Is the stablecoin no longer stable? USDT decoupling triggers fluctuations in the crypto market.

Editor's Note

As stablecoins gradually evolve into core liquidity tools within the cryptocurrency financial system, the stability of their price anchoring mechanisms is increasingly drawing attention from the market and regulatory bodies. Although stablecoins like Tether (USDT) aim to provide value stability, serve as transaction mediums, and act as value stores, the frequent occurrence of "de-pegging" during market turmoil reveals potential systemic risks. This paper constructs a robust framework for identifying de-pegging events and detecting price jumps based on high-frequency trading data, systematically assessing the impact pathways of USDT de-pegging events on Bitcoin price jumps and market co-jumps. The study finds that USDT de-pegging significantly increases the probability and magnitude of price jumps in the crypto market after the event occurs, and in certain scenarios, it is more likely to trigger investor expectations, leading to premature market volatility. Furthermore, the market instability triggered by downward de-pegging is particularly notable. The above results provide empirical evidence for identifying new sources of risk in the cryptocurrency asset market and offer important insights for stablecoin regulation and financial infrastructure design.

1. Research Background

Since the launch of the first stablecoin BitUSD, pegged to the US dollar, on the BitShares platform in July 2014, stablecoins have gradually become a key component of the crypto asset ecosystem. In the same year, Realcoin (later known as Tether) was launched on the Bitcoin blockchain, aiming to enhance the compatibility of fiat currencies in the global crypto asset market. Stablecoins anchor their value to reference assets like the US dollar, enabling faster value transfers at lower costs and bridging with the traditional financial system. Although the feasibility of stablecoins was already validated, their circulation scale only saw rapid growth from 2017 to 2018. For instance, Tether's market cap grew from about $10 million to $2.8 billion, and continued to climb, reaching a total market cap of $95 billion by early 2024.

The rise of stablecoins not only reflects the market's demand for low-volatility trading mediums but is also closely related to their multiple functions within the DeFi ecosystem. In the cryptocurrency market, stablecoins have replaced traditional fiat currencies, becoming the underlying asset for the vast majority of trading pairs, maintaining the daily liquidity and trade matchmaking of centralized and decentralized platforms. As of May 2022, stablecoins accounted for 45% of the liquidity in decentralized exchanges; Tether has long held the position of the world's highest trading volume, surpassing other major cryptocurrencies. Besides their role as trading mediums, stablecoins are also widely used in decentralized finance as collateral. The "collateral chain" mechanism they bring allows loans obtained through stablecoin collateral to be further leveraged for secondary collateralization, significantly amplifying leverage levels. Although each on-chain loan is typically over-collateralized, this mechanism introduces a multiplier effect similar to that in traditional finance, making stablecoins a core driving force for market leverage expansion.

However, several significant events have gradually revealed the vulnerabilities of the stablecoin system. Although the core goal of stablecoins is to suppress volatility and provide price anchoring, in actual operation, they have not been able to fully withstand the price deviation risks caused by market shocks. In recent years, Tether and other mainstream stablecoins have repeatedly experienced "depegging" behaviors, where market prices significantly deviate from their target price of $1. For example, the collapse of the Terra-Luna project triggered massive liquidations of DeFi protocols, and in 2023, USDC significantly depreciated due to the bankruptcy of Silicon Valley Bank; at the same time, geopolitical events such as the war in Ukraine led to upward premiums on stablecoins, also causing positive deviations. These "depeg" events, whether negative or positive, can be seen as external shocks that significantly impact the pricing mechanism and trading stability of the entire crypto market.

As pointed out by the Southeast Asian Central Banks Research and Training Centre (SEACEN), the decoupling of stablecoins not only affects their own value but can also trigger systemic risks through multiple channels such as collateral mechanisms, trading pair structures, and protocol execution paths. Once a large-scale decoupling occurs, it may lead to liquidity crises, platform interruptions, and imbalances in decentralized ecosystems. Automatic liquidation mechanisms can quickly release risks, leading to panic selling of crypto assets and severe price fluctuations. In addition, the impact of stablecoins can create network-level transmission paths between CeFi and DeFi platforms, further amplifying the initial shock through price misalignments and protocol defaults.

