WHEN ANALYSTS MEET CRYPTO: EVIDENCE FROM CORPORATE DISCLOSURE By Shuting Wu A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Business Administration - Accounting - Doctor of Philosophy 2024 ABSTRACT Improving the accounting and disclosure of crypto assets to provide decision useful information is the goal of FASB Accounting Standards Update (ASU) 2023-08. This study investigates whether firms’ voluntary disclosure about their exposure to cryptocurrency is informative to financial analysts. Applying textual analysis to 8-K filings, I find that the intensity of firms’ crypto-related disclosure is associated with more favorable target price revisions by analysts, but not near-term earnings forecast revisions. These findings imply that analysts perceive firms’ crypto exposure to have a long-term benefit on firm value, but no immediate impact on operational performance. Cross-sectional analyses further show that analyst target price revisions contain more optimism when: 1) firms’ crypto disclosure is related to corporate governance, and 2) firms’ stock returns comove more closely with the cryptocurrency market returns. Finally, I find that analysts’ target price revisions appear to enhance the stock market’s positive reaction to firms’ crypto-related disclosure in 8-K filings. My findings suggest the potential effectiveness of FASB’s ASU for crypto assets. ACKNOWLEDGEMENTS I am thankful for the members of my dissertation committee for their invaluable guidance and support: Professors Isabel Wang (chair), Ranjani Krishnan, John Jiang, Musaib Ashraf, and Jeffrey Wooldridge. They have all contributed to my PhD journey in enlightening me to approach academic research with both critical and creative thinking and training me to grow as an accounting researcher. Especially, I wish to express my sincere gratitude to my advisor, Isabel Wang, for her immense support and care, for her invaluable guidance on my research and career, and for her unwavering mentorship and encouragement every step of my doctoral program. I am thankful for Michigan State University for its financial support. I thank fellow PhD students who overlap with me in the program. I thank my friends for always being in my corner. I am deeply grateful to my family for their love and support throughout this journey, and for always believing in me and encouraging me to pursue my dreams. iii TABLE OF CONTENTS CHAPTER 1: INTRODUCTION ....................................................................................................1 CHAPTER 2: BACKGROUND AND RELATED LITERATURE ...............................................8 CHAPTER 3: RESEARCH DESIGN ............................................................................................16 CHAPTER 4: RESULTS ...............................................................................................................23 CHAPTER 5: CONCLUSION ......................................................................................................30 REFERENCES ..............................................................................................................................32 APPENDIX A: CRYPTO-RELATED KEYWORDS AND 8-K FILINGS .................................37 APPENDIX B: FREQUENT WORDS AROUND CRYPTO-RELATED KEYWORDS ...........39 APPENDIX C: VARIABLE DEFINITION ..................................................................................40 APPENDIX D: FIRM CHARACTERISTICS FOR 8-K FILINGS ..............................................42 APPENDIX E: TABLES OF EMPIRICAL RESULTS ................................................................43 iv CHAPTER 1: INTRODUCTION In this paper, I examine whether financial analysts respond to firms’ voluntary disclosure about their exposure to cryptocurrency in 8-K filings (crypto-exposure hereafter).1 Despite the dramatic surge of public interest in cryptocurrency and companies’ exposure to it, research on the informativeness of crypto assets disclosure remains limited. Currently, cryptocurrency is recorded as an intangible asset with indefinite life, subject to impairment write-down but not write-up, even if the cryptocurrency’s fair value later increases. Recognizing crypto assets as indefinite-lived intangible assets does not provide timely information for users. Recent FASB Accounting Standards Update, Intangibles—Goodwill and Other—Crypto Assets (Subtopic 350- 60) requires firms to apply fair-value accounting for crypto assets and disclose details about their cryptocurrency holdings (FASB 2023). Effective in December 2024, FASB ASU 2023-08 aims to refine crypto asset accounting and disclosure in providing decision useful information for users. Some firms have already provided voluntary disclosure about their crypto exposure. Assessing whether firms’ voluntary disclosure is informative to a sophisticated group of users— financial analysts—helps understand the potential effectiveness of mandatory disclosure with the upcoming FASB ASU 2023-08. The paper is also motivated by the unsettled debate on cryptocurrency as merely hype or getting real. In Management Science’s special issue on blockchains and crypto economics, Biais, Capponi, Cong, Gaur, and Giesecke (2023) calls for empirical studies that help understand whether the crypto sector is getting real or can get real. Some practitioners see the potential of blockchain-based cryptocurrency to change the way the economy operates, while others view it as hype. The proliferation of cryptocurrency coincides with a surge in public interest. 1 Throughout the paper I use the term “crypto exposure” to refer to firms’ 8-K disclosure of operating, investing, financing, and corporate governance activities that involve cryptocurrency. 1 “Dogecoin” and “Ethereum Price” are among the top 10 most searched news terms in 2021 (Google 2021). Google Trends reported “Bitcoin” as the second most searched global news term in 2017 (Google 2017). Perhaps not surprisingly, Deloitte’s Global Blockchain Survey finds that leaders of the global financial services industry “view digital assets and their underlying blockchain technology as a strategic priority now and in the near future” (Deloitte 2021a, page 2). Brian Armstrong, the founder of the cryptocurrency platform Coinbase, believes that a true breakout in the cryptocurrency economy would arrive if Wall Street took it seriously (Roberts 2020). On January 10, 2024, the U.S. SEC approved eleven Bitcoin spot ETFS, enabling investors an easy way to add Bitcoin exposure to their investment portfolios. However, some sophisticated investors remain skeptical of the value of cryptocurrency. Warren Buffett called Bitcoin “probably rat poison squared” ahead of the Berkshire Hathaway annual shareholder meeting in 2018 and reaffirmed his aversion to cryptocurrencies because “cryptocurrencies basically have no value, and they don’t produce anything” in an CNBC interview in 2020 (Bursztynsky 2020). Buffett’s long-time business partner, Charlie Munger, further stated that Bitcoin is “disgusting and contrary to the interests of civilization” (Li 2021). These controversial sentiments to cryptocurrency suggest the complexity and uncertainty underlying firms’ exposure to cryptocurrency. Financial sophistication and knowledge of analysts prove to be valuable in the traditional stock market, especially when the uncertainty is high (Frankel, Kothari, and Weber 2006; Loh and Stulz 2011; Bradshaw, Ertimur, and O’Brien 2017). Several studies examine the role of ICO analysts in the setting of initial coin offerings (ICOs) and find the ICO analysts’ ratings are associated with the functioning and the failure of ICOs (Bourveau, De George, Ellahie, and Macciocchi 2022; Florysiak and Schandlbauer 2022; Barth, Laturnus, Mansouri, and Wagner 2023). Studies on analysts’ role in the traditional stock 2 market and the ICO market shed light on the importance of analysts’ perspective on the potential impact of cryptocurrency on firm value, whereas empirical evidence remains limited on whether and how analysts incorporate firms’ crypto exposure into their research output. Ex-ante, it is unclear whether analysts view firms’ crypto exposure as value-enhancing, value-destroying, or irrelevant. On the one hand, analysts may respond favorably to firms’ crypto exposure. Cryptocurrency is one of the most recent FinTech innovations (Biais et al. 2023). A favorable view of innovation has been conveyed in early academic work, from Adam Smith’s pin factory to Joseph Schumpeter’s creative destruction, suggesting the importance of innovation for growth opportunities (Kempf and Spalt 2022). Recent studies have examined the value implications of FinTech innovations. Chen and Srinivasan (2023) find that adopting digital technologies increases firms’ market value. Chen, Wu, and Yang (2019) find that most FinTech innovations generate substantial market value for innovators, with cryptocurrency and the underlying blockchain technology being considered as particular valuable innovations.2 On the other hand, analysts may react unfavorably to firms’ crypto exposure. Cryptocurrency faces a highly uncertain regulatory environment. While some countries have adopted Bitcoin as official currency (e.g., El Salvador, Central African Republic), others have either banned (e.g., China) or plan to ban the trading and use of cryptocurrency (Corbet, Lucey, Urquhart, and Yarovaya 2019). In the U.S., the regulatory risk for cryptocurrency remains high. The SEC has significantly increased its resources and effort in the enforcement against initial coin offerings and heightened regulatory scrutiny on cryptocurrency (Sharma 2022). The potential pricing bubble is also a critical concern for the cryptocurrency market (Corbet et al. 2 For example, a survey report of the World Economy in 2015 suggests that 10% of global GDP will be stored on blockchain technology (https://www.weforum.org/reports/deep-shift-technology-tipping-points-and-societal- impact). 3 2019). Cheah and Fry (2015) demonstrate that Bitcoin exhibits speculative bubbles and suggest that the fundamental price of Bitcoin should be zero. Moreover, cryptocurrency can be used to hide illegal transactions (Foley, Karlsen, and Putninš 2019). The regulatory risk and the speculative use of cryptocurrency indicate that exposure to cryptocurrency can harm firm value. To empirically answer whether and how analysts respond to firms’ crypto exposure, I apply textual analysis to all 8-K filings filed between January 2014 and September 2021. I identify 1,112 8-K filings from 223 U.S. public firms that contain crypto-related keywords and use the keyword count to capture the intensity of firms’ crypto exposure. I examine analysts’ responses to these 8-K flings through analyst target price revisions within five calendar days of the 8-K filing date because target price forecast reflects analysts’ perceptions of firm value, making it a useful investment signal (Brav and Lehavy 2003; Bradshaw, Brown, and Huang 2013). If analysts interpret firms’ crypto exposure positively, analysts will revise their target price forecasts upward, and vice versa. I also examine analysts’ revision of one-year ahead annual earnings forecasts to assess whether analysts anticipate any immediate impact of crypto exposure on firms’ near-term performance.3 I find that the keyword count is significantly and positively associated with analysts’ target price forecast revisions, suggesting that analysts respond favorably to firms’ crypto exposure in forecasting firms’ stock prices. In the additional analysis, I extend the sample period to 2023 to include the recent crypto market downturn and continue to find a significantly positive association between analyst target price forecast revision and firms’ crypto exposure. In contrast, I find no significant association between analysts’ one-year ahead annual earnings forecast 3 Target price forecasts incorporate analysts’ expectations of future earnings in all periods, conveying information about firm value in the long run. In contrast, one-year annual earnings forecasts incorporate analysts’ expectations of earnings in the next period, conveying information about firms’ near-term performance. 4 revision and the crypto-related keyword count throughout the analysis, indicating that analysts do not perceive any immediate impact of crypto exposure on firms’ earnings. I then focus on cross-sectional analyses in the effect of firms’ crypto exposure on analyst target price revisions. Prior studies find certain corporate governance arrangements to be value- enhancing in technology adoption (Bresnahan and Greenstein 1996; Bloom, Sadun, and Reenen 2012; Ashraf, Michas, and Russomanno 2020). I manually code the 8-K filings with crypto- related disclosure and classify them based on whether the disclosure is related to corporate governance such as board member selection. I find that the positive association between analyst target price forecast revisions and firms’ crypto exposure is stronger when the firm has crypto- related corporate governance arrangement. This finding indicates that analysts view crypto- related expertise in top managers or board members as value-enhancing. My second cross-sectional analysis examines the impact of the co-movement between a firm’s stock returns and the returns in the cryptocurrency market on analyst target price revisions. This co-movement captures the degree to which a firm’s fundamentals relate to the aggregate trend in the cryptocurrency market. I find that analysts revise their target price forecasts upward to a greater degree for firms with higher degrees of return co-movement. This evidence suggests that analysts view firms’ crypto exposure as more valuable when firms’ stock returns are more closely related to the cryptocurrency market. Finally, I investigate whether analyst target price revisions have any incremental impact on investor reactions to the same 8-K filings with crypto-related disclosure. I find that analyst target price revisions strengthen the market’s positive reaction to firms’ crypto exposure. These findings imply that the market incorporates analysts’ interpretation of firms’ crypto exposure and reacts in the same direction. Various robustness tests alleviate the concerns that my results are 5 driven by confounding events such as earnings announcements, analysts’ riding of cryptocurrency mania, and firms’ disclosure styles. My study makes several contributions. First, my findings provide evidence of the potential effectiveness of the updated accounting standards about crypto assets. The debate among lawmakers regarding how to oversee the booming digital asset market highlights the need to evaluate whether firms’ voluntary disclosure of crypto assets is informative to the public (Schonberger 2022). I find a positive association between firms’ crypto-related keyword counts in 8-K filings and analyst target price revision but not one-year ahead annual forecast revision, suggesting that firms’ voluntary disclosure of crypto exposure is informative to analysts. However, the number of firms that voluntarily disclose crypto exposure is relatively small, and the content of disclosure varies across firms. The recent accounting standards of crypto assets can ensure a base level of transparency and comparability, offering more decision-useful information to capital market participants. The recent accounting standards of crypto assets mainly focus on firms’ crypto holdings whereas firms’ crypto exposure has various forms.4 Future accounting policies may expand the scope and provide guidance on how to quantify the impact of other types of crypto exposure on financial reporting. Second, my findings add to the emerging literature on cryptocurrency. The academic inquiry into crypto by finance and management scholars gets active since 2020 and research that answers the question whether crypto sector is hype or getting real is needed (Biais et al. 2023). Although studies in economics and finance offer some evidence on the factors relevant to the valuation of cryptocurrency (Cong, Li, and Wang 2021; Cong, He and Li 2021; Sockin and 4 In its most recent accounting standard update of the project “Accounting for and Disclosure of Crypto Assets” (FASB, 2023), FASB requires firms to measure certain crypto assets at fair value each reporting period with changes in fair value recognized in net income and to disclose significant holdings, contractual sale restrictions, and changes during the reporting period. 