Roadmap to 300M revenues end of year: A Balance Sheet approach to executing an Aggressive Growth Strategy

All models are wrong, but some are useful

Summary

Previously, we defined an Asset-Liability Management framework 13 for MakerDAO and analyzed some assets 15 in which MakerDAO could invest

This paper shows the importance of balance sheet design to define the path towards 300M revenues and prioritization.

1. Our current balance sheet is slightly underinvested and leaving ca. 9-11M DAI in annual revenues on the table

• We explore short-term bonds and ETH investment as those asset don’t depend on external factors
• Both would increase the MKR holder profitability but will be constrained by the Surplus Buffer size

1. Trying to model the Aggressive Growth Strategy, we learn that capital raise could be accretive and that scalable asset types are importants

• Even with a significant growth plan (+2B crypto-backed loans, +300M RWA, +1B crypto D3M), it is not enough to deploy all our asset base if we can’t rely on scalable asset types (short-term bonds, ETH, …)

• Assuming full deployment of our asset base (through the expected plan or with only short-term bonds), the capital raise (issuing MKR tokens) is accretive for MKR holders (i.e. increasing their earnings)
• Using a debt offering at 10% is accretive as well. Structure of the debt issuance is not discussed but assumed to be equity-like (which might not be the case)

1. Adopting a diversified and liquid balance sheet approach would lead to 300M revenues

• This would give 304M annual revenues vs the current 76M

We conclude that having a high-level model is good to focus the protocol on what balance sheet it wants to target, and how

• We should focus first on what mix of asset type are are targeting
• Then focus on starting with asset items that are scalable so we can allocate the balance sheet quickly in a focused way
• Then optimize the mix by adding non-scalable assets

Introduction

Previously, we defined an Asset-Liability Management framework 13 for MakerDAO and analyzed some assets 15 in which MakerDAO could invest. This paper will provide an analysis of different scenarios to provide some inputs for MakerDAO Governance decisions.

We will start by analyzing the current MakerDAO balance sheet, then providing ways to improve it without relying on external assumptions. Then we shift gears and move towards quantifying the proposed Aggressive Growth Strategy and proposing concrete targets. We will end with a top down balance sheet allocation based on some high level guidelines.

Methodology

The methodology is mainly based on the Asset-Liability Management framework 13. We divide the MakerDAO balance sheet in homogenous categories and assign for each a return over asset (RoA, the expected return of the asset over a 1-year period) and Capital at Risk metric (CaR, not very different from Value at Risk) which represents a significant worst case loss that such asset can generate (which will be discussed for each asset). 1/CaR is the maximum leverage that is safe to be taken based on the existing equity part of the balance sheet (which will be defined narrowly as the Surplus Buffer)

The Return over Equity, defined as RoA / CaR will provide the expected return of 1 DAI of equity. Assuming the Surplus Buffer is fixed, the goal is to optimize the RoE metric. Nevertheless, one should keep the following things in mind:

• This model only optimizes solvency and not liquidity of the protocol. This will be taken in account while designing the balance sheet but isn’t the main consideration here;
• Not all asset types can be expanded/contracted. The only asset types that can fluctuate in size without constraint or significant work are stablecoins (PSMs), short-term bonds and buying crypto assets (like ETH);
• The model doesn’t take concentration risk in account. All things being equal, a diversified exposure might be preferable.
• Operating costs are not taken into account in this analysis.

For simplicity we will discard the impact of liabilities. Indeed, currently the DAI Saving Rate (DSR) is only 1bps which has no meaningful impact. This shouldn’t be taken as a given and as competition intensifies in the sector, we should expect an increasing need to bring the DSR back up again. In this light, higher RoA opportunities are better to prepare the protocol to face the need for a higher DSR. Indeed, in this analysis we use RoE = RoA / CaR but in practice we should consider RoE = (RoA - DSR) / CaR. Hence, while a low RoA (e.g. 0.5%) but very low CaR could provide a high RoE, it would not be the case if the DSR is enabled (if RoA is below DSR, the RoE is negative).

