Next Article in Journal
National Governance Index, Corruption Index and Growth Rate—International Evidence from Sub-Saharan and MENA Countries
Next Article in Special Issue
Bubble in Carbon Credits during COVID-19: Financial Instability or Positive Impact (“Minsky” or “Social”)?
Previous Article in Journal
Residual State Ownership and Firm Performance: A Case of Vietnam
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:

Impact of Negative Tweets on Diverse Assets during Stressful Events: An Investigation through Time-Varying Connectedness

Symbiosis Institute of Business Management, Symbiosis International (Deemed University), Electronic City, Phase-I, Bengaluru 560100, Karnataka, India
Author to whom correspondence should be addressed.
J. Risk Financial Manag. 2022, 15(6), 260;
Received: 12 May 2022 / Revised: 31 May 2022 / Accepted: 31 May 2022 / Published: 9 June 2022
(This article belongs to the Special Issue Global Trends and Challenges in Economics and Finance)


Tweets seem to impact diverse assets, especially during stressful periods. However, their interrelations during stressful events may change. Cryptos are apparently more sensitive to the sentiment spread by tweets. Therefore, a construct could be formed to study such complex interrelation during stressful events. This study found an interesting outcome while investigating three major asset classes (namely, Equity, Gold and Bond) alongside negative sentiment (derived from tweets of Elon Musk) and Dogecoin (an emerging asset class) from 1 June 2015 to 20 February 2022. Negative sentiment emerged as the significant risk transmitter, while Gold emerged as the significant net recipient of shocks (risk). Interestingly, Dogecoin was found to be less impacted and not impactful (not transmitting shock and receiving tiny shocks) at the same time. In fact, the interconnectedness between negative sentiment (percolated through Twitter) and Dogecoin prices was found to be rather feeble. Further, the study showed that the COVID-19 breakout and Brexit referendum in 2016 were less stressful events compared to the Greek debt crisis back in 2015.

