Should BAT SF be reduced?

For the past few weeks the amount of BAT backed debt hovering around 1.5 MM (about 50% of the current debt ceiling)

Conversely we see that ETH holders are more willing to take advantage of a DAI loan with almost 100% of the available debt (~100MM) being utilized.

Is there a case to be made that the SF currently too high for BAT holders to be incentivized to borrow against their collateral?

If so should the BAT SF be lowered to help spur on the creation of new DAI and help grow the ecosystem?

3 Likes

My intuition would say that BAT is even more volatile/speculative than ETH so BAT should actually have a higher SF if anything. I understand the sentiment of growing the system by keeping SF lower to promote Dai minting but I don’t think that is the optimal solution, especially with our limited grasp on controlling the peg right now. Perhaps once more risk teams and modules become available we can see collateral types decouple.

3 Likes

My intuition would say that BAT is even more volatile/speculative than ETH

Yes, but do we have any data to back that up? This is just some napkin math here based on coinmarketcap data, but since jun 1 2017 BAT’s price has been decidedly less volatile (as measured by it’s coefficient of variation) than ETH.

Not to say that number in that spreadsheet really means anything. I just don’t think it is fair to dismiss the idea outright based on our intuition.

Are there other metrics that we are looking at beside DAI price to determine if the current stability fees are effective? When would we consider polling the MKR voting body to see if a change in SF is warranted for a given collateral type vs the SF of the system overall?

3 Likes

I don’t think it’s wise to look at risk only quantitatively. For example, BAT is not really necessary for the Brave project (on protocol level or something like that).

1 Like

Total ETH cap approx $19B - $100M borrowed approx .5% of cap.
Total BAT cap approx $300M - $1.5M borrowed approx .5% of cap.

Surprisingly about the same % of cap.

This was my point about adding ‘new’ asset classes to Maker. I would not bother adding other assets that only have caps of < $1B as they won’t add that much to the DAI borrow. ETH is growing by itself nicely and the BAT probably will grow with it as more people with multple assets find out about Maker.

In effect based on the above imo there is nothing amiss and having the BAT DC be 2x the % of available cap as compared to ETH is like having ETH DC at 200M and complaining it isn’t being utilized. The idea that Maker could or would claim significantly more than .5-1% of the total available caps in anything at this point really isn’t a realistic view. I’m thinking at the current rate of debt and deposit expansion that Maker could hit $200M DAI outstanding in approximately 6 months give or take and by then the $3M DC on BAT likely will be hit as well and need to be upped. By then Maker will have raised the ETH DC probably 3 more times to approx $200M all with ~1% of the total available ETH and BAT market cap value in DAI borrowed. FYI given the collateralization of > 200% this means there is at least 2-3x the collateral value deposited vs. the DAI minted which I’m looking at above and so this means Maker already has about 1-1.5% of the available caps as collateral deposited and by summer will have 2-3% of the available cap value deposited.

As to Risk Premiums on top of the SF’s I’m going to leave that for Cyrus. :wink:

BTW: Total DAI outstanding grew to $130M (1.85M in the past 24 hours). Folks just sit back because from what I’m seeing growth in DAI outstanding has been consistent and only slows during significant market pull backs (probably due to liquidations). My only annoyance is not seeing a chart of the total ETH and BAT deposited vs. the $$ amounts as I think this is a more important metric to keep track of along with collateralization ratios. One of my concerns here is that it took SFs of 16-20% to slow this growth, and about the same time we saw the market pull back. There is some concern here that the SF rates were not very effective in controlling debt facility growth and a more dynamic debt facility SF based on utilization curve would be more appropriate for managing this.

3 Likes

Thanks for the great response. I hadn’t looked at the amount of collateral as % of the available marketcap in said asset. This really sets things into perspective.

I think daily volatility data, while important, it not sufficient to justifiably quantify the risk. BAT was designed to be less volatile than ETH, for the Maker protocol, given one hour OSM delay, it will be interesting to see if there were periods of short-term (less than an hour) significant price drops. We will look at the coinbase one-minute candles as see what we can find from there.

2 Likes

While it is very interesting point, I think it determines Debt Ceiling not SF.

I believe that

Debt Ceiling should be set on level low enough to ensure that there is allways sufficient liuidity to sell off undercollateralized debt.

and

SF should reflect voliality of an asset.

1 Like

Will just add another metric here to potentially give some signal on ETH v BAT volatility.

I went through this morning and tried to calculate the beta coefficient for each of the collateral types. For this I compared the daily change in closing price for each asset vs the daily change in the CCMIX cypto index.

Found BAT’s beta coefficient to be 0.3017 and ETH’s to be 0.6425

You can check my work here.

