Insights from bitcoin trading

• Author: Thomas H. McInish,Jonathan L. Miller,Pankaj K. Jain
• Date: 01 December 2019
• Published date: 01 December 2019
• DOI: http://doi.org/10.1111/fima.12299

DOI: 10.1111/fima.12299

ORIGINAL ARTICLE

Insights from bitcoin trading

Pankaj K . Jain Thomas H. McInish Jonathan L. Miller

Department of Finance, Insurance,and Real

Estate, University of Memphis, Memphis,

Tennessee

Correspondence

ThomasH. McInish, Department of Finance,

Insurance,and Real Estate, University of Mem-

phis,108 Fogelman Executive Center,Memphis,

TN38152-3120.

Email:tmcinish@memphis.edu

Abstract

Weexamine commonality in returns and volume for Bitcoin–fiat cur-

rency pairs, each trading in a country with a single time zone. Bitcoin

has substantial volume and obeys the theory related to commonal-

ity,liquidity, and price discovery. Wefind evidence that one common

factor explains68% of the variance in hourly volume. Though trading

is higher on weekdays, there is substantial weekendtrading, reflect-

ing high retail participation. Volume is higher on exchanges during

local working hours, as seen in forex markets, supporting the view

that trading patterns depend on the location of traderather than the

location of the asset traded.

KEYWORDS

cryptocurrency, liquidity, microstructure

1INTRODUCTION

Satoshi Nakamoto created Bitcoin, the first and the most widely used cryptocurrency, in 2008 to facilitate and

reduce the cost of transferring funds.1In 10 years, the market capitalization of Bitcoin has grown spectacularly from

zero to more than $100 billion USD as of October 2018 and a daily 24-hr volume of over $3 billion USD (https://

coinmarketcap.com/currencies/bitcoin/[ason October 24, 2018]). Business and government leaders, such as Bill Gates

and Janet Yellen, have highlighted the importance of Bitcoin and blockchain technology for the future of the U.S.

economy (https://www.cnbc.com/2017/12/19/bill-gates-in-2014-bitcoin-is-better-than-currency.html; https://www.

coindesk.com/us-fed-yellen-blockchain-impact/).Bitcoin markets offer a unique opportunity to test several constructs

in microstructure theory related to commonality,liquidity, and price discovery.

Bitcoin merits study as it has severalinteresting features in terms of market participants, trading platforms, trading

hours, global fungibility,and the availability of international trading data that allows us to shed new light on microstruc-

ture noise, marketparticipant preferences, and biases that affect price discovery. First, unlike most financial assets that

are dominated by institutions, Bitcoin features high levels of retail participation from individuals who are major play-

ers in Bitcoin trading. As of October 2018, 48.4% of those engaged with Bitcoin are between the ages of 25 and 34

and about 25% are between the ages of 35 and 44.2The median transaction value on February 10, 2018 was $543

c

2019 Financial Management Association International

1SatoshiNakamoto, Bitcoin: A peer-to-peer electronic cash system (https://bitcoin.org/bitcoin.pdf).

2Consideringthose 18+only (https://coin.dance/stats/age).The opinions expressed are those of the authors and do not necessarily reflect those of the Office

ofFinancial Research in the US Department of the Treasury, where Jain is a fellow.

Financial Management. 2019;48:1031–1048. wileyonlinelibrary.com/journal/fima 1031

1032 JAIN ETAL.

USD (https://bitinfocharts.com/bitcoin/). Many individuals trade Bitcoins on peer-to-peer networks that call them-

selves exchanges. One exchange that trades Bitcoin, Coinbase/GDAX, has more than 500 employees, over 13 mil-

lion accounts, more than the brokerage firm, Charles Schwab (Cheng, 2017; https://en.wikipedia.org/wiki/Coinbase).