These phenomena challenge the basic assumption of stablecoins as "stabilizing the market." Previous research has mostly focused on the volatility characteristics of stablecoins or their role as a safe-haven asset, but there is still a lack of systematic quantitative research supported by high-frequency data regarding whether the decoupling of stablecoins will provoke price jumps and co-jumps in non-stablecoin assets. Therefore, this paper takes USDT as the research object, utilizing 5-minute high-frequency trading data covering 70 non-stablecoin assets to reveal how the instability of stablecoins triggers extreme risk events in the cryptocurrency market. In the context of increasingly close ties between traditional finance and crypto finance, understanding the transmission mechanisms of stablecoin risks and the pathways of market shocks is of great practical significance for the design of stablecoins, investor risk management, and policy regulation.

2. Research Data

This study analyzes the intraday price volatility behavior of unstable cryptocurrencies and the de-pegging behavior of stablecoins. The data used has a granularity of 5 minutes, covering the price, market capitalization, and 24-hour trading volume information of Tether and unstable cryptocurrency assets. The time range of the research sample is from January 1, 2022, to June 30, 2023, which encompasses the major stablecoin de-pegging events in recent years.

To ensure that the sample has broad market representation, this study constructs a research sample based on the top 100 cryptocurrencies by market capitalization as of July 1, 2023. During the screening process, wrapped tokens representing other assets, assets issued for the first time after 2022, and assets with a large number of missing observations were excluded. The final sample includes 70 non-stablecoin assets and 1 stablecoin (Tether), with each asset containing 157,248 observation data points.

In terms of stablecoin selection, this article focuses on Tether because it ranks first in both market capitalization and trading volume among stablecoins. Furthermore, Tether is widely present in trading pairs on both centralized and decentralized trading platforms, demonstrating a high degree of market relevance. In contrast, although other fiat-collateralized stablecoins like USDC are also representative, this study aims to focus on the stablecoin that has the greatest impact on the market. As for algorithmic and partially algorithmic stablecoins, they have not been included in the analysis due to their smaller market share.

Tether has disclosed its reserve structure and emphasized that its holdings are primarily in conservative and highly liquid assets to maintain stability with the US dollar. As of February 2023, Tether announced that it had completely eliminated its commercial paper exposure in 2022, reallocating to cash and cash equivalents, including over $39 billion in short-term US Treasury bonds, money market funds, reverse repurchase agreements, and bank deposits. In September 2024, an audit report by BDO Italia showed that Tether's total reserves amounted to $125.5 billion, supporting the circulating $119.4 billion USDT, corresponding to a collateralization ratio of 105%. Among these, 71% of the assets are short-term US Treasury bonds, 11% are reverse repurchase agreements backed by US Treasury bonds, 5% are money market funds, cash and bank deposits account for less than 0.5%, and another 17% of the assets are high-risk assets. This information indicates that Tether has a certain capability to cope with market pressures, maintain price stability, and ensure liquidity.

3. Key Variables

(1) Stablecoin Depeg Detection

Depegging of stablecoins refers to the situation where their market trading price continuously and significantly deviates from the established peg value (usually 1 USD). Taking USDT as an example, when the price is above or below 1 USD and exceeds the set volatility threshold, it can be recognized as a depegging event. Depegging may arise from various risk factors, including redemption risks caused by the bankruptcy of the custodian bank, delisting from trading platforms due to compliance pressure, and liquidity crunches leading to bank runs. For instance, USDC briefly fell below the pegged value due to some reserves being held in a failed bank; USDT also experienced price fluctuations due to new regulatory rules in the EU.

In practice, to eliminate the noise caused by daily minor price fluctuations, clear event identification criteria have been researched and established. Specifically, this includes: the price must first break through the set threshold range (such as 0.975 to 1.025) and remain for more than two 5-minute observation periods; if the price crosses the boundary again within 20 minutes after returning to the normal range, it is considered a continuation of the same event; otherwise, it is marked as a new decoupling event. In addition, to reflect different degrees of price deviation, a decoupling level indicator based on the degree of deviation has also been constructed (such as 0.5%, 1%, 1.5%, etc.), which facilitates subsequent layered analysis.

