Stablecoins; Not Quite Stable Yet ♂️28th March 2019
Stablecoins are a large part of the discussion in crypto-land, and probably for good reason. They currently represent ~2% of the whole market capitalization of cryptoassets, which unarguably puts them in the Top-5 clusters in the cryptoasset space (alongside currencies, smart contract platforms, privacy coins and exchange tokens) — an impressive feat, no less. Their utility is beyond any question, as in a universe of highly volatile assets, they stand out as a plausible solution to a huge adoption hurdle. As such, many have put stablecoins front and centre when considering what the key pivots that trigger the next wave of adoption, might be.
The cluster’s ecosystem has developed significantly over the crypto-winter (especially from 4Q2018 onwards), with new entrants challenging for Tether’s monopoly and older features continuously iterating and improving on their fundamentals (e.g. DAI). Be that as it may, stablecoins are still rough around the edges. Their claim to stability is repeatedly challenged, liquidity is oftentimes scarce, while the market leader’s viability has sent pundits and investors alike in a multi-year long “witch hunt”.
In the sections to follow I aim to provide an overview of how the space has developed in recent times, looking at key variables such as market shares and liquidity, provide some colour on stablecoin volatility and its various sources, and further, make a case for what the fundamental drivers of stability are and how these are some of the key areas for current issuers to focus on. Let’s dive in.
Relative market shares: Tether is being challenged
Let’s first have a look at how the competitive landscape has developed over the past couple of years in the stablecoin space. The following plot examines the relationships between the relativemarketcaps of the stablecoins examined. The focus here is on the most dynamic competitors in the space, and less so on more fringe players such as Digix Gold, bitUSD and Steem Dollars.
The all powerful Tether, started out at a whopping 92% market dominance and ended at ~75%, with the new-entrant insurgence becoming aggressive around November 2018, with the introduction of USDC and Gemini Dollars in the competitive landscape, that coincided with the BSV fork and the monumental sell-off in mid-November 2018. If Alex Jones was writing this piece, it would read something like “Winklevi, Armstrong, Jihan and Faketoshi, stage dump to dethrone Tether.” We’ll keep things sober and stick to the data. For more colour on what the sentiment in the stablecoin space was prior to November’s sell-off, see this post from October 4th 2018.
Things become even more interesting when plotting the relative market shares of the “challengers” over the past year. Where DAI started the year as the lone star, it has found itself occupying only ~15% of the challenger space, with TrueUSD, USDC and Gemini entering the space in a decisive way.
In absolute terms, the stablecoin market has never surpassed the $3B mark in the year past, with the new entrants occupying the space that Tether’s “unwinding” created, but not adding to total capitalization beyond that threshold. Perhaps this is a nod to the current SAM for the stablecoin contingent.
Since the beginning of 2019, relative shares in the challenger space have remained unchanged for the most part, with the exception of DAI reclaiming some of the lost ground.
Liquidity: Tether is far from being challenged
However, marketcaps are only a part of the story. Another, is arguably liquidity. Borrowing from Wes Levitt’s excellent analysis on the same topic, here’s how the contenders examined here stack up, as of January 4th, 2019.
While Tether has conceded some of its dominance to new entrants with respect to marketcap, it still holds the reigns as far as liquidity is concerned — and by a substantially large margin. On the flipside, DAI — for all its merits — is certainly hard to find in trading venues.
Stability: a pipe dream
Having looked at how the market has developed over the last two yeasr, with respect to marketcap, market shares and liquidity, I want to now divert your focus on stability — the grand promise of the category. Here’s how things have played out over the past 2 years.
Woah — ok so Steem Dollars have run extra wild for a significant amount of time (7 months to be exact). Removing that from the picture, gives us a more digestible view. The following graph and table outline what would have happened if one bought $1 worth of said stablecoin at any point that it became available, since March 2017.
bitUSD, sUSD and Digix Gold have been the least stable of the group, having not stomached highly volatile periods like January 2018 and December 2019 very well.
