Adverse Selection in Cryptocurrency Markets

1 Introduction
The notion of fundamental value for cryptocurrencies can be considered to be very different from
other financial assets such as stocks or fiat currencies. Cryptocurrencies have been found to provide
transactional benefits that money issued by central banks does not (Biais et al., 2020). For example,
cryptocurrencies may provide a platform to allow for the completion of transactions even when
some centralised currencies or banking systems face significant problems, while cryptocurrencies
can also provide flexibility to investors when transferring funds from one country to another, even
considering the presence of closed economies. Cryptocurrencies also lead to transaction costs in
the form of limited convertibility, a lower rate of acceptance, and fees embedded within the mining
process (Biais et al., 2020)
. The theoretical framework presented in Biais et al. (2020) indicates
the existence of a feedback loop from future prices to transactional benefits and current prices;
where investors who expect future prices to be high, anticipate elevated transactional benefits,
which results in high prices today. Biais et al. (2020) underline that the fundamental value of a
cryptocurrency is, therefore, a function of net transactional benefits that depend on its future price.
In practice, those with internal knowledge with regards to the technological processes surrounding
both cryptocurrency creation and the mining process may better assess net transactional benefits
and, therefore, could possess an informational advantage when attempting to understand the

fundamental value of the associated product. Therefore, investors who lack sufficient knowledge about the
underlying cryptocurrency product might face significant adverse selection problems. Such issues
can also arise for both retail and institutional investors of cryptocurrency. Institutional investors,
who are traditionally considered to be informed because of their ability to monitor market conditions

and process information at lower costs (Seppi, 1992; Hessel and Norman, 1992; Lang and
McNichols, 1997), have become increasingly prevalent in cryptocurrency markets in recent years
(Borri, 2019)
. While earlier attempts by Momtaz (2020) and Chod and Lyandres (2020) focus
on information asymmetry at the initial coin offering stage, the role of information asymmetry in
secondary cryptocurrency markets remains unexplored.
This research aims to complement the existing debate by examining the impact of information

asymmetry at the exchange level when considering the volatility, liquidity, market toxicity, and
returns of cryptocurrency. We proxy information asymmetry with the adverse selection component
of the effective spread. Specifically, we make use of the seminal models of Glosten and Harris (1988),
Huang and Stoll (1997), and Madhavan et al. (1997) to decompose the spread into two parts: an
adverse selection component related to permanent price change due to informed trading, and a
transitory component related to the temporary price change and order-processing costs. We also
apply the three-way decomposition of Huang and Stoll (1997) to account for the changes in the
transaction costs due to inventory risk. In a panel regression framework, we attempt to specifically
investigate the following questions: (i) does an increase in the adverse selection component of the
spread significantly impact the future volatility of cryptocurrency returns? (ii) does the adverse
selection component of the spread influence cryptocurrency liquidity? (iii) is the adverse selection
component related to a reduction in market toxicity? And finally, (iv) can the adverse selection
component act as a significant predictor of short-term cryptocurrency returns?
Accordingly, using the limit order book and trade data sourced from the Bitfinex Exchange, we

document that the adverse selection component of the effective spread is significant for the twelve most
frequently traded cryptocurrencies between August 2017 to June 2018. Adverse selection costs, on
average, correspond to ten percent of the estimated spread, suggesting that the proportion of the
spread attributable to adverse selection risk in cryptocurrency markets is also economically significant.

Moreover, we show that the adverse selection component of the spread is a significant predictor
of future volatility proxied by the standard deviation of cryptocurrency returns. Our results suggest

that informed trading activity increases the future levels of return volatility in cryptocurrency
markets. Furthermore, we present evidence that indicates the existence of an inverse relationship
between adverse selection costs and cryptocurrency liquidity. Through intraday panel regressions,
we initially show that informed trading activity increases the realized spread
for cryptocurrency
trading. We then document that an increase in adverse selection costs results in higher levels of
order-book slope Amaya et al. (2018) and Amihud (2002) illiquidity levels. However, our findings
indicate that in cryptocurrency markets, the adverse selection component of the spread, does not
have a statistically significant impact on future levels of order-book depth. The impact of adverse

Adverse Selection in Cryptocurrency Markets