In further analysis, the study also introduced a backtracking mechanism, using the moment when the price first deviated from 1 dollar as the precise starting point, thereby capturing the market response more accurately. This method ensures that only persistent and systematic decoupling phenomena are identified, providing a clear foundation for subsequent event studies and risk assessments.

(2) Jump detection

In addition to daily fluctuations, the prices of crypto assets often experience sudden and drastic changes, referred to as "price jumps", reflecting the market's nonlinear reaction to information or risks. To systematically identify this phenomenon, the research is based on high-frequency data of the BTC/USD trading pair and constructs a method for jump detection. First, log returns are calculated every 5 minutes, and the Bipower Variation is used to estimate the normal volatility level in the absence of jumps, thereby isolating abnormal volatility.

Due to the evident intraday trading rhythm in the crypto market (such as the alternating trading sessions of Asia, Europe, and America), after estimating the underlying volatility, further volatility structure fitting is performed for each time period using the Truncated Maximum Likelihood (TML) method to extract standardized factors. Next, the actual returns are standardized to construct test statistics, and the significance threshold is set using the Gumbel extreme value distribution to identify anomalies that exceed expected volatility. If the standardized statistic at a certain point in time exceeds this critical value, it is recognized as a price jump.

During the identification process, the time, direction (positive or negative), and magnitude of mutations are recorded simultaneously. All events are standardized based on a 5-minute window to ensure alignment with the timing of stablecoin de-pegging events. This method combines model-free volatility estimation, intraday rhythm adjustment, and extreme value theory calibration, making it applicable not only to the highly volatile cryptocurrency market but also enhancing the accuracy and robustness of identification. The resulting sequence of mutation events provides critical data support for subsequent analysis of how stablecoin risks affect market volatility.

(3) Cojump detection

Co-jumps refer to the simultaneous price jumps of multiple assets within the same time window, reflecting the synchronicity risk and systemic shocks at the market level. To identify such phenomena, the study adopts a time window of 5 minutes and standardizes the logarithmic returns of all cryptocurrency assets in the sample. The specific approach is to construct the MCP (Mean Cross Product) statistic on the cross-section by averaging the standardized return products of each pair of assets based on the covariance matrix of the daily returns, thereby measuring whether there is an abnormal co-movement structure in the market at that moment.

Due to the lack of a closed distribution for MCP statistics under finite samples, the study introduces the bootstrap method to simulate and construct its empirical distribution. Under the hypothesis of no mutations, a large number of diffusion paths are generated, and the 99.9% quantile of MCP is extracted as the determination threshold. When the actual MCP exceeds this value, it is considered that a coordinated mutation has occurred in the market. In addition, if the interval between two coordinated mutation events is less than 10 minutes, they are merged into a single event to avoid statistical redundancy.

This method not only identifies whether there are cross-asset synchronous jumps but also assesses the systemic intensity of mutations based on the amplitude of MCP values and the number of participating assets. This provides a quantitative basis for measuring whether the de-pegging of stablecoins triggers broader market shocks. The study uniformly adopts a 5-minute window combined with data smoothing techniques to enhance identification accuracy and control high-frequency noise interference, thereby ensuring the consistency of coordinated mutation identification in terms of time and method, supporting subsequent causal inference analysis.

(4) Event Research Design

To examine whether the instability of stablecoin prices significantly increases the probability of market price shocks or simultaneous shocks, a systematic assessment using event study design is conducted. Each Tether (USDT) de-pegging event is defined as a starting point for market shocks, and multiple symmetric time windows of varying lengths are constructed around this event to investigate the response changes in Bitcoin price shocks and overall market simultaneous shocks.

Specifically, set event windows of 5 minutes, 15 minutes, 30 minutes, 1 hour, 2 hours, and 4 hours centered around the occurrence of the decoupling event, and count the frequency of both abrupt events and co-mutation events within each window. Taking Bitcoin against the US Dollar (BTC/USD) as the subject, record whether there is a single price mutation within each window to construct a binary response variable; at the same time, record whether there are co-mutation events in that window to measure market-level reactions.