The following table lists all asset types under consideration for this study. We are adding a liquidity metric which represents the expected relative ease of an investment (from 1 to 10, the higher the more liquid) to meet a sudden liquidity need.

Stablecoins

Stablecoins represents the PSM that MakerDAO currently uses (USDC, USDP and GUSD). Capital at Risk is set to 0% as the main risk is a tail one (i.e. blacklisting, smart-contract bugs). There is no Return over Asset expected. While the idea to ask for profit sharing from those stablecoin issuers was floated, nothing seems concrete currently. Therefore, Return over Equity is 0%. Liquidity is intrablock as all the exposure can be used for peg support within a block. This asset type is fully scalable.

ETH

In Toward asset risk 15, we showed that investing in ETH could be of significant strategic value. The CaR used here is 94%. This might be viewed as quite conservative but it is up to the community to voice a disagreement (using post-2018 history or the change in nature of ETH with the institutions coming or the merger). RoA is set at 100% which again is welcome to be contested by the community. The current assumptions provide a 106% RoE. Liquidity is given at 4 as while it might be easy to sell ETH on the market, it is unclear if it would be wise while ETH is down significantly from the buy price. This asset type is fully scalable (at least for a few billions).

Short-term bonds

Short-term bonds are off-chain investments in public markets remaining under the control of MakerDAO governance. MakerDAO provided a declaration of intent on the topic. The CaR is taken from the last assets study 15 while the RoA was updated with the recent yield to maturity (YTM) of one treasury and one corporate ETF (50-50 mix). Those are SHY for treasuries with a 2.62% YTM and a 1.86 effective duration, and SLQD for corporates with a 3.39% YTM and a 2.34 duration. This leads to a 3.01% RoA, a 3.40% VaR and, therefore, a 89% RoE.

It should be mentioned that structure costs are not taken in account here (neither for the RWA benchmark). They could be between 10-50bps and remain to be defined.

Short-term bond liquidity can be easily achieved within a week beside some specific market-related issues. Furthermore, they shield protocol revenues from pure crypto-related market risk and volatility.The governance part being less clear for now as Signal Requests are quite slow. It should also be noted that transaction fees can be quite significant. Therefore, while short-term bonds are quite liquid, they can’t be the first line of defense. This asset type is fully scalable.

Crypto-backed loans

Crypto-backed loans are the current bread and butter of MakerDAO with vault facilities like ETH-A, WBTC-A and so on. The CaR is coming from Risk CU with a simpler version of their published CaR 3 which is using a 1.7% VaR. This shouldn’t be interpreted as a definitive measure as it depends on many external inputs like market liquidity. Nevertheless, it is a reasonable assumption for now.

RoA is the weighted average Stability Fees, i.e. 2.24% currently. Liquidity is good as increasing fees will remove much of the demand quickly. Yet, it is unwise commercially to increase SF quickly. Therefore, it is not expected to be a first line of defense of the peg and shouldn’t be used in normal times.

Crypto-backed loans are driven by borrower demand. It was illustrated in the past that elasticity of the product is quite limited regarding Stability Fees (SF). We can reduce the borrower demand by increasing SF but hardly generate demand even with SF quite below competition. There is a limited demand for DeFi crypto-backed lending. Nevertheless, the space is evolving with the introduction of Institutional Vaults and the introduction of TradFi bank through tri-party repo 2. It is unclear where we are headed. Moreover, it is not impossible that the space will be anchored on SOFR rates 2 plus a crypto premium at some point. SOFR is currently around 30bps, significantly lower than DeFi rates.

It should be noted that Liquidity Position-backed loans are not taken in account at all in this analysis… Most of the exposure is on DAI-USDC Gelato LP which are more an alternative to the PSM than anything else (SF are 0-5bps).