Once again, I won’t focus too much on Steem Dollars, even though one would have made a hefty profit had they bought in March 2017 and sold January 2018; weird flex, but ok. What is particularly interesting in the graph above, is the stdev (proxy for volatility) and BTC correlation metrics.
bitUSD and Digix Gold have been the most volatile assets in the cluster (most likely due to their weaker volume profiles, when compared to the rest), while USDC and Paxos have been the least volatile. It’s worth mentioning that DAI has kept a respectable volatility profile, especially when considering the fact that it is the only algorithmically driven solution AND the only crypto collateralized product amount the last 6 (bitUSD works under a similar model, with BTS being the collateral instead of ETH).
Let’s now have a look at the correlations with Bitcoin.
This is where things get really interesting! The least volatile stablecoins (TUSD, Paxos, USDC) display high-moderate negative correlation to BTC, which is a strong indication towards the nature of their utility, implying that they are mostly used to protect from Bitcoin volatility on trading venues. In other words, sell-offs in Bitcoin are likely to induce a break of the peg upwards, as demand for the stablecoins sharply increases in trading venues. Given that their capitalization stands at a much lower figure than Tether’s, once more, the lack of enough available supply along with their arguably more trustworthy brands, can provide a credible explanation with regards to the driver of this relationship.
Considering what happens on the flipside (Bitcoin’s price appreciates), I will avoid focusing on bitUSD and sUSD as their utility is relatively limited in trading venues and much of that correlation can be explained but their limited capital base and liquidity. Tether, on the other hand, is a curious case. Being fairly well established as a “trader’s tool”, one would expect that it would have a similar profile to TUSD, Paxos and USDC. This is not the case. Tether, in fact, shows a weak positive correlation to Bitcoin, implying that when Bitcoin’s price appreciates, demand for Tether increases (driving it off its peg — upwards), while the opposite happens when Bitcoin’s price depreciates. Does this imply that Tether has indeed propped up demand for Bitcoin in the 2017 bull-run? This is an interesting finding and certainly warrants more research.
Volatility: Improving — though really not
Ok, so we have established that stablecoins are far more volatile than one would hope, but has that improved at all? In this section we look into how stablecoin volatility profiles have panned out over time. The following chart plots out the standard deviation of the stablecoins in question over the past 1 to 6 months.
Most of the assets examined have displayed positive change, with Tether (!) and Paxos showing the biggest improvement. At the same time however, the latter part of the period in question has been a period of lower large-cap volatility. To that end, we can take the finding as an early indication that there is some positive correlation between the volatility of the space overall and the volatility of stablecoins.
In order to uncover the exact nature of that relationship, we turn to an analysis of the correlations of the volatility profiles of stablecoins, as well as how much of the volatility in each stablecoin can be explained by volatility in Bitcoin and Tether (Bitvol, USDTvol, etc hereon, through a simple linear regression model.
Note: The analysis has been done on daily closing prices, which is arguably more noisy than running the same analysis on continuous data. Further, implementing a Newey-West estimator would help control for autocorrelation and heteroskedasticity, ultimately improving the statistical significance of the results.
Given that we’ve established that the more fringe assets (bitUSD, Steem Dollars etc) introduce noise, I opted for not including them in the analysis and rather focus on Tether, Dai, USDC, TrueUSD and the Gemini Dollar — data on Paxos were unavailable at the time this analysis was performed.
What we will be examining here, is the annualized 30-day volatility of the assets in question, defined as the standard deviation of the daily returns of the past 30-days, times the square root of 365 (days of trading in a year).
< stdev (30-day daily returns) * sqrt (365) >
The sample for the analysis was based on the last 6 months of trading history, looking exclusively at daily closing prices. Here’s how the annualized 30-day volatility of the assets examined panned out over the period.
Consistent with the results presented in the beginning of the section, the volatility profiles of stablecoins (except for TUSD and DAI) are decreasing, while (thankfully!) all of the stablecoins in the group have volatility profiles lower than that of Bitcoin’s. Interestingly, however, it appears that recently the DAI’s volatility profile has been steadily increasing — to the point that it almost equals that of Bitcoin. When comparing the DAI to the rest of the group, however, one has to be cognizant of the fact that the DAI is rather a feature of the algorithmic credit facility Maker DAO, and not the core product itself.