To establish a reasonable comparative benchmark, a "baseline sample" is constructed from the entire sample period, which completely excludes all identified stablecoin de-pegging events and their adjacent influence intervals, ensuring that it is not affected by any explicit stablecoin fluctuations. Within this baseline period, time periods equal in number to the event windows are randomly selected to estimate the natural occurrence probability of price jumps and co-jumps under normal market conditions, serving as the baseline probability.

Using the non-parametric sign test as the primary statistical testing method, we compare the occurrence probability of mutations or co-mutations within each event window to see if it is significantly higher than the natural probability of the benchmark sample. Specifically, we calculate the proportion of price mutations or co-mutations that occurred within the event window among all disanchoring events, and compare it with the occurrence proportion in the corresponding control sample to test whether there is a statistically significant difference between the two.

4. Empirical Results

(1) Descriptive Statistics

The average and median values of the three main stablecoins are close to their $1 peg target, reflecting the overall effectiveness of the peg mechanism. However, the maximum and minimum prices reveal varying degrees of de-pegging phenomena. Taking USDT as an example, smaller price deviations (±0.5%) occurred most frequently during the sample period, with positive deviations (prices above $1) being dominant; however, when the deviation threshold rose to 1.5%, the proportion of negative de-pegging (prices below $1) significantly increased to 75%, indicating that larger magnitude de-pegging is more likely to manifest as a price drop.

In the non-stablecoin market, Bitcoin, Ethereum, and Aave generally exhibit price fluctuation behavior. The positive and negative fluctuation distributions of BTC and ETH are relatively symmetrical, while Aave shows a slightly higher tendency for negative fluctuations. BTC and ETH experience at least one fluctuation on more than 85% of trading days, indicating a high frequency of market volatility. Nevertheless, the probability of a single fluctuation occurring intraday across the entire sample is still below 1%, making it a relatively rare and severe change.

After USDT uncouples, the probability of a negative spike in BTC significantly increases in the short term, especially within a 1-hour window; a negative uncoupling is more likely to trigger a positive spike in BTC. This seemingly inverse price linkage may stem from the arbitrage mechanisms and capital reallocation behaviors in the market. For example, when the USDT price is at a premium, investors tend to exploit the price difference between BTC/USDT and BTC/USD for arbitrage, thus exerting downward pressure on BTC prices. Overall, uncoupling behavior is closely related to short-term spikes in Bitcoin, with a structural relationship between the direction of price deviation and market reactions.

(2) Event study test

Using event study methodology, statistical inference was conducted on the price fluctuations and changes in the probability of co-fluctuation in the market before and after the USDT de-pegging. The results show that in most observation windows, the conditional probability of Bitcoin price fluctuations before the USDT de-pegging is significantly higher than the unconditional probability during the baseline period, indicating that some investors may have anticipated the de-pegging risk of the stablecoin in advance, leading to adjustments in their positions and triggering market volatility. However, this expectation effect gradually weakens starting about 3 hours before the de-pegging occurs and is no longer significant within the 4-hour window.

After the event occurred, the probability of Bitcoin price volatility significantly increased, with the conditional probabilities in almost all time windows being markedly higher than during normal periods, indicating that the USDT de-pegging has a significant amplifying effect on market fluctuations. This trend continues with the extension of the observation window, with both the frequency and intensity of volatility events increasing simultaneously, reflecting a concentrated outbreak of market hedging responses following the de-pegging event.

Expanding the perspective to the market level, it is observed that coordinated mutation events exhibit similar patterns. Prior to the decoupling, there were signs of synchronized fluctuations in the market, particularly in the 1.5 hours leading up to the event, where the probability of coordinated mutations significantly increased. After the decoupling occurred, this characteristic of systemic fluctuations became more pronounced, with mutations appearing synchronously across multiple crypto assets, revealing that the instability of stablecoins could trigger broader market interconnected shocks.

