*2.2. How the COVID-19 Pandemic Intervened on the Economy Worldwide*

At the beginning of 2020, the economy of China started to be influenced by COVID-19, earlier than other countries. Moreover, as the world's hub for global manufacturing and trade, immediate adverse effects on the Chinese economy resulted in global impacts [16]. Different regulations have been applied to handle the disease, such as closing national borders as well as stopping business activities across the world, strongly influencing the global economy [16]. Eventually, the global financial panic in March 2020 took place. In [18], the authors pointed out that the similarity calculated by ACC and ADCC models between the US and Chinese markets increased dramatically during the pandemic. Regarding the stock prices, when the pandemic occurred, the prices of the US and Chinese stocks decreased but started to recover again since July 2020. This trend is also true for other emerging and developed stock markets in different countries from different continents such as Japan, Germany, Australia and Canada [34]. Likewise, even less risky assets such as gold were adversely affected [35]. The increase in the correlation between different financial markets in the presence of good and bad news has been observed for some decades. In [36], the authors stated that stocks are more affected by the presence of bad news, compared to good news. Moreover, bad news has a stronger correlation in traditional markets. These results align with what happened during the COVID-19 pandemic. Although the world continued facing different COVID-19 waves afterwards, its impact on different asset classes lessened significantly [37], stock prices increased and volatilities decreased again to their original values before the pandemic [38]. Furthermore, the connectedness between different assets also experienced a major decline [39].

In [19], the authors investigated the impact of the COVID-19 pandemic on the cryptocurrency market by using daily prices of 45 well-known cryptocurrencies between September 2019 and April 2020—the majority of which are also used in our present study. In particular, they measured the stability of cryptocurrency time series using Largest Lyapunov Exponent and Approximate Entropy. All time series are divided into two parts: the first part spans September to December 2019, considered normal time, while the second spans January to April 2020, considered a pandemic period. They revealed that the pandemic increases in cryptocurrency market uncertainty as prices fluctuated significantly. Moreover, the same experiment has also been carried out on the stock market, results indicating a lower level of price fluctuations in the stock compared to digital currencies. Also on the same topic, Drozdz et al. [21] compared the Pearson correlation between the cryptocurrency market and different asset classes including stocks, fiat currencies and commodities, revealing that these conventional markets easily influence the cryptocurrency market when they are in turbulent times, while there is no significant correlation between digital currencies and other markets in normal times, given the time resolutions they used are 10 and 360 mins.

Reactions of the general public to the COVID-19 outbreak were also observed to examine its relationship with cryptocurrencies' returns. For example, authors in [40] measured the fear of people by the frequency of occurrence of keywords *COVID-19* and *coronavirus* on Google Trends (https://www.thinkwithgoogle.com/, accessed on 4 August 2022). Thanks to the vector autoregressive (VAR) models, they compared the evolution of this fear with the stock market's expectation of volatility VIX index (the VIX index is a measure of constant, 30-day expected volatility of the US stock market, derived from real-time, mid-quote prices of *S*&*P*500. Normally, it is calculated using the Black–Scholes formula) as well as the Bitcoin returns. They found that increases of fear can lead to Bitcoin crashes, as the correlation coefficient is −0.9. Furthermore, negative sentiment generated by coronavirus news is associated with market volatility, which is in line with other findings such as in [41]. Interestingly, some studies on the relationship between news-based sentiment and cryptocurrencies showed that, although both bad and good news cause the change in the returns and volatilities of cryptocurrencies, positive news has more effect on the volatilities and returns of cryptocurrencies in comparison with negative news [42–44].

Recently, network analysis in the cryptocurrency market during the COVID-19 pandemic has been carried out, with the common result being that the pandemic, as well as the global downturn, actually caused a change in the network structure of the cryptocurrency market. Specifically, cryptocurrencies tend to form bigger groups during the downtime, i.e., the number of potential clusters found in the network decreases during the downtime, with a few cryptocurrencies acting as central nodes. This topic has only been explored in a few studies to date [21,22,45,46]. Moreover, there are some gaps: (1) the lack of deep investigation of the network structure as they only consider MSTs; (2) the noise and trend effects are not removed; (3) data limitation issues.

We will address these shortcomings by doing deeper experiments on the network structure of the cryptocurrency market before, during and after the COVID-19 pandemic via a longer dataset with the effect of noise and trend removed. In addition, we will look at the way cryptocurrencies form a group during turbulent times by considering their rankings (identified by its market capitalization, the larger its maket capitalization, the higher its rank). We believe that this research can propose a better understanding of interconnections between digital currencies during standard and unstable periods. Furthermore, we also aim at understanding the investment decision of investors in different market states based on the results of community detection.