In addition, Bitcoin’sunique trading and settlement procedures affect its price discovery. Many of the exchanges oper-

ate lit limit orders books (LOBs) that enforce price and time priority.Some exchanges act as dealers so that customers

can use market orders without interactingdirectly with the LOB.

Trades on these exchanges settle instantaneously because the exchange holds both Bitcoins and cash. Recipients

of Bitcoin on the blockchain often require that a set number of blocks, such as six, be posted subsequently before

the transaction is considered settled. Like foreign exchange, but unlike most financial assets, Bitcoin trades24 hr a

day.However, Bitcoin also trades 7 days a week, but there is very little weekend tradingin the forex market (Huang &

Masulis, 1999). The combination of 24/7 trading and young, individual traders, combined with the availability of LOB

data typically seen in equity markets,makes Bitcoin an attractive vehicle to revisit several issues previously considered

in the finance literature.

Moreover,several unique features of Bitcoin affect its commonality of price, returns, and volume across many fiat

currencies. Unlike equities, debt instruments, and forex, Bitcoin has no home market and is not issued by a specific

sovereign authority. Storagecosts are low and the cost of sending Bitcoins from one location to another around the

world is also low (Bitwise Asset Management, 2019).

Initially, we focus on the behavior of Bitcoin prices and returns. We find strong evidence for the propensity of

the law on one price to hold; Although Corwin-Schultz (2012) Bitcoin spreads are higher than for equities Amihud’s

(2002) illiquidity measure is significantly lower providing some indication that the Bitcoin market is tight as discussed

by Foucault, Kadan, and Kandel (2005). Forreasons we elaborate on below, all of the results are based on trading for

four currency pairs: GBP/BTC,JPY/BTC, KRW/BTC, and ZAR/BTC. We find no evidencethat Bitcoin prices depreciate

during local trading hours and appreciate during nontrading hours, as reported for foreign exchanges by Breedon

and Ranaldo (2013). We find strong support for the proposition of Goettler, Parlour, and Rajan (2005, 2009) that

microstructure noise acts as a volatility multiplier.

Wealso test severaladditional hypotheses related to the number of Bitcoins traded (volume),which plays an impor-

tant role in commonality, liquidity,and price discovery. Hasbrouck and Seppi (2001) and Korajczyk and Sadka (2008),

among others, have shown commonality in the order flows for equities. We find that the correlation of synchronous

trading volumebetween each of our Bitcoin–fiat currency pairs ranges from 46% to 72%. Using factor analysis, we find

that our four trading series load on one factor, providing further evidenceof commonality for the contemporaneous

Bitcoin volume. Nevertheless,while there is substantial commonality in synchronous trading volume across exchanges

around the world, there is also a substantial local element in the trading volume.

We test whether Bitcoin markets exhibit the properties of LOBdynamics as modeled in the price discovery litera-

ture for other markets. Parlour (1998) and Goettler et al. (2005, 2009) provide theoretical predictions regardingthe

dynamics of the LOB in equity markets.Microstructure noise is the deviation of transaction prices from the estimated

fundamental value. The need for models that address the unique feature of cryptocurrency markets is illustrated by

Goettler et al. (2009), who define fundamental value as the expectation of the present value of future cash flows. Bit-

coin does not have expectedfuture cash flows. Nevertheless, we expect many ideas developed for traditional markets

to apply to cryptocurrencies.

Next, we focus on the relation between volume and volatility.Specifically, we test and confirm that there is higher

trading volumewhen asset prices are more volatile, which, according to Goettler et al. (2009), could reflect that specu-

lators supply less liquidity and instead demand greater liquidity under these market conditions.

Microstructureliterature has also highlighted day-of-the-week and time-of-the-day effects in addition to the impor-

tance of location of trade for price discovery and commonality.In contrast to the findings of Bollerslev and Domowitz

(1993) for forex markets, we reject the hypothesis of equality of mean hourly volume across weekdays.We find sig-

nificantly more trading on weekdays than on weekends, which is the same pattern found in the forex markets.But,