6 Xiong 2023), accounting research on cryptocurrency is limited. Cheng, Franco, Jiang, and Lin (2019) provide early evidence on the market’s reaction to firms’ initial crypto-related disclosure in 8-K filings. Luo and Yu (2022) find that US firms recognize cryptocurrencies as intangible assets at cost and revalue them to reflect impairment loss without later upward adjustment, but Anderson, Fang, Moon, and Shipman (2022) find that US firms use both fair value accounting and accounting for indefinite-lived intangible assets for crypto assets. I find that firms’ crypto- related disclosure in 8-K filings is informative to analysts. My findings also suggest that firms’ crypto exposure is value-enhancing, adding evidence to the view that crypto is getting real. Third, my study contributes to the literature on the role of analysts in corporate innovations (He and Tian 2013). Palmon and Yezegel (2012) document that analysts’ stock recommendation revisions are more informative to investors for high R&D firms than for low R&D firms, suggesting that analyst research provides value to the valuation of innovative firms. Given the controversial views on cryptocurrency, it is not clear how analysts incorporate their perceptions of firms’ crypto exposure into their research output. Using firms’ voluntary disclosure of crypto exposure, I find that analysts respond favorably through their target price revisions, but not through their earnings forecasts. My findings suggest that analysts see long- term benefits from crypto exposure, which could encourage firms to engage in more innovations. The rest of the paper proceeds as follows. Section 2 discusses the background of cryptocurrency and related literature. Section 3 describes the research design. Section 4 reports descriptive statistics, main results, and additional analyses. Finally, section 5 concludes. 7 CHAPTER 2: BACKGROUND AND RELATED LITERATURE The earliest and the most popular cryptocurrency, Bitcoin, was originally introduced in a whitepaper by Nakamoto (2008) and came into existence in 2009 (Makarov and Schoar 2020). Since then, blockchain-based cryptocurrency has evolved rapidly, growing from nearly nothing to over $1 trillion in market capitalization in the past decade.5 The proliferation of cryptocurrency coincides with surging public interests. Crypto terms such as “Bitcoin,” “Dogecoin,” and “Ethereum Price” became top searched news terms in recent years (Google 2017; Google 2021). As of March 2024, about 460 million unique Bitcoin wallets have been created (Duarte 2024). Cryptocurrency has inspired a debate. Some see cryptocurrency and the underlying blockchain technology as the “greatest technological breakthroughs since the Internet” (Chavez- Dreyfuss and Connor 2014). Deloitte’s (2021a, p2) Global Blockchain Survey finds that financial industry leaders view cryptocurrency and the underlying blockchain technology as “a strategic priority now and in the near future.” The leaders are concerned that “their organization will lose an opportunity for competitive advantage” if they fail to adopt cryptocurrency and blockchain technology. Goldman Sachs has already backed their lending facility with Bitcoin (Yang 2022). An emerging literature in economics and finance has offered some rationale for understanding the value of cryptocurrency (Cong, Li, and Wang 2021; Cong, He, and Li 2021; Sockin and Xiong 2023). Theoretical papers demonstrate that cryptocurrency has fundamental values which are anchored by the underlying utility values (Cong, Li, and Wang 2021; Sockin 5 See https://coinmarketcap.com/. 8 and Xiong 2023).6 Liu, Sheng, and Wang (2024) find that the technological sophistication of cryptocurrency drives the successes and valuations in initial coin offerings (ICO), suggesting that the value of cryptocurrency comes from their underlying innovative technologies. Therefore, having a stake in cryptocurrency can be an investment in the future of its innovative technologies. Taken together, these studies suggest that cryptocurrency has the potential to positively change the way the economy operates. In contrast, others call blockchain-based cryptocurrency a “black hole” where investors’ money disappears (Freifeld and Chavez-Dreyfuss 2015). Warren Buffett calls Bitcoin “rat poison squared” and believes that “cryptocurrencies basically have no value, and they don’t produce anything” (Bursztynsky 2020). His long-time business partner, Charlie Munger, further stated that Bitcoin is “disgusting and contrary to the interests of civilization” (Li 2021). Also, many remain skeptical of the genuine innovativeness of cryptocurrency, not to mention their association with illegal activities such as money laundering and drug dealing (Narayanan and Clark 2017). Based on the estimation by Foley, Karlsen, and Putninš (2019), 46% (amounts to $76 billion per year) of Bitcoin transactions are serving illegal activities and approximately 25% of Bitcoin users are involved in illegal activities. Similarly, Amiram, Jørgensen, and Rabetti (2022) find evidence that cryptocurrency is used to finance terrorist attacks such as Sri Lanka Easter bombing. Given the conflicting views, it is not surprising that cryptocurrency faces a highly uncertain regulatory environment. Although El Salvador and Central African Republic have adopted Bitcoin as legal tender, China has banned the use of cryptocurrency and other countries plan to ban it (Corbet et al. 2019). In the U.S., the regulatory risk for cryptocurrency is also high 6 Cryptocurrencies can be the means of payment on digital platforms that support certain economic transactions. For example, Filecoin is a digital platform that allows users to exchange data storage space for its tokens. 9 because the SEC has significantly increased its resource and effort in the enforcement against initial coin offerings and heightened the scrutiny on stablecoins and other tokens (Sharma 2022). Moreover, price manipulation is frequent in the cryptocurrency market, which can substantially distort the valuation of cryptocurrency (Griffin and Shams 2020). Griffin and Shams (2020) document that the growth of Tether, the largest pegged cryptocurrency, is supplied to investors as part of a scheme to inflate Bitcoin prices. Li, Shin, and Wang (2023) find that pump-and-dump schemes (P&Ds) infuse the cryptocurrency market and are harmful to cryptocurrency liquidity and prices. Cheah and Fry (2015) demonstrate that Bitcoin exhibits speculative bubbles and argue that the fundamental price of Bitcoin should be zero. Despite the debate, an increasing number of companies are using cryptocurrency for operations, transactions, and investment (Deloitte 2021b). One telling example is MicroStrategy, a business intelligence company, which has raised the company’s Bitcoin holding units to 214,246 as of March 18, 2024 (Shen 2024). While more and more companies dabble into the crypto space in the surge of cryptocurrency, whether and how such crypto exposure affects financial analysts’ perceptions of firms’ valuation in both the short-and the long-term remains underexplored. As market intermediaries, financial analysts gather, interpret, and produce information, and they are central to the flow of information between firms and investors (Kothari, So, and Verdi 2016; Bradshaw, Ertimur, and O’Brien 2016). In the traditional stock market, analysts provide valuable research to the investment community, especially when either macro-level or firm-level uncertainty is high (Frankel et al. 2006; Loh and Stulz 2011, 2018). Recent few studies examine the role of ICO analysts on ICO markets and find that ICOs with higher ICO 10 analyst ratings are more successful (Bourveau et al. 2022; Florysiak and Schandlbauer 2022).7 Given that analysts’ view matters in both the traditional stock market and the ICO market, whether analysts support, discount, or ignore firms’ exposure to cryptocurrency is likely to affect investors’ beliefs and wealth allocation. Prior literature finds mixed evidence on whether analysts embrace or dismiss corporate innovations. Amir, Lev, and Sougiannis (2003) use R&D intensity to measure corporate innovation and find that analyst earnings forecasts are more optimistically biased for firms with high R&D intensity. Li (2016) finds that analysts revise their long-term earnings growth forecasts upward when firms obtain new patents or trademarks. These findings suggest that analysts can convey favorable information about corporate innovations to investors. In contrast, He and Tian (2013) find that firms covered by a larger number of analysts generate fewer patents or produce patents with lower impact because managers exhibit myopia behavior and sacrifice firm innovation when analysts focus on short-term earnings and do not fully account for information in the long-term in their stock recommendations. They interpret this as the “dark side” of analyst coverage on corporate innovations because analysts’ unfavorable views of corporate innovations can hinder firm value in the long run. When it comes to cryptocurrency, one of the most recent FinTech innovations, it is also unclear whether analysts hold favorable views of firms’ exposure to crypto-related activities. A favorable view of innovation has been conveyed in early academic work, from Adam Smith’s pin factory to Joseph Schumpeter’s creative destruction, suggesting the importance of innovation for growth opportunities (Kempf and Spalt 2022). FinTech innovations increase firm value by 7 ICO analysts are experts in the ICO market who voluntarily provide their ratings based on multiple dimensions (e.g., team, vision, and product) in evaluating the overall prospects of ICOs to rating platforms such as ICObench (Bourveau et al. 2022; Barth et al. 2023). 11 facilitating growth opportunities and improving productivity (Brynjolfsson, Rock, and Syverson 2017; Cockburn, Henderson, and Stern 2018; Goldstein, Jiang, and Karolyi 2019; Chen et al. 2019). Chen et al. (2019) construct a measure of firm value that combines stock market responses with predicted firm-level innovation intensity by identifying FinTech innovations and classifying them based on their underlying technologies, using patent filings during 2003-2017. They find that FinTech innovations yield substantial value to firms, with cryptocurrency and blockchain technology being particularly valuable. In the meantime, the regulatory risk and the speculative use of cryptocurrency suggest that dabbling into the crypto space is likely to heighten firms’ risks and jeopardize firm value. The objective of this paper is to provide empirical evidence by examining whether and how analysts revise beliefs when firms disclose their crypto exposure in 8-K filings. Economist Robert Solow said in 1987 that “the computer age was everywhere except for the productivity statistics,” (known as the Solow Paradox), suggesting that the benefits of IT adoption were not evident in the data yet because they took so long to realize (Krishnan, Mischke, and Remes 2018). Frictions (e.g., the requirement of developing organizational capabilities) in adopting new technology can also delay the benefits (Bresnahan and Greenstein 1996). More recently, Chen and Srinivasan (2023) find that adopting digital technologies increase long-term productivity (as reflected in the asset turnover three-years after digital disclosure) even if it decreases sales growth. Similarly, even if cryptocurrency benefits the firm, such benefits may take a long time to realize. Target price forecasts “provide market participants with analysts’ most concise and explicit statement on the magnitude of the firm’s expected value” (Brav and Lehavy 2003, page 1933). Compared to near-term earnings forecasts, target price forecasts incorporate analysts’ 12 long-term assessment of firms’ earnings and risks (Bilinski, Lyssimachou, and Walker 2013). The nature that target price forecasts reflect analysts’ beliefs of firms’ prospects in the long run makes target price forecasts suitable for examining whether and how analysts respond to firms’ crypto exposure. I also examine whether analysts revise their one-year ahead earnings forecasts when firms disclose their crypto exposure. This comparison allows me to examine whether crypto exposure affects analysts’ expectations of firms’ short-term or long-term performance. Given the lack of regulation or rules governing firms’ disclosure of their involvement in cryptocurrency, information regarding firms’ exposure to cryptocurrency is limited. Firms’ 8-K filings offer analysts a channel to assess firms’ exposure to cryptocurrency in a timely fashion. A study by Cheng et al. (2019) examines the stock market reaction to firms’ initial disclosure of crypto exposure. They find that investors react positively in the seven-day event window to 8-K filings that speculatively mention blockchain, cryptocurrency, or Bitcoin, but such reaction is reversed in the following 30 days. Their study provides early evidence on the market’s reaction to the very first disclosure related to cryptocurrency. A few other studies examine how firms account for their holdings of cryptocurrency. Luo and Yu (2022) compare the financial reporting for Bitcoin holding of U.S firms and international firms. They find that US firms recognize Bitcoin as intangible assets at cost and estimate the impairment loss without later upward adjustment while most firms under IFRS recognize Bitcoin as intangibles and inventories at fair value. Anderson et al. (2022) focus on U.S. firms’ crypto holdings. They find that US firms use both fair value accounting and accounting for indefinite- lived intangible assets for crypto assets and that firms tend to make fair value disclosures when cryptocurrencies are more liquid. 