Obviously this is still just another volatility metric though but i thought it was more useful than the previous one I posted since it compares the two against the volatility of the overall crypto market.

And again based on this metric we see that ETH could potentially be viewed as a higher risk asset.

2 Likes

To add a little more context to those beta numbers I came up with a little CAPM graph to show what those numbers vs their historical return. To calculate the security market line in this graph i defined the “risk free rate” = DSR. All of the return numbers are based on the closing price change between 07/01/2017 - 01/26/2020

chart

DAI could easily be 10% of ETH cap, but today’s ETH price is not sustainable - think $1000/ETH in 2020/21. 10% of that is about 10B. That is realistic. If you have a vision to look further - you’ll see a chance that ETH becomes bigger than BTC at $10k+/coin.

As a preface, I want to say that my understanding of DAI is that it gains its value through the value of the collateral offered to mint it. That said:

I’ve given this a bit more thought over the past couple of days and at this point I’m at least fairly certain that at least conceptually BAT SF should not equal ETH SF because its rate of return (RoR) is not equal to the ETH RoR.

I haven’t seen much in this thread to explain to me why the BAT SF == ETH SF. IMO, we only find ourselves in this place today because setting the BAT SF == ETH SF was the simplest course of action when launching MCD.

Having thought a bit more about the SF the past few days I would like to put out a strawman argument that the SF should be more or less be defined by the green line in the following graph for each asset.

chart

That is, that the SF should equal the difference between the expected RoR for a particular asset and the desired RoR at the given risk level times the desired collateralization ratio. Whether or not beta is an adequate way to quantify risk is probably up to debate, but i think conceptually the idea is there i.e risk v return.

As i see it, when demand dynamics are ignored, the SF is simply a mechanism to protect DAI against the inflation caused by the DSR which ensures that each and every DAI token remains backed by the appropriate amount of collateral. To drive this point home I’ll provide a quick example. Let’s imagine a world where the SF of all assets is 0%. All dai is backed by 2 assets types A and B. The DAI DSR is 8.75%. Asset A has a return rate of 7.75% and asset B has a return rate 9.75%. On Jan 1 I mint 100 DAI by offering up $100 of asset A and $100 of asset B. My collateralization ratio at this point is 2. I immediately lock the 100 in the DSR to begin drawing interest. At the end of the year my DAI has grown and I now should have 108.75 DAI (54.38 DAI / vault). The value of my collateral also grew over this time the $100 worth of asset A is now worth $107.75 and the $100 in asset B is now worth $109.75, but now my collateralization ratio is not 2. The 54.38 DAI that is backed by the collateral in vault A now only has a collateralization ratio of ~1.98. For vault B that number is ~2.02. However, if we change this example and charge asset A a SF of 2% we should find that at the end of the year that the DAI minted by that vault should still be backed by 2x the collateral.

I’ll leave refuting my strawman as homework for the reader, but here is where I will segue into my point. Whether or not the past 3 years of price appreciation is going to be representative of the future is unknown. ETH may prove to be the better asset to hold but that too is unknown. What i do know is that ETH RoR over the past 2.7 years is not equal to BAT RoR over the past 2.7 years, and I’d be willing to bet that BAT RoR will not equal to ETH RoR over the next 2.7 years, so it seems a bit off to me that we would keep the SF for these two assets in sync.

BAT over the past 2.7 years has been both a higher return and a lower volatility asset, so MAYBE I’m justified in saying that the SF should be lower. That said, I think it is ultimately up to the MKR holders to decide if I am right or not, so what i am trying to figure out at this point is what is stopping us from letting the MKR holders vote on the SFs separately. Is there a tech limitation or are we just looking for some signal to justify polling separately?

1 Like

Several remarks:

  1. besides SF and DC there is also third risk parameter - Collateralization Ratio which can be used to mitigate both voliality and potential lack of liquidity (which will result in price slippage)

  2. My personal thinking about SF is more as it being a “price of DAI”
    Like central bank interest rates and If that is a case then it should be common for all assets since DAI is fungible and independent from it’s origin. In that case risk of an asset should be handled by Debt Ceiling and Collateralization Ratio

  3. I must say I do not understand Your example. This one in particular would allow risk free rent - so it would operate on loss of MKR holders.
    I do not see correlation between expected returns on an asset and
    SF and if there is one, it is positive not negative. If asset is expected to increase in value then owners of this asset will be ready to pay higher
    SF in order to generate DAI and leverage on this asset with newly generated DAI. Historically SF was higher during bull market (to keep DAI peg) not lower.
    I believe You are making false asumtion that by generating DAI against collateral You somehow give up ownership of this collateral. This is not the case. If collateral increases in value, you still 100% benefit from that increase, so You do not give up any of your profits (you can eve increase them if You used generated DAI in order to buy more collateral not to put them in DSR)

  4. This thread looks very interesting. Are You aware of
    Signaling Guidelines ?
    Consider registering Signal Request for your proposal.

