This article is part of a series on bitcoin. This series started with:
And will be followed by articles on:
- Bitcoin Series #3 – Security
- Bitcoin Series #4 – The Bitcoin Arms Race
- Bitcoin Series #5 – Altcoins And Forks
- Bitcoin Series #6 – Other Considerations
- Bitcoin Series #7 – The Endgame
The present article will delve into bitcoin usage. This is a very important theme, because:
- Usage will be a measure of bitcoin’s adoption. Bitcoin can only fulfill its investors’ dreams if it becomes widely used and thus becomes a widely accepted “currency”.
- Usage will also determine whether it can be a “store of value”. Bitcoin fans’ fallback position to the possibility that bitcoin won’t be widely accepted, is that it doesn’t matter anyway. That it will become a “store of value” instead.
Furthermore, bitcoin usage data doesn’t come up all that often. This article will both aggregate existing data and try to estimate its actual usage. This will be an extremely long article – I expect other articles in this series to be shorter, so bear with me.
Bitcoin transactions will happen, broadly, for two reasons:
- Speculative trading.
- Actual usage in commerce.
When I talk about “usage” in this article, I mean actual usage – in commerce.
Bitcoin then presents another difficulty. We can separate actual usage in:
- Licit usage in regular commerce. This is ultimately what concerns us, as for bitcoin to establish itself as a currency/means of exchange, it needs wide adoption in licit activities.
- Usage for illicit activities. Bitcoin’s (and other cryptocurrencies’) anonymization and difficulty in tracking who spent what and who got what payment from whom, as well as the ability to easily cross borders, makes it alluring for usage in illicit activities ranging from crime to tax evasion. Were bitcoin to find its usage mostly in illicit activities, and over time it would be regulated and shut down into irrelevance.
On Illicit Usage
It’s very hard to come up with statistics regarding illicit bitcoin usage. However:
- Blockchain Intelligence Group estimated that from more than 50% of usage, illicit usage fell to ~20% during 2016, and is now a lot lower than that.
- Europol conservatively estimated that $1 billion was transacted through AlphaBay since its launch back in 2014, until it was shut down on July 20, 2017. AlphaBay was the “largest criminal marketplace on the Dark Web”.
What we can say here, is that initially bitcoin really found its mojo serving as a currency for illegal activity.
However, over time the usage mix has shifted towards more legitimate commerce. This has happened both because legitimate uses have expanded and because other cryptocurrencies like Monero or Ethereum have been adopted by criminals as “being safer” (for them).
Usage On Regular Commerce
The first thing I should note here is that bitcoin isn’t really accepted by commerce. That is, nearly no actual commercial endeavor actually accepts bitcoin. It just seems that way.
So how can you find so many businesses which accept bitcoin? The answer is “bitcoin payment processors”. These are companies which provide merchants with the ability to accept bitcoin. When a merchant decides to work with one of these companies and provides a bitcoin payment option, what happens is the bitcoin payment processor guarantees the merchant is instantly (when the transaction is validated) paid on regular currency (dollars, euros) into his bank account.
As a result, the merchant neither has to worry with all of bitcoin’s technical arcana, nor with its massive volatility. He doesn’t really accept bitcoin, he just sees “dollars and euros”. For the merchant, accepting bitcoin turns into the same thing as accepting PayPal.
So who are the largest of these payment processors? As of early 2016, and as estimated by Cubit’s blog based on several sources, the main were:
- BitPay, with around 40-49% of the market.
- Coinbase, with around 28-34% of the market.
- And a plethora of others, including Coinify (which has acquired several bitcoin payment processors), representing the rest.
As we can see, BitPay is the largest bitcoin payment processor by far. And we’re pretty lucky, as it’s also a company that often puts out interesting statistics (which we are about to use).
Estimating Usage Through Transactions
If we assume BitPay has roughly kept its market share, then we can take a stab at estimating current usage as a percentage of total bitcoin trading activity (which is entirely dominated by speculative trading).
