First Release | Analysis of Coinbase’s “Pull-up Effect”

Note: On Thursday night, the world's leading cryptocurrency exchange Coinbase announced that it will be directly listed on the Nasdaq on April 14 with the stock code "COIN". Based on previous private market transactions, Coinbase's valuation is likely to exceed $100 billion, which will naturally trigger a new market storm.
In addition to the stock market, Coinbase's listing is also expected to boost the influence of the exchange itself. In this regard, Messari researcher Roberto Talamas analyzed the "Coinbase effect."

The following is the translation:

It has become a well-known phenomenon that when Coinbase exchanges list a new asset, this event will have a very positive impact on the asset. We call it the "Coinbase effect". The economics behind this effect are simple. When a crypto asset is listed on a popular exchange, it is immediately exposed to a new group of market participants. When we study this effect, Coinbase is the main subject of study. In addition, in this report, we also expand the scope of research to 5 other well-known exchanges to compare the effects of listing crypto assets on different exchanges.

Additionally, given that Coinbase is the largest retail onramp for crypto assets, the research includes an event study using a sample of Coinbase listings from the Messari Enterprise API. Event analysis provides a better understanding of how listing events affect token returns, taking into account the impact of market conditions.

The effect of listing coins on different exchanges

In the first part of our analysis, we used Messari’s intelligence database to collect listing information from six different centralized crypto exchanges:

1. Coinbase (listed 28 assets)

2. Binance (listed 22 assets)

3. FTX (listed 19 assets)

4. Gemini (listed 19 assets)

5. Kraken (listed 11 assets)

6. OKEx (listed 14 assets)

For each exchange, we use the distribution of cumulative returns over the first 5 days after listing as a rough proxy to see how effective it is. Not surprisingly, Coinbase has the highest average return at 91%, but also has the widest distribution of returns, ranging from -32% to 645%.

Given the above results, we can easily conclude that listing on Coinbase will bring higher returns than other exchanges. However, in the Coinbase sample, there are some outliers that have received huge returns after the announcement of its listing, which significantly skews the mean to the right of the return distribution.

Among these outliers, District0x (a platform for creating decentralized markets and communities) and Civic (an identity verification solution) saw their token prices increase by 645% and 493%, respectively. By controlling for outliers in the Coinbase sample, we can more clearly compare the distribution of gains across exchanges.

After removing the outliers, it is clear that Coinbase listings still have the largest impact, with an average 5-day return of 29%. On the surface, these results suggest that exchange listings (especially Coinbase) tend to lead to increased asset returns.

However, another important factor we need to consider when analyzing the impact of events on asset prices is the market environment in which these events occur. To address this issue, we conducted an event study on a sample of Coinbase listings to explore market conditions at the time of the announcements, in an attempt to decompose the reactions attributed to the market from the impact of the listing events.

Introduction to Event Study Method

The Efficient Market Hypothesis (EMH) states that all publicly available information is immediately incorporated into the price of an asset. While a full explanation of the validity (or lack thereof) of the theory is beyond the scope of this report, we can build on this premise to study how a particular event changes the market perception of an asset by quantifying the impact of that event on the asset's returns. Academic researchers have proposed event study methodologies to address this type of analysis, focusing on understanding the statistical properties of how events affect asset returns.

The purpose of an event study is to quantify the economic impact of an event on “abnormal returns.” Abnormal returns are calculated as the difference between the return that would have been achieved if the event had not occurred (called the normal return) and the actual return.

Although actual returns are easy to observe, normal returns need to be estimated. For this part of the analysis, the expected return model is usually used, which is also a common valuation tool in other financial research fields.

The market model is the most commonly used method for estimating normal returns in event studies. It uses the actual returns of the reference market and the correlation of the asset with the reference market to estimate the normal return. Formally, the abnormal return on a particular day is expressed as the difference between the actual return of the asset and the normal return, which is estimated based on the typical relationship and correlation between the asset and the reference market (expressed as ? and ?) and the actual return of the reference asset.

So what’s the point of going through all these extra steps if we only look at the returns of an asset before and after an exchange listing to assess the impact of an event? Looking at the intuition behind the model can help clarify this question.

Let’s assume that Coinbase announces a new token listing on the exchange. As we would expect, the token experiences a dramatic price increase after listing on the exchange, which is not surprising at this point. Let’s also assume that on the same day, Bitcoin (a currency that represents the entire market) achieves a 10% gain on some macro news. In this case, how can we be sure that the token’s price increase is due to the exchange listing and not due to the overall market trend? This is where market models come into play.

Using historical price data, we can study the potential relationship between them by observing how their returns move together. In other words, if Bitcoin rises 5% on a given day, what kind of returns should we generally expect the coin to achieve given their historical relationship? This relationship can be better visualized by plotting the daily returns of the two assets.

In this case, regressing the daily returns of Ethereum (our token of interest) onto the returns of Bitcoin (our reference market) yields the fitted line shown above, which can then be used as a simple linear model to estimate the normal returns required to perform an event study. Following this approach, we can isolate the reaction to exchange listing events by deducting the effect of the entire market from the asset's returns on the day of listing.

Experimental methods and results

To conduct our exchange listing event study, we used Messari’s intelligence database to collect data on tokens that were listed on Coinbase between May 14, 2020 and March 23, 2021, for a total of 28 crypto assets. Since the results of the study rely heavily on accurate estimates of normal returns, we only list assets that have at least one hundred days of data before listing to avoid introducing bias into the model. In addition, the study focuses on analyzing abnormal returns of assets using a 15-day window before and after listing (called an event window), so assets listed on Coinbase after March 8, 2021 are excluded from the study. After filtering according to these conditions, only 14 assets met the research requirements.

To measure the overall impact of a listing event, we add up all individual abnormal returns to create a cumulative abnormal return chart. As an example, the following chart depicts Maker’s abnormal and cumulative abnormal returns during the listing event window.

In the chart, the vertical lines represent the abnormal returns experienced on each day of the event window, the solid line represents the cumulative abnormal returns, and the gray area represents the 99% confidence interval. On the day of Coinbase listing, Maker experienced significant abnormal returns associated with the event, indicating that Coinbase listing had a strong economic impact on asset returns.

In our observational study of 14 assets, we found that 9 assets showed statistically significant abnormal returns during the event window. Among these 9 assets, 4 assets continued to have positive cumulative abnormal returns after listing, indicating that Coinbase listing has a lasting impact on the returns of the asset.

Two of the assets saw abnormal returns on the day Coinbase announced the listing. In these cases, the asset’s price incorporated all the information of the listing and did not see any additional reaction on the day of the listing.

The remaining three assets all showed high positive abnormal returns around the event date but quickly returned to normal, indicating that listing had no significant impact on returns.

in conclusion

Our analysis shows that, in terms of exchange listings, Coinbase’s listing announcements have the largest positive impact on the average return of an asset. This result can be attributed to Coinbase’s popularity among retail investors, making it the most prominent cryptocurrency trading platform.

Furthermore, an event study of Coinbase listings shows that the listing market has a strong impact on asset returns, but not all observations are consistent. Of the 14 assets in the initial sample, only 9 assets showed statistically significant returns during the event window, and only 6 assets achieved higher positive returns.

Given these results, we can conclude that while Coinbase listings have the potential to positively impact asset returns, it does not affect all tokens in the same way.

Original article: https://messari.io/article/analyzing-the-crypto-exchange-pump-phenomenon

By Roberto Talamas