Fubon Center for Technology, Business and Innovation | Franz Hinzen

Why You Don’t Buy Coffee with Bitcoin: The Micropayments Myth
Sean Foley, Franz Hinzen, Kose John and Fahad Saleh

The first blockchain, Bitcoin, was conceptualized as an alternative system for electronic payments that would not require financial institutions serving as trusted intermediaries. The Bitcoin white paper argues that by removing financial institutions this payment system would decrease transaction costs and thereby allow for small value transactions. In our work, we evaluate this argument and ask how the design of Bitcoin shapes the transaction size distribution observed on this blockchain.

When studying the transaction size distribution over time, we find that Bitcoin has experienced a remarkable shift. In its early days, most transactions were small in value with the median transaction size being less than $2. However, by 2014 this number had increased more than eight-fold to approximately $16, comparable to transactions via cash or Visa, and increased again more than four-fold to $73 by 2018. This shift in the transaction distribution has been particularly pronounced for very large transactions. While very few transactions exceeded $1 million in size in Bitcoin’s early days, those type of transactions are now a common occurrence. In contrast, small value transactions represent a diminishing fraction of payments on Bitcoin today casting doubt on Bitcoin’s ability to better support small value payments.

Subsequently, we investigate the transaction costs experienced by Bitcoin users. During Bitcoin’s first years, transaction costs were indeed low. Fees rarely exceeded a few cents – many transactions paid fees less than one cent or even no fee at all – and most transactions were processed on the next block. In contrast, the median transaction in 2018 included a fee of $2.37 and experienced a significant increase in its transaction delay. During some recent periods, transactions even carried $100s in fees and wait times could be several hours sometimes taking more than one day.
To understand these patterns, we collect historic data on transactions prior to them being recorded on Bitcoin. In contrast to the information publicly recorded on the blockchain, these data allow us to measure the degree to which Bitcoin is congested due to the number of transactions awaiting confirmation being large relative to the rate at which transactions can be processed. Using transaction-level data, we provide evidence that increases in congestion lead to increases in transaction costs both in terms of fees and wait times. Lastly, we show that congestion ultimately leads to the transaction size distribution becoming tilted towards larger value transactions.

This result arises due to Bitcoin’s design: Bitcoin faces a capacity constraint at approximately 2500 transactions every 10 minutes. When submitting their transactions, users choose their fees. Transactions are processed in descending order of those fees which determines wait times. When transaction demand rises and the blockchain is more congested, transaction costs increase since Bitcoin’s capacity limit imposes a supply constraint. Importantly, the increase in transaction costs is independent of transaction sizes because the order in which transactions are processed is not a function of transaction sizes. This is different from fees in many traditional payment systems in which a large part of transaction costs is set relative to transaction values. Facing the same increase in fees, users with small size transactions are the first to transact via an alternative traditional payment platform or abandon their transaction altogether. In sum, Bitcoin becomes less attractive for small value transactions as transaction demand increases. In our work, we formalize this intuition in an economic model of a Proof-of-Work blockchain.

Our findings highlight that, in contrast to its initial vision, Bitcoin’s design implies a trade-off between the ability to serve small value transactions and being widely adopted as digital currency. Our results have implications for other blockchains, such as Ethereum, which have innovated on top of Bitcoin’s design to allow for a broad set of financial applications. Understanding the limitations and economic consequences of Bitcoin’s benchmark technology provides insights for future private and public blockchain ventures.

WEBSITE: https://franzhinzen.com/