure let the protocol just hoard DAI to cover it’s potential losses. That only makes up about 7% of the current DAI supply, and under normal circumstances I doubt that amount of DAI out of circulation would have much effect on peg.

That said if the VaR is more like $50MM then we have another situation altogether we would be talking about hoarding almost half the DAI supply just to cover the potential losses of the protocol, and that doesn’t even begin to bring in operating expenses. Definitely going to have peg issues in that case.

All of this is to say that depending on what we think VaR is we may need to start to consider complex and crazy ideas just to make sure we have enough capital on hand to cover the potential losses of the protocol.

But the way I see this working, is we only push money towards this DAI reserve fund when DAI is trading below the peg. If DAI is above peg, we pause funneling any money towards the reserves.

The reserves act as both a safety net, and an additional lever to balance the peg.


I think we can cross this bridge when we come to it. In the meantime, adding more forms of collateral is likely our best path to reduce peg volatility. We are pretty close to peg now.

I totally agree that the primary focus should be trying to close the gap on the peg, but I find this comment a bit dismissive.

I mean is it so crazy to suggest that we might want to consider what our options are when it pertains to the protocol holding storing value when given that currently there is > 1MM MKR issued. Over the course of SAI + MCD lifespan, the protocol bought back ~14.5k MKR @ around $500 / MKR. That is ~$7MM the protocol spent on MKR buybacks over it’s lifespan. That alone would have been enough to cover “Black Thursday” losses had the protocol hoarded say USDC instead.

Frankly, I think we (MKR holders) have “crossed this bridge” and we took a haircut in the process, so given our experience, I find a bit offensive to be so casually dismissed.

All I am really saying here is let’s consider our options on how we might endeavor to be better prepared for the next rough patch we encounter, and that DEPENDING on how much money we actually need to set aside we MIGHT have to consider options outside of DAI.

Sorry, I didn’t mean to offend. I guess I just think that once we start losing properties like censorship resistance, we’ve already lost. These should be last ditch efforts when we feel that the ecosystem would otherwise collapse. But admittedly, I’m probably more strongly ideological in this way than most.