In Search of Cryptocurrency Failure

Abstract

The failure probability and economic losses are astonishingly high in the cryptocurrency market. We perform a comparative analysis of a dynamic logit model and machine learning methods for the predictors for cryptocurrency failure and the pricing of crypto failure risk. We document different significant market- and characteristic-based predictors for coin and token failures. Moreover, we document a significantly positive relation between failure risk and returns, which cannot be explained by the common pricing factors and arbitrage costs in the cryptocurrency market. The high failure risk premium suggests that investors require extra returns for bearing high failure risk of crypto assets.