Next, let’s take a look at how the volatilities of each of the assets examined correlate with one another. Bear in mind that what we’ll get out of this is not directional data, but rather an indication of the likelihood of volatility in one asset will coincide with volatility on another (breaking the peg upwards or downwards).
Bitvol x Stablevol Correlations
A few interesting observations here. Volatility in Bitcoin coincides with volatility in GUSD, USDT and USDC, while TUSD is relatively unaffected and the DAI is wholly unaffected — further indication that at this scale, the DAI is its own beast, with its own woes. The other noteworthy finding is a moderate negative correlation between volatility in USDT and volatility in DAI — meaning that when USDT is volatile, the DAI is likely not. Once more, probably an indication of the limited liquidity of DAI and the difference in the core use cases of the two assets.
Now, in order to shed some more light in the relationships between Bitvol and Stablevol, we take a look at a few simple linear regressions with Bitcoin as the independent variable (the influencer) and the stablecoins as the dependent variables. The results are shown below;
Here’s what we know; with DAI, the test was statistically insignificant. For TUSD and Bitcoin, the test had low statistical significance. In all others there are significant relationships in Bitvol and Stablevol (Gemini < Tether < USDC), with USDC being the one where the largest part of volatility is explained by volatility in Bitcoin (82.5%). The regression plots are quite telling with respect to the relationships.
Curious as to what’s causing this? It doesn’t take much, but a quick look into the Stablecoin/BTC pairs and the liquidity thereof to understand why. In this example, I’m taking only GUSD (weak relationship) and USDC (strong relationship) to illustrate the point.
Given that Fatbtc and BW.com have been found to report fake volumes (see findings of the TIE report here), we can relatively safely arrive to the conclusion that it is Binance traders that are mostly driving the relationship between USDCvol and Bitvol. Considering that there is a negative correlation between Bitcoin and USDC, it looks as if the USDC might just be THE WORST safe heaven for traders in periods of Bitcoin volatility- without even considering the exit/arb options that traders might get in USDC/stablecoin pairs.
According to TIE, LAToken is another venue that mis-reports volumes consistently, and thus doesn’t warrant attention. What does though, is the USDC/USDT pair in Binance, that is recording volumes are equivalent to the USDC/BTC pair in the same venue. With Tether being an exit option for USDC holders — and a few others, having a look at the volatility relationship between Tether and other stablecoins, might shed a little more light here.
Bitcoin price vs Stablecoin price correlations
The results are consistent with the price correlations presented earlier — bar Tether, that in the past 5–6 months seems to be highly negatively correlated with Bitcoin (remember, when looking at the past 2-years we found a weak positive correlation). Within the stablecoin group, the most outstanding correlation, is in fact NOT the one between USDC and Tether — where there is NO correlation, but rather the one between USDC and TrueUSD. Perhaps since there is low correlation with Tether, USDC might not be so bad after all, right?
Wrong! As outlined earlier (and affirmed by the regression below), there is high correlation between USDCvol and USDTvol, which implies that Tether is not a credible exit from USDC, as according to the data, it really is a coin toss whether it will be in parity, above or below.
The regression results above provide confirmation to the fact that the DAI is indeed its own beast (only 20% of the variability in Tethervol explains that in DAIvol), while out of the rest, the strongest relationship between USDCvol and Tether vol (with 60% of the variability explained).
So, what we have here is Tethervol <> USDCvol, and an unknown result with respect to directionality; in other words, total chaos 👹
So we have established a lot of the interplay of relationships within the stablecoin space, along with the fact that the volatility profiles of stablecoins are tied to that of Bitcoin’s (and somewhat Tether’s), and thus are not fundamentally improving. The 6-month window might be too short to arrive in robust conclusions, but this is what we have for now. In the following section, we take a look at what might be the underlying causes of stablecoins’ poor adjustment to demand and supply forces.