13 My study differs from prior research (Cheng et al. 2019; Luo and Yu 2022; Anderson et al. 2022) on crypto-related disclosure in several aspects. Cheng et al. (2019) focus on the firm’s initial mentioning of “blockchain”, “bitcoin”, or “cryptocurrency(ies)” in 8-K. However, the field of cryptocurrency is evolving rapidly over time, and so is firms’ exposure to it. I expand the set of keywords to include both initial and subsequent crypto-related disclosure in 8-K filings to provide a broad and dynamic picture of firms’ crypto exposure over time.8 Unlike Cheng et al. (2019), who manually classify a firm’s crypto-related disclosure as speculative or not speculative, I count the crypto-related keywords in 8-K to proxy for the intensity of the firm’s crypto exposure. More importantly, I focus on how analysts react to firms’ crypto-related disclosure because analysts are sophisticated market participants whose reactions can affect a broader set of market participants. By examining both the long-term evaluation of firm performance (e.g., target price forecast) and short-term evaluation of firm performance (e.g., one-year ahead annual earnings forecast) from analysts, I further shed light on whether the impact of exposure to cryptocurrency on firm performance is immediate or takes a longer period to realize, while such inferences cannot be disentangled via the market reaction test alone. Luo and Yu (2022) manually collect 40 financial statements of public companies that hold Bitcoin for the year 2020 from bitcointreasuries.net. Anderson et al. (2022) search financial statements and manually collect 438 firm-quarter observations of U.S. firms that hold crypto during 2013-2021. Luo and Yu (2022) and Anderson et al. (2022) focus only on firms’ crypto holdings—one specific type of crypto exposure. To have a comprehensive view on firms’ extent of crypto exposure, I examine a broader spectrum of firms’ crypto exposure that ranges from crypto holding to any of companies’ involvement or risks associated with crypto. 8 See Appendix A for details. 14 Whether analysts find crypto-related disclosure in 8-Ks useful is unclear. Easton, Ben- Repahael, Da, and Israelsen (2021) find that institutional investors may learn about information by means other than 8-K while retail investors seek information there. In this regard, as sophisticated market participants, financial analysts may find 8-K filings uninformative. On the other hand, information about firms’ exposure to cryptocurrency is scarce because firms do not have to provide such disclosure. Ex-ante, whether and how analysts respond to firms’ crypto- related disclosure in 8-K filings is an empirical question. 15 CHAPTER 3: RESEARCH DESIGN 3.1 Sample Selection I use Python to extract all 8-K filings issued from the beginning of 2014 to September 2021 from the SEC EDGAR database. The sample period starts from 2014 because trading data of Bitcoin, the earliest cryptocurrency, became available in the last week of 2013. Following Cheng et al. (2019), I filter out crypto-related 8-K filings based on mentions of crypto-related keywords in both the body of the 8-K and the exhibits (where firms attach files such as investor presentation slides and press releases). Tables A.1 and A.2 in Appendix A reports the list of crypto-related keywords in my sample selection process.9 The initial sample of 8-K filings is 560,612. I then apply textual analysis techniques to identify crypto-related 8-K filings and require firms to have necessary data from Compustat, I/B/E/S, and CRSP. This process leaves me with 1,112 8-K filings with crypto-related disclosure from 223 unique U.S. public firms. Table E.1A presents the annual number of crypto-related 8-K filings and firms from the beginning of 2014 to the third quarter of 2021. In 2014, 41 8-K filings of 10 firms contained crypto-related keywords. As of the end of September 2021, 246 8-K filings of 82 firms mentioned crypto-related keywords. Both the number of firms and the number of crypto-related 8-K filings have grown over the sample period. I use keyword counts in each crypto-related 8-K filing to measure the intensity of firms’ exposure to cryptocurrency. Table E.1B reports the annual distribution of keyword counts by categories. In total, there are 10,675 keyword hits in the 1,112 8-K filings with crypto-related 9 Prior studies such as Chen, Wu, and Yang (2019) and Cheng, Franco, Jiang, and Lin (2019) do not isolate cryptocurrency from blockchain, so I include blockchain as one of the keywords in this paper. I select 10 cryptocurrencies based on their ranking at several points of time in 2021. See https://www.fool.com/the- ascent/cryptocurrency/articles/10-biggest-cryptocurrencies-of-2021/, https://www.cointracker.io/blog/top-10-crypto, and https://www.nasdaq.com/articles/decoding-crypto%3A-the-10-most-popular-cryptocurrencies-2021-08-05. To make the count of keywords less noisy, I do not include “tether” and “ripple” because these two words have multiple meanings and usually do not refer to cryptocurrency in 8-K filings based on manual verification. 16 keywords. The keyword count of blockchain is 5,279, consisting of 49.5% of the total keyword counts. The keywords in the category of cryptocurrency were mentioned 2,646 times and take 24.8% of the total keyword counts. The keyword count of Bitcoin is 2,574 which consists of 24.1% of the total keyword counts. These descriptive statistics are visualized in the word cloud presented in Figure B.1 of Appendix B.10 To shed light on the content of firms’ crypto exposure disclosed in 8-K filings, I examine the most frequent words used within 5 words before and within 5 words after each crypto-related keyword. The word cloud in Figure B.2 of Appendix B shows that crypto-related keywords are most frequently accompanied by “mining,”, “technology,” “platform,” “customer,” and “revenue,” suggesting that firms’ crypto exposure disclosed in 8-K filings and their exhibits are more (less) likely to be associated with crypto mining, new technology and operating activities (speculation).11 To answer the question of whether and how analysts respond to firms’ exposure to cryptocurrency disclosed in 8-K filings, I obtain analyst target price forecasts and annual earnings forecasts from the I/B/E/S database. I first examine the analyst target price forecast revision issued within five calendar days of the 8-K filing date. I only keep the first target price forecast revision and drop the later ones if an analyst issues multiple target price forecasts for the same firm within the five-day window. After requiring non-missing value for my main variables, among 1,112 crypto 8-K filings for 223 firms, 118 firms (394 8-K filings) have at least one target price forecast revision within the five-day window of the 8-K filing date. Firms without any target price revision during the window are dropped. For the target price analysis, the sample contains 1,880 analyst target price forecast revisions. To shed light on whether analysts see any 10 Appendix B presents word clouds for the most frequent words used surrounding crypto-related keywords in 8-K filings and their exhibits. Texts within 5 words’ distance of a crypto-related keyword in each 8-K filing form the basis for the word clouds. 11 Figure A.1 of Appendix A provides an example of firms’ crypto-related 8-Ks and their exhibits. 17 near-term impact of firms’ crypto exposure on their earnings, I also examine analysts’ one-year ahead annual earnings forecast revisions issued within the same five-day window. Similarly, I use the first annual earnings forecast revision if an analyst issues multiple annual forecasts for the same firm. After dropping observations with missing values for the required variables, the sample for annual earnings forecast revisions contains 2,692 observations consisting of 349 8-K filings from 114 firms. I further use data from Compustat, CRSP, and I/B/E/S to construct variables for other firm- or analyst-level characteristics. To mitigate the influence of potential outliers, I winsorize all the continuous variables at 1% and 99%. 3.2 Model Specification 3.2.1 Analyst Belief Revision and Firms’ Crypto Exposure To empirically test whether analysts revise their beliefs about firms’ performance following crypto-related disclosure in 8-K filings, I estimate the following model: AnalystRevisionijt = α + β1LnKeywordCountjt + Controls + Fixed Effects + εjit (1) I construct two measures to capture analyst belief revision (AnalystRevisionijt). The first measure is ChgTP, defined as the difference between the current and prior target price forecast issued by the same analyst for the same firm, divided by the firm’s stock price two trading days prior to the current target price forecast issuance date. Following the prior literature that examines analysts’ response to firms’ public disclosure (Kross and Suk 2012), I require the current target price forecast to be issued within five calendar days after an 8-K filing that contains crypto-related disclosure. The second measure is RevAF, defined as the difference between the current and prior one-year ahead annual earnings forecast issued by the same analyst for the same firm, divided by the firm’s stock price two trading days prior to the current earnings 18 forecast issuance date. I require the current annual earnings forecast to be issued within five calendar days following an 8-K filing that contains crypto-related disclosure. The main variable of interest is LnKeywordCount, which is the natural logarithm of one plus the number of crypto-related keywords in an 8-K filing and its exhibits. The list of crypto- related keywords is available in Appendix A. Rather than simply capturing the presence of crypto-related keywords, I use LnKeywordCount to capture the intensity of firms’ crypto exposure. The higher the LnKeywordCount, the more crypto exposure the firm has. I control factors known to affect analyst responses to public disclosures following prior research. Specifically, I control stock turnover (Turnover) because stock turnover is directly related to trading commissions and analysts have incentives to issue optimistic research output to facilitate trading and help their brokerage houses generate more trading commissions (Irvine 2004). Analysts incorporate firms’ general risk (RetSD) into target price forecasts (Bilinski, Lyssimachou, and Walker 2013; Dechow and You 2020). Past winners may continue to perform well, and past losers may continue to perform poorly (Hong, Lim, and Stein 2000). To isolate the effect of momentum on analyst target price forecast, I control a firm’s buy-and-hold abnormal returns in the past 12 months (PastRet). I control firm size (LnAT) because it is often used as a proxy for the amount of information publicly available to a firm (Zhang 2006) and firms’ information environment is likely to affect analysts’ output. I include revenue growth (RevGrowth) and market-to-book value of equity to capture a firm’s growth opportunity (Bamber and Cheon 1998). I also control the number of analysts following the firm (LnAnalyst) because Bradshaw et al. (2019) find a positive association between the number of analysts following the firm and analysts’ target price forecast optimism. 19 The individual analyst characteristics I control for include analysts’ firm-specific forecasting experience (Exp_Firm), general forecasting experience (Exp_General), the number of firms an analyst covers (LnNumFirms), and the size of the brokerage house the analyst works for (BrokerSize). Clement (1999) finds that analysts’ experience and brokerage size are positively associated with, while analysts’ firm coverage is negatively associated with their forecast accuracy. Finally, I include year, industry, and analyst fixed effects to account for the time trend, and the impact of time-invariant industry and analyst characteristics that may not be fully captured by the control variables. 