1. Introduction

When Elon Musk wanted to become a full-time influencer and expressed the same through Twitter on 10 December 2021, it became viral.1 Elon Musk is one of the active users of Twitter, and many times, his messages have given a shock to the market (Jawad et al. 2022). Former US President Donald Trump, too, did not lag far behind. A recent paper proved the time-varying dependence of sentiments owing to Trump’s untimely tweets (Huynh). Some cardinal research questions emerged during our study, leading toward a thematic literature review. Can tweets emit negativity? Intuitively, negative emotions often leave a lasting imprint. Negative messages through social media platforms such as tweets accelerates four times quicker than positive messages (Scott 2019). The negative news is typically long-lasting, and it produces stress (Park 2015). Negative emotions have more impact than positive emotions, and negative news influences the investors more than positive news (Deeney et al. 2018). Similar to negative word of mouth, negative tweets are given importance in decision making, and they are considered to be more reliable (Hennig-Thurau and Wiertz 2015). The negative information influences human feelings, emotions and decisions, and it also results in quick responses (Park et al. 2016). One daunting question looms large: does this negative sentiment percolate down to volatile asset classes such as crypto?
Previous studies have explored how Twitter can impact the cryptocurrency market (Kraaijeveld and De Smedt 2020; Naeem et al. 2021; Öztürk and Bilgiç 2021; Shen et al. 2019). Though Bitcoin and Ethereum are the major players in the crypto market, Dogecoin, introduced as a joke back in 2013, found an influential follower in Elon Musk (Ren and Lucey 2022). The value of meme-based Dogecoin shot up nearly 20% within an hour after Elon Musk’s tweet.2 Based on the recent trends and studies, Musk’s tweets significantly impact the efficiency of the Dogecoin market (Jawad et al. 2022). The price explosiveness of Dogecoin is intensified based on Musk’s crypto-related tweets. Positive tweets increase Dogecoin’s trading volume and price. Further, they influence the volatility of Dogecoin compared with other cryptocurrencies. A significant price hike and more trading volume of Dogecoin during tweets shared on social media were found (Hamurcu 2022). Intuitively, tweets may not be the reasons for volatility, changes in price and trading volume. Some interesting hidden factors could have impacted Dogecoin. Elon Musk’s posts on social media are always astonishingly influencing and have emitted outlandish shocks in financial markets. In fact, Scott Paul offered to sell his house through Dogecoin with a 10% discount in December 2021.3 Interestingly, Dogecoin was one of the most volatile cryptocurrencies in the asset market, and after Musk’s posts, it has got hasty growth from a joke to a force to reckon with in the crypto universe (Hamurcu 2022). It is currently the 11th most valuable token as per and has a market capitalization of over USD 26 billion, according to CoinGecko.4 However, investigations into the effect of Musk’s tweets on Dogecoin have not been conducted. Despite being in highlight, serious research is yet to be conducted on some cardinal questions, such as: is there a time-varying relationship between a negative tweet and Dogecoin? Would Dogecoin move according to Twitter, or is there some other hidden pattern? Often, crypto is considered as a hedge against possible calamities.5 Usually, Gold serves as a universal hedge. Typically, Gold is a tangible asset. It has been found to be a safe-haven asset for centuries due to transferability, durability and its physical characteristics (Wen et al. 2022). Gold is a safe investment for domestic and international investors, particularly when the share market falls down badly (Gözde and Ünalmış 2014). However, identifying the best possible safe asset becomes a challenge during the financial crisis period, and some studies doubt the reasons for the intensity in investing the Gold commodity market being extremely high during the global economic and financial crisis (Baur and Glover 2010; Klein 2017; Stelios et al. 2017). However, before and after global financial crisis, Gold lost its attraction and appeared riskier to invest in (Shahzad et al. 2019). Some studies have examined cryptocurrencies, Gold and crude oil to measure the safe haven and found that Gold had much more negative skewness than cryptocurrency. Thus, Gold received fewer gains and some extreme losses among the investors (Shahzad et al. 2019). However, investors in the developed nations prefer Gold as a safe asset, and emerging markets such as China consider Bitcoin as one of the best choices than Gold, leaving behind the curbs by the government of China. Therefore, Gold (tangible asset) is considered a safe investment traditionally. In the digital world, there is a possibility of alternative e-investments (cryptocurrencies) such as Dogecoin to occupy the investment trend. Another question that emerges from the existing literature is: which is the better hedge, Gold or Dogecoin? Traditionally, Gold is the eternal hedge against most financial catastrophes.
Any stressful situation (such as a financial crisis or pandemic) emits shocks. It has an impact and fluctuation in the traditional assets (Gold, Equity and Bonds), real estate or cryptocurrencies. In the case of developed nations’ stock markets, Gold is exhibited as a weak and safe-haven asset because extreme downturn stocks in developed nations are not correlated with the best returns in Gold in the upcoming periods (Shahzad et al. 2019). Similarly, evidence showed that Bitcoin or traditional assets are not weak/strong safe-haven assets. However, in the emerging economies, traditional assets acted as weak safe haven, and Bitcoin was not in the picture. Therefore, for any such stressful events, Bitcoin received shocks in the global scenario. However, in the cryptocurrency market, especially the Dogecoin context, this type of textual analysis is from the perspective of network connectedness. Financial markets, like any other, are a complex network of agents with multiple agendas. Therefore, the risk is transmitted in a unique fashion through this network. Further, it has been observed that the total connectedness index or TCI (representing the summation of net shocks) shoots up rapidly (Bouri et al. 2021b) during microstructure disturbance caused owing to stressful situations (read as a financial crisis as well as a pandemic).
Financial investments such as Gold and Equity are interdependent by nature. Market participants could ideally differentiate between various asset classes as per EMH (Fama 1970). However, it is far from reality, where the interconnected dynamics between asset classes are extremely difficult to decipher (Shahzad et al. 2019). The connectedness among these investment options is apparent and complex to understand in any market in a developing economy during a financial crisis (Baur and Glover 2010; Klein 2017; Stelios et al. 2017). These investment options are highly interconnected with one another; if any investment receives the shock from any other external source, it impacts other connected investments (Ando 2019). Whether traditional asset classes are connected to cryptocurrency remains a puzzle to be solved. The cryptocurrency market has recently emerged as an attractive investment avenue for individual and institutional investors (Bouri et al. 2021b). The addition of cryptocurrency to the existing connectedness nexus has made it even more complex (Kraaijeveld and De Smedt 2020). Connectedness exists during calmer periods; however, it would be curious to examine whether network connectedness surged during extreme events such as COVID-19 outbreak. Lastly, the movement of TCI in an asset mix during extreme circumstances needs to be carefully studied. Questions regarding the same emerged as well from the review so far. Network connectedness in extreme circumstances, does it really surge? Typically, connectedness (TCI) surges rapidly during extreme events.

2. Literature Review

We further delved deep into three specific themes, namely Twitter and sentiments, crypto (especially Dogecoin) and connectedness, and impact on other assets. The first theme is to investigate the embedded sentiments of Musk’s tweets on different investment options. The second theme revolves around the impact of Twitter on the cryptocurrency market. The third is to understand the interconnectedness of different assets and how they impact each other during stressful events.

