This paper presents a novel methodology—leveraging a combination of AI and machine learning to estimate the geographic distribution of international stablecoin flows, overcoming the “anonymity” of crypto assets. Analyzing 2024 stablecoin transactions totaling $2 trillion, our findings show: (i) stablecoin flows are highest in North America ($633bn) and in Asia and Pacific ($519bn). (ii) Relative to GDP, they are most significant in Latin America and the Caribbean (7.7%), and in Africa and the Middle East (6.7%). (iii) North America exhibits net outflows of stablecoins, with evidence suggesting these flows meet global dollar demand, increasing during periods of dollar appreciation against other currencies. Further, we show that the 2023 banking crisis significantly impeded stablecoin flows originating from North America; and finally, offer a comprehensive comparison of our data to the Chainalysis dataset.
Methodology for estimating the geographic region of self-custodial wallets
We divide the world into five regions: Africa and the Middle East, Asia and the Pacific, Europe, North America, and Latin America and the Caribbean. To estimate the geographic region of self-custodial wallets, our methodology involves obtaining geographic information for a subset of wallets through two distinct approaches. First, we leverage domain names assigned to wallets purchased through systems such as the Ethereum Name System (ENS). We employ a LLM to infer linguistic and cultural markers—such as language, script, or regional references—that suggest a wallet’s likely region. Second, we identify wallets that frequently transact with centralized exchanges targeting specific regional markets, assuming that a wallet predominantly interacting with, for example, a Latin American-focused exchange is likely from that region. These two methods provide an ad hoc regional classification for a subset of wallets, which we then use as labeled training data to train a machine learning model for classification of arbitrary wallets.
The core of our approach lies in leveraging this training data to train a machine learning model to recognize patterns in on-chain activity that are indicative of a wallet’s geographic origin. We construct features capturing wallets’ behavioral and transactional characteristics, including time-of-day activity patterns, adherence to daylight savings time, interactions with certain centralized exchanges, and engagement with popular ERC-20 tokens and smart contracts. By learning region-specific patterns the trained model can estimate the geographic region of any arbitrary self-custodial wallet. The identifying assumption of the methodology is, that conditional on the features we selected to train the model, wallets that are in the training set and those outside the training set exhibit the same patterns. The methodology then enables us to map the geographic distribution of wallets, which we can then leverage to map international stablecoin flows.
We then train a Gradient Boosted Decision Tree model on this dataset, which achieves an overall accuracy of 65% in predicting geographic regions. For context, random guessing with five regions yields 20% accuracy. We then apply the model to predict the region of any arbitrary self-custodial wallet. Using these predictions, we map international stablecoin flows for 2024. Our analysis captures approximately 138 million transactions totaling $2,019 billion, with an average transaction size of $14,630.
Conclusion
Contrary to prevailing misconceptions, we find that measuring international crypto asset flows, while complex, is not impossible. We develop a novel methodology to estimate the geographic allocation of crypto wallets and employ this approach to quantify international stablecoin flows. We determine that stablecoin flows in 2024 total $2 trillion, the majority of which are international. In absolute terms, we observe the highest volumes in the Asia and Pacific region and North America, whereas we find the lowest volumes in Africa and the Middle East, alongside Latin America and the Caribbean. However, relative to GDP, we find the volumes in these regions to be the most substantial. We establish a correlation between net stablecoin inflows into regions and the relative weakness of domestic currencies against the U.S. dollar, either suggesting that stablecoins serve as a mechanism to fulfill global demand for dollar-based assets for people that seek a hedge against currency depreciation, or that stablecoin flows could possibly be sizable enough to drive exchange rate dynamics. Furthermore, we present evidence of the interconnection between stablecoins and the banking system, highlighting disruptions in stablecoin flows precipitated by the banking crisis of March 2023. We believe that our methodology facilitates a wide range of prospective applications for future research, including the derivation of more granular country-level estimates, the assessment of the geographic distribution of the stock of crypto assets in addition to flows, and the examination of the geographic patterns of decentralized finance application usage.
Leave a Reply