Clearing, Settlement and T+1

Execution is not ownership (yet)

When your order executes you have the **price** and the **economic exposure**: if the stock moves, the gain or loss is yours. But the legal transfer of shares and cash happens at **settlement**, and in between, the trade is **cleared**. The National Securities Clearing Corporation (**NSCC**) steps into the middle — the buyer to every seller and the seller to every buyer — which is what a central counterparty means. It then **nets**: millions of trades collapse into one net obligation per firm, so only a small fraction of the shares and dollars actually has to move. On T+1 the Depository Trust Company (**DTC**) moves the securities by **book entry**. You almost certainly hold your shares in **street name**: your broker is the registered owner on DTC’s books, and your account shows you as the **beneficial owner**. That distinction sounds like paperwork until something goes wrong, at which point it is the entire protection. Ownership therefore has two clocks running. The price clock starts the moment you fill. The ownership clock does not start until settlement — and between the two there is a window in which a counterparty can fail, which is exactly what the clearing system exists to absorb. How it absorbs one is a lesson in itself: a fixed default waterfall, where the failing firm absorbs the first loss and the clearing fund the second. Netting in miniature (one broker, one stock, one day) — Customers bought: 1,000,000 shares · Customers sold: 950,000 shares · Net shares the broker must receive: 50,000 ←

Why settlement speed matters

Between the trade and settlement, either side could fail. Clearinghouses protect the system by collecting **margin (collateral)** from brokers, and the requirement grows with volatility, because the chance of a failure inside that window grows with it. In January 2021 that is exactly what forced some brokers to restrict buying in meme stocks: the collateral demanded against a violently volatile book rose faster than the brokers could fund it (M21). Settlement also sets the rules for cash accounts. If you buy with the proceeds of a sale that has not settled yet and then sell before those proceeds settle, you commit a **good-faith violation**. Repeat violations restrict the account, and the restriction is the point: you were trading with money you did not have yet. If your broker fails, customer securities are held separately from the firm’s own assets, and **SIPC** covers missing assets up to $500,000, including $250,000 in cash. It does not cover market losses — nothing does — only assets that go missing when a firm goes under. This is the plainest way to see what T+1 actually changed: shortening the cycle shrank the window in which either side could fail, which lowered the collateral the whole system has to hold.

How the system contains a failure

A central counterparty only removes counterparty risk if it can survive one of its own members failing, so the question to ask is not “does clearing work?” but “who pays when it does not?”. The answer is a **default waterfall**, and its order is the whole design. The failing firm's own margin goes first, then its contribution to the **clearing fund**, then the fund itself — which is mutualised, meaning every surviving member has pre-funded a share — and only after that do further assessments on members come into play. That sequence is why a single broker's failure does not become a chain. The defaulter absorbs the first loss, the pooled fund absorbs the second, and the system keeps settling. It is also why the fund's size is a live number: it is sized against plausible member failures in plausible markets, and the members who are most exposed pay the most into it. Netting is the other half of the answer, and its scale is easy to underestimate. The **notional** value of a day's trading is enormous; the value that actually changes hands at settlement is a small fraction of it, because offsetting obligations cancel before anyone moves anything. Without netting, the same day would require moving gross amounts that no settlement system could process — which is why every innovation in this area since the 1970s has been about collapsing exposures rather than moving more paper. Shortening to T+1 also changed the plumbing upstream. Because the clearinghouse cannot net what it has not been told, allocations and **affirmations** for institutional trades were moved to trade date, with a deadline the evening of the trade — around 9:00 pm ET — rather than the following morning. A system that settles in one day has no room for a confirmation that arrives late. The default waterfall, in order — 1. The failing member's own margin: absorbed first · 2. That member's contribution to the clearing fund: next · 3. The mutualised clearing fund: shared by surviving members ← · 4. Further member assessments: the last line, and rarely reached SIPC and the clearing fund answer different questions, and it is worth keeping them apart. SIPC protects **your assets** when your broker fails. The clearing fund protects **the system** when a member fails to settle. Neither one covers a stock you bought that went down. Settlement risk is not a theoretical worry that only matters to specialists. Every broker restriction on trading you have ever read about — buying blocked in a volatile stock, a position you cannot open — traces back to the collateral this system requires and how fast that requirement can move.