3.2.2 Cross-Sectional Analyses of Analyst Belief Revision It is natural to expect that firms’ crypto-related disclosure in 8-K filings is not equally informative to analysts. To explore possible cross-sectional variations in the effect of crypto- related disclosure, I focus on two dimensions, including the type of firms’ crypto-related exposure and the comovement of firms’ stock returns with cryptocurrency market returns. While my focus is mainly on analyst target price revisions, I also use analysts’ annual earnings forecast revisions for the cross-sectional analyses. Bresnahan and Greenstein (1996) highlight the importance of developing complementary organizational capabilities in technology adoption. Bloom et al. (2012) find that better people- management practices (e.g., performance evaluation, reward-punishment practices) in U.S. firms quite explained the productivity gap in information technology (IT) adoption between U.S. and European firms. Ashraf et al. (2020) find that IT expertise on the audit committee of a firm enhances both the reliability and timeliness of financial reporting, suggesting the benefits of board expertise in IT adoption. I manually code the crypto 8-K filings and classify them based on whether the underlying crypto exposure is associated with corporate governance such as board 20 member selection (CorpGov).12 The variable of interest in this cross-sectional test is the interaction between LnKeywordCount and CorpGov. Next, I examine whether analyst response to crypto-related disclosure varies with the degree of comovement between firms’ stock returns and cryptocurrency returns. Cheng et al. (2019) find that stock returns of companies that have crypto exposure comove with Bitcoin returns, suggesting that the return comovement reflects how closely a firm’s fundamentals are related to the trend in the cryptocurrency market. I construct two measures for the return comovement between the firm’s stock returns and the returns of the cryptocurrency market. To start with, I collect trading data of cryptocurrencies available on https://coinmarketcap.com/ for the period 12/31/2013-9/30/2021. Following Liu, Tsyvinski, and Wu (2022), I exclude the cryptocurrency whose market capitalization is less than $1 million. To calculate cryptocurrency returns, I require the cryptocurrency to have price information. I measure the initial cryptocurrency market returns as the value-weighted returns of all the underlying cryptocurrencies and construct the cryptocurrency market returns as the difference between the initial cryptocurrency market returns and the risk-free rate (i.e., the one-month Treasury bill rate). I measure return comovement during the period from 252 trading days before to 6 trading days before the crypto-related 8-K filing issuance date (i.e., [-252, -6]). The first measure of return comovement (Comove_Coef) is the coefficient on the cryptocurrency market returns when I regress the firm’s risk-free rate adjusted returns on the Fama-French three factors and the cryptocurrency market returns. The second measure of return comovement (Comove_Corr) is the 12 Examples include the appointment of managers or board members with expertise in cryptocurrency and blockchain technologies, the use of Bitcoin as part of top management’s compensation, and the change of the corporate name to make it crypto-related. 21 Pearson correlation between the firm’s risk-free rate adjusted returns and the cryptocurrency market returns. 3.2.3 The Impact of Analyst Belief Revision on Market Reaction to Firms’ Exposure To understand the impact of analyst target price revisions surrounding firms’ crypto- related disclosure, I assess whether analyst target price forecast revisions affect investors’ reactions to crypto-related 8-K filings. I estimate the following model: BHAR_FF3jt = α + β1LnKeywordCountjt + β2LnKeywordCountjt × AvgChgTPjt + Controls + Fixed Effects + εit (2) Following Cheng et al. (2019), I use buy-and-hold abnormal returns (BHAR_FF3) adjusted for the Fama-French three-factor model to capture the market reaction from three trading days before to three trading days after the crypto-related 8-K filing date (i.e., [-3, 3]). AvgChgTP is the average target price forecast revisions issued during the trading-day period [-3, 3] around the crypto-related 8-K filing date by any analysts who follow the same firm. I code AvgChgTP as zero if there is no analyst target price forecast revision during the period [-3,3]. The variable of interest is the interaction between LnKeywordCount and AvgChgTP. 22 4.1 Summary Statistics and Main Results CHAPTER 4: RESULTS Table E.2 presents the summary statistics for the target price revision sample, annual earnings forecast revision sample, and the sample for the market reaction test, respectively. Table E.2A shows that the mean (median) of analysts’ target price revision (ChgTP) is 0.041 (0.051), suggesting that on average analysts revise their target price forecasts upward within five days of firms’ crypto-related 8-K filing date. Table E.2B shows that the mean (median) of analysts’ annual earnings forecast revision (RevAF) is 0.000 (0.001). The summary statistics for the rest of the variables in Table E.2A and Table E.2B are similar, indicating that firms in the target price revision sample and the annual earnings forecast revision sample are similar. In Table E.2C, the mean of BHAR_FF3 is 0.004 and the median is -0.014. Table E.3 reports the Pearson correlations for the variables in the target price forecast revision sample, the main sample for the subsequent analyses. The Pearson correlation between ChgTP and LnKeywordCount is significantly positive, providing some evidence that analysts view firms’ crypto disclosure favorably by revising their target price forecast upward. Table E.4 presents regression results of model (1), where I focus on the analysts’ belief revision in response to firms’ crypto disclosure in 8-K filings. Column (1) of Table E.4 shows that the coefficient on LnKeywordCount is significantly positive. Holding constant other firm- and analyst-level characteristics, a 1 percentage point increase in the number of crypto-related keywords in 8-K flings will increase analysts’ target price revision by 1.7 percentage points. This finding suggests that analysts respond favorably to firms’ crypto-related disclosure by incorporating the positive view into their assessment of firms’ long-term performance (i.e., target price forecast). As a contrast, column (2) estimates the effect of firms’ crypto-related disclosure 23 on analysts’ one-year ahead annual earnings forecast revision. Interestingly, there is no significant association between LnKeywordCount and RevAF, suggesting that analysts do not see any immediate impact of firms’ crypto exposure on earnings. 4.2 Results of Cross-sectional Analyses Tables E.5 and E.6 report the results of cross-sectional analyses that focus on variations in analysts’ target price forecast revision conditional on the characteristics of 8-K filings and the firm. As a contrast, I also present cross-sectional results of analysts’ annual earnings forecast revision. Table E.5 reports the results of the cross-sectional analysis conditional on whether the firm’s crypto exposure disclosed in 8-K filings is related to corporate governance. In column (1), it shows that the coefficient on the interaction between LnKeywordCount and CorpGov is significantly positive, suggesting that analysts respond more favorably to firms’ crypto exposure associated with corporate governance (e.g., the appointment of board members with expertise in cryptocurrency and blockchain) compared to other types of exposure. Not surprisingly, the coefficient on the term on the interaction between LnKeywordCount and CorpGov is insignificant in column (2). This finding indicates that analysts do not see any immediate impact of crypto exposure on firms’ earnings regardless of the types of exposure. Table E.6 reports the results of the cross-sectional analysis based on the comovement between the firm’s stock return and the cryptocurrency market return. I focus on analyst target price forecast revision in columns (1) and (2). I find that the coefficient on the interaction between LnKeywordCount and Comove_Coef is significantly positive in column (1). In column (2), the coefficient on the interaction between LnKeywordCount and Comove_Corr is also significantly positive. These findings suggest that when firms are more fundamentally related to cryptocurrency, analysts perceive their crypto-related disclosure in 8-K filings as more valuable 24 and revise target price forecasts upward to a greater degree. In columns (3) and (4), I report the results of analysts’ earnings forecast revision. The coefficient on the interaction between LnKeywordCount and the return comovement measure is insignificant in columns (3) and (4), suggesting that analysts do not consider firms’ crypto exposure to have an immediate impact even if such exposure is fundamental. 4.3 Results of Market Reaction Test So far, my analyses show that analysts revise target price forecasts upward following firms’ crypto disclosure in 8-K filings, suggesting analysts perceive firms’ crypto exposure to have a positive impact on firm performance in the long run. In this section, I investigate whether analysts’ favorable view about firms’ crypto exposure affects investors’ reaction to firms’ exposure. As a baseline, columns (1) and (2) in Table E.7 report the market reaction to firms’ crypto-related disclosure in 8-K filings. The coefficient on LnKeywordCount is significantly positive in both columns, suggesting that the greater the firms’ crypto exposure, the more positive the market reaction is. More importantly, as shown in columns (3) and (4), the coefficients on LnKeywordCount and the interaction between LnKeywordCount and AvgChgTP are both significantly positive. This means that when analysts revise target price forecast to a larger degree, market reaction to firms’ crypto disclosure is greater, suggesting investors incorporate analysts’ favorable view about firms’ crypto disclosure in decision making and react in the same direction as analysts. As robustness tests, I replace the dependent variable BHAR_FF3 with three alternative measures: BHAR_Madj (i.e., buy-and-hold abnormal return adjusted for the value-weighted market return), BHAR_M (i.e., buy-and-hold abnormal return adjusted for the market model), and BHAR_FF3M (i.e., buy-and-hold abnormal return adjusted 25 for the model with Fama-French three-factors and the momentum). I rerun the regressions of the market reaction test. As shown in Table E.8, the results are quantitatively the same.13 4.4 Robustness Tests In this section, I conduct multiple robustness tests to further check the validity of my inferences. Specifically, I run multiple tests to mitigate the concerns that my results are driven by (1) confounding events such as earnings announcements, (2) analysts’ riding of crypto mania, (3) the firm’s disclosure style in 8-K filings and their exhibits, and (4) the firm’s blockchain exposure. To alleviate the concern that analysts’ favorable target price forecast revision in response to firms’ crypto exposure disclosed in 8-K is driven by confounding events such as earnings announcements, I further control the earnings news around the 8-K filing date (EarningNews), whether the 8-K filing contains the item 2.02 Results of Operations and Financial Condition (Item202), and the information content of the 8-K filing (ItemCount). Table E.9 shows that both my main results and cross-sectional results hold when controlling for confounding events. An alternative explanation to analysts’ favorable target price forecast revision in response to firms’ crypto exposure is that analysts are riding the price mania of cryptocurrency. To rule out this alternative explanation, I control the momentum of crypto market return prior to the analysts’ target price revisions (CAR_Crypto and BHAR_Crypto). As shown in Table E.10, my inferences remain. This analysis suggests that my results are less likely to be driven by analysts riding the cryptocurrency’s price mania but more likely reflect the possibility that analysts view firms’ crypto exposure as value-enhancing. 13 The inferences remain the same when I replace the buy-and-hold-abnormal return with the cumulative abnormal return (CAR). The results are untabulated. 26 Instead of proxying for the intensity of firms’ crypto exposure, the number of crypto- related keywords in 8-K flings may capture firms’ disclosure styles of simply providing information in detail. If the latter is the case, a greater amount of crypto-related keyword counts reveals more information about firm’s crypto exposure but not necessarily the extent of the firm’s crypto exposure. Detailed crypto-exposure information is likely to help analysts better predict firms’ prospects both in the short-term and in the long run, and therefore, is likely to improve the accuracy of analysts’ target price forecasts and annual earnings forecasts. However, in Table E.11, there is no significant association between analyst forecast accuracy and the count of crypto-related keywords in both columns, suggesting that the number of crypto-related keywords is more likely to capture firms’ crypto exposure rather than firms’ disclosure styles. Finally, to further isolate analysts’ reaction to firms’ crypto-exposure, I conduct the main analysis within a subsample where the 8-K filings contain all the other crypto-related keywords except for blockchain. In Table E.12, I continue to find that analysts revise their target price forecasts upward in response to firms’ crypto exposure but do not do so with their annual earnings forecasts, suggesting that my results are explained by firms’ crypto exposure rather than simply driven by analysts’ reaction to blockchain technology. 4.6 Exploratory Analysis Using the Extended Sample The cryptocurrency landscape has undergone rapid transformations since 2022, characterized by a prolonged bear market in the cryptocurrency industry and dramatic events such as the collapse of FTX. To have a more complete picture of analysts’ views, I investigate analysts’ perspective for firms’ crypto exposure over a longer sample period that includes crypto market downturns. Examining this difference is helpful for understanding analysts’ rationale in responding to firms’ crypto exposure. If analysts’ views hold throughout crypto market 27 fluctuations, analysts are more likely to view crypto exposure as having profound strategic implications on firm value. Otherwise, analysts are more likely to view firms’ crypto exposure as speculative and ride the crypto mania in their forecasts. To answer this question, I extract additional SEC 8-K filings for the period 10/1/2021 – 12/31/2023. Using the same set of keywords as in the original sample selection process, I filtered out 1,130 crypto-related 8-K filings that can be matched with firm identifiers (e.g., GVKEY, PERMNO, TICKER) in Compustat, I/B/E/S, and CRSP. This leaves me with 2,242 crypto- related 8-K filings from 1/1/2014 to 12/31/2023 in total.14 I obtain analyst target price forecasts and one-year ahead annual earnings forecasts issued during 1/1/2014 – 12/31/2023 from I/B/E/S. After requiring the sample to have non-missing value for control variables, the full sample consists of 3,646 target price revisions issued within five calendar days following 809 crypto- related 8-K filings from 217 firms. As a comparison, there are 4,932 one-year ahead annual earnings forecasts revisions issued within five calendar days following 731 crypto-related 8-K filings from 208 firms. Table E.13 presents regression results using the full sample. Column (1) shows that the coefficient on LnKeyword is significantly positive and the magnitude of the coefficient on LnKeyword is comparable to that in Table E.4. Overall, analysts respond favorably to firms’ crypto exposure in their target price forecast revision during 2014-2023. Column (2) shows that 14 As shown in Appendix D, there are 445,870 SEC 8-K filings during 1/1/2014 – 12/31/2023 that can be matched with firm identifiers in Compustat, I/B/E/S, and CRSP and have non-missing firm level control variables. 