(3) mutation amplitude

Statistical analysis of the Bitcoin price fluctuations after the USDT de-pegging event found that the magnitude of the fluctuation returns was significantly higher than the non-event period (control sample) across multiple time windows. Whether it was a positive or negative fluctuation, the average returns and medians were significantly higher in the post-de-pegging period compared to the control period, and the Wilcoxon rank-sum test rejected the hypothesis of consistent return distributions between the two groups at a 1% significance level, indicating that the price fluctuations triggered after the stablecoin de-pegging were more severe.

Further improve the recognition criteria, retaining only the most significant mutations (α=0.01%), defined as "ultra-large mutations". The results show that large mutation events are still concentrated during the decoupling period. Although the total number of observations in the control samples is nearly 10 times that of the decoupled samples, the large mutations are only 3.8 times that of the decoupled samples. Among all large mutations, 17% occurred within a 4-hour window after the decoupling event, indicating that the market is more prone to violent reactions after stablecoins deviate from their pegged prices.

In addition, the relative proportion and probability of significant deviations in the de-pegged samples are significantly higher than those in the control samples, with the likelihood of large deviations being 2.7 times that of the control samples. This result emphasizes that the de-pegging of USDT not only increases the frequency of price fluctuations but also significantly amplifies the volatility of its returns, demonstrating that the instability of stablecoins has an amplifying effect that can trigger strong reactions in the cryptocurrency market.

(4) Anchor Release Direction

Data shows that in Tether's de-pegging events, upward de-pegging (price above $1) is more common, occurring 196 times, while downward de-pegging (price below $1) has only occurred 41 times.

To this end, a separate event group was constructed that contained only downward decoupling events, and the event study was conducted again. The results showed that whether it was the price mutation of the BTC/USD exchange rate or the coordinated mutation of the entire cryptocurrency market, there was a significant increase within multiple time windows after the downward decoupling. Compared with the combined analysis of upward and downward decoupling, the conditional probability values and odds ratios derived from focusing solely on downward decoupling were higher, indicating that the downward decoupling had a stronger effect in triggering severe market volatility.

Further analysis of the directional differences in the magnitude of Bitcoin price mutations reveals that both positive and negative mutations show a significant deviation from the control period in their return distribution after decoupling events. Specifically, downward decoupling events typically lead to larger positive mutation magnitudes, with the maximum positive return reaching 2.9%, while the corresponding value for upward decoupling is 1.9%. In terms of negative mutations, the average and median mutation returns after downward decoupling are also slightly higher than those after upward decoupling. However, the largest absolute negative mutation during the sample period occurred after an upward decoupling event, indicating that individual extreme cases can still break general trends.

Additionally, in further examination of the probability of "large jumps," after controlling for sample size differences, it was found that although there are more upward decouplings (resulting in a higher total number of large jumps), the relative proportion and odds ratio of large jumps are significantly higher in downward decouplings. This indicates that the market reaction intensity triggered by the price decline of stablecoins is generally stronger and more likely to cause systemic fluctuations.

5. Further Analysis

The depegging event of the stablecoin USDT not only triggered market fluctuations after it occurred, but may also have prompted investor expectations in advance. An analysis of the conditions before the event revealed that prior to the USDT depegging, the conditional probability of BTC price fluctuations and market collaborative fluctuations showed a significant increase compared to the unconditional probabilities in the control sample, indicating that some market participants may have anticipated the depegging risk of the stablecoin and adjusted their positions in advance to avoid potential losses. This anticipatory behavior could lead to severe price volatility before the event. However, after replacing the depegging identification, this anticipatory effect was no longer significant, and the BTC/USD response to the event's prior fluctuation was no longer valid under the adjusted settings, resulting in a more stable market performance before the event. This conclusion also applies to the analysis of market collaborative fluctuations.