Looking at transactions committed to the blockchain, bitcoin has been bumping against the theoretical limit of ~288,000 transactions/day.
This limit results from the 1MB block limit which allows ~2,000 transactions per block. So, with 1 block per each ~10 minutes, 144 blocks per day, we get 2,000 x 144 = 288,000 a transactions/day limit. Since there’s some variability in both block size, timing and transactions per block, the limit varies but we can say the system is working at capacity.
Indeed, a measure of the system being at capacity is the mempool size – which represents the size of the waiting transaction queue. Here we can see how the mempool size has behaved in terms of number of transactions waiting for confirmation:
There is something interesting here. While the system is still at capacity, things were much worse back in May-July. Right now there are 25,000-50,000 transactions waiting, whereas during that period there were as many as 175,000. Transaction volume isn’t that different now versus then, so what might have changed?
While it’s easy to establish how many transactions were committed to the bitcoin blockchain, it’s harder to establish exactly how many bitcoin transactions actually existed. The largest sources of such transactions are the bitcoin exchanges.
In terms of transactions, bitcoinity.org puts the trading volume at around 250-400 trades per minute over the most important bitcoin exchanges. This comes to 360,000-576,000 transactions per day.
Of note already, the estimated transaction volume on the bitcoin exchanges exceeds the number of transactions committed to the blockchain. And these transactions represent merely speculative/investment trading volume. How can that be?
The reason is simple. Exchanges internalize a lot of the volume, so that many bitcoin trades are settled internally and never committed to the blockchain. From time to time, exchanges will issue transactions to the blockchain settling net amounts due to wallets not under their control, etc. The internalization practice goes by the name of off-chain transactions, and improves both the speed and cost of those transactions subjected to it. However, as we’ll see in the next chapter dedicated to security, that’s not all off-chain transactions do.
Payment Processor Transactions
As of December 2016, BitPay crossed 200,000 transactions per month:
Source: TheAtlas.com, BitPay
It’s fair to say that it might now be processing as many as 250,000 transactions per month, though the company has reported transactions are slowing down versus payment volume, as more B2B activity (higher dollar amount per transaction) is taking place:
“We’ve seen a smaller than normal increase in the number of transactions from last year. We’ve also seen a large increase in [business-to-business] transactions, which are around $200,000 per transaction.”
These 250,000 transactions/month represent actual commerce use. These come to about 8,333 per day.
Now, BitPay represents only 40-49% of the total market coverage when it comes to bitcoin payment processors. If we assume transactions are proportional to market share, then this would mean the actual usage for all payment processors would be around 17,000-20,800 transactions/day.
Actual Usage Estimate, Based On Transactions
Given the above we can say:
- Based on blockchain-committed transactions, actual commerce usage would represent 5.9%-7.2% of all transactions. Not a shabby number at all, but there will be more on this later.
- Based on exchange transactions, which exceed blockchain-committed transactions because of off-chain transactions, then actual commerce usage would represent 2.9-5.5% of all (exchange+payment processor) transactions.
However, here’s the thing:
- Investment/Speculative trades are intrinsically higher value than commerce transactions.
As a result, we should also look at usage from a (dollar) volume perspective.
Estimating Usage Through Volume
Again, we follow the same process.
Blockchain Estimated Volume
Here things are made easy for us. Blockchain.info estimates the 7-day daily average dollar volume is running ~$1.4 billion/day. This is the same as a ~$42 billion/monthly volume.
Exchange Estimated Volume
As for the exchanges, Coinmarketcap.com estimates 30-day dollar volume at ~$68.2 billion, which comes to roughly $2.27 billion/day.
Payment Processor Volume
Again, BitPay comes to our help. Very recently, BitPay disclosed that its annualized volume is now on pace for $1 billion/year. If we optimistically use a $1.2 billion run rate/year, then we get a payments volume of ~$100 million/month.