2,032 are crypto-related and 443,838 are non-crypto-related 8-K filings. There are significant differences in firm characteristics between crypto-related and non-crypto related 8-K filings evidenced by t-test for difference in mean. Crypto-related 8-K filings are more likely to be growth firms that have a smaller size, a lower number of analyst following, and a higher growth rate. Considering the significant difference between two groups of firms, parallel consumption does not hold for a DID like research design that includes non-crypto related 8-K filings as a control group in the analyses. Instead, I focus on crypto-related 8-K filings throughout the analyses and focus on the variation in firms’ crypto exposure captured by the number of crypto-related keyword counts. 28 there is no significant association between firms’ crypto exposure and analysts’ one-year ahead annual earnings forecast revision. 29 CHAPTER 5: CONCLUSION This study examines whether and how financial analysts respond to firms’ crypto exposure. With the surge in public interest, cryptocurrency has stimulated a debate. On the one hand, cryptocurrency has the potential to change the way the economy operates. On the other hand, it can be hype where bubbles and fraud pervade. Despite the conflicting views, more and more firms are dabbling into the crypto space. While research in economics and finance has examined both the potential value of cryptocurrency and the associated downside risks, the impact of crypto exposure on firm value is not obvious. Given the uncertain and complex nature of cryptocurrency, it is important to understand analysts’ interpretation of firms’ crypto exposure because analysts’ research output is more valuable in circumstances of high uncertainty (Loh and Stulz 2011, 2018). Applying the textual analysis to 8-K filings filed between January 2014 and September 2021, I identify 1,112 8-K filings that contain crypto -related keywords from 223 unique U.S. public firms. I find a positive association between analyst target price forecast revisions and the count of crypto-related keywords in the 8-K filings, suggesting that analysts view a higher level of crypto exposure as having a positive effect on firm value in the long run. In the additional analysis, I extend the sample period to 2023 to include the recent crypto market downturn and continue to find a significantly positive association between analyst target price forecast revision and firms’ crypto exposure. In contrast, crypto exposure does not affect analysts’ near-term earnings forecasts because analyst earnings forecast revisions are not associated with the intensity of firms’ crypto exposure. Further cross-sectional analyses show that the positive association between analysts’ target price revision and firms’ crypto exposure is stronger when firms’ crypto exposure is 30 related to corporate governance such as appointing the board of directors with crypto-related expertise. The positive association between target price revision and crypto exposure is also stronger when firms’ stock returns comove more closely with cryptocurrency market returns. Interestingly, when I conduct similar cross-sectional analyses using analysts’ one-year ahead annual earnings forecast revisions, I do not find any cross-sectional variations. My final analysis examines the incremental impact of analyst target price forecast revisions on the market’s reaction to firms’ crypto exposure. I find that analyst target price revisions appear to enhance the market’s positive reaction to firms’ crypto exposure disclosed in the 8-K filings. This study enhances our understanding of how firms’ exposure to cryptocurrency affects analysts’ perceptions of firm value and the role analysts play in affecting investor reactions to firms’ crypto exposure. Collectively, my paper provides evidence that firms’ crypto exposure disclosed in 8-K filings is informative to financial analysts, suggesting that carefully designed disclosure rules from regulators can offer more decision-useful information to capital market participants. My findings also have implications for investors’ wealth allocation. To have a stake in growth opportunities associated with crypto-related FinTech innovations, investors may consider holding shares of public companies that have crypto exposure. 31 REFERENCES Amir, E., B. Lev, and T. Sougiannis. 2003. Do financial analysts get intangibles? European Accounting Review 12 (4): 635–659. Amiram, D., B. N. Jørgensen, and D. Rabetti. 2022. Coins for Bombs: The Predictive Ability of On-Chain Transfers for Terrorist Attacks. Journal of Accounting Research 60 (2): 427–466. Anderson, C. M., V. W. Fang, J. Moon, and J. E. Shipman. 2022. Accounting for Cryptocurrencies. Working paper. ASC 350 Intangibles—Goodwill and Other. https://asc.fasb.org/1943274/2147482650. Ashraf, M., P. N. Michas, and D. Russomanno. 2020. The Impact of Audit Committee Information Technology Expertise on the Reliability and Timeliness of Financial Reporting. The Accounting Review 95 (5): 23–56. Bamber, L. S., and Y. S. Cheon. 1998. Discretionary Management Earnings Forecast Disclosures: Antecedents and Outcomes Associated with Forecast Venue and Forecast Specificity Choices. Journal of Accounting Research 36 (2): 167–190. Barth, A., V. Laturnus, S. Mansouri, and A. F. Wagner. 2023. Conflicted Analysts and Initial Coin Offerings. Management Science 69 (11): 6641–6666. Ben-Rephael, A., Z. Da, P. D. Easton, and R. D. Israelsen. 2022. Who Pays Attention to SEC Form 8-K? The Accounting Review 97 (5): 59–88. Biais, B., A. Capponi, L. W. Cong, V. Gaur, and K. Giesecke. 2023. Advances in Blockchain and Crypto Economics. Management Science 69 (11): 6417–6426. Bilinski, P., D. Lyssimachou, and M. Walker. 2013. Target Price Accuracy: International Evidence. The Accounting Review 88 (3): 825–851. Bloom, N., R. Sadun, and J. V. Reenen. 2012. Americans Do IT Better: US Multinationals and the Productivity Miracle. American Economic Review 102 (1): 167–201. Bourveau, T., E. T. De George, A. Ellahie, and D. Macciocchi. 2022. The Role of Disclosure and Information Intermediaries in an Unregulated Capital Market: Evidence from Initial Coin Offerings. Journal of Accounting Research 60 (1): 129–167. Bradshaw, M., Y. Ertimur, and P. O’Brien. 2017. Financial Analysts and Their Contribution to Well-Functioning Capital Markets. Foundations and Trends® in Accounting 11 (3): 119–191. 32 Bradshaw, M. T., L. D. Brown, and K. Huang. 2013. Do sell-side analysts exhibit differential target price forecasting ability? Review of Accounting Studies 18 (4): 930–955. Brav, A., and R. Lehavy. 2003. An Empirical Analysis of Analysts’ Target Prices: Short-term Informativeness and Long-term Dynamics. The Journal of Finance 58 (5): 1933–1967. Bresnahan, T., and S. Greenstein. 1996. Technical Progress and Co-Invention in Computing and in the Uses of Computers. Brookings Papers on Economic Activity. Microeconomics 1996: 1–83. Brynjolfsson, E., D. Rock, and C. Syverson. 2017. Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics. Working Paper. Bursztynsky, J. 2020. Warren Buffett: Cryptocurrency “has no value” – “I don’t own any and never will.” CNBC. https://www.cnbc.com/2020/02/24/warren-buffett-cryptocurrency-has-no- value.html. Chavez-Dreyfuss, G., and M. Connor. 2014. ANALYSIS-All the rage a year ago, bitcoin sputters as adoption stalls. Reuters. https://www.reuters.com/article/markets-bitcoin-adoption- idUKL1N0SN2RK20141211. Cheah, E.-T., and J. Fry. 2015. Speculative bubbles in Bitcoin markets? An empirical investigation into the fundamental value of Bitcoin. Economics Letters 130: 32–36. Chen, M. A., Q. Wu, and B. Yang. 2019. How Valuable Is FinTech Innovation? The Review of Financial Studies 32 (5): 2062–2106. Chen, W., and S. Srinivasan. 2023. Going digital: implications for firm value and performance. Review of Accounting Studies. Cheng, S. F., G. D. Franco, H. Jiang, and P. Lin. 2019. Riding the Blockchain Mania: Public Firms’ Speculative 8-K Disclosures. Management Science 65 (12): 5901–5913. Clement, M. B. 1999. Analyst forecast accuracy: Do ability, resources, and portfolio complexity matter? Journal of Accounting and Economics 27 (3): 285–303. Cockburn, I. M., R. Henderson, and S. Stern. 2018. The Impact of Artificial Intelligence on Innovation. Working Paper. Cong, L. W., Z. He, and J. Li. 2021. Decentralized Mining in Centralized Pools. The Review of Financial Studies 34 (3): 1191–1235. Cong, L. W., Y. Li, and N. Wang. 2021. Tokenomics: Dynamic Adoption and Valuation. The Review of Financial Studies 34 (3): 1105–1155. 33 Corbet, S., B. Lucey, A. Urquhart, and L. Yarovaya. 2019. Cryptocurrencies as a financial asset: A systematic analysis. International Review of Financial Analysis 62: 182–199. Dechow, P. M., and H. You. 2020. Understanding the Determinants of Analyst Target Price Implied Returns. The Accounting Review 95 (6): 125–149. Deloitte. 2021a. Deloitte’s 2021 Global Blockchain Survey. Deloitte Insights. https://www2.deloitte.com/us/en/insights/topics/understanding-blockchain-potential/global- blockchain-survey.html. ———. 2021b. The Business Benefit of Using Cryptocurrency. https://www2.deloitte.com/us/en/pages/audit/articles/corporates-using-crypto.html. FASB. 2023. Accounting for and Disclosure of Crypto Assets. https://www.fasb.org/page/PageContent?pageId=/projects/recentlycompleted/accounting-for- and-disclosure-of-crypto-assets.html. Florysiak, D., and A. Schandlbauer. 2022. Experts or charlatans? ICO analysts and white paper informativeness. Journal of Banking & Finance 139: 106476. Foley, S., J. R. Karlsen, and T. J. Putniņš. 2019. Sex, Drugs, and Bitcoin: How Much Illegal Activity Is Financed through Cryptocurrencies? The Review of Financial Studies 32 (5): 1798– 1853. Freifeld, K., and G. Chavez-Dreyfuss. 2015. New York regulator issues final virtual currency rules. Reuters. https://www.reuters.com/article/us-bitcoin-regulation-new-york- idUSKBN0OJ23X20150603. Goldstein, I., W. Jiang, and G. A. Karolyi. 2019. To FinTech and Beyond. The Review of Financial Studies 32 (5): 1647–1661. Google. 2017. Google’s Year in Search. Google Trends. https://trends.google.com/trends/yis/2017/GLOBAL/. ———. 2021. Google’s Year in Search. Google Trends. https://trends.google.com/trends/yis/2021/GLOBAL/. Griffin, J. M., and A. Shams. 2020. Is Bitcoin Really Untethered? The Journal of Finance 75 (4): 1913–1964. He, J. (Jack), and X. Tian. 2013. The dark side of analyst coverage: The case of innovation. Journal of Financial Economics 109 (3): 856–878. 34 Hong, H., T. Lim, and J. C. Stein. 2000. Bad News Travels Slowly: Size, Analyst Coverage, and the Profitability of Momentum Strategies. The Journal of Finance 55 (1): 265–295. Duarte, F. 2024. How Many People Own Bitcoin? Exploding Topics. https://explodingtopics.com/blog/blockchain-stats. Irvine, P. J. 2004. Analysts’ Forecasts and Brokerage‐Firm Trading. The Accounting Review 79 (1): 125–149. Kempf, E., and O. Spalt. 2022. Attracting the Sharks: Corporate Innovation and Securities Class Action Lawsuits. Management Science. https://doi.org/10.1287/mnsc.2022.4388. Kothari, S. P., E. So, and R. Verdi. 2016. Analysts’ Forecasts and Asset Pricing: A Survey. Annual Review of Financial Economics 8 (1): 197–219. Krishnan, M., J. Mischke, and J. Remes. 2018. Is the Solow Paradox back? McKinsey Quarterly. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/is-the-solow-paradox- back. Kross, W. J., and I. Suk. 2012. Does Regulation FD work? Evidence from analysts’ reliance on public disclosure. Journal of Accounting and Economics 53 (1): 225–248. Li, Q. 2016. Do Analysts Understand Innovation? Evidence from Patents and Trademarks. Working Paper. Li, T., D. Shin, and B. Wang. 2023. Cryptocurrency Pump-and-Dump Schemes. Working Paper. Li, Y. 2021. Charlie Munger calls bitcoin “disgusting and contrary to the interests of civilization.” CNBC. https://www.cnbc.com/2021/05/01/charlie-munger-calls-bitcoin-disgusting- and-contrary-to-the-interests-of-civilization.html. Liu, Y., J. Sheng, and W. Wang. 2024. Technology and Cryptocurrency Valuation: Evidence from Machine Learning. Working Paper. Liu, Y., A. Tsyvinski, and X. Wu. 2022. Common Risk Factors in Cryptocurrency. The Journal of Finance 77 (2): 1133–1177. Luo, M., and S. Yu. 2022. Financial reporting for cryptocurrency. Review of Accounting Studies. Loh, R. K., and R. M. Stulz. 2011. When Are Analyst Recommendation Changes Influential? The Review of Financial Studies 24 (2): 593–627. ———. 2018. Is Sell-Side Research More Valuable in Bad Times? The Journal of Finance 73 (3): 959–1013. 35 Makarov, I., and A. Schoar. 2020. Trading and arbitrage in cryptocurrency markets. Journal of Financial Economics 135 (2): 293–319. Maurer, M. 2022. FASB Lays Out Disclosure Requirements for Companies’ Crypto Holdings Ahead of Potential Proposal. Wall Street Journal. https://www.wsj.com/articles/fasb-lays-out- disclosure-requirements-for-companies-crypto-holdings-ahead-of-potential-proposal- 11671052966. Nakamoto, S. 2008. Bitcoin: A Peer-to-Peer Electronic Cash System. https://bitcoin.org/bitcoin.pdf. Narayanan, A., and J. Clark. 2017. Bitcoin’s academic pedigree. Communications of the ACM 60 (12): 36–45. Palmon, D., and A. Yezegel. 