Source-linked AI summary
Do the rich get richer? An empirical analysis of the BitCoin transaction network
Dániel Kondor, Márton Pósfai, István Csabai, Gábor Vattay
TL;DR
Because individual monetary transactions are rarely available for study, the paper analyzes Bitcoin’s public transaction records. It reconstructs the transaction network and examines both its evolving structure and money flows, finding distinct attachment patterns for network growth and wealth distribution, plus a degree–wealth scaling relation.
Problem
Everyday monetary transactions are difficult to analyze because such data are scarce and sensitive, while econophysics models often compare only macroscopic outcomes with real-world data.
Method
The authors compile Bitcoin’s complete public transaction history, reconstruct the transaction network, and analyze network characteristics, payment timing, amounts, wealth, and microscopic growth statistics.
Results
Linear preferential attachment drives network growth, while sublinear preferential attachment governs wealth-distribution evolution; node degree and wealth follow a scaling relation.
Takeaways & Limitations
Bitcoin provides unusually detailed empirical access to monetary transactions, linking transaction-network structure with money flow and wealth accumulation.
Takeaways & Limitations
The analysis observes Bitcoin addresses rather than users, and the user population may not represent society because some illegal transactions may be overrepresented.
Abstract
from arXiv · showhide
The possibility to analyze everyday monetary transactions is limited by the scarcity of available data, as this kind of information is usually considered highly sensitive. Present econophysics models are usually employed on presumed random networks of interacting agents, and only macroscopic properties (e.g. the resulting wealth distribution) are compared to real-world data. In this paper, we analyze BitCoin, which is a novel digital currency system, where the complete list of transactions is publicly available. Using this dataset, we reconstruct the network of transactions, and extract the time and amount of each payment. We analyze the structure of the transaction network by measuring network characteristics over time, such as the degree distribution, degree correlations and clustering. We find that linear preferential attachment drives the growth of the network. We also study the dynamics taking place on the transaction network, i.e. the flow of money. We measure temporal patterns and the wealth accumulation. Investigating the microscopic statistics of money movement, we find that sublinear preferential attachment governs the evolution of the wealth distribution. We report a scaling relation between the degree and wealth associated to individual nodes.
I. INTRODUCTION
The paper uses Bitcoin’s publicly accessible transaction records to study monetary exchange at unprecedented detail. It reconstructs the transaction network and combines structural network analysis with payment timing and amount data.
- Everyday monetary transactions are usually private, limiting empirical analysis of how money moves between individuals.
- Bitcoin is a decentralized peer-to-peer digital cash system whose transactions are publicly accessible.Users transact through Bitcoin addresses rather than bank accounts maintained by a central authority.
- The system expanded from a small enthusiast experiment into a more widely used currency after MtGox and mainstream media attention.MtGox began in 2010, while wider public attention from 2011 coincided with rapidly rising and fluctuating prices.
- The authors reconstruct a directed network in which nodes are Bitcoin addresses and links represent transactions between addresses.They also extract the time and amount of every payment.
- The study examines network structure, transaction dynamics, and microscopic growth statistics to investigate preferential attachment.Structural measures include degree distributions, degree correlations, and clustering; dynamic measures include wealth statistics and transaction timing.
A. Evolution of the transaction network
The Bitcoin transaction network grows through distinct early and trading phases, with heterogeneous degree distributions and persistent higher-order structure. Linear preferential attachment fits link creation, while the network also exhibits disassortative correlations and elevated clustering.
- Network growth: After five years, the network reaches N = 13, 086, 528 nodes and L = 44, 032, 115 links, following distinct initial and trading phases.The initial phase had low activity and large fluctuations; network measures converged by mid-2011 and changed little afterward.
- Degree distributions: The indegree and outdegree distributions are heterogeneous and approximately power-law, with trading-phase exponents 2.18 and 2.06, respectively.The exponents do not change significantly during the trading phase.
- Degree distributions: The trading phase stabilizes degree inequality at Gin ≈0.629 and Gout ≈0.521, while both distributions remain highly heterogeneous.In the initial phase, indegree inequality is close to 1 and outdegree inequality is much lower.
- Network correlations: The degree correlation coefficient remains negative after mid-2010, reaching r ≈−0.014 by 2013, and average-neighbor analysis confirms disassortative behavior.Nodes with high outdegree tend to connect to nodes with low indegree.
- Higher-order structure: Trading-phase clustering settles around C ≈0.05, exceeding the random-network benchmark Crand ≈0.0037(9).The initial phase fluctuates around C ≈0.15 before the clustering coefficient reaches its trading-phase stationary value.