Again, applying BitPay’s market share, this translates to $204-$250 million/month volume for all payment processors. It also translates to an yearly volume run rate of $2.45-$3.0 billion for all bitcoin payment processors.
Actual Usage Estimate, Based On Volume
With the numbers above, we can estimate:
- Based on blockchain estimated volume, actual commerce usage would represent 0.5%-0.6% of all bitcoin dollar volume transactions.
- Based on exchange estimated volume, which again exceeds blockchain-committed volume because of off-chain transactions, actual commerce usage would represent 0.3%-0.4% of all exchange volume.
These are much lower figures than those derived from transaction numbers. This, though, is to be expected. The average ticket in commerce is certainly much lower than the average speculative transaction. Therein, as we’ll see, lies another problem.
How Does Bitcoin Compare?
To put things in context, we can compare the above-estimated bitcoin payment processing transactions and dollar volume (all of the bitcoin payment processors put together) with those of another well-known digital payments processor … alone. I’m talking about PayPal (PYPL).
So here’s what we know about PayPal:
- In its latest quarter, it processed 1.9 billion payment transactions. If we just annualize this on a straight line, it would come to ~7.6 billion payment transactions/year (this likely greatly under-represents PayPal volume, since we’re not including Q4 seasonality).
- Also in its latest quarter, it processed $114 billion in payments volume. Again annualizing this on a straight line would come to $456 billion (this would again under-represent actual PayPal volume due to Q4 seasonality).
As a result of these numbers we can say all bitcoin payment processing represents:
- Less than 0.1% of the payment transactions handled by PayPal.
- 0.5-0.7% of the payment volume handled by PayPal.
Of course, PayPal is just one of many payment processors, though arguably the most representative one of the new digital generation. Bitcoin payment processing pales in comparison.
It gets worse, though:
- PayPal currently has ~$89 billion in market capitalization.
- Bitcoin has ~$125 billion in market capitalization.
But here’s the thing, these two market capitalizations aren’t really comparable. Here’s why:
- When you buy PYPL stock, you’re buying a share of its economic activity. This includes the internal tokens to represent accounts and account balances (which are represented digitally) and are arguably comparable to bitcoin (though they don’t change in price). But it also includes a share on all the economic activity and commissions deriving from that economic activity. Which, after costs, represents profit. PayPal profits amount to ~$2.25 billion per year.
- Indeed, if you were to compare PYPL to the Bitcoin ecosystem, buying PYPL shares would amount to buying an economic interest in the entire Bitcoin ecosystem. That is, it would amount to owning not just bitcoin, but also a share of all bitcoin exchanges, payment processors and miners. And thus, a share on all their profits.
- Now, here’s the problem. When you buy bitcoin, you buy nothing of that sort. You just buy an interest on the bitcoin token. A token which, on the PayPal side, is basically worthless – since all the value is in the economic activity. As a result of this, when you buy bitcoin you are not buying a share of any bitcoin economic activity.
- As a corollary, a bitcoin success would theoretically not need bitcoin itself to be worth tens or hundreds of billions of dollars. Bitcoin could easily fulfill the currency function even with bitcoin just being worth a fraction of that. As we see, PayPal tokens are basically worthless yet the payments scheme works anyway. This is, of course, conceptually. There are structural barriers to this being so because of how bitcoin is structured (and because of who gets paid on the generated economic activity).
Now think about it:
- When you buy bitcoin you don’t buy a share of any economic activity and any resulting profits.
- When you buy bitcoin you’re buying a tiny otherwise-worthless slice of a payments system that’s 200-1000x smaller than PayPal.
- And yet, when you buy bitcoin (for speculative purposes), you’re paying a market capitalization that’s 40% higher than PayPal’s!! And this for a token which gives no right to any underlying economic activity.
From here alone, we already know that bitcoin cannot but be mispriced. However, there’s a lot more to say, both on this article and the following ones.
