Where do we stand in cryptocurrencies economic research? A survey based on hybrid analysis

1 Introduction

• The bibliometric stage analyzes titles, abstracts, keywords, authors, and journal titles; close reading then examines 98 papers from journals identified in that stage.The close reading focuses on methodologies, data sources, and results.Source: §1 · pp. 1–2
• The survey combines bibliometric analysis with close reading to build a comprehensive landscape of cryptocurrency research exclusively within economics.The authors describe this as the first paper to combine the two approaches in one paper, to the best of their knowledge.Source: §1 · p. 2
• The survey includes articles in press alongside papers indexed in Scopus or Web of Science, and aims to identify emerging research lines.The authors present the hybrid method as useful when a large paper count makes detailed reading of every paper unrealistic.Source: §1 · pp. 1–2

2.1 First step: distant reading by means of bibliometric analysis

• The bibliometric stage retrieves 626 Web of Science records, narrows them to 438 articles, and uses the bibliometrix R package for analysis.The authors exclude conference proceedings, book chapters, and Forbes articles.Source: §2.1 · p. 2
• The bibliometric results include 38 Highly Cited Papers, with 15 in Economics Letters and 12 in Finance Research Letters.The paper also reports that, among the 20 most-cited papers, four appeared in Finance Research Letters and six in Economics Letters.Source: §2.1 · p. 2, Table 2 · p. 3
• Keyword analysis identifies Bitcoin as the predominant object and clusters covering informational efficiency, safe-haven properties, volatility, hedge properties, and price bubbles.The paper presents the keyword tables as inputs to the groups developed in the following section.Source: §2.1 · pp. 2, 4, Table 3 · p. 4

2.2 Second step: close reading of cryptocurrency literature

• The authors caution that bibliometric analysis complements rather than substitutes for comprehensive review, and that Web of Science indexing can lag online publication.The lag can leave accepted papers absent from the indexed dataset.Source: §2.2 · p. 4
• For close reading, the authors select Economics Letters, Finance Research Letters, and International Review of Financial Analysis, plus two papers from highly ranked journals.They cite the three main journals’ 26% share of cryptocurrency papers and 30 of 38 Highly Cited Papers as reasons for selection.Source: §2.2 · p. 4, Table 4 · p. 5

3 Close reading findings

• The close-reading dataset contains 98 articles selected from 116 published across five journals.Its source distribution is reported in Table 4.Source: §3 · p. 5
• The review records each paper’s cryptocurrencies, data frequency and source, quantitative method, research aim, and main results.The paper directs readers to Table 8 in the Appendix for these details.Source: §3 · p. 5
• Böhme et al. (2015) provide an early overview of Bitcoin and blockchain, including pros and cons, monetary-policy challenges, risks, and regulation needs.The authors characterize it as an introductory paper for beginning study of the field.Source: §3 · p. 5

3.1 Data sources

• 61% of the reviewed papers use data from Coinmarketcap, Coindesk, or Bitcoincharts.The authors suggest API access and broad coverage of trading platforms and cryptocurrencies as reasons for their use.Source: §3.1 · p. 5
• Except for three papers, the reviewed studies rely on only one data source.Source: §3.1 · p. 5
• The three websites generate price indices by averaging across platforms, so data are not homogeneous across papers and comparisons may be weakened.The authors note that different index-weighting approaches can produce different results in stock markets, with a similar situation possible in cryptocurrency markets.Source: §3.1 · p. 5

3.2 Data frequency

• Cryptocurrency markets provide free, real-time information and trade 24/7, allowing researchers to choose different data granularities.Source: §3.2 · p. 5
• 81% of empirical studies use daily data, while intradaily data is used by 14%.Source: §3.2 · p. 6, Table 6 · p. 6
• Daily sampling leaves high-frequency issues unexplored, which could yield new insights and reveal stylized facts at ultra-high frequency.Source: §3.2 · p. 6

3.3 Main research topics

• Informational efficiency (26%), price discovery (16%), and price clustering (3%) are prominent topics, alongside volatility (15%), portfolio formation (11%), and safe-haven properties (7%).The authors assign each paper its main topic even when it covers more than one.Source: §3.3 · p. 6

3.3.1 Monetary economics and overview of bitcoin ecosystem

• In the overview literature, suspicious trading is linked to market vulnerability to manipulation, while Bitcoin demand is attributed to short-run speculation and long-run expectations of future utility as a medium of exchange.Gandal et al. (2018) examine suspicious trading on one platform; de la Horra et al. (2019) model demand.Source: §3.3.1 · pp. 6–7

