Where you are. Last lesson’s fund borrowed nine million against one and got a margin call on an unremarkable Tuesday. Somebody lent it that money, held its collateral, and made the call. This lesson is that somebody: the bank on the other side of every levered position, which is also the fund’s landlord, custodian and settlement agent.
One relationship, everything bundled
A new fund needs a great deal before it can trade. Somewhere to hold its securities. Somebody to settle its trades. Cash borrowed against its holdings. Shares borrowed so it can sell what it does not own. Reporting good enough to show its investors.
It could assemble that from five providers. Instead it buys the bundle from one bank, and that bank becomes the fund’s counterparty in nearly everything: prime broker, custodian, lender, clearer, and - not incidentally - the party with the clearest real-time view of the fund’s positions in the entire market.
The idea in one paragraph
A prime broker sells hedge funds the infrastructure of leverage as a package: custody, clearing, securities lending and margin financing. The bundling is commercially sensible and structurally significant, because it makes the prime broker the fund’s largest creditor while also making it the party that holds the fund’s assets. That combination is why prime brokers watch client risk continuously rather than periodically: they are lending against collateral they themselves hold, so they can see the cushion thinning in real time, and they can act on it. The mechanism they act with is the margin call, which is arithmetic - the loan may not exceed a set fraction of the collateral’s value, and when the value falls, the excess must be repaid.
The call, as arithmetic
The exercise makes it concrete. A fund posts collateral worth 1,000 and draws a loan of 800. The rule is that the loan may not exceed eighty percent of the collateral’s value, so at those prices it is exactly at its limit.
The collateral falls fifteen percent, to 850. Eighty percent of 850 is 680. The loan is 800. The excess - 120 - must be repaid or covered with more collateral, immediately.
Notice who the cushion protects. Twenty percent of the collateral’s value was never lent against, and that gap exists so the lender can sell the collateral in a falling market and still recover its money. It is not a buffer for the borrower’s convenience. It is the lender’s seatbelt, and this is exactly the point module 1 made about repo haircuts, now seen from the other end of the same relationship.
The concentration nobody chose
Because the bundle is worth buying and few banks can supply it, prime brokerage concentrates. A handful of banks serve most large funds, which means a handful of banks hold a large share of the industry’s collateral, extend a large share of its credit, and see a large share of its positions.
That produces the risk this module keeps circling. A prime broker with many similar clients does not have many independent exposures; it has one exposure held several times over. When those clients face calls simultaneously, its collateral is falling in value at the same moment its clients are least able to top it up, which is precisely the scenario module 5’s clearing machinery and its ordered loss-absorbing stack exist to survive.
Check yourself
1. Why does a fund buy custody, clearing, lending and financing from one bank rather than four specialists?
Because the services interlock: the collateral being lent against is the same collateral being held, and settlement, financing and stock borrow are all easier when one party sees the whole book. The bundle is cheaper and simpler for the fund, at the cost of concentrating its relationships in one counterparty.
2. Collateral worth 1,000 supports a loan of 800 under an eighty percent rule. It falls fifteen percent. What is the call, and why exactly that number?
The collateral is now worth 850 and supports 680, so the call is 120 - the amount by which the existing loan exceeds what the reduced collateral can support. It is arithmetic against a pre-agreed ratio, not a negotiation about the position’s merits.
3. Who does the twenty percent cushion protect, and against what?
The lender, against the collateral falling further while it is being sold. It exists so the loan can still be recovered in a market that has moved, which is why the borrower gets no benefit from it and why the same idea appeared as the repo haircut in module 1, seen there from the other side.
4. Meeting a call in cash brings the loan back within its limit. Why is the fund nevertheless worse off than before?
Because it now controls a smaller position with the same insight. The call deleveraged it, usually by forcing a sale into a falling market, so the fund realises losses and holds less of whatever it believed in. Being back inside the limit is a solvency statement, not a recovery.
Do this
Compute the call, then meet it on the ledger.
python3 code/margin_loan.pyThe starter leaves you margin_call, which returns the shortfall when the loan exceeds what the collateral supports and zero when it does not - never a negative number, because a broker does not refund over-collateralisation. The assertions check that the opening position is exactly at its limit, that the fifteen percent markdown produces a call of 120, and that repaying it on the miniledger restores both the ratio and the world’s invariants. Success prints the loan’s lifecycle and the line:
the call is arithmetic, not judgement - and it arrives on exactly the day the collateral is hardest to sellThe completed version is in solutions/margin_loan.py.
What you can now do. You can compute a margin call, say who the cushion protects, and explain why a prime broker’s client risk is correlated rather than diversified. One group of the cast remains: the brokers most readers actually use, and the question of who pays for a trade that is advertised as free.