A Problem – Not Growing Sustainably
When talking about transactions and volume above, I could also add that BitPay is seeing tremendous growth. Its transaction payments volume is up 328% year-on-year.
Had I said as much, and you’d quickly scream “growth”, and excuse any and all possible valuation arguments (up to and including you not sharing on any economic benefits from the possible success of the Bitcoin ecosystem).
However, there’s more to this growth which most people don’t know. We know, from past history, that this growth is tied to the massive increase in the bitcoin price.
How do we know this? Again, because of BitPay. Here are two charts depicting what happened during and after the previous bitcoin bubble (annotations are mine):
Source: TheAtlas.com, BitPay
Source: TheAtlas.com, BitPay
As you can see, the tremendous increase in usage was a function of the bitcoin bubble itself. Then as bitcoin prices crashed and stagnated, usage also decreased. This is unlike a truly sustainable payments system which, when presenting obvious advantages to its users, sees consistent increases in adoption (at least barring a large recession).
Moreover, the declines above understate the actual organic declines in usage. This is so because BitPay spent the period expanding its reach.
As a result of the above, we can expect bitcoin usage to again drop significantly just as soon as the bitcoin price crashes. This will make it rather obvious that bitcoin is not seeing widespread adoption for commerce.
Bitcoin, though, might still have a niche where it might see wider adoption. We’ll see that later.
Another Problem – Cost And Time
For bitcoin to see wide adoption it needs to have clear advantages over existing digital currencies. It needs to have these advantages for those planning on using bitcoin for licit commerce.
Of course, we know that for those trying to pull off illegal activities, bitcoin’s anonymization, difficulty in tracking payers/receivers, easy crossing of borders and the impossibility to roll back transactions are major attractions.
However, for non-paranoid regular users, these are hardly factors at all. Instead, to see wider adoption, bitcoin would need advantages on things like convenience, safety, speed or cost. But as it turns out, not only doesn’t bitcoin have advantages there but it instead often has disadvantages. Worse still, some of the disadvantages are structural. Let’s see.
As we’ve seen in my prior “Basics” article, bitcoin transactions are validated by being committed to the blockchain, as new blocks are mined (found).
The thing is, bitcoin is structured so that the average time to find a block approaches 10 minutes. As a result, in the best of cases any transaction can take up to 10 minutes to be validated, versus “instantaneous” for regular digital currencies.
It’s not a coincidence that the average time to confirm a transaction sits slightly above 10 minutes:
Source: Blockchain.info, for transactions with fees only. A transaction without fees could theoretically sit un-confirmed forever.
Of course, there are off-chain alternatives to make this speedier. Someone is going to take the risk of the confirmation ultimately not happening. But then again, “off-chain” is another way of saying “no longer bitcoin”.
When it comes to bitcoin, the cost of transacting is deeply tied with the time it takes for a transaction to be validated.
As we’ve seen in the prior “Basics” article, when each block is found by the miners, the miners include transactions into it. Obviously, the miners will include transactions paying the highest fees first. So as a user, if you want a quick confirmation, you have to pay up.
Even in the best of times, paying so as to be sure your transaction gets included quickly, can mean a “quickly” up to 10 minutes. As for the cost “to be sure it’s confirmed quickly”, it floats all the time. Currently it sits around $5 to make sure your transaction is validated within 10-30 minutes (the next block to the next 3). Average transaction costs, though, are all over the place and have been rising rapidly:
Now, those fees are just the most basic, underlying, marginal cost which needs to be paid to a miner for him to include (and thus confirm) the transaction in a blockchain block. These fees ignore all other possible costs including:
- Overhead at the payments processor.
- Margin required by the payments processor.
- Further costs to convert to and from regular currency.
Why are those costs exploding? As we saw earlier, the Bitcoin network is operating at capacity when it comes to transactional capacity. Increased demand – for speculative or commerce reasons, both of which are inflated by bitcoin rallying – versus limited supply (to validate transactions) necessarily leads to scarcity pricing.