DeFi Money Markets

DeFi Money Markets are defined as liquid and riskless onchain investments of DAI. Aave and Compound D3M fall into this category (only Aave being live currently). CaR could be derived from the Gauntlet Aave risk dashboard 2. The risk on DAI is currently $0 which would lead to a CaR of 0%. While liquidations are safer at Aave (lack of OSM delay), it is still unconstructive to favor Aave over MakerDAO crypto loans. Moreover, using Aave is exposing MakerDAO to external risks. Therefore, a made up 2% CaR is used in this analysis. It is slightly higher than our own vault as it is an external protocol even if the lack of OSM delay makes them possibly safer. It doesn’t impact the analysis too much. RoA is taken with the Aave rewards which stands at 2.57% as of writing. The resulting RoE is 129%.

Liquidity of DeFi Money Markets is quite high as all the exposure is accessible within a block (with the exception of specific conditions). Yet, it can’t be used for liquidity directly as it needs a governance vote (which will take a few days). The D3M should probably be updated to reduce exposure when liquidity inside the PSM is below a given threshold. This would probably allow liquidity within an hour.

DeFi Money Markets are not scalable and depend on the needs of borrowers. The possible end of liquidity mining might decrease their usage.

RWA

Real-World Assets are defined as illiquid loans given to real-world businesses. They can be formalized on-chain with an ERC20 or off chain. They are currently quite heterogeneous in type, underlying, stability fees (3->7%) and risk. The CaR is currently set at 5%. The suggested RoA, 4.79%. This 4.79% is decomposed with 3.39% which is the SLQD ETF (short-term corporate bonds) yield to maturity (2.34 effective duration) and a 140bps illiquidity premium taken from diverse sources (1, 2, 3 1). RWF will have the latitude to provide more meaningful numbers. Nevertheless, the outcome RoE is 96% which is a bit higher than short-term bonds showing the value of pursuing RWA. Being above short-term bonds is quite important because otherwise there would be no reason to invest in RWA. There are also other benefits of RWA like the Clean Money initiative which are not accounted for in this model.

RWA are by design quite illiquid, and while selling the underlying is possible for some RWA, it is not clear yet on how it will be done and at which cost. On scalability, there is plenty of RWA in the real-world, yet each onboarding is currently taking quite some time and is tailored to each borrower.

When we model the balance sheet with the current exposure, we will use 3.5% as RoA as it is the current number.

The following balance sheet data can be found in this spreadsheet 7.

1. The current balance sheet is slightly underinvested and leaving 8-10M DAI in revenue on the table

We will start by modeling the current balance sheet. As you can see, the main exposure (62%) is currently stablecoins which don’t need capital nor bring revenues. The next item is crypto-backed loans that are bringing 90% of revenues while being only 35% of the exposure and 88% of the needed capital. DeFi Money Markets are far behind as well as RWA.

Using the current methodology, the CaR is 59M currently while we have a 69M Surplus Buffer. This is below, but not far, from the actual Surplus Buffer. Crypto-loans demand fluctuates over time, therefore having an unused buffer is not dramatic. Moreover, the model is not perfect either, therefore having additional safety is good.

Nevertheless, we will suggest two scenarios, keeping the same Surplus Buffer but optimizing it at 100%. Both don’t rely on external demand.

1a. Optimize current activity by investing excess capital in short-term bonds

We will now model what would happen if we invest the excess capital (i.e. current Surplus Buffer - Capital at Risk of the previous model) in short-term bonds. As shown below, it would allow a 373M DAI investment in short-term bonds. This investment would provide 9M DAI of additional revenues, push the RoE to 122% and Price to Sales to 5.3%.

1b. Optimize current activity by investing excess capital in ETH

Another scenario that we can explore is to allocate the current excess Surplus Buffer to investing in ETH. ETH is quite intensive in equity, therefore we would be able to invest only 13M in ETH to keep the CaR within the current Surplus Buffer. Yet, as the RoA is quite big, it would lead to a 11M DAI increase in annual revenues. The reader should keep in mind that the ETH RoA will be quite volatile compared to short-term bonds.

Overall, it would lead to an RoE increase to 126% and a price to sale of 5.4%.

What asset mix is optimal for investing excess capital?

As we have seen the impact of investing the excess capital into short-term bonds and ETH, it is time to have a discussion of which is best.

