Commonality In Liquidity

A risk manager notices that the bid-ask spread on one bank stock has doubled over the course of a morning. Should she worry only about that name, or check the whole portfolio? If liquidity shocks were purely stock-specific — driven by that one company's earnings, or a single large seller — the answer would be to isolate the name. But liquidity does not usually behave that way. Chordia, Roll and Subrahmanyam showed in 2000 that spreads and depth across large baskets of stocks share a common factor: when one name's liquidity dries up for reasons that have nothing to do with fundamentals, many other unrelated names tend to dry up at the same time. This is commonality in liquidity, and it means a spread widening in one stock is often a symptom of a market-wide condition, not an isolated event.

Why it happens

The mechanism runs through the market makers themselves rather than through the stocks. The same handful of dealers and proprietary trading firms supply liquidity across hundreds of names, funded from a shared balance sheet and constrained by a shared risk budget. When that funding gets more expensive or scarcer — a margin call, a spike in the dealer's own funding costs, a risk-management directive to cut inventory — the dealer doesn't shrink its exposure to one stock, it shrinks its exposure everywhere at once, because the constraint binds at the level of the firm, not the security. Brunnermeier and Pedersen formalized this as the link between funding liquidity (a dealer's access to capital) and market liquidity (the tightness of the spreads that dealer quotes): squeeze the former and the latter degrades across the board, simultaneously.

A worked example

Suppose a proprietary market-making firm runs a risk model that caps total inventory value-at-risk (VaR) across its 300-name book at $10 million. On a calm day, that constraint doesn't bind — the firm is running $4 million of VaR and can quote tight, deep markets everywhere. A sudden volatility spike (say, an unexpected rate decision) roughly doubles the realized volatility of every stock the firm trades. VaR on the existing book jumps from $4 million to $8 million purely from the volatility repricing, with no change in position sizes. The firm is now much closer to its $10 million cap, so its risk system automatically widens quoted spreads and cuts quoted size across all 300 names — not because any individual stock's fundamentals changed, but because the shared VaR constraint tightened. A researcher who only watched the ten most volatile names in that basket would see spreads widen and might attribute it to name-specific news; watching all 300 reveals a single common cause.

Where this gets used

• Portfolio risk management: a fund that assumes it can exit each position independently is implicitly assuming zero liquidity commonality; stress tests that shock every position's transaction cost at once, correlated, are far closer to how real drawdowns unfold.
• Systematic strategy design: strategies that trade many names simultaneously (stat arb, index rebalancing) need to model that their aggregate execution costs spike together in the same regimes their positions are most likely to be unwound, compounding drawdowns exactly when it hurts.
• Detecting regime shifts: a rise in the average pairwise correlation of bid-ask spreads across a universe is itself a usable signal, distinct from and often leading price-based volatility measures.

This matters most for exactly the trades a risk manager plans to rely on in a crisis. A hedge that looks cheap to unwind under normal spreads can turn out to be several times more expensive the day it's actually needed, because that is precisely the day the common liquidity factor has moved against every name in the book at once. Backtests that price hypothetical unwinds using historical average spreads, rather than spreads conditioned on stress, systematically understate this cost — which is one reason realized transaction costs during a real drawdown routinely surprise desks that modeled liquidity name-by-name instead of as a shared, time-varying factor.

In interviews

If asked why an unrelated stock's spread widened alongside the one you were asked about, resist naming company-specific stories first. Ask whether a market-wide liquidity factor — a funding shock, a volatility spike, a risk-limit breach at a major dealer — could explain both simultaneously. That's the Chordia-Roll-Subrahmanyam insight in one sentence, and it's the answer this question is built to elicit.