Already, you should see that the whole promise of widespread bitcoin adoption is dead on arrival. Micropayments, or even regular payments, cannot cope with an underlying $5 cost before all overheads and margins. For instance, the average ecommerce ticket should be between $80-$100 (Amazon.com had a $84 average during 2015). At those levels, $5 would be 5-6.25% of the whole ticket before all other payment overheads and margins. Those are already levels at which it basically cannot compete with any other payments system.
Even BitPay is already saying as much, by:
- Claiming it doesn’t make sense to make transactions below $20.
Talking to Singh, though, it seems the company would love to be able to help both, but, with transaction fees currently rising, it’s just not practical to make transactions under $20, he said.
- Adding a network fee to the 1% commission paid by the merchant.
- Having users pay the transaction (miner) fee (which basically insulates BitPay from the horrendous cost but makes it uneconomic to use bitcoin for small payments).
There are ways to minimize these costs. They consist in:
- Internalizing payments where possible (but then it’s no longer bitcoin).
- Use lower fee amounts in hopes of getting confirmed, only later. Of course, this basically extends how long it will take to confirm the transaction and makes it uncertain. It could be hours. It could be days. It’s not a way to run a business. It’s also not a coincidence that the transactions waiting for confirmation are all bunched up trying to pay less (remember, the mempool represents transactions waiting to be confirmed):
From time and cost, we also know that bitcoin lags in convenience. These are things that both need to be economic for any transaction size, and which the user does not want to be bothered with – the user just wants to pay and have it be instantaneous. So does the merchant.
There is, however, one niche where bitcoin might still be viable over and beyond illegal activity. That’s the field of larger international payments.
Here, both the time constraint and the cost constraint are removed. Here’s why:
- Since cost is fixed no matter what the size of the transaction, the user can pay more to have the transaction confirm quickly – and still have that cost be an incredibly small portion of a large transaction.
- As a result, a large international transaction can be expected to confirm within 10 minutes (which is much more competitive than a wire transfer, though still slower than PayPal) and at a low cost.
Given this, competitive solutions for international transfers can indeed be built upon bitcoin. However, those can’t be remunerated as a % of value, as their competitors (wire transfers, mostly) aren’t priced that way either. Moreover, the ultimate cost will have to include currency to bitcoin and bitcoin to currency costs, which might again level the field versus traditional transfer methods.
Still, while we can say that bitcoin is structurally in trouble when trying to penetrate regular commerce, the same cannot be said for large international transfers – especially those between different currencies (where the bitcoin to currency conversions might compete well with currency to currency conversions, at least for small/medium customers not having access to currency market spreads).
The True Cost
There’s yet another problem, which is harder to portray. We already saw above that the marginal cost to validate a single transaction is now sitting at $5-$7. However, this actually isn’t the true cost. Let me explain:
- The transaction fee (which was the cost we talked about) is just part of the miner compensation.
- The value of the new bitcoins found (the block reward) is the other part.
The two taken together are the true cost (per transaction) needed to keep the network running. This is more evident because over time the block reward will trend to zero. Hence, over time the entire miner revenue will have to come from transaction fees … so to keep current revenues the transaction fees would have to trend to the sum of the block reward plus current transaction fees. This sum looks like this:
As we can see, the true cost per transaction now hovers around $50. This is incredibly high and, again, would invalidate most commerce uses other than for buying big ticket items. This would also beg the question of why one would go to the trouble of buying bitcoins just to then buy the big ticket items (thus incurring currency to bitcoin conversion fees on the user side and bitcoin to currency conversion fees on the merchant side).
Some might think that just increasing the blockchain capacity to handle transactions, by increasing the block size, will solve most problems. Alas, it won’t for a very obvious reason:
- As we saw, the transaction fee is close to 1/10th of the actual cost to transact in bitcoin (which includes all miner revenue).
