It's fascinating that the financial system has evolved this kind of layered abstraction (humans owning stocks is a fiction maintained by brokers owning stocks, which is a fiction maintained by DTC owning stocks, which is a fiction maintained...) reminiscent of hallowed engineering kludges like TCP/IP. Its especially interesting since, unlike data packets, stocks are purely an abstraction to begin with, but the systems around them have become sufficiently complicated that it can't function without this kind of structure.

A blockchain likely wouldn't help this issue either- high-speed traders would inevitably be dissatisfied with blockchain latencies and re-invent brokers that abstracted over blockchain transactions at a faster pace. These brokers would wind up causing bureaucratic meltdowns and become managed by some overarching structure similar to the DTC. Etc. If anything, a blockchain-based system would become more complicated, because it reduces the task of adding another financial sub-mechanism to an engineering problem.

Something about large distributed systems of non- or imperfectly-cooperative agents breed complexity. (There's probably already formalized versions of that principle, but none are presently coming to mind.)

Levine is such fun to read!

The solution to the problem he ends up with seems pretty easy to me - this is just counterparty risk:

When you lend your shares (or let them be lent on on your behalf) there are counterparty risks. You charge interest and accept collateral as compensation for these risks. Usually these risks are gone after the shares are returned or the position closed and settled out - but not always. Tracking down the short and making him cough up the extra $2.74 is the broker's job. If the broker can't find him, then if it was a case of a normal margin account the broker is on the hook. If it was a case of the client directly accessing the stock lending market then the broker is probably (fine print time!) not liable for the counterparty failure.

No, a "blockchain" wouldn't help. It wouldn't help finding the holders of three years ago. Also, putting all stock transactions for the entire market into one blockchain would have huge traffic and synchronization issues. All the players have to agree on transaction order. You'd have one giant file that was petabytes long. If you used one of those schemes that "summarizes" the blockchain, you wouldn't have an authoritative record of who owned something years ago.

Bitcoin only works because transactions are few and slow transaction commit is acceptable. Most schemes for scaling Bitcoin involve "off-chain transactions".

DTIC predates high-speed trading. It was designed when it was assume that if you bought a stock, you'd probably hold it for weeks or months, not seconds or minutes.

  1. Mr. A owns a share of stock.
  2. Mr. B borrows Mr. A's share of stock. 
  3 .Mr. B sells the share to Mr. C.

well now, and the article says that Mr B borrows Mr A's stock without asking or telling him and sells Mr C a share.

I understand that this is the way it has worked for some time, but franky I don't see how this is anything but fraud.

Eventual consistency strikes again.

If you own an equity ETF in a retail brokerage, you actually own a claim on a claim on a claim on a fund that owns a claim on a claim on a claim on corporate stock which is itself a secondary or tertiary claim (behind debt and/or preferred stock) on tangible and intangible assets.

It is a common & legal practice for the fund to lend its claim on a claim on a claim on corporate stock to other parties, for a small amount of interest, to help reduce its fees. See https://personal.vanguard.com/pdf/ISGSL.pdf

If the potential for fraud and error in this system doesn't give you a bit of a pause, it probably should.

"That system has worked pretty well for 40 years." - Wait, so I'm saving for retirement (40 years from now) in a system that isn't even that old?

Somehow I don't trust it to still be around when I need to get money back.

Ultimately I don't think it's fair to have BOTH short selling AND resolution of sale price YEARS later.

I actually don't like either of those as features, but both together is clearly incompatible.

I read an article recently on "Patterns for dealing with uncertainty". It never ceases to amaze me how often these concurrency patterns apply to more antiquated business models.

> took it private for $13.50 a share

How do you take something private? What does it mean?

Unfortunately I just keep staring at the woman in the picture at the top of the page.

That was way more interesting than I thought it might be. I really liked the writing style, kinda snarky, but still explained well:

""But what if ...," you start to ask, and I reply: Shh, shh. It just works."

Three words... Counter Party Risk

The problem is that the courts can amend a buyout, which goes against the principles of free market capitalism ..a buyout represents a significant gain above the market price even if the offer is low relative to what it 'should' be. Matt is blaming the wrong target.

This is why eShares or someone like them will disrupt this entire ecosystem.

ComputerShares should be killed off too.

The entire system is amazingly broken.

> There are little cracks that give us brief glimpses of the abyss below. Why not cover them up with a fresh, cool coat of blockchain?

This is a great example of Wall Street's continued but lessening snark towards blockchain and vicariously Bitcoin. That much of the complex system you know and revere and spent so long learning can be replaced by new fangled technology is still a hard pill to swallow. The snark is a coping mechanism.

tl;dr cloud computing has yet to reach stock offerings.

Probably too much regulatory burden for a startup though: https://www.sec.gov/divisions/marketreg/mrclearing.shtml

In before someone mentions blockchain as a solution (IBSMBAAS).

There already exists a database that works and has been well tested over the past 30+ years of securities settlement.

The issues was the database wasn't updated properly. So if a blockchain was used ti would stand to reason that that also wouldn't have been updated properly