Spending money ≠ growth, what has Arbitrum's 85 million ecosystem incentive program brought?

Author: Kerman Kohli

Compiled by: Deep Tide TechFlow

Imagine if you were a business launching a promotional campaign that promised $3 worth of value for every $1 spent. Moreover, anyone could claim this offer, no strings attached. Whether it's your grandmother, a homeless person on the street, a wealthy executive, or an ordinary middle-class individual, everyone qualifies for this offer.

What do you think would happen? Well, those who need money the most, often the least likely to become repeat customers, would flock to you, quickly depleting your funds or inventory until you can no longer sustain this offer.

The good news is that the real world doesn't work this way; the free market would swiftly eliminate such businesses.

The bad news is that the crypto industry does work this way, with the free market continuously driving funds into it.

Introduction

The scenario described above is essentially what Arbitrum has done, involving $85 million in funds, ultimately leading to a loss of $60 million. Let's delve into what this plan is, how it was constructed, and what we can learn from it.

Arbitrum DAO constructed this scheme in such a way that specific industries and their corresponding applications could receive ARB tokens to incentivize users to engage on their platform. The ultimate goal was to generate more fees for the Arbitrum network by encouraging the use of these platforms, benefiting the final protocols. It turns out one side wins here while the other does not (I believe you already know who the loser is).

The quality of this analysis is quite high, and the complexity of the measurements is well done, thanks to the Blockwork team for clearly articulating the reasons, content, and methods of their approach.

You can check the results here.

Method

From a top-level perspective, you can break this activity down into two main components:

1. Establishing a baseline to understand what percentage of the incentives can be attributed to spending compared to the baseline. They call this a "synthetic control" method, which employs some complex mathematics. This isn't too important because regardless of what our final numbers are, we need to adjust downward since not all outcomes can be attributed to this single effort. You can learn more about it in the original forum post.

2. Incentivizing end-users of various applications on Arbitrum by giving them ARB tokens to enhance their metrics. Three sectors were chosen (perpetual contracts, decentralized exchanges, liquidity aggregators). Each application was informed on how best to utilize these incentives.

I did find some interesting excerpts that I thought I could share for your own judgment:

• "Many protocols missed several bi-weekly reports or did not publish reports at all. About 35% of STIP recipients did not publish a final report."

• "Protocols rarely specified strictly why they should be allocated a certain amount of incentives when applying for STIP. Instead, the final allocations were often the result of back-and-forth communication between the protocol and the community, often leading to allocations that resembled 'we think this request is too big/small.'"

In summary, next, I have attached screenshots from different categories showing the amounts spent and the mechanisms (there are no methodological screenshots for decentralized exchanges, but essentially they just incentivized liquidity). The key to remember here is that 1 ARB is roughly equal to 1 dollar. So yes, millions of dollars were distributed here.

Results

I have divided the results into two parts because this experiment aimed to understand two aspects:

1. The impact of these incentives on applications

2. The impact of these incentives on sorter revenue

We will start analyzing from the first aspect, as it presents a slightly more pleasant story. If we start from basic principles, if someone gives you free money to promote your business, what do you think would happen? Typically, business improves—at least for a while. This is the overall situation we observed in this experiment.

First, let's look at Spot DEXs, whose results seem quite good on the surface:

Basically, we see that for every dollar spent, TVL (Total Value Locked) ranges from $2 to $24, which sounds good. However, we need to ask a real question—how much of that is retained? This is where it gets a bit tricky. Balancer's TVL essentially dropped after the rewards ended, which is evident in this chart:

However, Camelot successfully retained that portion of TVL! I'm not sure why there is a difference in retention between these two protocols, but if I had to guess, I think it might relate to how they ran their incentive programs and the types of users they attracted. This is something I have marked and will analyze in future articles.

Now that you understand some micro details, let's zoom out to comprehend the effectiveness of this for applications and three important top-level categories (spot trading volume, perpetual contract trading volume, and loans). I will show you key charts. To aid understanding, I have made some annotations above, so let's take a look together.

I drew two red vertical lines marking the start and end of the program. This will help us understand the timeframe involved.

Then, I drew several horizontal lines to understand different metrics and visualize the program's impact on these metrics throughout its lifecycle.

• The first blue line shows a significant spike in TVL (undoubtedly), but then it almost drops back to the levels at the start of the program, indicating almost no stickiness!
• The second line represents spot trading volume. I want to pause here to mention that unlike TVL (supply side), spot trading volume represents demand. As we can see, demand is at best stable, but actually decreases at the end of the program!
• The third line is outstanding loans, which is also a demand driver but shows no change. While no lending protocols were incentivized, I think this is another strong indicator of demand. In fact, this was declining throughout the program!

So what conclusions can we draw from all of this? Essentially, Arbitrum spent $85 million on these other businesses to enhance their supply-side metrics (which clearly worked), but without corresponding demand to absorb this TVL and tighter liquidity, these efforts became futile. Essentially, you could say that this money was wasted, given to those chasing short-term gains. At least some protocols had higher TVL and token prices, making some people wealthier in the process.

Speaking of demand-side metrics, these activities surely benefited the chain and led to increased revenue from all these transactions, right?

In reality, that's not the case.

The reality is quite different.

This is the sorter revenue chart from January 2022 to July 2024. The significant fluctuations around April were when cryptocurrencies started to rise sharply, and the synthetic control helped us adjust for this.

On the surface, we can see revenue rising, peaking at $400,000 per day in certain months. Here is a clearer chart showing the impact solely on Arbitrum, considering synthetic control:

So what is the area under the curve? $15.2 million. If we remove synthetic control, the total revenue for the sorter is $35.1 million. Considering they spent $85 million, we are still far from expectations!

Learning Summary

To summarize all of the above:

• Arbitrum decided to spend $85 million to incentivize activity on its network to increase market share and revenue.

• They achieved this by providing free tokens to applications and protocols, which would be distributed to end-users.

• Analysis revealed that these free tokens primarily benefited supply-side drivers, while demand-side showed almost no change.

• Further analysis found that all these activities resulted in sorter revenue that was $60 million less than the expenditure.

My conclusion from this is: Supply-side incentives equate to burning money, and should not be taken lightly unless you are facing supply-side issues (which usually, the real problem lies with demand).

The second point is a premise I mentioned at the beginning of the article: if you randomly distribute funds to strangers without discerning their identity and background, the outcome will be very poor. Protocols that continuously provide funds to users without understanding who they are and what their intentions are will ultimately find themselves in the situation described at the beginning of this article.

Imagine if this incentive program could identify the recipients of these tokens through wallet permissionless identities and set the following criteria:

• Is this user genuinely using DEX, or is it a brand new wallet?

• What is the net worth of this wallet, and are they a potentially valuable wallet?

• How much has this wallet spent on fees? Are they sticking with the platform they are using?

• Is this address engaging with all upcoming token projects? They might be yield farmers.

What do you think the final outcome would be?