The Numeraire Effect in Initial Coin Offerings

Abstract

There is a high correlation between the returns of cryptocurrencies and those of utility tokens, which are claims to products and services yet to be developed that are issued through ICOs and traded on crypto-exchanges. I demonstrate the presence of a numeraire effect in the pricing of these tokens and present evidence that it is driven by a combination of group thinking and representativeness bias. Investors mistakenly overestimate the probability that a cryptocurrency-denominated token is issued by a blockchain firm, and thus believe the fundamental value of the token is correlated with that of Bitcoin. I show that a 1% increase in the return on Bitcoin during the month before a token first lists on a crypto exchange predicts a 5% higher ICO return for a cryptocurrency-denominated token than for a fiat-currency-denominated token. If a token is denominated in a cryptocurrency on one exchange and its otherwise identical twin is denominated in a fiat currency on another exchange, then a 1% increase in the cryptocurrency return relative to the fiat currency predicts a 60 bp divergence in their prices. Consistent with being driven by a combination of group thinking and representativeness bias, the numeraire effect is more pronounced for tokens with more complex business plans. Experimental evidence corroborates these empirical findings and suggests that the numeraire effect is present in other asset prices as well and can explain home-currency bias. The combination of high volatility and numeraire effects undermines the ability of cryptocurrencies to serve as units of account.

Keywords: ICOs, utility tokens, Bitcoin, cryptocurrency, numeraire, representativeness bias, group thinking, home bias

JEL Classification: E42, G12, G30, G41, L26