2.1. Twitter and Sentiments

Elon Musk’s negative tweets could be the most influential on Dogecoin and other interconnected assets, as he has more than 81 million tweet followers.6 Elon Musk is known for his active participation in Twitter. His tweets on cryptocurrencies are emitting shocks among the investors; subsequently, investors register positive and negative responses to various cryptocurrency options. Elon Musk’s followers and investors highly consider his tweets, and they have shocked the investment market, including cryptocurrency. Empirical findings unearth that people undergo negative emotions as a result of negative news from Twitter (Park 2015).
A couple of studies of late have found the predicting capacity of Twitter-based sentiments, making it an interesting proposition as a predictor variable. The number of tweets has been a proven proxy for investor attention (sentiment); it further determines both volume and volatility for crypto (Shen et al. 2019). In addition, Twitter-based sentiment is found to be quite a strong predictor for various crypto returns (Kraaijeveld and De Smedt 2020). Therefore, Twitter-based sentiments (especially negative) can be used as a justified proxy for understanding the market microstructure through connectedness during stressful events.

2.2. Crypto (Especially Dogecoin)

A recent study (Ante 2022) provides empirical evidence that Elon Musk’s referral content on social media (i.e., Twitter) could affect the crypto market. Another study (Huynh 2022) examined rigorously and identified that tone of his content drives the Bitcoin market. Recent studies confirmed that social networking sites influences cryptocurrency in the market and found that (Urquhart 2016) it is possible to predict the Bitcoin price, volatility and trading volumes based on using the number of hashtags on the Twitter platform. Elon Musk tweeted in early February 2021 that Tesla had spent USD 1.5 billion to purchase Bitcoin.7
Noticeably, the value of cryptocurrencies surged, especially Bitcoin, after the tweet. Bitcoin holds the top position among the cryptocurrencies (Yuan and Wang 2018). Recently (Huynh 2022), the link between Elon Musk’s tweet content and Bitcoin market was investigated. Investors may or may not perceive the messages as investment advice; however, they blindly followed Musk’s tweets to gain an advantage from the information in crypto trading (Huynh 2022). Musk intends to create positive statements on social media, but sometimes, his posts emit negative sentiments as well (Kraaijeveld and De Smedt 2020). The market observes shocks whenever tweets emit negativity at the receiver’s hand (Al Guindy 2021).
Our focus on this paper rests solely on Dogecoin, principally because of the patronage of Elon Musk, SpaceX and Tesla chief. On the advent of winter way back in 2013, a meme surfaced and was considered a joke. A meteoric spike of 800% in January 2021, followed by another strong upsurge of 400% in three months’ time, took it to the limelight. Critiques still questioned its apparently weak fundamentals. As per recent studies, tweets have impacted the value of Dogecoin transactions (Lansiaux 2022). However, its phenomenal growth has been strongly attributed to Twitter-driven sentiments, especially from Musk (Tjahyana 2021). Noticeably, the price of Dogecoin witnessed a steady decline whenever Musk’s perception was negative (Cary 2021).

2.3. Connectedness and Impact on Other Assets

To understand the complexity of the context, the study adopted VAR models at upper and lower percentiles through quantile regression, which allows it to capture the network connectedness associated with extreme negative shocks. This remains the same even during the economic crisis because the changes in the market interdependencies and return had significant connectedness during extreme events (Mensi et al. 2016). The interesting pivotal factor is uncovering the connectedness of shocks in a predetermined network through the approach of connectedness (Shahzad et al. 2019). The COVID-19 shock led to a large disruption in the global financial markets (Corbet et al. 2020). Recent studies proved that the COVID-19 outburst had a considerable effect on financial markets, including the cryptocurrency market; the total connectedness index (TCI) has been observed to spike rapidly around the declaration of the COVID-19 outbreak in early 2020 (Bouri et al. 2021a; Naeem et al. 2021). Further, extreme circumstances such as COVID-19 add more complexity to the network connectedness among the variables. The financial system is typically an interconnection of various assets consisting of Gold, Bonds, Equity, cryptocurrencies and sentiment, which often cannot be explained by EMH but can be explained by both AMH and FMH (Fama 1970; Lo 2004; Peters 1994). The remaining of the paper goes as follows: Section 3 presents the data and methodology, Section 4 depicts results and illustrates the interpretations, Section 5 includes the conclusive remarks and, finally, Section 6 is about limitations and scope for future studies.