2012. R&D Intensity and the Value of Analysts’ Recommendations. Contemporary Accounting Research 29 (2): 621–654. Roberts, J. J. 2020. Kings of Crypto: One Startup’s Quest to Take Cryptocurrency Out of Silicon Valley and Onto Wall Street. Harvard Business Press. Schonberger, J. 2022. SEC’s Gensler wants crypto exchange regulation in 2022, warns on stablecoin. yahoo!finance. https://finance.yahoo.com/news/se-cs-gensler-wants-crypto-exchange- regulation-in-2022-warns-on-stablecoin-risks-130254530.html. Sharma, R. 2022. How SEC Regs Will Change Cryptocurrency Markets. Investopedia. https://www.investopedia.com/news/how-sec-regs-will-change-cryptocurrency-markets/. Shen, M. 2024. MicroStrategy Owns About 1% of All Bitcoin With Latest Purchase. Bloomberg.com. https://www.bloomberg.com/news/articles/2024-03-19/microstrategy-owns- about-1-of-all-bitcoin-with-latest-purchase-mstr. Sockin, M., and W. Xiong. 2023. A Model of Cryptocurrencies. Management Science 69 (11): 6684–6707. Yang, Y. 2022. Goldman Offers Its First Bitcoin-Backed Loan in Crypto Push. Bloomberg. https://www.bloomberg.com/news/articles/2022-04-28/goldman-offers-its-first-bitcoin-backed- loan-in-crypto-push. Zhang, X. F. 2006. Information Uncertainty and Analyst Forecast Behavior. Contemporary Accounting Research 23 (2): 565–590. 36 APPENDIX A. CRYPTO-RELATED KEYWORDS AND 8-K FILINGS Table A.1 List of Crypto-related Keywords List of crypto-related keywords “blockchain” The search of keywords in 8-K filings is case-insensitive and considers plurals when applying the regular expression in Python. Category Blockchain Cryptocurrency “cryptocurrency”, “crypto asset”, “digital currency” Bitcoin Ethereum Litecoin Other coins “bitcoin” “ethereum” “litecoin” “cardano”, “polkadot”, “bitcoin cash”, “chainlink”, “binance coin”, “dogecoin”, “usd coin” Table A.2. Selection of Crypto-related 8-K Filings Total number of 8-K filings issued during 1/1/2014-9/30/2021 Excluding: 8-Ks without crypto-related keywords 8-Ks with an irrelevant mention of crypto-related keywords (e.g., mention Bitcoin in the section of SEC header) Firms with unavailable Compustat Identifier (i.e., GVKEY) Firms with unavailable I/B/E/S Identifier (i.e., TICKER) Firms with unavailable CRSP Identifier (i.e., PERMNO) Combining 8-K filings released on the same day of the same firm into one record Number of crypto-related 8-K filings 560,612 (557,955) (55) (861) (490) (123) (16) 1,112 37 Figure A.1 Crypto Excerpts from Mechanical Technology, Incorporated (“MTI”) 8-K and Exhibit Excerpts from Mechanical Technology, Incorporated (“MTI”) crypto-related 8-K and its exhibit filed on 4/12/2021. 38 APPENDIX B. FREQUENT WORDS AROUND CRYPTO-RELATED KEYWORDS Texts within 5 words’ distance of a crypto-related keyword in 8-K filings and their exhibits for period 1/1/2014 – 9/30/2021 are the basis of word clouds. Word size represents a word’s number of occurrences in the texts. In plotting the word cloud, Figure B.1 includes crypto-related keywords and Figure B.2 excludes crypto-related keywords. Figure B.1 Word Cloud with Crypto-related Keywords Figure B.2 Word Cloud without Crypto-related Keywords 39 APPENDIX C. VARIABLE DEFINITION Table C.1 Variable Definition Variable BHAR_FF3 BHAR_FF3M BHAR_M BHAR_Madj ChgTP RevAF TPAccuracy AFAccuracy AvgChgTP BrokerSize BHAR_Crypto CAR_Crypto Comove_Coef Comove_Corr CorpGov EarningsNews Exp_Firm Definition Buy-and-hold abnormal return adjusted for the Fama-French three-factor model for the [-3, 3] trading days relative to the 8-K filing date. Buy-and-hold abnormal return adjusted for the model with Fama-French three factors and the momentum for the [-3, 3] trading days relative to the 8-K filing date. Buy-and-hold abnormal return adjusted for the market model for the [-3, 3] trading days relative to the 8-K filing date. Buy-and-hold abnormal return adjusted for the value-weighted market return for the [-3, 3] trading days relative to the 8-K filing date. The difference between the current and prior target price forecast issued by the same analyst for the same firm, scaled by stock price two trading days prior to the current target price forecast announcement date. The difference between the current and prior one-year ahead annual earnings forecast issued by the same analyst for the same firm, scaled by stock price two trading days prior to the current earnings forecast announcement date. Negative one times the absolute value of the difference between current target price and the forecast horizon end stock price, scaled by stock price two trading days prior to the current target price announcement date. Negative one times the absolute value of the difference between current one-year ahead annual earnings forecast and the actual earnings per share, scaled by stock price two trading days prior to the current earnings forecast announcement date. The average of analyst target price forecast revisions (i.e., ChgTP) for the same firm issued during [-3, 3] trading days relative to the 8-K filing date. AvgChgTP is coded as zero if there is no analyst target price forecast revision for the firm during [-3, 3] trading days relative to the 8-K filing date. The natural logarithm of one plus the number of analysts in a year working for the brokerage that the analyst is associated with as recorded in I/B/E/S. The market adjusted buy-and-hold abnormal cryptocurrency market return for the [-21, -6] trading days relative to the 8-K filing date, where the cryptocurrency market return is the value-weighted return of all the underlying cryptocurrencies adjusted by risk-free rate. The market adjusted cumulated abnormal cryptocurrency market return for the [-21, -6] trading days relative to the 8-K filing date, where the cryptocurrency market return is the value-weighted return of all the underlying cryptocurrencies adjusted by risk-free rate. The coefficient on the cryptocurrency market return when regressing the firm’s risk-free rate adjusted return on Fama-French three factors and the cryptocurrency market return for the [- 252, -6] trading days relative to the 8-K filing date, where the cryptocurrency market return is the value-weighted return of all the underlying cryptocurrencies adjusted by risk-free rate. The Pearson correlation coefficient between the firm’s risk-free rate adjusted return and the cryptocurrency market return for the [-252, -6] trading days relative to the 8-K filing date, where the cryptocurrency market return is the value-weighted return of all the underlying cryptocurrencies adjusted by risk-free rate. An indicator that equals one if the firm’s crypto exposure disclosed in the 8-K filing is related to corporate governance and zero otherwise. The difference between the actual earnings per share announced within 5 days prior to the 8- K filing date and the median of analyst forecasts issued within 30 days before the earnings announcement date, scaled by the stock price at the beginning of the quarter of the 8-K filing date. EarningsNews is set to zero if there is no earnings announcement within 5 days prior to the 8-K filing date of the firm. Time interval in years since an analyst provides the first earnings forecast for the underlying firm as recorded in I/B/E/S. 40 Table C.1 (cont’d) Exp_General Item202 Time interval in years since an analyst provides the first earnings forecast as recorded in I/B/E/S. An indicator that equals one if the 8-K filing contains the item 2.02 Results of Operations and Financial Condition and zero otherwise. Number of different items in the 8-K filing. The natural logarithm of one plus the number of analysts following the firm. The natural logarithm of total assets. ItemCount LnAnalyst LnAT LnKeywordCount The natural logarithm of one plus the number of crypto-related keywords in an 8-K filing and LnNumFirms MB PastRet RetSD RevGrowth Turnover its exhibits. The natural logarithm of one plus the number of firms an analyst covers. The ratio of the market value of equity to the book value of equity. Buy-and-hold abnormal return adjusted for the value-weighted market return for the [-252, - 6] trading days relative to the 8-K filing date. The standard deviation of daily stock returns of a firm for the [-252, -6] trading days relative to the 8-K filing date. The annual growth of total revenues in the past year. The average of the daily stock turnovers of a firm for the [-252, -6] trading days relative to the 8-K filing date. 41 APPENDIX D. FIRM CHARACTERISTICS FOR 8-K FILINGS Table D.1 Firm Characteristics For 8-K Filings The table below presents summary statistics of firm characteristics of 8-K filings issued during 1/1/2014 – 12/31/2023 that can be matched with firm identifiers in Compustat, I/B/E/S, and CRSP and have non-missing firm level control variables. Two sample t tests are used to examine the main difference between firm characteristics of crypto-related and non-related 8-K filings. Variables are defined in Appendix C. Non-crypto-related 8-K filings Crypto-related 8-K filings Turnover RetSD PastRet LnAnalyst LnAT MB RevGrowth N 443,838 443,838 443,838 443,838 443,838 443,838 443,838 Mean 0.01 0.03 -0.03 2.00 21.12 3.28 0.22 N 2,032 2,032 2,032 2,032 2,032 2,032 2,032 Mean Mean difference t value 0.04 0.05 0.07 1.61 20.52 3.82 0.60 -0.03 -0.02 -0.10 0.39 0.60 -0.54 -0.37 -74.90 -46.20 -9.00 19.50 12.25 -3.30 -19.50 42 Table E.1 Distribution of Crypto-related 8-K Filings and Keyword Counts APPENDIX E: TABLES OF EMPIRICAL RESULTS This table presents the distribution of crypto-related 8-K filings and keyword counts for the period from January 2014 to September 2021. Table E.1A shows the distribution of 1,112 8-K filings containing crypto-related keywords by firms over time. Table E.1B exhibits the number of crypto-related keywords by category over time. The category of crypto-related keywords is defined in Appendix A. Table E.1A Distribution of 8-K Filings Containing Crypto-related Keywords by Firms Over Time Year 2014 2015 2016 2017 2018 2019 2020 2021 Number of 8-K Filings Containing Crypto-related Keywords 41 25 54 93 293 191 169 246 Number of Firms 10 13 19 31 100 68 44 82 Table E.1B Distribution of Crypto-related Keyword Counts Disclosed in 8-K Filings Year 2014 2015 2016 2017 2018 2019 2020 2021 Total Blockchain 21 21 312 652 1,275 521 944 1,533 5,279 Cryptocurrency 66 115 37 269 524 226 576 833 2,646 Bitcoin 309 146 42 139 296 163 336 1,143 2,574 Ethereum 0 0 0 39 27 5 3 36 110 Litecoin 2 0 0 9 27 4 1 3 46 Other coins 6 0 0 1 9 3 1 0 Total 404 282 391 1,109 2,158 922 1,861 3,548 20 10,675 43 Table E.2 Summary Statistics This table reports summary statistics of variables used in the analyses. Table E.2A reports the summary statistics of variables for the target price revision sample. Table E.2B reports the summary statistics of variables for the annual earnings forecast revision sample. Table E.2C reports the summary statistics of variables for the sample of the market reaction test. Continuous variables are winsorized at 1% and 99%. All variables are defined in Appendix C. Table E.2A Summary Statistics of the Target Price Revision Sample ChgTP LnKeywordCount Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth CorpGov Comove_Coef Comove_Corr N 1,880 1,880 1,880 1,880 1,880 1,880 1,880 1,880 1,880 1,880 1,880 1,880 1,880 1,880 1,877 1,877 Mean 0.041 1.351 0.019 0.027 0.398 5.580 16.255 3.066 2.956 3.900 23.132 6.296 0.196 0.012 0.013 0.081 SD 0.187 0.932 0.020 0.015 0.921 4.722 10.787 0.898 0.507 1.017 1.876 35.393 0.263 0.110 0.030 0.104 Min -0.613 0.693 0.002 0.009 -0.619 0.000 0.170 0.000 1.099 1.099 17.276 -283.393 -0.199 0.000 -0.039 -0.101 p25 -0.029 0.693 0.007 0.015 -0.091 1.892 7.659 2.773 2.773 3.219 22.190 3.183 -0.010 0.000 -0.006 0.003 Median 0.051 1.099 0.011 0.025 0.128 4.240 13.952 3.401 3.045 4.078 22.970 8.041 0.150 0.000 0.007 0.070 p75 0.120 1.609 0.025 0.033 0.527 8.241 23.248 3.689 3.277 4.718 24.091 16.934 0.309 0.000 0.030 0.137 Max 0.731 4.394 0.107 0.088 5.984 22.151 37.827 3.989 3.892 5.485 28.561 48.348 1.438 1.000 0.152 0.359 44 Table E.2B Summary Statistics of the Annual Earnings Forecast Revision Sample RevAF LnKeywordCount Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth CorpGov Comove_Coef Comove_Corr N 2,692 2,692 2,692 2,692 2,692 2,692 2,692 2,692 2,692 2,692 2,692 2,692 2,692 2,692 2,692 2,692 Mean 0.000 1.322 0.018 0.027 0.360 5.961 17.060 3.048 2.960 3.885 23.215 3.812 0.188 0.013 0.013 0.084 SD 0.009 0.901 0.020 0.015 0.984 5.224 10.819 0.934 0.501 0.938 1.965 42.326 0.252 0.113 0.028 0.099 Min -0.046 0.693 0.002 0.009 -0.655 0.005 0.318 0.000 1.099 1.099 17.276 -283.393 -0.199 0.000 -0.040 -0.101 p25 -0.001 0.693 0.007 0.015 -0.092 1.956 8.175 2.862 2.773 3.219 22.227 2.185 -0.010 0.000 -0.005 0.012 Median 0.001 1.099 0.011 0.026 0.086 4.588 15.248 3.401 3.045 3.932 22.970 7.685 0.137 0.000 0.008 0.073 Table E.2C Summary Statistics of the Sample for Market Reaction Test BHAR_FF3 LnKeywordCount Turnover RetSD LnAnalyst LnAT MB RevGrowth AvgChgTP N 1,087 1,087 1,087 1,087 1,087 1,087 1,087 1,087 1,087 Mean 0.004 1.577 0.049 0.054 1.478 20.165 4.550 1.031 0.007 SD 0.205 1.027 0.072 0.039 1.192 2.830 9.008 4.951 0.184 Min -0.450 0.693 0.001 0.009 0.000 15.563 -35.891 -0.882 -1.179 p25 -0.089 0.693 0.006 0.022 0.693 17.479 1.286 -0.053 0.000 Median -0.014 1.099 0.012 0.043 1.099 19.949 3.253 0.054 0.000 p75 0.002 1.609 0.022 0.032 0.421 8.645 25.393 3.638 3.258 4.691 24.662 16.845 0.296 0.000 0.030 0.147 p75 0.048 2.079 0.066 0.077 2.303 22.359 6.925 0.287 0.010 Max 0.033 4.394 0.107 0.098 6.272 27.403 37.868 3.989 3.892 5.485 28.561 48.348 1.438 1.000 0.146 0.324 Max 1.009 4.754 0.294 0.195 3.850 26.824 49.466 30.765 0.636 45 Table E.3 Pearson Correlations of the Target Price Revision Sample This table reports Pearson correlations of the variables for the target price revision sample. ***, **, and * denote significance at 1%, 5%, and 10% levels. (4) (9) (8) (3) (2) (7) (1) (5) (6) (1) ChgTP (2) LnKeywordCount (3) Turnover (4) RetSD (5) PastRet (6) Exp_Firm (7) Exp_General (8) LnAnalyst (9) LnNumFirms (10) BrokerSize (11) LnAT (12) MB (13) RevGrowth (14) CorpGov (15) Comove_Coef (16) Comove_Corr 1.000 0.192*** -0.010 0.092*** 0.230*** -0.010 0.047* 0.057* -0.036 -0.008 -0.003 0.061** 0.083*** -0.006 0.118*** 0.195*** 1.000 0.296*** 0.324*** 0.318*** -0.270*** -0.053* 0.063** -0.016 -0.024 -0.122*** 0.143*** 0.444*** -0.058* 0.319*** 0.245*** 1.000 0.675*** 0.562*** -0.195*** 0.010 -0.007 -0.007 -0.103*** -0.432*** 0.084*** 0.345*** -0.014 0.283*** 0.187*** 1.000 0.553*** -0.226*** 0.038 -0.247*** -0.015 -0.173*** -0.500*** 0.090*** 0.309*** 0.060** 