An increase in block size would likely collapse transaction fees. However, the total revenue wouldn’t collapse (the block reward is basically a function of time and the bitcoin price).
As a result, over time as the block reward got lower, again the same cost per transaction would have to come from somewhere. A panacea would be greatly increased transaction volumes (in terms of number of transactions, not value) … but how would that happen if clearly the allure to hold bitcoin comes from its increasing quote, and not from using it in commerce?
In commerce, bitcoin will always be slower or not be bitcoin at all (off-chain trade). That’s even if the cost converges with regular digital currencies and the intricacies of dealing with bitcoin are all solved transparently.
There are many conclusions we can draw here:
- Bitcoin’s usage for commerce is a tiny fraction of the usage for speculation. This is especially evident when it comes to transaction value.
- Bitcoin usage for commerce has been growing rapidly, but so did it during the previous bubble. Bitcoin usage depends on the bitcoin price, so clearly wealth effects are playing a role that’s more relevant than the underlying attractiveness of using bitcoin for commerce. Bitcoin usage for commerce will likely collapse as soon as the bitcoin price also collapses.
- It’s easy to see why wealth effects are more alluring that any underlying attractiveness for commerce. Bitcoin takes more time to confirm and is too expensive (in terms of transaction fees) to use as a regular currency for online commerce other than for more expensive purchases.
- This is even more evident when considering the true, total cost implied in transacting with bitcoin, instead of just the transaction fees.
- Arguably, bitcoin can find a niche in international transfers by small/medium customers between two different currencies. Under those conditions, it looks to be cost and time competitive as the fixed fees are diluted. The requirement of it being between two different currencies comes from the added cost of currency/currency conversion on regular payment systems. Under bitcoin, these conversions are replaced by currency/bitcoin and bitcoin/currency conversions which are likely to be cheaper for small/medium customers.
- Bitcoin/currency and currency/bitcoin conversions can be cheaper for small/medium users because they’ll be more easily exposed to (bitcoin) market spreads than foreign exchange spreads. For a larger corporation, which will have closer-to-market foreign exchange spreads, bitcoin is likely not competitive even on the “international transfer between two currencies” scenario. This is so because bitcoin spreads are larger than say, EUR/USD, USD/GBP, etc., forex pair spreads. This can be turned on its head when working with exotic pairs, though, like USD/CNY.
- For regular commerce usage, bitcoin is at a cost and time disadvantage versus typical digital payment systems like PayPal. Both are structural. The time disadvantage comes from the block generation time (~10 minutes). The cost disadvantage comes from the total revenues enjoyed by miners. Even if the transaction fee problem is solved by a bitcoin fork or upgrade, the actual cost to transact will always come to the fore with time (as less bitcoins are awarded per block).
- We’ve seen the problems bitcoin has in becoming widely accepted as a currency for commerce. It faces an even worse uphill battle to become a “store of value”. The reason is simple: The entire trading taking place on-blockchain (and which also supports the off-blockchain trade) has to support the whole Bitcoin system. The cost to support the whole Bitcoin system is running at a ~$5 billion/year pace (~$14 million/day, 9/10ths of which come from the “disappearing over time” block rewards and 1/10th of which comes from transaction fees) or ~4% of the entire bitcoin market capitalization. An asset which has intrinsic costs of 4%/year even before other system costs, overheads and margins is an intrinsically poor store of value.
- Finally, bitcoin is wildly overvalued in relative terms. When someone buys bitcoin he only buys a token used in a much larger system (miners, exchanges, payment processors). The token confers no rights to economic returns enjoyed within that system. Compare that to PayPal: When you buy a PayPal share you get the right to economic returns from the entire similar system – a system which is 200-1000x larger, and growing consistently (as opposed to bitcoin’s reliance on bubbles to propel activity). Yet, in spite of representing a fraction (of the economic system, arguably zero) of a fraction (of the actual payments activity) of PayPal, bitcoin trades for 40% more than PayPal’s market capitalization.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.