3. Research Methodology

3.1. TVP-VAR

We have followed Antonakakis and Gabauer’s research work while deploying the TVP-VAR model to measure the time-varying connectedness amongst the target variables (Antonakakis et al. 2018). This methodology is typically an extension of the model proposed and perfected by Diebold and Yilmaz for analyzing dynamic connectedness (Diebold and Yilmaz 2012, 2014). The undue subjectivity around a certain window size was never a problem again with the advent of TVP-VAR. Further, relatively smaller sample sizes could well be accommodated by this method. The equations below describe the TVP-VAR model:
Z t = B t Z t 1 + u t u t ~ N ( 0 , S t )
v e c ( B t ) = v e c ( B t 1 ) + v t v t ~ N ( 0 , R t )
where Z t , Z t 1 and the error term u t are vectors having dimension k × 1, B t and S t are matrices having dimensions n × n; v e c ( B t ) , is of dimension k2 × 1. Here, all information that is available until t − 1 is given by p t 1 . The other error term v t has dimensions k2 × 1 while R t has a dimension k2 × k2. St and R t represents the variance–covariance matrices, which vary with time.
Scaled generalized forecast error variance decomposition (GFEVD) was calculated as the next step following Koop, Pesaran, etc. (Koop et al. 1996; Pesaran and Shin 1998). Unlike the pre-existing model, GFEVD is entirely invariant to the ordering of variables. Further, to obtain GFEVD, TVP-VAR is transformed to vector moving average representation or TVP-VMA using the Wold theorem by the following transformation:
Z t = i = 1 p B i t Z t i + u t
φ i j , t g ( H ) , is the unscaled GFEVD, which is further normalized to the scaled version. This would ensure that the summation in each row is unity. Again, this implies the pairwise directional connectedness from variable j to variable i. To compute the above terms, the following step is conducted:
φ ˜ i j , t g ( H ) = φ i j , t g ( H ) / j = 1 k φ i j , t g ( H )
In addition, the connectedness measures are derived as suggested by Diebold and Yilmaz (2012, 2014).
T O j t = i = 1 , i j k φ ˜ i j , t g ( H )
F R O M j t = i = 1 , i j k φ ˜ j i , t g ( H )
N E T j t = T O j t F R O M j t
T C I t k 1 j = 1 k F R O M j t
We find Equation (5) as a measure of total directional connectedness from j TO all others in the network, whereas Equation (6) is a measure of total directional connectedness to j FROM all others. Moreover, Equation (7) is obtained as the difference between (5) and (6) and indicates the net total direction of connectedness associated with j. For instance, if N E T j t > 0, shows j is a net driver who is involved in transmitting shocks. Equation (8) is an aggregate measure of the total connectedness amongst all and serves as a proxy to the overall interconnectedness. Typically, a higher TCI would indicate a shock in a particular variable affecting the network. Usually, TCI is in a higher range during extreme events and subsequently reduces as the stress gets relieved.

3.2. Textual Analysis

We collected all 14,206 tweets posted to Elon Musk’s account (@elonmusk) from 1 June 2015 to 20 February 2022. We selected specific negative words (see Table 1), adopting Loughran and McDonald (2011) since it specifies sentiment words used by Tetlock (2007) for financial context analysis (Tetlock 2007; Loughran and McDonald 2011). We carefully went through the context of using those words. Therefore, we followed existing research to create a proxy for the negative sentiment (Huynh 2021).
ρ = n N × 100
where ρ represents Ngsnt or negative sentiment, n is the number of negative words and N is the total number of words.

3.3. Data Details

We considered daily observations from 1 June 2015 to 20 February 2022. We had a total of five variables, namely Gold (GC.CMX), Bond (pimco), Equity (GSPC), Dogecoin (Doge) and Negative sentiment or Ngsnt (calculation is shared in the previous section). The first four variable information were gathered from BLOOMBERG, whereas Ngsnt was calculated based on the Twitter feed (see representation in Figure 1 and Figure 2).