0.469*** 0.485*** 1.000 -0.210*** -0.045* -0.112*** -0.045 -0.036 -0.211*** 0.088*** 0.341*** -0.029 0.248*** 0.234*** 1.000 0.392*** 0.179*** 0.120*** 0.114*** 0.290*** -0.120*** -0.174*** -0.052* -0.110*** -0.061** 1.000 0.067** 0.337*** -0.010 0.032 -0.060** -0.031 -0.012 0.032 0.058* 1.000 0.066** 0.078*** 0.374*** 0.080*** 0.187*** -0.128*** -0.006 0.069** 1.000 0.064** 0.019 -0.012 -0.018 0.017 0.007 0.017 46 Table E.3 (cont’d) (10) BrokerSize (11) LnAT (12) MB (13) RevGrowth (14) CorpGov (15) Comove_Coef (16) Comove_Corr (10) 1.000 0.121*** -0.021 -0.050* -0.049* -0.058* -0.067** (11) (12) (13) (14) (15) (16) 1.000 -0.018 -0.213*** -0.192*** -0.231*** -0.127*** 1.000 0.262*** 0.017 0.048* 0.018 1.000 -0.029 0.267*** 0.363*** 1.000 -0.047* -0.032 1.000 0.779*** 1.000 47 Table E.4 Analyst Revision and Crypto Exposure This table reports regression results of analysts’ revision within five calendar days following firms’ crypto exposure disclosed in 8-K filings. I estimate the following model: AnalystRevisionijt = α + β1LnKeywordCountjt + Controls + Fixed Effects + εjit. The dependent variable used in column (1) is ChgTP, which is the difference between the current and prior target price forecast issued by the same analyst for the same firm, deflated by stock price two trading days prior to the current target price forecast announcement date. The dependent variable used in column (2) is RevAF, which is the difference between the current and prior one-year ahead annual earnings forecast issued by the same analyst for the same firm, deflated by stock price two trading days prior to the current earnings forecast announcement date. LnKeywordCount is the natural logarithm of one plus the number of crypto-related keywords in an 8-K filing and its exhibits. All variables are defined in Appendix C. Standard errors are clustered at the analyst level, and t-statistics (two-tailed test) are shown in the parenthesis. ***, **, and * denote significance at 1%, 5%, and 10% levels. Variables LnKeywordCount Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared (2) RevAF 0.000 (0.66) -0.017 (-0.85) -0.061 (-1.10) 0.001*** (3.74) -0.000 (-1.23) 0.000 (0.49) 0.000 (0.22) -0.001 (-1.09) 0.000 (0.50) -0.000 (-0.11) 0.000* (1.74) 0.002 (0.99) -0.005 (-0.32) Yes Yes Yes 2,692 0.181 (1) ChgTP 0.017** (2.45) -2.090*** (-2.84) 0.174 (0.20) 0.078*** (5.70) 0.002 (1.13) -0.025 (-1.56) 0.010 (0.97) -0.009 (-0.39) -0.026* (-1.70) -0.002 (-0.50) 0.000*** (5.10) -0.081*** (-2.96) 0.580* (1.94) Yes Yes Yes 1,880 0.260 48 Table E.5 Cross-sectional Analysis: Crypto-related Disclosure on Corporate Governance This table reports the results of the cross-sectional analysis based on whether firms’ crypto exposure disclosed in 8- K filings related to corporate governance. I estimate the following model: BHAR_FF3jt = α + β1LnKeywordCountjt + β2LnKeywordCountjt × AvgChgTPjt + Controls + Fixed Effects + εit. The dependent variable used in column (1) is ChgTP, which is the difference between the current and prior target price forecast issued by the same analyst for the same firm, deflated by stock price two trading days prior to the current target price forecast announcement date. The dependent variable used in column (2) is RevAF, which is the difference between the current and prior one-year ahead annual earnings forecast issued by the same analyst for the same firm, deflated by stock price two trading days prior to the current earnings forecast announcement date. LnKeywordCount is the natural logarithm of one plus the number of crypto-related keywords in an 8-K filing and its exhibits. CorpGov is an indicator that equals 1 if the firm’s crypto exposure disclosed in the 8-K filing is related to corporate governance and zero otherwise. All variables are defined in Appendix C. Standard errors are clustered at the analyst level, and t-statistics (two-tailed test) are shown in the parenthesis. ***, **, and * denote significance at 1%, 5%, and 10% levels. Variables LnKeywordCount LnKeywordCount × CorpGov CorpGov Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared (1) ChgTP 0.016** (2.40) 0.249*** (3.41) -0.210** (-2.52) -1.938*** (-2.73) 0.215 (0.24) 0.077*** (5.63) 0.002 (1.16) -0.028* (-1.66) 0.012 (1.09) -0.006 (-0.26) -0.024 (-1.58) -0.002 (-0.45) 0.000*** (5.13) -0.082*** (-2.95) 0.593* (1.95) Yes Yes Yes 1,880 0.263 49 (2) RevAF 0.000 (0.61) -0.005 (-0.76) 0.009 (1.28) -0.012 (-0.61) -0.062 (-1.10) 0.001*** (3.73) -0.000 (-1.27) 0.001 (0.67) 0.000 (0.08) -0.001 (-1.14) 0.000 (0.58) 0.000 (0.28) 0.000* (1.67) 0.002 (1.05) -0.009 (-0.62) Yes Yes Yes 2,692 0.183 Table E.6 Cross-sectional Analysis: Degree of Comovement between Stock Returns and Cryptocurrency Market Returns This table reports the results of the cross-sectional analysis based on the degree of comovement between firms’ stock returns and the cryptocurrency market returns. I estimate the following model: AnalystRevisionijt = α + β1LnKeywordCountjt + β2LnKeywordCountjt × Comovejt + β3Comovejt + Controls + Fixed Effects + εjit. The dependent variable used in columns (1) and (2) is ChgTP, which is the difference between the current and prior target price forecast issued by the same analyst for the same firm, deflated by stock price two trading days prior to the current target price forecast announcement date. The dependent variable used in columns (3) and (4) is RevAF, which is the difference between the current and prior one-year ahead annual earnings forecast issued by the same analyst for the same firm, deflated by stock price two trading days prior to the current earnings forecast announcement date. LnKeywordCount is the natural logarithm of one plus the number of crypto-related keywords in an 8-K filing and its exhibits. Comove is proxied by Comove_Coef and Comove_Corr. Comove_Coef is the coefficient on the cryptocurrency market return when regressing the firm’s risk-free rate adjusted return on Fama- French three factors and the cryptocurrency market return for the [-252, -6] trading days relative to the 8-K filing date. Comove_Corr is the Pearson correlation coefficient between the firm’s risk-free rate adjusted return and the cryptocurrency market return for the [-252, -6] trading days relative to the 8-K filing date. The cryptocurrency market return is the value-weighted return of all the underlying cryptocurrencies adjusted by the risk-free rate. All variables are defined in Appendix C. Standard errors are clustered at the analyst level, and t-statistics (two-tailed test) are shown in the parenthesis. ***, **, and * denote significance at 1%, 5%, and 10% levels. ChgTP RevAF Variables LnKeywordCount LnKeywordCount × Comove_Coef Comove_Coef LnKeywordCount × Comove_Corr Comove_Corr Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB (2) 0.004 (0.52) 0.104*** (2.60) 0.186** (2.00) -1.590** (-2.20) -1.546 (-1.54) 0.075*** (5.37) 0.001 (0.87) -0.045*** (-2.82) 0.005 (0.42) -0.012 (-0.50) -0.027* (-1.73) -0.004 (-1.06) 0.001*** (1) 0.006 (0.80) 0.362** (2.47) -0.364 (-1.06) -2.039*** (-2.72) -0.214 (-0.24) 0.076*** (5.45) 0.002 (1.06) -0.025 (-1.59) 0.008 (0.74) -0.006 (-0.27) -0.028* (-1.87) -0.001 (-0.30) 0.000*** 50 (3) 0.001 (1.21) 0.003 (0.16) -0.038 (-1.35) -0.016 (-0.82) -0.051 (-0.90) 0.001*** (3.84) -0.000 (-1.24) 0.001 (0.58) 0.000 (0.66) -0.001 (-0.90) 0.000 (0.65) -0.000 (-0.03) 0.000* (4) 0.001 (1.55) -0.002 (-0.67) -0.015*** (-2.65) -0.024 (-1.25) -0.052 (-0.93) 0.001*** (4.06) -0.000 (-1.23) 0.001 (0.64) 0.000 (0.85) -0.001 (-0.77) 0.000 (0.59) 0.000 (0.26) 0.000** Table E.6 (cont’d) RevGrowth Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared (4.91) -0.084*** (-3.04) 0.584** (2.01) Yes Yes Yes 1,877 0.263 (5.13) -0.114*** (-4.18) 1.029*** (3.40) Yes Yes Yes 1,877 0.274 (1.91) 0.001 (0.63) -0.008 (-0.55) Yes Yes Yes 2,692 0.186 (2.22) 0.002 (1.27) -0.010 (-0.67) Yes Yes Yes 2,692 0.189 51 Table E.7 Market Reaction Test This table reports the results of market reaction to firms’ crypto-related 8-K filings. I estimate the following two models for columns (1)-(2) and columns (3)-(4), respectively: BHAR_FF3jt = α + β1LnKeywordCountjt + Controls + Fixed Effects + εjt, and BHAR_FF3jt = α + β1LnKeywordCountjt + β2LnKeywordCountjt × AvgChgTPjt + Controls + Fixed Effects + εit. The dependent variable BHAR_FF3 is the buy-and-hold abnormal return adjusted for the Fama- French three-factor model for the [-3, 3] trading days relative to the 8-K filing date. LnKeywordCount is the natural logarithm of one plus the number of crypto-related keywords in an 8-K filing and its exhibits. AvgChgTP is the average of analyst target price forecast revision (i.e., ChgTP) for the same firm issued during [-3, 3] trading days relative to the 8-K filing date. AvgChgTP is coded as zero if there is no analyst target price forecast revision for the firm during [-3, 3] trading days relative to the 8-K filing date. All variables are defined in Appendix C. Standard errors are clustered at the firm level in columns (1) and (3) and are clustered at the industry level in columns (2) and (4). t-statistics (two-tailed test) are shown in the parenthesis. ***, **, and * denote significance at 1%, 5%, and 10% levels. BHAR_FF3 (3) 0.030*** (2.65) 0.066** (2.10) 0.042 (0.56) -0.881* (-1.68) -0.511 (-0.55) 0.010 (0.27) 0.033 (1.19) -0.000 (-0.19) 0.002 (0.83) -0.656 (-1.17) Yes No Yes 1,087 0.005 (4) 0.025*** (2.95) 0.058** (2.38) 0.063 (1.21) -0.408** (-2.10) -0.245 (-0.77) 0.010** (2.34) -0.006*** (-3.02) -0.001** (-2.27) 0.001 (0.67) 0.114* (1.98) Yes Yes No 1,087 0.059 Variables LnKeywordCount LnKeywordCount × AvgChgTP (1) 0.034*** (2.98) (2) 0.030*** (4.26) AvgChgTP Turnover RetSD LnAnalyst LnAT MB RevGrowth Constant Year Fixed Effects Industry Fixed Effects Firm Fixed Effects Observations Adjusted R-squared -0.945* (-1.75) -0.408 (-0.45) 0.001 (0.02) 0.037 (1.34) -0.000 (-0.08) 0.003 (1.01) -0.730 (-1.30) Yes No Yes 1,087 -0.020 -0.457** (-2.22) -0.157 (-0.47) 0.009** (2.52) -0.005** (-2.13) -0.001* (-1.70) 0.001 (0.77) 0.074 (1.22) Yes Yes No 1,087 0.036 52 Table E.8 Robustness of Market Reaction Test This table reports the results of market reaction to firms’ crypto-related 8-K filings with alternative market reaction measures. Panel A and Panel B estimate the following models, respectively: BHAR_FF3jt = α + β1LnKeywordCountjt + Controls + Fixed Effects + εjt, and BHAR_FF3jt = α + β1LnKeywordCountjt + β2LnKeywordCountjt × AvgChgTPjt + Controls + Fixed Effects + εit. In both Table E.8A and Table E.8B, the dependent variable in columns (1) and (2) is buy-and-hold abnormal return adjusted for the value-weighted market return (BHAR_Madj). In columns (3) and (4), the dependent variable is buy-and-hold abnormal return adjusted for the market model (BHAR_M). In columns (5) and (6), the dependent variable is buy-and-hold abnormal return adjusted for the model with Fama-French three factors and the momentum (BHAR_FF3M). All three dependent variables are measured during the [-3, 3] trading days relative to the 8-K filing date. LnKeywordCount is the natural logarithm of one plus the number of crypto-related keywords in an 8-K filing and its exhibits. AvgChgTP is the average of analyst target price forecast revision (i.e., ChgTP) for the same firm issued during [-3, 3] trading days relative to the 8-K filing date. AvgChgTP is coded as zero if there is no analyst target price forecast revision for the firm during [-3, 3] trading days relative to the 8-K filing date. All variables are defined in Appendix C. Standard errors are clustered at the firm level in columns (1), (3) and (5), and are clustered at the industry level in columns (2), (4) and (6). t-statistics (two-tailed test) are shown in the parenthesis. ***, **, and * denote significance at 1%, 5%, and 10% levels. Table E.8A Market Reaction to Firms’ Crypto Exposure BHAR_Madj BHAR_M BHAR_FF3M Variables LnKeywordCount Turnover RetSD LnAnalyst LnAT MB RevGrowth Constant Year Fixed Effects Industry Fixed Effects Firm Fixed Effects Observations Adjusted R-squared (1) 0.038*** (3.22) -0.859 (-1.56) 0.257 (0.28) -0.021 (-0.55) 0.053** (2.08) -0.000 (-0.30) 0.003 (1.03) -1.044** (-2.07) Yes No Yes 1,087 -0.030 (2) 0.035*** (5.30) -0.459* (-1.99) 0.352 (1.22) 0.004 (0.88) 0.000 (0.16) -0.001 (-1.10) 0.001 (0.61) -0.047 (-0.71) Yes Yes No 1,087 0.031 (3) 0.035*** (3.42) -0.998* (-1.82) -0.218 (-0.24) -0.006 (-0.17) 0.042 (1.57) 0.000 (0.00) 0.003 (1.06) -0.827 (-1.54) Yes No Yes 1,087 -0.016 (4) 0.031*** (4.84) -0.531** (-2.71) -0.027 (-0.07) 0.008** (2.28) -0.004 (-1.52) -0.001 (-1.45) 0.001 (0.77) 0.051 (0.71) Yes Yes No 1,087 0.040 (5) 0.034*** (2.94) -0.939* (-1.75) -0.561 (-0.66) -0.001 (-0.04) 0.033 (1.14) 0.000 (0.06) 0.003 (0.97) -0.635 (-1.09) Yes No Yes 1,087 -0.019 (6) 0.029*** (4.05) -0.467** (-2.31) -0.260 (-0.88) 0.009*** (3.17) -0.006*** (-2.83) -0.001* (-1.77) 0.002 (0.78) 0.103** (2.07) Yes Yes No 1,087 0.039 53 Table E.8B The Effect of Analyst Target Price Revision on Market Reaction to Firms’ Crypto Exposure Variables LnKeywordCount LnKeywordCount × AvgChgTP AvgChgTP Turnover RetSD LnAnalyst LnAT MB RevGrowth Constant Year Fixed Effects Industry Fixed Effects Firm Fixed Effects Observations Adjusted R-squared BHAR_Madj BHAR_M BHAR_FF3M (1) 0.034*** (2.89) 0.062** (2.08) 0.075 (0.93) -0.790 (-1.47) 0.148 (0.16) -0.011 (-0.28) 0.048* (1.90) -0.000 (-0.46) 0.003 (0.85) -0.964* (-1.90) Yes No Yes 1,087 0.002 (2) 0.030*** (3.77) 0.053** (2.35) 0.107* (1.91) -0.405* (-1.82) 0.250 (0.90) 0.005 (0.96) -0.002 (-0.90) -0.001* (-1.71) 0.001 (0.50) 0.003 (0.04) Yes Yes No 1,087 0.063 (3) 0.032*** (3.01) 0.062* (1.96) 0.059 (0.72) -0.934* (-1.75) -0.321 (-0.34) 0.003 (0.09) 0.038 (1.41) -0.000 (-0.11) 0.002 (0.87) -0.752 (-1.40) Yes No Yes 1,087 0.010 (4) 0.026*** (3.29) 0.052** (2.21) 0.085 (1.49) -0.481** (-2.59) -0.119 (-0.33) 0.009** (2.12) -0.006** (-2.35) -0.001* (-1.89) 0.001 (0.64) 0.094 (1.37) Yes Yes No 1,087 0.065 (5) 0.030*** (2.62) 0.065** (2.06) 0.044 (0.59) -0.875* (-1.68) -0.664 (-0.76) 0.008 (0.22) 0.029 (1.01) -0.000 (-0.05) 0.002 (0.78) -0.561 (-0.97) Yes No Yes 1,087 0.006 (6) 0.025*** (2.79) 0.057** (2.33) 0.064 (1.24) -0.417** (-2.20) -0.347 (-1.25) 0.010*** (2.85) -0.008*** (-3.85) -0.001** (-2.27) 0.001 (0.68) 0.142*** (3.14) Yes Yes No 1,087 0.062 54 Table E.9 Analyst Revision and Crypto Exposure after Controlling Confounding Events This table reports regression results of analysts’ revision within