Summary

• Prominent macroeconomists Kenneth Rogoff and Carmen Reinhard co-author a new NBER working paper that is the first to quantify Bitcoin’s use as a vehicle for transactions.
• They believe 7.4% of all Bitcoin trades are used for (cross-border) wealth transfers and payments, although the 2019 power outage in Venezuela suggests the share could be far larger.
• Domestic Bitcoin payments are most popular in Russia and China, while countries with strict capital controls, such as Nigeria and Argentina, use Bitcoin more for international transfers.

Introduction

In a new NBER working paper, Harvard University professors Kenneth Rogoff and Carmen Reinhart provide perhaps the strongest quantitative evidence to date of demand for Bitcoin as a vehicle for transactions. As previously discussed, understanding Bitcoin’s use, and the ratio of users to speculators, is crucial for its long-term valuation.

To quantify transaction (rather than investment) demand, the authors analyse high-frequency, ‘off-chain’ data to distinguish between investment purchases and trades used as a vehicle for wealth transfers and payments. This contrasts past research that analyses ‘on-chain’ data, representing as little as a tenth of all Bitcoin trades, and that has focused primarily on the users, rather than the uses, of crypto.

By building an algorithm, the professors can tag four years of trades on the world’s largest centralised peer-to-peer exchange markets as either ‘crypto vehicle transactions’ or ‘investment vehicle transactions’. But because the data includes the currency used in transactions, they can also distinguish between cross-border capital flows and domestic payments. Overall, they find:

• At a conservative estimate, 7.4% of all Bitcoin trades, equivalent to 3.8% of trade volumes, are crypto vehicle transactions. Of those, 20% represent international currency movements.
• The average domestic transfer using Bitcoin takes 44 minutes and has a value of $115 per transaction, while cross-border movements take around 83 minutes with an average value of $139.
• A key determinant of Bitcoin activity is capital control evasion and hyperinflation in the fiat currency, making South America a hive of Bitcoin transactions.

Data

The core dataset runs from 2017 to 2021 and uses data from LocalBitcoins.com. This is the world’s largest peer-to-peer (P2P) Bitcoin exchange. Such an exchange facilitates trades of Bitcoin between two pseudonymous private wallets, typically in direct exchange for goods or fiat currency. On LocalBitcoins, Bitcoin can be traded in 135 different fiat currencies.

The model works, and has been copied by other P2P exchanges, because it only offers deposit services for members’ Bitcoins. It does not involve national fiat currencies, thereby sidestepping national banking regulations. Trades that occur within the exchange are off-chain in that individuals buy and sell only their claims on Bitcoins that the intermediary houses.

Crucially, it is possible to buy Bitcoin from an account holder in country A using country A’s currency and sell it to a third account holder in country B in exchange for country B’s currency – all within LocalBitcoins and without ever registering on the blockchain. This is what the authors aim to quantify by using an algorithm and exploiting information on the time stamp, trade size, fiat currency used, and the price paid in fiat currency.

Methodology

One of Bitcoin’s functionalities is as a vehicle for transactions. To explain it intuitively, the paper uses an example (Chart 1).

Imagine an Argentine citizen wants to convert pesos in her Buenos Aires account to dollars in her Miami account but evade Argentine capital controls. Instead of using traditional interbank markets, she can simply trade pesos for Bitcoin through an exchange or P2P platform (presumably from an Argentine resident), then sell the same amount in exchange for dollars (presumably from an American resident). She can then use the dollars to make a payment or deposit them in an American bank.

The process has one key fundamental characteristic – converting one currency to Bitcoin, then Bitcoin to another currency in a very short period. By matching two identically sized trades in and out of Bitcoin in a brief window, the authors can estimate trades used as a vehicle for transactions (called a ‘vehicle trade’).

For the algo, the authors choose a five-hour window in which to match transactions. This balances two considerations: a longer window increases the probability that two unrelated transactions of identical size match, and a smaller window reduces the chance of a true match. For this method to work, the authors must assume the matching trades are sufficiently unique, which would not be so if most trades were of similar values. Reassuringly, most are different sizes.