4. Results and Interpretation

This section exhibits all the outcomes and tries to explain the reason underneath. Interpretation and implication also extend to the next section.
We found a leptokurtic (heavy-tailed distribution) trail in all out variables, however, with varying degrees. The negative sentiment (Ngsnt) was significantly leptokurtic compared to others (see Table 2). This seems rather logical, though. Bond is second in the list (referring to Kurtosis), having a substantially higher skewed nature over the rest. The presence of fat-tails or heavy tails in all the variables under consideration proves significant deviation from EMH due to behavioral traces (represented by Ngsnt). The stationarity of the variables has been proved by ERS test (Elliot et al. 1996).
The negative sentiment (Ngsnt) was found to be 3.3 times stronger net emitter over GSPC (S&P 500) (see Table 3). Results remain consistent with the fundamental intuitive understanding. Tweets (negative) of Elon Musk emitted the shocks toward all other asset classes. This disapproves EMH and establishes the fundamental premise of behavioral finance. Moreover, it is substantially stronger than that of shocks generated from Equity markets (S&P 500 or GSPC). Further, Ngsnt impact (risk emission) on Gold, Bond and Equity are almost identical (namely, 0.41, 0.43 and 0.47). Therefore, we can infer that negative sentiment percolated by Twitter messages impacts similarly to the traditional asset classes.
Gold has received a shockwave three times over that of Dogecoin (see Table 3). We observed that both Gold (represented by GC.CMX) and Bond (represented by pimco) received shocks; however, their quantum varied. While Gold received the bulk of it, Bonds received at a moderate level. Interestingly, Dogecoin received low levels of shocks, which can be explained by the unexplored nature of cryptocurrencies barring Bitcoin and perhaps Ethereum. Gold and Bond being traditional and age-old asset classes, typically find significantly higher traction.
Another interesting observation (see Figure 3 and Figure 4) exhibits that dynamic PCI between Equity (GSPC) and Bond (pimco) surged sharply to 75% (from 25%) around the COVID-19 outbreak.
Typically, stressful events increase connectedness (TCI). We found that TCI was more than 70% during 2015–16 (see Figure 5) and nosedived during 2017–19. TCI was stable until the COVID-19 breakout (around 20%). Spike was observed at 38% (see Figure 5 and Table 4) as COVID-19 surfaced globally (early 2020). Perhaps the most interesting observation was yet to come. We found COVID-19 breakout was surprisingly less impactful (see Table 4) when compared to the Greek debt crisis back in 2015 (half of it). TCI during the Greek debt crisis was around double the mark than in the COVID-19 outbreak.
Typically, TCI is supposed to surge when stress increases in its underlying assets. Further, TCI decreases as stress reduces. Therefore, excess TCI indicates a stressful situation. Figure 5 identifies two zones (see Table 4) with higher levels of TCI values.

5. Conclusive Remarks

We found four research queries during our extensive literature review (which is written all across this paper). First, we wanted to empirically confirm whether Twitter emits negative sentiment. Interestingly, it was found that the Twitter induced negative sentiment (Ngsnt) was found to be three times stronger net emitter over Equity, which is represented by GSPC (or S&P 500) (see Table 3 and Table 5). Second, our query was regarding the negative tweet (Ngsnt) and Dogecoin prices interconnectedness. Their interconnectedness was found to be quite feeble in comparison to other asset classes (see Table 3). The relative newness of Dogecoin over Bitcoin and Ethereum could justify this outcome. Third, we checked whether Gold as a traditional hedge wins over Dogecoin or not. It was found that Gold has received three times (3.68 vs. 1.21) shockwave over that of Dogecoin (see Table 3 and Table 5). This proves that Gold remains a better hedge over Dogecoin by a margin. Last, we have empirically checked the ‘stylized fact’ of connectedness (TCI) shooting up during extreme events. We found that it does; however, the COVID-19 breakout was surprisingly less impactful (see Table 4) when compared to the Greek debt crisis back in 2015 (in fact, half of it). Shockingly, Brexit (23 June 2016) did not seem to have much impact on TCI, as it stayed around 17% levels. This further proves that all stressful events do not have a similar impact on time-varying connectedness. Therefore, many useful findings emerged during this humble attempt, which are perfectly in sync with the latest studies around the same domain (Ante 2022; Huynh 2021, 2022). In addition to that, these outcomes can be of some use to the policymakers and portfolio managers alike.

6. Limitations and Scope for Future Studies

Twitter was purchased just after this paper was sent for publication. Therefore, new research can be carried forward in a different direction altogether. Instead of Dogecoin, other coins can be investigated as well. In addition to that, other methods such as DCC-GARCH can be used to check the connectedness.

Author Contributions

All authors equally contributed to completing this article successfully. N.L.B. contributed to conceptualization, data collection, writing the introduction and literature review; B.G. contributed concept development, methodology, use of appropriate software analysis and conclusive remarks; S.M. contributed towards data curation, validation and editing of the entire manuscript. All authors have read and agreed to the published version of the manuscript.


This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Available on request.


We hereby acknowledge Toan L D Huynh for providing a large part of Twitter data.

Conflicts of Interest

The authors declare no conflict of interest.


4 (accessed on 18 April 2022).