- Network growth: The best-fit attachment exponent is α ≈1, indicating that preferential attachment agrees with the observed overall network growth.The authors caution that preferential attachment alone does not explain the network’s disassortative correlations and high clustering.
B. Dynamics of transactions
Bitcoin transactions show heterogeneous temporal activity and wealth, with rich addresses typically gaining more and wealth linked to transaction-partner count. The wealth dynamics are better described by sublinear preferential attachment and a stretched-exponential distribution than by a simple power law.
- Temporal patterns: The inactivity-time distribution is broad and approximated by P(T) ∼1/T.Inactivity time is the elapsed time between consecutive outgoing transactions from a node.
- Wealth distribution: 6.28% of addresses possess 93.72% of total Bitcoin wealth, while the trading-phase Gini coefficient stabilizes near G ≈0.985.The balance distribution is stable over time and highly heterogeneous.
- Wealth distribution: The balance distribution has a power-law tail with exponent −1.984 above approximately 50 BTC, but the overall distribution is better approximated by a stretched exponential.The stretched-exponential form is P(b) ∼b^−γe^−(ab)^(1−γ), with γ = 0.873 and a = 8014 BTC^−1.
- Preferential attachment: Average monthly balance growth rises with starting balance as ∼b^0.857, indicating a “rich get richer” pattern.For decreasing balances, many addresses lose all wealth within one month, partly because users move unspent bitcoins to new addresses for privacy.
- Preferential attachment: Balance increments are best approximated by sublinear preferential attachment with exponent α ≈0.8, rather than a single simple power law.The analysis treats each unit of balance increase as a separate simultaneous event when an address receives payment.
- Degree–wealth relation: The average balance scales with indegree as b ∼k^0.617 for degrees from 1 to 3000, covering over 99.99% of nodes with nonzero balance.This links accumulated wealth to the number of distinct transaction partners.
A. The Bitcoin network
Bitcoin is a decentralized peer-to-peer payment system whose public transaction records enable network analysis, but addresses do not directly identify users. The system’s user population and transaction mix also limit how broadly its observed patterns can be generalized.
- System structure: Bitcoin uses a peer-to-peer network, proof-of-work validation, and a blockchain composed of blocks referencing prior blocks.Blocks are formed at approximately one-block-per-10-minutes, and altering past transactions becomes practically infeasible after several blocks.
- Public transaction records: All historical transactions are publicly available, revealing sending and receiving addresses, transferred sums, and approximate transaction times.This makes unusually detailed transaction-level financial data available for analysis.
- Address–user distinction: Each user can create unlimited Bitcoin addresses, which appear as separate nodes, making large-scale identification of user-level entities difficult.Constructing a network of users would require joining addresses belonging to the same individual.
- Scope boundary: The representativeness of Bitcoin users is uncertain, and illegal commerce may be overrepresented among Bitcoin transactions.Comparisons with data from other sources are needed to assess which observed phenomena generalize beyond Bitcoin.
B. Data
The study used a blockchain snapshot and supplementary timestamp archive to reconstruct Bitcoin transaction data. The resulting dataset contains millions of blocks, transactions, and addresses, with balances calculated from transaction flows.
- Data acquisition: The researchers downloaded the blockchain on May 7, 2013 and extracted all transactions using a modified open-source bitcoind client.More precise transaction timestamps came from the blockchain.info archive, and the data and source code were made available.
- Dataset scale: The dataset contains 235,000 blocks, 17,354,797 transactions, and 13,086,528 addresses.Of these addresses, 1,616,317 were active during the last month.
- Balance reconstruction: Address balances were calculated from the total bitcoins received and sent because the Bitcoin network does not store balances directly.The reconstructed balances support analysis of wealth and money accumulation.
IV. DISCUSSION
The paper uses Bitcoin’s publicly available transaction data to study both the structure and evolution of its transaction network and the flow of bitcoins. It identifies distinct phases in the system’s development and characterizes their differing network properties.
- IV. DISCUSSION: Bitcoin’s open transaction records enabled analysis of monetary transactions that are usually kept private.The authors compiled the complete transaction list and extracted payment times and amounts.
- IV. DISCUSSION: A schematic transaction links each input address to each output address in the reconstructed network.The example contains four input addresses and three output addresses.
- IV. DISCUSSION: The study examined transaction-network structure, evolution, and bitcoin flow using payment times and amounts.The analyses included network characteristics and the dynamics of money movement.
- IV. DISCUSSION: Two phases emerged: an early experimental period with fluctuating characteristics, heterogeneous indegree, and homogeneous outdegree distributions, followed by wider public attention.The supplied passage identifies the first phase and begins describing the later phase, but does not provide its full characterization.