The authors must also assume people who want to transfer money or make payments via Bitcoin want to minimise their holding times. Bitcoin’s price volatility is probably a good incentive for both buyers and sellers to do so. Indeed, for many trades, the authors find the time between the trade legs of a vehicle trade is typically only a few hours or less. It is longer when it involves fiat currencies in banking systems with no fintech alternatives to the interbank market to make domestic transactions.

Results

Of the 45 million trades that occurred on the exchange, 10.5% had an exact match in terms of size, but the algorithm could not guarantee a high enough probability (95%) of uniqueness for all of them. The authors conclude 7.4% of all trades represent crypto vehicle transactions in which Bitcoin is used to make payment in fiat currency. These trades represent 3.8% of total Bitcoin volumes.

Furthermore, under a fifth of all trades are cross-currency/cross-border. That is, where two different fiat currencies are used in the two transactions of exactly matching size. The rest are domestic transfers or payments. The average time to transfer money domestically is 44 minutes and 83 minutes to make international transactions, with the average trade size of $115 and $139, respectively.

By origin, Russia and China have an exceptionally high share of same-currency trades. That is, their respective currencies are both the origin and destination currency in 89.8% and 84% of all trades. This may be due to US financial sanctions in the case of Russia and to avoid state surveillance in China (and Russia). In the US, EU and UK, domestic payments represent 60%, 76.5% and 76.9% of all trades, respectively.

The largest international flows are US dollars to Nigerian naira and Colombian pesos to Venezuelan bolivars, both likely capturing remittance flows. In 2020, the most active markets, measured by annual transactions per capita, were Venezuela, Russia, Belarus, Colombia and Chile. According to the authors, this suggests a large determinant of Bitcoin’s popularity and use is linked to capital control evasion. One such example is Argentina where, after the government imposed capital controls in 2019, the Bitcoin market expanded rapidly.

Finally, the authors break down the share of transactions used as a vehicle trade by cross-border and domestic transfers across countries (Chart 3). Again, countries with significant capital controls throughout the period, like Argentina and Nigeria, have the highest share of cross-country transactions. With the unfolding hyperinflation and strict bank withdrawal limits in Venezuela, Bitcoin’s attractiveness as an alternative to the local currency means domestic transactions are as significant as international ones.

Why Is 7.4% a Conservative Estimate?

On 7 March 2019, Venezuela experienced an unexpected three-day power outage, impairing cross-border transfers using Bitcoin. This had significant spillover effects on Bitcoin trading volumes in Venezuela, but also on the number of trades effectuated in the Mexican peso, the Peruvian sol, the Chilean peso and the Colombian peso.

To illustrate why the 7.4% is a conservative estimate of vehicle trades, the authors use the following example. For Peru, the algorithm showed 7.5% of all Bitcoin trades were used for wealth transfer and payments domestically or abroad. Of those, 38% involved the Venezuelan bolivar, which likely captures cross-border transactions to Venezuela. And so the power outage should have led to around a 3% reduction in total Bitcoin trades that happened in Peru around those days. Instead, the actual drop in trade volume was 65%.

In essence, this says cross-border transactions using Bitcoin represent a far larger share of all Bitcoin trades than the algorithm suggests. While the power outage in Venezuela may have disincentivised Peruvian investors from trading Bitcoin in the immediate aftermath, an inability to transfer money across the border more likely explains the drop in trade volumes.

Bottom Line

If the event in Venezuela can be applied globally, far more than 7.4% of all trades in Bitcoin could be transacted with the purpose of transferring money. However, even if it is not, 7.4% as a conservative estimate is still a sizeable amount when you factor in that Bitcoin also has other uses, like as a store of value. And so, we are starting to get quantifiable and robust evidence from very reputable researchers that demand for Bitcoin extends beyond the allure of making money.

Citation

Luckner, Reinhart & Rogoff (2021) Decrypting New Age International Capital Flows, NBER, https://www.nber.org/papers/w29337