  1. Al Guindy, Mohamed. 2021. Cryptocurrency price volatility and investor attention. International Review of Economics and Finance 76: 556–70. [Google Scholar] [CrossRef]
  2. Ando, S. 2019. International Financial Connection and Stock Return Comovement. In IMF Working Papers. Washington, DC: Inernational Monetary Fund, vol. 16. [Google Scholar] [CrossRef]
  3. Ante, Lennart. 2022. How Elon Musk’s Twitter Activity Moves Cryptocurrency Markets. SSRN Electronic Journal 16: 1–28. [Google Scholar] [CrossRef]
  4. Antonakakis, Nikolaos, David Gabauer, Rangan Gupta, and Vasilios Plakandaras. 2018. Dynamic connectedness of uncertainty across developed economies: A time-varying approach. Economics Letters 166: 63–75. [Google Scholar] [CrossRef][Green Version]
  5. Baur, B. Dirk Glover, and Kristoffer J. Glover. 2010. The Destruction of a Safe Haven Asset? pp. 8–15. Available online: (accessed on 13 March 2022).
  6. Bouri, Elie, Oguzhan Cepni, David Gabauer, and Rangan Gupta. 2021a. Return connectedness across asset classes around the COVID-19 outbreak. International Review of Financial Analysis 73: 101646. [Google Scholar] [CrossRef]
  7. Bouri, Elie, Tareq Saeed, Xuan Vinh Vo, and David Roubaud. 2021b. Quantile connectedness in the cryptocurrency market. Journal of International Financial Markets, Institutions and Money 71: 101302. [Google Scholar] [CrossRef]
  8. Cary, Michael. 2021. Down with the # Dogefather: Evidence of a Cryptocurrency Responding in Real Time to a Crypto-Tastemaker. Journal of Theoretical and Applied Electronic Commerce Research 16: 2230–40. [Google Scholar]
  9. Corbet, Shaen, Charles James Larkin, and Brian Lucey. 2020. The contagion effects of the COVID-19 pandemic: Evidence from gold and cryptocurrencies. Finance Research Letters 35: 1015–54. [Google Scholar] [CrossRef]
  10. Deeney, Peter, Mark Cummins, Michael M. Dowling, and Alan Smeaton. 2018. Negativity Bias in the European Emissions Market: Evidence from High-Frequency Twitter Sentiment. SSRN Electronic Journal. Available online: (accessed on 11 March 2022).
  11. Diebold, Francix X., and Kamil Yilmaz. 2012. Better to give than to receive: Predictive directional measurement of volatility spillovers. International Journal of Forecasting 28: 57–66. [Google Scholar] [CrossRef][Green Version]
  12. Diebold, Francix X., and Kamil Yilmaz. 2014. On the network topology of variance decompositions: Measuring the connectedness of financial firms. Journal of Econometrics 182: 119–34. [Google Scholar] [CrossRef][Green Version]
  13. Elliot, Graham, Thomas J. Rothenberg, and James H. Stock. 1996. Efficient Tests for an Autoregressive Unit Root. Econometrica 64: 813–36. [Google Scholar] [CrossRef][Green Version]
  14. Fama, Eugene. 1970. Efficient Capital Markets: A Review of Theory and Empirical Work. Journal of Finance 25: 383–417. [Google Scholar] [CrossRef]
  15. Gözde, Gurgun, and İbrahim Ünalmış. 2014. Is gold a safe haven against equity market investment in emerging and developing countries? Finance Research Letters 11: 341–48. [Google Scholar]
  16. Hamurcu, Cagri. 2022. Can Elon Mask’s Twitter Posts About Cryptocurrencies Influence Cryptocurrency Markets by Creating a Herding Behavior Bias? Fiscaoeconomia 6: 215–28. [Google Scholar] [CrossRef]
  17. Hennig-Thurau, Thorsten, and Caroline F. Wiertz. 2015. Does Twitter matter? The impact of microblogging word of mouth on consumers’ adoption of new movies. Journal of the Academy of Marketing Science 6: 375–94. [Google Scholar] [CrossRef]
  18. Huynh, Toan Luu Duc. 2021. Does Bitcoin React to Trump’s Tweets? Journal of Behavioral and Experimental Finance 31: 100546. [Google Scholar] [CrossRef]
  19. Huynh, Toan Luu Duc. 2022. When Elon Musk Changes his Tone, Does Bitcoin Adjust Its Tune? Computational Economics 13: 1–23. [Google Scholar] [CrossRef]
  20. Jawad, Syed, Hussain Shahzad, Muhammed Anas, and Elie Bouri. 2022. Price explosiveness in cryptocurrencies and Elon Musk’s tweets. Finance Research Letters 47: 102695. [Google Scholar] [CrossRef]
  21. Klein, Tony. 2017. Dynamic Correlation of Precious Metals and Flight-to-Quality in Developed Markets. Finance Research Letter 23: 283–90. [Google Scholar] [CrossRef]