What T+1 broke for the rest of the world

Shortening settlement from two days to one is a large operational change, and the industry’s own post-mortems listed the same handful of problems everywhere. The first is **funding in a different time zone**. A manager in Asia or Europe buying U.S. shares now has to have dollars in place by the next U.S. business day, which for a fund whose cash sits in another currency and another market means arranging foreign exchange a day earlier than before. Firms that made the change described retiming funding cycles, keeping more prefunded dollars, and running staff on U.S. hours — costs that never appear in a retail account and do appear in an institution’s mandate. The second is that T+1 shrank the window rather than removing the friction. With one day to work, exceptions become urgent: a failed foreign exchange trade, a mismatched instruction or a late affirmation no longer has a spare day to resolve itself. The industry’s response was to move the matching and confirmation work earlier, which is why the confirmations that used to be exchanged on the evening of trade date now often happen within minutes. The retail-facing consequence is smaller but real, and it is the one worth carrying: **entitlement is decided on the record date, not on settlement**. If you buy a stock and hold it through the record date, you are entitled to the dividend even though your shares may not have settled. Brokers extend that credit to you and call it being “short the dividend” on their own books. In practice this is invisible until a broker’s systems get it wrong, and then it shows up as a dividend you expected and did not receive. The T+1 change moved all of those dates one day closer together, which is why the operational cost of any error is now paid sooner. • Non-U.S. buyers must fund in dollars a day earlier, which forces prefunding and FX timing changes. • Fails and mismatches have half as long to resolve, so matching and affirmation moved earlier. • Dividend entitlement follows the record date, not settlement — brokers bridge the gap on their books. • A shorter window concentrates operational risk rather than eliminating it.

When settlement fails anyway

The machinery described above is designed to make a broken delivery rare, and rare is not never. Knowing what happens when a seller does not hand over the shares on the settlement date matters for two reasons: it is how the system stops one failure becoming everyone’s problem, and it is a risk that sits quietly inside every short position a retail account takes. A **fail to deliver** is exactly what it sounds like. The cash leg settles; the securities leg does not. The buyer’s account is debited and carries a receivable, the seller carries a payable, and the obligation stays open and is marked to market until the shares arrive. The most common cause is a short sale where the locate was never properly secured, so there is no share to borrow and none to hand over — which is why fails cluster in thinly traded names and in stocks with high short interest. The rules push against that. A firm that has a fail open for more than a few consecutive settlement days loses the ability to satisfy a short sale with a mere locate: it must locate and actually deliver before it may accept further short orders in that name. Those names are published, and their appearance on the list is a signal about how thin the borrow is. The backstop is the **buy-in**: the clearing side or the broker goes into the open market, buys the shares that were never delivered, and charges the cost to whoever failed to deliver them. A forced buy-in is not a gentle process — it is a market purchase executed on someone else’s behalf, at whatever price the market is asking, and the failing party pays the difference. That is the mechanism a short seller should care about, and it is the reason a short position in an illiquid, heavily shorted name carries a risk that a short in a large liquid one does not. If a buy-in is triggered against your broker because the shares could not be sourced, the covering trade is executed for you, at a time and a price you did not choose, and it can land on a day the stock is up sharply. Short interest, borrow fee and days-to-cover are the data that let you estimate this exposure before it becomes somebody else’s decision. It is worth separating the real phenomenon from an internet legend. Fails are an accounting state with a published aggregate value, and they are routinely cited as proof that the same share is being counted twice in a way that breaks the market. What the data actually shows is a normal operating feature of a settlement system: a small value of trades failing to settle for a day or a few days, worked down by the close-out rules and the buy-in process. The system’s job is to keep each firm’s net obligation correct even while individual deliveries run late, so a fail shows up as a cost and a delay, not a hole in ownership. The shorter settlement cycle sharpened this rather than softening it. With one day to work, a failing delivery has less time to cure itself, and the market’s response was to move matching and confirmation earlier so that fewer trades ever reach the point of failing. So the direction of travel is fewer fails, each one resolved more urgently — which is exactly the trade a shorter cycle makes everywhere. For a long-term investor, a fail is invisible: the receivable is a claim on the shares and the position behaves normally. For a short seller, it is a contingent event with a market price, and the cheaper the borrow looks, the more worth checking whether it is real.

Sources

• Shortening the settlement cycle to T+1SEC — T+1 settlement, effective May 28, 2024
• Clearing, netting and the central counterpartyDTCC / NSCC — Clearing and settlement
• Cash accounts and good-faith violationsFINRA — Cash account trading and settlement
• SIPC protection: what is covered and what is notSIPC — Investor protection