In contrast, the market reaction following the USDT depeg event was significantly enhanced. Within a few minutes to 4 hours after the depeg began, the probability of a BTC price spike increased dramatically, especially within 5 minutes of the event, where the spike probability exceeded 25%, far higher than the less than 1% level in the control sample. Even when the observation window is extended to 2 hours or 4 hours, this effect remains significantly present. For example, within 5 minutes of the depeg event, the likelihood of a Bitcoin price spike was 35 times that of normal periods, and even after 2 hours, it was still 3 times. The impact of the depeg event on market co-movement spikes was also extremely significant, with the probability of co-movement spikes occurring in the relevant window being 2 to 39 times that of normal periods.

The robustness analysis indicates that these effects remain statistically significant, but the magnitude of the impact diminishes after controlling for exogenous factors such as macro news, and the duration of the event impact correspondingly shortens to less than 4 hours. However, under various model specifications, the decoupling of USDT significantly increases the probability of BTC price jumps and market co-jumps, exhibiting stable statistical significance.

In addition to frequency, the magnitude of the mutations after the event has also significantly increased. Compared to the control period, there are significant differences in the distribution of positive and negative mutation returns of BTC price after the decoupling event, with both the mean and median of the absolute value of mutations significantly rising. This result indicates that when stablecoins fail to maintain their pegged status, it not only increases the probability of price mutations in non-stablecoin assets but also exacerbates their volatility. Further analysis shows that the direction of price deviation does not have a significant difference on this effect, whether it is positive or negative decoupling, the degree of market impact is comparable. Furthermore, large mutations are significantly more common during the decoupling period, indicating that the instability of stablecoins is an important trigger for severe market turbulence.

These findings reveal the dual characteristics of stablecoins. When the market operates normally, stablecoins indeed provide liquidity support and valuation anchoring for transactions, serving as a "stabilizer"; however, during extreme events or confidence shocks, stablecoins themselves can become a source of instability. Once stablecoins lose their peg, they no longer act as a buffer mechanism in the market, and may instead become a conduit for systemic risk due to their critical position and widespread use (especially as collateral in the DeFi ecosystem).

The instability of Tether indicates that when stablecoins fail to maintain their price stability objectives, it can trigger broader market volatility. The structural characteristic of this type of stablecoin being "normally stable but extremely unstable" in the cryptocurrency asset ecosystem highlights the potential systemic risks carried by stablecoins under the current market structure and institutional arrangements.

6. Conclusion

This article systematically examines the impact of the depeg events of the stablecoin Tether (USDT) on the jumps and co-jumps behavior of the cryptocurrency market based on high-frequency data. The study reveals the disruption mechanism of stablecoin instability on extreme volatility patterns in the market by constructing a robust depeg identification framework, incorporating event study and conditional probability analysis.

The results show that the de-pegging of USDT significantly increases the probability of a sudden change in Bitcoin prices, particularly in the short term following the de-pegging. This effect is observed not only in individual assets but also in the coordinated sudden change behavior at the market level, meaning that de-pegging events increase the likelihood of multiple assets experiencing price changes simultaneously. Further analysis indicates that the direction of de-pegging is an important factor influencing the intensity of market reactions: downward de-pegging (when the USDT price falls below the pegged value) typically causes a higher level of market disruption than upward de-pegging.

In addition, the statistical test of the mutation amplitude shows that after the depegging event, whether the price rises or falls, the distribution of mutation returns significantly deviates from the control sample and exhibits higher average and median amplitudes. This means that the instability of stablecoins not only increases the frequency of sudden fluctuations but also amplifies the amplitude of fluctuations, exacerbating market instability.

These findings have important policy and practical implications. On one hand, stablecoins have become key liquidity intermediaries in the crypto financial system, and the effectiveness of their anchoring mechanism directly affects the operational stability of the entire market. On the other hand, existing regulations tend to focus more on the asset coverage and fiat currency peg mechanisms of stablecoins, while lacking sufficient assessment of the systemic market responses triggered by price deviations. The quantitative evidence provided in this paper indicates that the decoupling of stablecoins is not only a failure of their own mechanisms but may also become a trigger point for extreme market fluctuations.

In summary, the risk identification and impact path assessment of stablecoins from a high-frequency perspective provide forward-looking warning basis for regulatory agencies, trading platforms, and DeFi protocol designers, and offer empirical foundation for building a more robust crypto financial infrastructure in the future.