  22. Koop, Gary, M. Hashem Pesaran, and Simon M. Potter. 1996. Impulse response analysis in nonlinear multivariate models. Journal of Econometrics 74: 119–47. [Google Scholar] [CrossRef]
  23. Kraaijeveld, Olivier, and Johannes De Smedt. 2020. The predictive power of public Twitter sentiment for forecasting cryptocurrency prices. Journal of International Financial Markets, Institutions & Money 65: 101188. [Google Scholar] [CrossRef]
  24. Lansiaux, Edouard. 2022. Community Impact on a Cryptocurrency: Twitter Comparison Example between Dogecoin and Litecoin Community Impact on a Cryptocurrency. Frontiers in Blockchain 5: 829–65. [Google Scholar] [CrossRef]
  25. Lo, Andrew. 2004. The adaptive market hypothesis: Market efficiency from an evolutionary perspective. The Journal of Portfolio Management 30: 15–29. [Google Scholar] [CrossRef]
  26. Loughran, Tim, and Bill McDonald. 2011. When Is a Liability Not a Liability? The Journal of Finance 66: 35–65. [Google Scholar] [CrossRef]
  27. Mensi, Walid, Shawkat Hammoudeh, Duc Khuong Nguyen, and Sang Hoon Kang. 2016. Global financial crisis and spillover effects among the US and BRICS stock markets. International Review of Economics & Finance 42: 257–76. [Google Scholar]
  28. Naeem, Muhammad Abubakr, Imen Mbarki, Muhammed Tahir Suleman, Xuan Vinh Vo, and Syed Jawad Hussain Shahzad. 2021. Does Twitter Happiness Sentiment predict cryptocurrency? International Review of Finance 21: 1529–38. [Google Scholar] [CrossRef]
  29. Öztürk, Serda Selin, and M. Emre Bilgiç. 2021. Twitter & Bitcoin: Are the most influential accounts really influential? Applied Economics Letters 29: 1–4. [Google Scholar]
  30. Park, Chang Sup. 2015. Applying “Negativity Bias” to Twitter: Negative News on Twitter, Emotions, and Political Learning. Journal of Information Technology and Politics 12: 342–59. [Google Scholar] [CrossRef]
  31. Park, Seunghyun Brian, Jang Jichul, and Micheal O. Chihyung. 2016. Analyzing Twitter to explore perceptions of Asian restaurants. Journal of Hospitality and Tourism Technology 7: 405–22. [Google Scholar] [CrossRef]
  32. Pesaran, H. Hashem, and Yongcheol Shin. 1998. Generalized impulse response analysis in linear multivariate models. Economics Letters 58: 17–29. [Google Scholar] [CrossRef]
  33. Peters, Edgar. E. 1994. Fractal Market Analysis. Applying Chaos Theory to Investment & Economics. Hoboken: Wiley. [Google Scholar]
  34. Ren, Ballis, and Brain Lucey. 2022. Do clean and dirty cryptocurrency markets herd differently? Finance Research Letters 47: 102795. [Google Scholar] [CrossRef]
  35. Scott, Kim V. 2019. The Impact of Negative Twitter Feeds on Washington Metropolitan Area Transit Authority Ridership. Hampton: Hampton University. [Google Scholar]
  36. Shahzad, Syed Jawad Hussian, Elie Bouri, David Roubaud, Ladislav Kristoufek, and Brain Lucey. 2019. Is Bitcoin a better safe-haven investment than gold and commodities? International Review of Financial Analysis 63: 322–30. [Google Scholar] [CrossRef]
  37. Shen, Dehua, Andrew Urquhart, and Pengfel Wang. 2019. Does twitter predict Bitcoin ? Economics Letters 174: 118–22. [Google Scholar] [CrossRef]
  38. Stelios, Bekiros, Sabri Boubaker, Duc Khuong, and Gazi Salah Uddin. 2017. Black swan events and safe havens: The role of gold in globally integrated emerging markets. Journal of International Money and Finance 73: 317–34. [Google Scholar]
  39. Tetlock, Paul. 2007. Giving Content to Investor Sentiment: The Role of Media in the Stock Market. The Journal of Finance 62: 1139–68. [Google Scholar] [CrossRef]
  40. Tjahyana, Lady Joanne. 2021. Brand Monitoring for Dogecoin Cryptocurrency on Twitter. Paper presented at 2nd International Conference on Social Sciences and Humanities, Surabaya, Indonesia, July 30; pp. 131–21. [Google Scholar]
  41. Urquhart, Andrew. 2016. The inefficiency of Bitcoin. Economics Letters 148: 80–82. [Google Scholar] [CrossRef]
  42. Wen, Fenghua, Xi Tong, and Xiaohang Ren. 2022. International Review of Financial Analysis Gold or Bitcoin, which is the safe haven during the COVID-19 pandemic ? International Review of Financial Analysis 81: 102121. [Google Scholar] [CrossRef]