If we start by looking at the stablecoin exposure, we can see that both solutions are keeping quite a large exposure in stablecoin (58% for short-term bonds and 61% for ETH). There is a strong advantage on liquidity for short-term bonds but it isn’t that important in such a case.

The return profile of ETH and short-term bonds is different. While it is predictable with short-term bonds, ETH is by construction more speculative and volatile. Some of that can be improved by using stETH or an ETH-DAI LP investment, but it would still be quite volatile and unpredictable.

We have a preference for a mixed allocation of short-term bonds and ETH. As with all strategic choices in the DAO, the final mix would be up to community preference. Different mixes are all possible and would all represent a substantial improvement on current asset productivity.

2. Trying to model the Aggressive Growth Strategy, we learn that capital raise could be accretive and scalable asset types are the key to success

A proposal was made to raise capital and debt to finance growth for MakerDAO but left open the question of how much growth to finance and what amount to raise to achieve it. The proposal suggests increasing exposure with three levers:

• Institutional Vaults: Those are tailored crypto vaults (ETH, WBTC) for institutions. We will categorize those in Crypto-backed loans.
• RWA
• Multi-chain expansion: This is about deploying Maker MCD on L2 or other L1 and/or providing D3M style liquidity to crypto-lending platforms. We will put this initiative under crypto-backed loans in our model.

We will assume that the Growth scenario is applied for the end of 2022. We keep the balance sheet size constant (as an approximation of keeping the DAI outstanding constant). One could challenge this assumption but it feels the safest to start with. Indeed, DAI demand can increase, some significant liquidity farming like 3pool on Curve or Aave on Avalanche can end.

We will start by using the following hypothesis to model

• Increase of 2B DAI in crypto-backed loans through Institutional Vaults and L2 MCD deployments
• Increase of 1B DAI in DeFi Money Markets with new D3M for Compound, Fuse and others
• Increase of 300M DAI in RWA
• Keeping a 15% buffer in stablecoins
• The remaining is invested in short-term bonds.
• The gap between the current surplus buffer and the Capital at Risk of this allocation is raised on the market at the current MKR price.

As we can see in the table above, such an increase in risky exposure will push the required Surplus Buffer to 157M DAI, meaning 87M DAI to be raised.

If we raised it 100% in equity, it would cost the protocol around 48k MKR (assuming the price stays the same) which corresponds to a 5.4% dilution for current MKR holders.

Meeting these targets would increase revenues to 187M DAI annually. This represents a Sales to Price of 11%, where previous scenarios only achieved up to 5.4%. The Return over Equity is down to 119% which is decreasing as we rely less on high RoE crypto-backed loans.

The apparent divergence between Sales to Price and RoE is due to the fact that the Price to Book (MKR Market cap / Surplus Buffer) is around 23x making additional equity quite cheap in this model. Another way to look at it is that the MakerDAO equity yields 119% (RoE, again without operating cost in this analysis), yet investors are assigning such a premium over equity that they are buying those revenues at a 11.6% capitalization rate.

From a liquidity perspective, the highly liquid exposure (Stablecoin, DeFi Money Markets and Short-term bonds) would be below the 50% threshold (39%). While this is still a good liquid position, it is something to be discussed.

This capital raise could also be done with debt (in part or in full). The next table illustrates the same scenario as the previous one but with raising debt at 10% instead of issuing MKR tokens. We have updated the RoE, Sales to Price and P/S to make use of the cost of the debt (17M DAI annually).

The RoE is obviously lower than before as we add the cost of debt. Yet, the Sales to Price (and the inverse P/S) is higher than before. Raising debt instead of equity is beneficial (financial leverage).

Assuming significant growth going forward, the cost of debt (as ratio of all revenues) will become less important while the MKR dilution would remain the same. This is another positive for debt.

Yet, it is unclear at this stage how debt could provide the same guarantees to DAI holder safety as equity. Same goes for the rate the market will command. The community’s view should prevail here over the ideal financing mix.

