Leverage is a fair-weather friend. It smiles in calm seas and abandons you in the squall. If margin borrowing is now north of $1 trillion by some measures, ask the only question that matters in a selloff: Who is the buyer of last resort when the exits get crowded? This is not a moral argument against borrowing. It is an engineering one. Systems fail at their weakest joint, and leverage makes the joint invisible until it snaps.
Wall Street’s official margin debt clocks in near recent records, with one major tally around $800 billion and rising. Include securities-based loans, derivatives financing, and private credit lines, and the stack clears $1 trillion. The exact figure is less important than the direction and structure. Margin is not just borrowing. It is a speed amplifier. If a three-standard-deviation day happens more often than textbooks predict, and if margin calls hit at 20 to 30 percent drawdowns, the path of prices matters more than the destination. Forced sellers do not negotiate. When dealers, funds, and retail accounts face the same thresholds at the same time, microstructure takes over. Liquidity that looks deep at noon can be a mirage by 3 p.m. This is how orderly markets become feedback loops: lower prices beget more selling, which begets lower prices. It is not doom. It is basic probability meeting balance sheets.
Study the peaks and you will see a pattern that is suggestive, not conclusive. Margin balances rose into 2000 and 2007. They also rose for years after 2012 without an immediate crash. The lesson is not to call tops. The lesson is that high leverage narrows the margin for error. Investors explain away each surge as a sign of confidence. That is true until confidence meets a catalyst. Sometimes it is policy. Sometimes it is earnings. Often it is nothing at all, just a shift in volatility that turns options hedging and risk limits into sellers. Hyman Minsky sketched this long ago: Stability breeds complacency, complacency invites leverage, leverage ensures instability. The surface can stay placid for a long time. Then the smallest stone triggers a landslide. In markets, the load-bearing pillar is always liquidity. You only learn its rating on the day you exceed it.
The cleanest number on a regulator’s website is never the full picture. Family offices, hedge funds, and structured notes run borrowings through total return swaps and prime broker credit that do not print in headline margin figures. Recall the blowup of a single family office that vaporized more than $50 billion in bank credit exposure and left prime brokers scrambling. It was not unprecedented. It was a reminder. When a string is pulled from the fringe of the system, it often connects to the core. Correlation climbs toward one when it matters. Banks hedge. Brokers call. Funds de-risk. The point is not to name villains. It is to note the architecture: lightly regulated channels allow leverage to flourish out of sight, and when the tide goes out, the plumbing links back into the banking system. What looks isolated is connected.
Zoom out and the macro backdrop is no antidote. Global debt has pushed above $300 trillion, climbing faster than growth in many regions. The United States runs large fiscal deficits at full employment. As rates rose, the easy math of the last decade died. Debt service is no longer a rounding error. It is a line item. Rollovers face higher coupons. Investors cheer disinflation, but the sovereign balance sheet now carries rate sensitivity like a long-duration asset. You do not need a crisis for this to bite. You need time. As maturities reset, the cash flow drain crowds out risk appetite and narrows policy choices. This is the dull, grinding version of fragility. The sharp version comes when a funding market hiccups and policymakers realize the optionality they thought they had is already spent.
Liquidity is like oxygen. You notice it only when it is gone. ETFs promise intraday liquidity, but they deliver the liquidity of their underlying assets, minus friction, in a stress event. Options provide cheap convexity until everyone rushes to the same strikes and dealers flip from dampening moves to chasing them. Zero-day contracts are hailed as harmless entertainment until they are not. The physics are simple: if many players rely on the same stop-loss, VAR model, or daily rebalance rule, then the crowd’s risk management becomes the risk. This is a coordination problem, straight out of game theory. If I know you will sell on a 3 percent drop, my best move is to sell at minus two and front-run you. Multiply by thousands of actors and the equilibrium is a stampede.
A bridge does not fail because traffic believes in it. It fails when load exceeds design and maintenance lapses. Markets are no different. We live in a long drought of drawdowns, punctuated by brush fires that are quickly suppressed. Suppression breeds fuel. When a real fire comes, it is larger. We cut rates, we buy bonds, we extend and pretend. It works until the remedy weakens the patient. The story we tell ourselves is that each time is unique. The mechanism is depressingly similar: rising leverage during calm, a trigger, forced selling, and policy intervention that sets the stage for the next cycle. None of this argues for panic. It argues for the right question: What breaks when the system is stressed, and how quickly does that stress spread?
Recency bias is not a quirk. It is a costly input to portfolio construction. People extrapolate the last two years. They learn to buy the dip with borrowed money because it worked last time. They mistake low realized volatility for low risk. They seek yield without asking about path dependence. This is how securities-based lending becomes a wealth management feature, not a bug. This is how carry trades seduce institutions that should know better. In poker, leverage is like playing short-stacked at a full table. You cannot ride out the cards. You are all-in more often than you think. When the deck runs cold, time horizon shrinks to one hand. That is not investing. That is variance with a margin clerk.
Resilience comes from redundancy, cash flow, and the ability to be a buyer when others cannot. That means less reliance on procyclical financing. It means sizing positions to survive big gaps. It means funding tails when they are cheap and knowing you will look wrong until you look right. It is not glamorous. It does not backtest well in calm regimes. But it is the only way to turn volatility into opportunity instead of ruin. The market does not reward bravado. It rewards staying power. The investor with dry powder and no covenant clock is the one who names the price in a panic.
Everyone wants a forecast. The better lens is path analysis. If leverage is high, liquidity is thin by design, and hidden borrowings exist outside the clean data, then the distribution of outcomes is wider than it looks. The mean can be fine while the tails get fatter. That is the scenario that hurts the most participants the fastest. Borrowing to buy exposure is one convexity. Borrowing to buy optionality is double convexity. In quiet markets, both feel brilliant. In a selloff, both become accelerants. Nothing breaks until it does. When it does, the bids disappear first, and the margin calls arrive on time. The number attached to margin debt is a headline. The structure behind it is the story.