five calendar days following firms’ crypto exposure disclosed in 8-K filings after controlling confounding events. Columns (1) – (3) add EarningsNews, Item202, ItemCount, respectively, to control for confounding events. Table E.9A reports the baseline regression results. Table E.9B reports the cross-sectional results related to corporate governance. Table E.9C and Table E.9D report the cross-sectional results related to the degree of comovement between stock return and the cryptocurrency market return. All variables are defined in Appendix C. Standard errors are clustered at the analyst level, and t-statistics (two-tailed test) are shown in the parenthesis. ***, **, and * denote significance at 1%, 5%, and 10% levels. Table E.9A Analyst Revision in Response to Crypto Exposure Variables LnKeywordCount Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth EarningsNews Item202 ItemCount Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared ChgTP (2) 0.017** (2.44) -2.102*** (-2.84) 0.189 (0.22) 0.078*** (5.70) 0.002 (1.10) -0.026 (-1.60) 0.010 (0.97) -0.010 (-0.41) -0.026* (-1.70) -0.002 (-0.46) 0.000*** (5.23) -0.082*** (-2.93) 0.005 (0.38) 0.579* (1.93) Yes Yes Yes 1,880 0.260 (1) 0.018*** (2.69) -2.127*** (-3.01) -0.300 (-0.36) 0.080*** (6.12) 0.002 (1.00) -0.030* (-1.85) 0.008 (0.76) -0.013 (-0.54) -0.027* (-1.87) -0.004 (-0.96) 0.000*** (5.32) -0.087*** (-3.17) 9.984*** (4.48) 0.726** (2.44) Yes Yes Yes 1,880 0.271 55 (3) 0.016** (2.32) -2.170*** (-2.91) 0.242 (0.28) 0.079*** (5.74) 0.002 (1.08) -0.024 (-1.49) 0.011 (1.02) -0.008 (-0.32) -0.026* (-1.74) -0.001 (-0.31) 0.000*** (5.08) -0.081*** (-2.97) 0.007 (0.84) 0.526* (1.74) Yes Yes Yes 1,880 0.260 Table E.9B Analyst Revision in Response to Crypto Exposure Conditional on Corporate Governance Variables LnKeywordCount LnKeywordCount × CorpGov CorpGov Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth EarningsNews Item202 ItemCount Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared ChgTP (2) 0.016** (2.39) 0.247*** (3.41) -0.206** (-2.46) -1.944*** (-2.73) 0.222 (0.25) 0.077*** (5.62) 0.002 (1.15) -0.028* (-1.68) 0.012 (1.09) -0.007 (-0.27) -0.024 (-1.58) -0.002 (-0.41) 0.000*** (5.28) -0.083*** (-2.93) 0.003 (0.24) 0.590* (1.93) Yes Yes Yes 1,880 0.262 (1) 0.018*** (2.64) 0.221*** (2.93) -0.184** (-2.08) -1.988*** (-2.90) -0.250 (-0.29) 0.079*** (6.04) 0.002 (1.03) -0.032* (-1.91) 0.009 (0.88) -0.010 (-0.41) -0.025* (-1.75) -0.004 (-0.89) 0.000*** (5.33) -0.088*** (-3.15) 9.653*** (4.41) 0.731** (2.43) Yes Yes Yes 1,880 0.273 56 (3) 0.016** (2.30) 0.240*** (3.06) -0.204** (-2.37) -1.989*** (-2.73) 0.252 (0.29) 0.078*** (5.63) 0.002 (1.13) -0.027 (-1.61) 0.012 (1.11) -0.005 (-0.23) -0.025 (-1.60) -0.002 (-0.35) 0.000*** (5.11) -0.083*** (-2.96) 0.004 (0.47) 0.564* (1.81) Yes Yes Yes 1,880 0.262 Table E.9C Analyst Revision in Response to Crypto Exposure Conditional on Return Comovement Coefficient Variables LnKeywordCount LnKeywordCount × Comove_Coef Comove_Coef Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth EarningsNews Item202 ItemCount Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared (1) 0.009 (1.16) 0.310** (2.11) -0.287 (-0.84) -2.088*** (-2.90) -0.622 (-0.70) 0.079*** (5.89) 0.002 (0.95) -0.030* (-1.89) 0.006 (0.57) -0.010 (-0.45) -0.029** (-2.03) -0.003 (-0.77) 0.000*** (5.13) -0.090*** (-3.26) 9.821*** (4.28) 0.730** (2.52) Yes Yes Yes 1,877 0.273 57 ChgTP (2) 0.006 (0.77) 0.359** (2.33) -0.361 (-1.03) -2.042*** (-2.70) -0.208 (-0.23) 0.076*** (5.42) 0.002 (1.06) -0.025 (-1.61) 0.008 (0.74) -0.007 (-0.28) -0.028* (-1.87) -0.001 (-0.29) 0.000*** (5.06) -0.084*** (-2.98) 0.001 (0.10) 0.583** (2.00) Yes Yes Yes 1,877 0.262 (3) 0.006 (0.75) 0.344** (2.25) -0.321 (-0.90) -2.106*** (-2.74) -0.168 (-0.18) 0.077*** (5.46) 0.002 (1.02) -0.024 (-1.54) 0.008 (0.77) -0.005 (-0.22) -0.028* (-1.89) -0.001 (-0.16) 0.000*** (4.88) -0.084*** (-3.05) 0.006 (0.66) 0.542* (1.83) Yes Yes Yes 1,877 0.262 Table E.9D Analyst Revision in Response to Crypto Exposure Conditional on Return Comovement Correlation Variables LnKeywordCount LnKeywordCount * Comove_Corr Comove_Corr Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth EarningsNews Item202 ItemCount Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared (1) 0.005 (0.66) 0.106*** (2.65) 0.161* (1.74) -1.654** (-2.37) -1.863* (-1.83) 0.077*** (5.74) 0.001 (0.76) -0.048*** (-3.01) 0.003 (0.27) -0.015 (-0.62) -0.029* (-1.92) -0.006 (-1.42) 0.001*** (5.30) -0.118*** (-4.33) 9.474*** (4.11) 1.135*** (3.75) Yes Yes Yes 1,877 0.284 58 ChgTP (2) 0.004 (0.43) 0.105** (2.37) 0.185* (1.94) -1.584** (-2.15) -1.555 (-1.53) 0.075*** (5.28) 0.001 (0.89) -0.045*** (-2.84) 0.005 (0.42) -0.012 (-0.49) -0.027* (-1.73) -0.004 (-1.09) 0.001*** (5.29) -0.114*** (-4.11) -0.002 (-0.11) 1.030*** (3.38) Yes Yes Yes 1,877 0.274 (3) 0.004 (0.51) 0.094** (2.17) 0.209** (2.10) -1.690** (-2.28) -1.488 (-1.47) 0.076*** (5.38) 0.001 (0.81) -0.045*** (-2.76) 0.005 (0.47) -0.011 (-0.44) -0.027* (-1.74) -0.004 (-0.89) 0.001*** (5.09) -0.114*** (-4.17) 0.008 (0.93) 0.976*** (3.16) Yes Yes Yes 1,877 0.274 Table E.10. Analyst Revision and Crypto Exposure after Controlling Crypto Momentum This table reports regression results of analysts’ revision within five calendar days following firms’ crypto exposure disclosed in 8-K filings after controlling the momentum of cryptocurrency market returns. Columns (1) and (2) add CAR_Crypto and BHAR_Crypto, respectively, to control the momentum of cryptocurrency market returns. Table E.10A reports the baseline regression results. Table E.10B reports the cross-sectional results related to corporate governance and the degree of comovement between stock return and the cryptocurrency market return. All variables are defined in Appendix C. Standard errors are clustered at the analyst level, and t-statistics (two-tailed test) are shown in the parenthesis. ***, **, and * denote significance at 1%, 5%, and 10% levels. Table E.10A Analyst Revision and Crypto Exposure ChgTP Variables LnKeywordCount Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth CAR_Crypto BHAR_Crypto Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared (2) 0.017** (2.38) -2.082*** (-2.83) 0.058 (0.07) 0.079*** (5.90) 0.002 (1.12) -0.023 (-1.42) 0.011 (1.05) -0.006 (-0.25) -0.026* (-1.76) -0.002 (-0.57) 0.000*** (5.05) -0.084*** (-3.08) -0.000 (-1.61) 0.549* (1.83) Yes Yes Yes 1,880 0.261 (1) 0.017** (2.38) -2.085*** (-2.84) 0.055 (0.06) 0.079*** (5.92) 0.002 (1.13) -0.023 (-1.42) 0.011 (1.04) -0.006 (-0.26) -0.026* (-1.75) -0.002 (-0.57) 0.000*** (5.06) -0.084*** (-3.08) -0.000 (-1.61) 0.548* (1.83) Yes Yes Yes 1,880 0.261 59 Table E.10B Cross-sectional Variation Results of Analyst Revision and Crypto Exposure (1) 0.016** (2.33) 0.244*** (3.30) -0.204** (-2.45) (2) 0.016** (2.33) 0.244*** (3.30) -0.204** (-2.45) Variables LnKeywordCount LnKeywordCount * CorpGov CorpGov LnKeywordCount * Comove_Coef Comove_Coef LnKeywordCount * Comove_Corr ChgTP (3) 0.006 (0.77) (4) 0.006 (0.77) (5) 0.004 (0.52) (6) 0.004 (0.51) 0.356** (2.43) -0.352 (-1.03) 0.357** (2.43) -0.351 (-1.02) Comove_Corr Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT -1.934*** (-2.72) 0.099 (0.11) 0.078*** (5.86) 0.002 (1.16) -0.026 (-1.53) 0.012 (1.15) -0.003 (-0.14) -0.025 (-1.63) -0.002 (-0.51) -1.932*** (-2.72) 0.102 (0.12) 0.078*** (5.83) 0.002 (1.16) -0.026 (-1.53) 0.013 (1.16) -0.003 (-0.13) -0.025 (-1.63) -0.002 (-0.51) -2.035*** (-2.72) -0.329 (-0.37) 0.078*** (5.65) 0.002 (1.06) -0.023 (-1.45) 0.009 (0.81) -0.004 (-0.15) -0.028* (-1.92) -0.002 (-0.37) 60 -2.033*** (-2.71) -0.327 (-0.36) 0.078*** (5.63) 0.002 (1.06) -0.023 (-1.45) 0.009 (0.82) -0.003 (-0.14) -0.028* (-1.93) -0.002 (-0.37) 0.101** (2.54) 0.191** (2.04) -1.589** (-2.20) -1.658* (-1.65) 0.076*** (5.55) 0.001 (0.87) -0.043*** (-2.68) 0.005 (0.49) -0.009 (-0.38) -0.028* (-1.77) -0.005 (-1.13) 0.102** (2.56) 0.190** (2.03) -1.585** (-2.19) -1.661* (-1.65) 0.076*** (5.53) 0.001 (0.87) -0.043*** (-2.68) 0.005 (0.50) -0.009 (-0.37) -0.028* (-1.78) -0.005 (-1.13) Table E.10B (cont’d) MB RevGrowth CAR_Crypto BHAR_Crypto Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared 0.000*** (5.08) -0.086*** (-3.07) -0.000 (-1.54) 0.561* (1.85) Yes Yes Yes 1,880 0.264 0.000*** (5.07) -0.086*** (-3.07) -0.000 (-1.54) 0.562* (1.85) Yes Yes Yes 1,880 0.264 0.000*** (4.87) -0.087*** (-3.15) -0.000 (-1.57) 0.553* (1.90) Yes Yes Yes 1,877 0.264 0.000*** (4.85) -0.087*** (-3.15) -0.000 (-1.58) 0.554* (1.90) Yes Yes Yes 1,877 0.264 0.001*** (5.09) -0.117*** (-4.25) -0.000 (-1.61) 0.999*** (3.30) Yes Yes Yes 1,877 0.275 0.001*** (5.07) -0.117*** (-4.25) -0.000* (-1.65) 0.999*** (3.29) Yes Yes Yes 1,877 0.275 61 Table E.11 Analyst Forecast Accuracy and Crypto Exposure RetSD PastRet Turnover Exp_Firm Variables LnKeywordCount This table reports regression results of the accuracy of analysts’ forecasts within five calendar days following firms’ crypto exposure disclosed in 8-K filings. I estimate the following model: AnalystAccuracyijt = α + β1LnKeywordCountjt + Controls + Fixed Effects + εjit. The dependent variable used in column (1) is TPAccuracy, which is negative one times the absolute value of the difference between current target price and the forecast horizon end stock price, scaled by stock price outstanding two trading days prior to the current target price announcement date. The dependent variable used in column (2) is AFAccuracy, which is negative one times the absolute value of the difference between current one-year ahead annual earnings forecast and the actual earnings per share, scaled by stock price outstanding two trading days prior to the current earnings forecast announcement date. LnKeywordCount is the natural logarithm of one plus the number of crypto-related keywords in an 8-K filing and its exhibits. All variables are defined in Appendix C. Standard errors are clustered at the analyst level, and t-statistics (two-tailed test) are shown in the parenthesis. ***, **, and * denote significance at 1%, 5%, and 10% levels. (2) AFAccuracy 0.000 (0.65) 0.074*** (4.05) -0.297*** (-7.33) 0.003*** (3.35) -0.000 (-1.23) 0.007*** (6.45) 0.001 (1.42) 0.000 (0.24) 0.001 (1.13) 0.001*** (3.73) -0.000 (-0.86) 0.001 (0.38) -0.140*** (-7.80) Yes Yes Yes 2,158 0.537 (1) TPAccuracy -0.023 (-0.62) -0.786 (-0.42) -2.911 (-1.03) 0.334*** (5.94) 0.007 (1.10) 0.098** (2.22) 0.067* (1.94) 0.094 (1.09) -0.009 (-0.15) 0.046*** (2.78) -0.001*** (-2.94) 0.174 (1.14) -3.545*** (-4.39) Yes Yes Yes 1,485 0.471 Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared LnNumFirms Exp_General RevGrowth BrokerSize LnAnalyst Constant LnAT MB 62 Table E.12 Test of the Subsample without Blockchain Mentioning This table reports regression results of analysts’ revision within five calendar days following firms’ crypto exposure disclosed in 8-K filings that do not contain the mentioning of blockchain. I estimate the following model: AnalystRevisionijt = α + β1LnKeywordCountjt + Controls + Fixed Effects + εjit. The dependent variable used in column (1) is ChgTP, which is the difference between the current and prior target price forecast issued by the same analyst for the same firm, deflated by stock price two trading days prior to the current target price forecast announcement date. The dependent variable used in column (2) is RevAF, which is the difference between the current and prior one-year ahead annual earnings forecast issued by the same analyst for the same firm, deflated by stock price two trading days prior to the current earnings forecast announcement date. LnKeywordCount is the natural logarithm of one plus the number of crypto-related keywords in an 8-K filing and its exhibits. All variables are defined in Appendix C. Standard errors are clustered at the analyst level, and t-statistics (two-tailed test) are shown in the parenthesis. ***, **, and * denote significance at 1%, 5%, and 10% levels. Variables LnKeywordCount Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared (1) ChgTP 0.041*** (3.81) -7.034*** (-5.71) -0.273 (-0.30) 0.087*** (5.26) -0.003 (-1.49) -0.031 (-1.23) 0.020 (1.49) 0.004 (0.12) -0.025 (-1.28) -0.034*** (-3.05) 0.000*** (2.74) -0.159*** (-4.40) 1.464*** (2.93) Yes Yes Yes 998 0.311 63 (2) RevAF -0.000 (-0.08) -0.276*** (-5.01) 0.259*** (3.28) 0.001* (1.72) -0.000 (-0.55) 0.003* (1.86) -0.001 (-1.46) -0.000 (-0.36) 0.000 (0.18) 0.000 (0.12) -0.000** (-2.13) 0.007*** (3.72) -0.045 (-1.57) Yes Yes Yes 1,447 0.138 Table E.13 Analyst Revision and Crypto Exposure Using the Extended Sample This table reports regression results of analysts’ revision within five calendar days following firms’ crypto exposure disclosed in 8-K filings issued during 1/1/2014-12/31/2023. I estimate the following model: AnalystRevisionijt = α + β1LnKeywordCountjt + Controls + Fixed Effects + εjit. The dependent variable used in column (1) is ChgTP, which is the difference between the current and prior target price forecast issued by the same analyst for the same firm, deflated by stock price two trading days prior to the current target price forecast announcement date. The dependent variable used in column (2) is RevAF, which is the difference between the current and prior one-year ahead annual earnings forecast issued by the same analyst for the same firm, deflated by stock price two trading days prior to the current earnings forecast announcement date. LnKeywordCount is the natural logarithm of one plus the number of crypto-related keywords in an 8-K filing and its exhibits. All variables are defined in Appendix C. Standard errors are clustered at the analyst level, and t-statistics (two-tailed test) are shown in the parenthesis. ***, **, and * denote significance at 1%, 5%, and 10% levels. Variables LnKeywordCount Turnover RetSD PastRet Exp_Firm Exp_General LnAnalyst LnNumFirms BrokerSize LnAT MB RevGrowth Constant Year Fixed Effects Industry Fixed Effects Analyst Fixed Effects Observations Adjusted R-squared (1) ChgTP 0.0226*** (2.82) -1.1437 (-1.17) -0.8104 (-0.81) 0.1380*** (8.89) 0.0049** (2.09) 0.0395** (2.09) 0.0092 (1.02) -0.0024 (-0.08) -0.0170 (-0.72) 0.0080 (1.37) 0.0002 (1.04) -0.0819*** (-6.93) -0.8468** (-2.39) Yes Yes Yes 3,646 0.372 64 (2) RevAF -0.0001 (-0.17) 0.1099 (1.28) -0.1066 (-1.43) 0.0029*** (4.08) -0.0002* (-1.85) -0.0041 (-1.49) 0.0024*** (3.84) 0.0003 (0.17) 0.0008 (0.57) -0.0000 (-0.04) 0.0000* (1.75) -0.0107*** (-6.25) 0.0614 (1.27) Yes Yes Yes 4,932 0.139