Stability fundamentals: slow supply-side, weak demand side
Let’s for a moment consider what are the determinants of stability in stablecoins. At a very high level, it is demand and supply. In the existing models, the supply side lends itself to more aptly influence for stability — more so for fiat collateralized stablecoins and less so for algorithmic, crypto-collateralized stablecoins such as the DAI. In the former model, provided that there is enough collateral in the bank, the issuer can expand or contract the supply through direct market operations, in order to counterpoint demand action and maintain the peg. In the latter, things get a little more complicated as the lag between peg break recognition and subsequent action is significantly longer, moderated by a slower decentralized governance process. I won’t go into too much detail on how this transpires here — if you’re interested this recent piece from Trustnodes is a good starting point. The concept is actually pretty similar to the lags that fiscal and monetary stabilization policy is subject to — a very well documented area of econ literature.
On the demand side, things get a little more complicated. For fiat collateralized stablecoin issuers, the mechanisms are equivalent to those described above, with the added benefit of profit minded, arbitrageur market makers, that can intervene in volatile periods and help bring the peg back to balance (see here for more colour). For issuers like Maker DAO, things get even more complicated, as there is no 1:1 convertibility between the DAI and the USD, and therefore little to no ability to influence the peg with a demand-side approach.
Overall, given the wild fluctuations that stablecoins often undergo, it would appear that no meaningful progress has been made on optimising for either side of the market. Again, at a high level, this might imply a few things; most notably that (i) collateral has not increased in excess of demand — at least enough to allow for enough of a volatility buffer via supply side action, (ii) not enough market makers are operating in the space and that (iii) on aggregate, the issuers and market makers operating in the space have not gotten more sophisticated over time — with regards to their processes in the
To emphasize this point, let’s take a look at the price action in the least volatile stablecoins examined above, over the past year.
At a glance, it seems that even the most competitive players, still have ways to go in achieving true stability. Despite the criticism surrounding it, Tether seems to be the one trailing the dollar more closely. USDC appears to experience more excess demand than all others — displaying the highest mean of the assets examined ($1.013), Paxos has been the least volatile while DAI has been the most volatile, showing its weakness in lacking a demand-side stability mechanism. Further, consistent with the results from the analysis on the relationship between stdev and BTCcorrel, Gemini and USDC are the top performers in keeping things tidy on the “below the peg zone”, getting to minimums of $0.982 and $0.993 respectively — above most of their competitors.
Rounding off the marathon, let’s recap what seem to be the learnings here:
- Stablecoins are not stable, and don’t look like they are getting to it.
- Tether is still king, though it has weakened significantly over 2018.
- $3B seems to be the SAM for stablecoins in their current form and set of use cases.
- Given the high correlation of Stablevol with Bitvol, a lot of the reducing volatility in recent months can be explained by the coincidently reducing Bitvol.
- Off the bunch, USDC and Paxos have been the most stable — though USDC has been stable above the peg.
- There is a weak negative correlation between a stablecoin’s volatility profile and its correlation with Bitcoin — the more correlated to Bitcoin a stablecoin is, the less volatile it’s likely to be — perhaps a testament to their liquidity?
- USDC is a poor safe heaven in times of increasing Bitcoin volatility, while TUSD might just be the best option of all for one seeking flight from BTC — both weakly negative correlated to Bitcoin’s price (the only one) and negatively correlated to Bitvol (when Bitcoin gets volatile, TUSD is likely not).
- The DAI is it’s own beast, and relatively insular from the rest — providing some validation to the “credit facility first — stablecoin second” approach.
- Stablecoin issuers are likely not (yet) deploying enough resources towards maintaining the peg — both in direct market operations and working to attract market makers in their respective markets.
Ultimately, with great power, comes great responsibility; stablecoins have amassed great power in a relatively short amount of time. Their utility is beyond question and they will likely play a pivotal role in further driving adoption in Dapps and crypto as a whole. That said, in order to cross the chasm, stablecoin issuers need to assume the responsibility that the power bestowed upon them demands. Do that, or stand to be disrupted by a new entrant that has cracked both sides of the market, with a product that is a true bastion of value preservation.
Stablecoins; Not Quite Stable Yet 🤷🏻♂️ was originally published in Hacker Noon on Medium, where people are continuing the conversation by highlighting and responding to this story.