You are correct in this assertion (although governance also voted to keep them the same, it wasn’t decided by the Foundation, see here: Signal Request / Poll: How should we structure monetary policy polls in MCD?). To be honest, the only reason BAT is in MCD at all is that we wanted to launch with two collateral types in order to test the Multi-Collateral functionality and for PR reasons (it’s in the name!)

That said, I don’t think that’s a good argument for changing it separately to the ETH SF at the current time. We know it doesn’t accurately represent the risk for BAT, we know the same thing for ETH, which has also not yet had risk parameters defined for it.

At this stage, we are effectively waiting for risk models to be produced by the interim risk team (@cyrus) and risk parameters proposed for both BAT and ETH. Neither of these assets is currently being priced correctly according to their risk. Both assets are being priced solely due to the requirements of monetary policy. Risk isn’t even a component yet, which is why I think your arguments here are premature.

All that said, I’m fantastically encouraged that more people are contributing to the discussion around risk, so please don’t think of this as a criticism. We need people to work as part of risk teams! If you don’t mind me asking, is this a general interest of yours or is your interest limited to BAT?

To be completely honest risk analysis goes over my head at this point, so you’ll have to forgive me for not engaging directly with your argument above.

2 Likes

Let me try and clear up some of the confusion in my example. I was only trying to point out that as long as the RoR > DSR DAI can remain appropriately collateralized even in the complete absence of the stability fee. Which IMO should in theory mean that 1 DAI == $1 again ignoring the demand dynamics of the marketplace.

It doesn’t really bring in factors such as the desired profit of the maker holder. You are correct in saying that maker would be operating at a loss or at best break even in that example. Nor does it bring in risk factors which also should be included in the whatever model you use to set SF.

I would argue that if you expand upon that idea a bit and instead of trying break even (by making the sure you are charging enough interest to cover the DSR) you set the SF to make up the difference between the RoR of the asset and some desired RoR for the MKR holder (maybe that security market line??) then things start to change.

I don’t really have any particular interest in BAT per se. I’ve only downloaded the browser since ive started looking into this :stuck_out_tongue:. I just noticed something that seems off in the numbers and started trying to dig into the data.

Where my continued interest lies is that it seems to be a fairly high return on investment question to answer. I know BAT doesn’t represent that much liquidity to bring onto the platform, but we have support for it. There are not code changes that need to happen and there isn’t a million executive votes that need to pass, so if we spend a few hours (or even days) to come up with a better number and bring on say 2-3x the liquidity that’s potentially ~ $270k this year in interest. Seems like a win to me!

I hear a lot about “risk models” in these forums, but I don’t have clarity on what exactly we are trying to produce. Is there a forum post or even an academic paper where I can go to learn more about what we are looking for from that?

3 Likes

The best person to talk to is @Cyrus, I only have a vague idea of what these are and there were some presentations at some point. Let me see if I can dig up the appropriate governance and risk meetings.

Edit: Some detail here, possibly more in the meetings around this one Scientific Governance and Risk - Thursday, June 27 9 AM PST (4:00 PM UTC)

This is a fair point, although it’s worth noting that we couldn’t reduce the BAT fee below the DSR without causing problems.

1 Like

Sweet thank you so much! I’ll probably be back in a few days once i dig through this some.

Hey @Andy_McCall thanks for that really great analysis. Certainly exploring betas and daily volatilities is one avenue towards assessing risk. As others mentioned, however, we are opting to take a slightly different approach.

While it is counterintuitive, a market risk approach (such as the one you have used using CAPM and such) is less appropriate given the CDP nature of MakerDAO. Instead, a credit risk analysis is (in our view) more applicable.

So daily volatility becomes much less of a factor as opposed to risk of a CDP defaulting, which occurs due to sudden and sharp drawdowns in a collateral asset. Daily volatility is less useful for anticipating these sudden drawdowns, versus some other approaches (such as just basic fundamental analysis). So while BAT may have a lower daily vol over the past 3 years, or a higher expected return, it is (likely) to have significantly higher risk of a sudden drawdown. (In a credit risk model, you might liken this drawdown to a probability of default and the loss given default).

A simple counterexample that I like to use a lot is Tether. USDT has incredibly low volatility, but that does not make it a safe collateral type for MakerDAO.

4 Likes

Hey @LongForWisdom - is there a reason why the protocol does not allow an asset SF to be less than the DSR? I can imagine a number of conditions where we would want a particular asset to have a SF lower than the DSR.