3.3.2 Informational Efficiency

• Studies test random-walk and independence properties; findings include early Bitcoin inefficiency followed by movement toward greater efficiency, and weak efficiency for some power transformations of returns.Urquhart (2016) reports the movement toward efficiency; the passage also describes tests of serial correlation, independence, and long memory.Source: §3.3.2 · pp. 7–8

3.3.3 Price discovery

• Predictability findings differ by target: prior high realized volatility and volume influence Bitcoin attention, while Google searches predict Bitcoin trading volume but not returns.Urquhart (2018) studies search queries, realized volatility, volume, and returns; Aalborg et al. (2019) report the volume and return findings.Source: §3.3.3 · p. 8

3.3.4 Price volatility

• Volatility studies commonly find clustering; Katsiampa (2017) selects AR-CGARCH, while Ardia et al. (2019) find MSGARCH outperforms single-regime GARCH specifications in Value-at-Risk forecasting.Katsiampa’s selected model incorporates short- and long-run conditional-variance components; Bitcoin and Ether returns are also reported as stationary with volatility clustering.Source: §3.3.4 · pp. 8–9

3.3.5 Assets correlation and portfolio optimization

• Portfolio studies report varied findings: Guesmi et al. (2019) identify diversification and hedging benefits, Liu (2019) finds cross-cryptocurrency diversification improves investment results, and other studies find relative isolation from traditional assets.Dyhrberg (2016a) classifies Bitcoin between gold and the US dollar, whereas Baur et al. (2018a) report distinct return, volatility, and correlation characteristics.Source: §3.3.5 · p. 9

3.3.6 Safe-haven characteristics

• Safe-haven evidence is mixed: some studies report hedging at shorter investment horizons and at the tails, while others find Bitcoin lacks distinctive gold properties or offers protection only against Asia Pacific stocks.The review notes that regulatory uncertainty, cyberattacks, high volatility, and lower liquidity contribute to uncertainty about safe-haven status.Source: §3.3.6 · p. 10

3.3.7 Bubble formation

• Bubble studies report speculative bubbles before 2017, a negative bubble in Bitcoin and Ripple from 2014 with Ripple-to-Bitcoin spillover, and multidirectional co-explosivity across cryptocurrencies.The co-explosivity finding does not imply that interactions run from larger to smaller or younger markets.Source: §3.3.7 · pp. 10–11

• Most cryptocurrency research focuses on financial topics such as informational efficiency, volatility, portfolio optimization, and bubble behavior.Source: §4 · p. 11
• Most studies use daily data, leaving the information value of high-frequency analysis underexplored; such analysis could also offer hints about traditional assets.Source: §4 · p. 11
• Economic studies rarely estimate blockchain technology’s overall environmental impact or assess its positive and negative externalities.Source: §4 · p. 11
• The review finds no paper examining how consensus protocols affect cryptocurrency price formation, returns, or volatility.Source: §4 · p. 11
• The authors identify a lack of theoretical work on national or supranational regulation and of an institutional-economics perspective.Source: §4 · p. 11
• Prior research mostly studies Bitcoin or a handful of leading coins, despite thousands of active cryptocurrencies with differing technologies, liquidity, and age.Source: §4 · p. 11

• The study combines machine-learning bibliometrics and close reading, treating bibliometric analysis as a complement rather than a substitute for specialized researchers’ insight.Source: §5 · pp. 11–12
• The hybrid approach uses bibliometric findings to inform which papers are read closely, supporting an updated review for researchers in the field.Source: §5 · p. 12
• The close review covers 98 papers, classifies research topics, identifies leading papers and journals, and highlights gaps and future research paths.Source: §5 · p. 12

• Bitcoin is described as moving toward market efficiency, while return predictability diminishes as cryptocurrency liquidity increases.Source: §6 · p. 21
• Reported cryptocurrency characteristics include long memory, leverage, stochastic volatility, and heavy-tailedness; some markets also show heterogeneous informational inefficiency.Source: §6 · p. 21
• Studies report a Taylor effect in Bitcoin time series, declining average price delay, and day-of-the-week anomalies in returns and volatility.Source: §6 · p. 21
• Herding evidence varies by setting: one study reports time-varying herding, while another finds herding during down markets.Source: §6 · p. 21