2a. Extending exposure to short-term bonds is favorable even under a no-growth scenario

A worst case modelization below shows a scenario where the only thing we achieve is investing in short term bonds and essentially limit ourselves to investing our excess float for our current debt capacity. Short-term bonds don’t depend on external factors and can be scaled almost limitlessly relative to the size of our balance sheet. Failure to even implement that would obviously show that the capital raise (equity or debt) would be destroying value for MKR holders.

As you can see in the table below, that would lead to a 2.8B exposure in short-term bonds. The newly raised capital would still be fully used. It would even still need more capital to be able to decrease Stablecoin exposure to the targeted 15% (it leave 29% in Stablecoins)

It is interesting to note that, even in this case, the capital raise would be favorable for current MKR holders (Sales to Price of 9.6%). 53% of the revenues would come from short-term bonds which is probably not ideal as nothing prevents the rates from going down to 0% again in the future (but that would give us more than a year to find a solution).

2b. Adding ETH to the mix is even more favorable

The last model we will study is starting from a successful aggressive growth strategy and adding a 100M DAI ETH investment.

Therefore the following hypothesis to model will use the following assumptions:

• Increase of 2B DAI in crypto-backed loans through Institutional Vaults and L2 MCD deployments
• Increase of 1B DAI in DeFi Monet Markets with new D3M for Compound, Fuse and others
• Increase of 300M DAI in RWA
• Investing 100M DAI in ETH
• Keeping a 15% buffer in stablecoins
• The remaining is invested in short-term bonds.
• The gap between the current surplus buffer and the Capital at Risk of this allocation is raised on the market at the current MKR price.

As you can see in the table above, adding 100M DAI worth of ETH is increasing the capital raise by 91M.

An interesting feature of this targeted balance sheet is that no single line of asset type is above 50% in terms of revenues or above 50% of CaR. Nevertheless, the amount of liquid assets, Stablecoin, DeFi Money Market and Short-term bonds, while still good, is again below 50% (38%).

In the table below, we perform an impact analysis of some environmental changes. We are analyzing 3 environmental factors:

• an increase/(decrease) of the ETH price by 10%. This is impacting only ETH investment;
• an increase/(decrease) of the DeFi rates (SF for MakerDAO and supply rate for DeFi Money Markets) by 50bps;
• an increase/(decrease) of the TradFi rates (impacting short-term bonds and RWA) by 50bps.

Some are impacting future expected revenues (DeFi and TradFi rates) while the ETH price fluctuation is impacting directly the balance sheet (one can argue that a 10% fluctuation affects the expected future return of ETH but that would be quite hard to model). We are not impacting the interest rate risk on the Surplus Buffer as it would be far lower than revenues.

3. Further optimization of the balance sheet by diversifying more broadly and strong liquidity

In this section, we will target a balanced balance sheet allocation between factors (ETH price, DeFi rates and TradFi rates).

• 50% allocated to liquid items (Stablecoins, DeFi Money Markets and Short-term bonds)
• Diversified revenues sources between ETH price (29%), TradFi (37%) and DeFi (34%)

.

As you can see in the table below, the impact of macro environment changes are more balanced with this scenario.

Such a scenario would require a 214M DAI capital increase which is more than the previous scenario, a better diversification and higher expected revenues. The RoE is still at 107%. Sales to Price is increasing to 16.6%.

Conclusion

As we have seen, modeling a targeted balance sheet can provide significant insight. It allows us to focus on the global outcome for MakerDAO instead of being narrowly focused on a specific project. At the end of the day, our goal is to maximize MKR token-holder value and protect DAI holders.

Currently, most of the asset side of the balance sheet is uninvested (stored in stablecoins). Simultaneously, the Surplus Buffer is underutilized (from a CaR perspective) and too small (as its size doesn’t allow a complete investment of the balance sheet).

Next Steps

• Refine the model with updated input from CUs (Risk, RWF)
• Define objectives of growth for each CU (Growth, PE, RWF)
• Define capital need to implement those growth targets safely (SF)
• Implement scalable asset types (short-term bonds and ETH)
• Raise capital (debt or equity depending on community wants)
• GROW TO 300M REVENUES