3. Leverage and liquidity: how a small loss turns into a cascade

Macro 5: Debt, credit and crises

Almost every financial crisis has the same internal structure, and it is not the one usually told. Nobody needs to have got things badly wrong: it is enough to have got them slightly wrong while heavily borrowed, holding an asset that is hard to sell. Leverage and liquidity are the two variables that turn an ordinary mistake into a historic event.

What leverage does to the numbers

Leverage means operating with more capital than you have, borrowing the difference. It is measured as the ratio between the size of the position and your own capital.

At leverage of 25 to 1, you control 100 with 4 of capital. A 1% rise in your positions gives you 25% on your capital: spectacular. A 1% fall takes that same 25% away. And a 4% fall does not leave you with less capital: it leaves you with none. The ruin threshold is simply the inverse of the leverage.

Two consequences follow that people underestimate:

• Leverage does not change the expected return, it changes the distribution. It multiplies gains and losses alike, but only losses have an absorbing floor: once capital reaches zero the game is over and there is no coming back, however right your analysis was.
• The higher the leverage, the less room for error in timing you have. A correct position that takes six months to prove you right is irrelevant if you are liquidated in week six.

Two different things both called liquidity

The word is used for two concepts that feed each other and are worth keeping apart:

• Market liquidity. How much of an asset you can sell without moving the price against you. It is high in index futures or short-dated government debt, and low in individual corporate bonds, the debt of small countries or any large position in a thin market.
• Funding liquidity. How much capacity you have to raise money to hold your positions up: credit available, collateral you can pledge, cash in hand.

The connection between the two is what drives crises. When funding liquidity dries up you have to sell; selling illiquid assets pushes the price down; as the price falls your collateral is worth less and your funding shrinks further still. A loop that feeds itself without anyone behaving irrationally: every participant is doing the sensible thing for themselves.

The exact mechanics of the cascade

It is worth walking through it step by step, because it is always the same:

• The price of an asset falls a little for whatever reason. Nothing dramatic.
• Whoever holds it with leverage gets a margin call: their broker demands additional cash or collateral to cover the loss.
• Meeting the call means selling something, and since the problem asset can no longer be shifted at a decent price, what gets sold is whatever has a market: government debt, gold, large liquid shares, precisely the healthy assets in the portfolio.
• Those sales push down assets that had nothing wrong with them, and infect other participants who were holding them.
• The broker, seeing markets in turmoil, raises the haircuts on collateral: it demands more security for the same loan. In 2008 the haircuts applied to securitized collateral went from almost nothing to tens of percentage points, which amounted to a wholesale withdrawal of funding without anyone taking a penny out of a branch.
• Everyone with leverage cuts positions at once, with the same risk models telling them the same thing on the same day.

That last point is the key one: forced selling is not a decision, it is a contractual obligation. And obligations are not argued out of with points about fundamental value.

What no individual risk measure captures

Every participant can measure their own risk with reasonable precision and the system can still be fragile, because what matters is how many people hold the same position with the same leverage and the same liquidation thresholds. That information sits on nobody's balance sheet. It is why opacity matters so much: if the exposure is taken through derivatives that never show up as holdings, not even the brokers financing the trade know the real size of what they are propping up.

LTCM in 1998 and Archegos in 2021

Long-Term Capital Management was, on paper, the best-equipped fund in history: two Nobel laureates on the board and a first-rate trading team. Its positions were relative value bets, designed to earn a little very often. For that little to matter, leverage was needed: it went into 1998 with some 4.7 billion dollars of capital and a balance sheet above 100 billion, of the order of twenty-five times, plus notional derivative positions running past a trillion dollars. When Russia defaulted on its debt in August 1998, correlations between trades that were supposed to be independent shot up, all of them lost at once and the fund gave up most of its capital in weeks. The Federal Reserve Bank of New York gathered fourteen institutions to recapitalize it with some 3.6 billion dollars, not to save the partners but because unwinding that balance sheet in one go would have dragged the banks down with it. Roger Lowenstein reconstructed the episode in *When Genius Failed*.

Archegos Capital Management, in 2021, repeated the structure with a different technology. It was a family office with enormous positions in a handful of shares taken through total return swaps, a contract that gives the economic exposure without appearing as ownership. Each bank saw its own slice and none saw the total. When the shares fell the margin calls came, the collateral fell short and the banks liquidated at the same time positions concentrated in stocks that could not absorb them. Credit Suisse lost around 5.5 billion dollars and Nomura of the order of 2.9 billion. Twenty-three years on, the same film: concentration, leverage and an asset that could not be sold quickly.