No, they aren't, because the price of BTC is growing exponentially without necessarily having dollar deposits grow at the same time. Here's a worked example.

In a closed system:

Day 1: $1 buys 1000 Marmotcoin

Day 2: $1000 buys 1 Marmotcoin

Day 3: 1000 Marmotcoin attempts to sell for $1,000,000

That's what the initial phases of a liquidity shock will look like. I can almost guarantee you that retail operations all have bank liquidity facilities to deal with sudden upswings in withdrawal demand, otherwise they have to commit house money to cover it.

Your article assumes that to sell on Day 3, the exchange has to borrow $1,000,000 from a bank.

That's now how an exchange works.

For the price to be 1,000,000 USD for 1,000 BTC, there must exist buyers with 1,000,000 USD and sellers with 1,000 BTC with orders to trade at that price.

There's no "house money" since exchanges aren't casinos where the players bet against the house. Buyers and sellers are transacting with each other, matching buy orders with sell orders.

The exchange doesn't have to borrow the USD from a bank, the buyers bring the USD to the trade.

There is no "liquidity shock" unless the exchange steals or loses the buyers' USD. And that's a different argument.

You're still missing the point. If the exchange is providing a fiat on/offramp, then it will have to manage the possibility that the amount of USD deposits is massively exceeded by the amount of Bitcoin deposits which may one day call on the USD deposits to be redeemed.

For example. If I bought $1000 for 1 BTC from Coinbase on January 1st 2017 and try to exit that position on Jan 1 2018, once I sell that $10,000 bitcoin to Coinbase I'll be withdrawing $9000 more than I put in. Coinbase has to get the $9000 from somewhere, whether that be from other deposits which are more recent than mine, or from its own trading operations.

If you have a large number of people trying to withdraw a large number of dollars at the same time, your offramp could run out of dollars or, in the alternative, be required to (a) draw down a facility with a bank or (b) sell assets. In an environment where the asset price is falling rapidly it may not be able to avail itself of either option.

As occurred in 2008, with dire consequences.

Except for the pathological example of Bitfinex (another story), every $1 of coin sold is backed by the $1 that a depositor has made. There cannot be a 'bank run' like you describe unless the exchange is running a fractional fiat reserve. This is not the case is which you describe. If there a massive withdrawal of billions from coinbase, it means that there is a massive deposit of billion TO coinbase to make the trade. In the case of bitfinex, they appear to be doing peer to peer bank transfers to get round their own banking problems. (ie: you want 10m out, i want 10m in, we trade A->B, rather than through the exchange as a middle man).

You simply cannot sell it if there is no buyer to take it. You never 'sell' to coinbase -- this is a front for GDAX, their exchange. Every buy is taken from a seller on their market. The money is exchanged between you and their account, and they take fees.

>Coinbase has to get the $9000 from somewhere

yeah, from the person that just bought BTC for $10,000.

Seriously, three comments in a row and he is still repeating the same fundamental misunderstanding. It's not a complicated concept, I don't get the confusion.

I have a share of AAPL stock. I wish to sell it.

I place the sell order with my brokerage, and I get fiat USD deposited into my account. This is exactly how Bitcoin is exchanged.

Where exactly lies the fundamental liquidity problem for my brokerage/exchange?

Exchanges don't provide fiat on/offramp. Exchange users do, by making deposits.

Every dollar that a user withdraws from an exchange is a dollar that was (already!) deposited by another user.

Exactly. Here's a revised "worked example".

Day 1: Alice buys 1,000 Bitcoin from Bob for 1 USD on a street corner

Day 2: Market price on the street corner is 1 Bitcoin for 1,000 USD

Day 3: Alice sells 1,000 Bitcoin to Charlie for 1,000,000 USD on a street corner

The street corner doesn't have to borrow 1,000,000 USD from a bank.

Charlie brings the USD to the trade.

You just proved my point. For this to keep working you need Charlie to show up with $1,000,000 new dollars.

In a panic, you need to flip this scenario on its head. Charlie is not going to show up and Diana, who got into Bitcoin in 2009 and hasn't done anything since, decides to take her profits on 1,000 BTC. So this system has $1,001 to meet $2,000,000 in deposit demand, assuming neither Alice nor Bob has withdrawn their USD and some Bitcoin hodlers pile in.

Particularly popular liquidity pools eg BitPesa or Coinbase may then be put under pressure to either cease operations, draw down liquidity facilities to buy BTC, or start selling their own assets in order meet demand of BTC holders who want to get out. Those are not good choices.

In this panic, the price of bitcoin drops as Diana dumps her 1k BTC market sell on the book. The USD comes from whomever purchases - there is no liquidity problem.

Call me back after the bubble crashes and we can have a discussion about where the dollars ran off to in the middle of it.

Why does he need to call you back? Stock market crashes are nothing new. Bitcoins are basically the same thing as stocks in terms of the way they're traded.

No one is saying a price crash isn't possible. But it won't be caused by some liquidity issue, it'll just be people selling off.

Diana can't just "decide to take her profits" at whatever price she'd like. If she wants $1m but Charlie isn't interested, she will have to lower her price until he is. This is how exchanges work, cryptocurrency or otherwise. If a stock falls to zero, do you think that Etrade dips into its own assets to provide an "off ramp" to its customers that weren't able to sell in time?

You're essentially saying that the exchange should artificially prop up the price of bitcoin with its own money during a panic.

Bids on an exchange's order book are definitely backed by dollars. I don't know why you insist they aren't.

You use strange language. Maybe the scenario you mean to describe is not about lack of deposits or lack of float, but is a classic panic sale (ie. order books not deep enough to absorb all asks). Yes, this could happen with Bitcoin, or basically any financial instrument: stocks, forex, etc.

As someone who has been trading bitcoins since 2010, I will retort that market depths have definitely been growing over time. I don't have precise data to show you, but for example GDAX's BTC/USD order book accounts for 1/20th of the worldwide BTC trading volume. Right now selling 3600 BTC on GDAX would net $30M and dip the price -20%. Assuming the same depth at other exchanges, this means traders could globally sell 72000 BTC at the same instant for $600 million and dip the price by only 20%. That's a pretty decent market depth.

If market depth grew proportionally to Bitcoin's price, it would mean that 7 years ago when Bitcoin was trading at $0.25 (1/36000th its current price: http://bitcoin.zorinaq.com/price/) then a sale of $16k worth of Bitcoins would have dipped the price by 20%. I don't know if we can find archives of MtGox's trading data from 2010, but I roughly remember selling blocks of 1000 BTC at a time on Mtgox for $0.70-1.00/BTC in February 2011 and each of my sale would dip the price by a few percents.

So my (vague) recollection seems to indicate I'm probably right that market depth increased proportionally to Bitcoin's price. Therefore Bitcoin would have been at an equal risk of a panic sale at any point in the last 7 years. And the fact there has never been a long-term panic sale is a testament to Bitcoin's resilience. The markets are deeper and are more resistant than you think.

That's not a worked example, it's ignoring everything they just explained.