  43. Yuan, Yong, and Feiyue Y. Wang. 2018. Blockchain and cryptocurrencies: Model, techniques, and applications. IEEE Transactions on Systems, Man, and Cybernetics: Systems 48: 1421–28. [Google Scholar] [CrossRef]
Figure 1. Raw data representing all the five variables.
Figure 1. Raw data representing all the five variables.
Jrfm 15 00260 g001
Figure 2. Transformed data representing all the five variables.
Figure 2. Transformed data representing all the five variables.
Jrfm 15 00260 g002
Figure 3. Dynamic pairwise connectedness index (PCI).
Figure 3. Dynamic pairwise connectedness index (PCI).
Jrfm 15 00260 g003
Figure 4. Net pairwise directional connectedness (NPDCI).
Figure 4. Net pairwise directional connectedness (NPDCI).
Jrfm 15 00260 g004
Figure 5. TCI over the years (from June 2015 to February 2022).
Figure 5. TCI over the years (from June 2015 to February 2022).
Jrfm 15 00260 g005
Figure 6. Network plot analysis depicting the net emitter of shocks (blue) and the net receiver of shocks (yellow).
Figure 6. Network plot analysis depicting the net emitter of shocks (blue) and the net receiver of shocks (yellow).
Jrfm 15 00260 g006
Table 1. Showcasing the negative words which are taken into consideration.
Table 1. Showcasing the negative words which are taken into consideration.
S. NoNegative WordsOccurrences
Context was paid attention to while identifying the negative words, following Loughran and McDonald (2011).
Table 2. Descriptive statistics of all five variables.
Table 2. Descriptive statistics of all five variables.
Skewness20.47 ***2.99 ***2.76 ***7.79 ***4.94 ***
Kurtosis671.57 ***9.6 ***11.82 ***120.8 ***43.94 ***
JB46,325,450 ***132,289 ***17,365 ***1,518,545 ***207,602 ***
ERS−18.609 ***−2.795 ***−12.080 ***−10.884 ***−10.853 ***
Q (10)14.7 ***12,817.7 ***105.3 ***701.31 ***934.93 ***
Q2(10)0.07710,720.5 ***89.8 ***620.8 ***1216.2 ***
*** indicate statistical significance at 1% level.
Table 3. Depicting average dynamic connectedness.
Table 3. Depicting average dynamic connectedness.
Ngsnt Doge GC CMXPimcoGSPCFrom
TO 7.21.3520.0425.3826.7580.72
Average dynamic connectedness is depicted in this table for all the variables using TVP-VAR model.
Table 4. Events linked with relatively higher TCI values.
Table 4. Events linked with relatively higher TCI values.
TCIDate RangeEventInterpretations
75%June–July 2015Greek debt crisisStressful events pushed the TCI to an extreme limit
17%June–July 2016Brexit referendum Stressful events pushed the TCI to a relatively smaller limit
38%March–April 2020COVID-19 breakoutStressful events pushed the TCI to a relatively higher limit
This table depicts the intensity of TCI during two stressful events (namely Greek debt crisis and COVID-19).
Table 5. Network connectedness table (showcasing net emitters/receivers).
Table 5. Network connectedness table (showcasing net emitters/receivers).
BlueLargeNgsntSignificant net transmitter of shocks; high weighted average net total directional connectedness (see Figure 6)
SmallGSPCLow net transmitter of shocks; low weighted average net total directional connectedness (see Figure 6)
YellowLargeGC CMXSignificant net receiver of shocks; high weighted average net total directional connectedness (see Figure 6)
MediumpimcoModerate net receiver of shocks; moderate weighted average net total directional connectedness (see Figure 6)
SmallDogeLow net receiver of shocks; low weighted average net total directional connectedness (see Figure 6)
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Balasudarsun, N.L.; Ghosh, B.; Mahendran, S. Impact of Negative Tweets on Diverse Assets during Stressful Events: An Investigation through Time-Varying Connectedness. J. Risk Financial Manag. 2022, 15, 260.

AMA Style

Balasudarsun NL, Ghosh B, Mahendran S. Impact of Negative Tweets on Diverse Assets during Stressful Events: An Investigation through Time-Varying Connectedness. Journal of Risk and Financial Management. 2022; 15(6):260.

Chicago/Turabian Style

Balasudarsun, N. L., Bikramaditya Ghosh, and Sathish Mahendran. 2022. "Impact of Negative Tweets on Diverse Assets during Stressful Events: An Investigation through Time-Varying Connectedness" Journal of Risk and Financial Management 15, no. 6: 260.

Article Metrics

Back to TopTop