How Liquidity Stress Travels from One Balance Sheet to Many
The IMF's systemwide stress-testing approach becomes most informative when liquidity is treated as a problem that can occur simultaneously across institutions, not merely as a private buffer inside one bank. This chapter traces how a stylized shock moves through banks, nonbank financial intermediaries (NBFIs), clients, and core markets by way of runs, redemptions, margin and collateral calls, forced sales, and price effects. The goal is to give the reader a causal model for interpreting those mechanisms—and a way to see where the chain depends on assumptions rather than inevitability.
The distinction between liquidity and solvency is a useful starting point. Solvency concerns whether an institution's assets ultimately cover its obligations. Liquidity concerns whether it can obtain cash or acceptable collateral when a payment, withdrawal, or funding need arrives. An institution can be solvent in the long run and still be under immediate pressure if the assets it owns cannot be sold quickly without a large price concession. Conversely, a price decline may be manageable if funding remains stable and sales can wait for orderly conditions.
A systemwide stress test is interested in what happens when those conditions change for many actors at once. The IMF's public description says that its approaches examine systemic liquidity risks affecting numerous institutions simultaneously. That framing matters because the same action—raising cash, posting collateral, reducing a position, or selling an asset—can be tolerable when one participant takes it and destabilizing when many participants take it together.
When liquidity becomes simultaneous
Start with a common stress scenario: the value or financing conditions of an asset deteriorate. The asset might be held directly by a bank, inside an investment fund, or by a client whose borrowing depends on it as collateral. The initial shock need not be identical for every participant. What matters is that it changes the conditions under which several participants must meet obligations or satisfy investors.
For a bank, the pressure may appear as a withdrawal of funding, a demand for repayment, or a need to replace financing that has become more expensive or less available. For an investment fund, the pressure may arrive through investor redemptions. A fund that promises to return cash must either use cash already available, borrow, or sell portfolio assets. For a client, the problem may be a tighter financing condition or a request for additional collateral. These are different institutional events, but each turns a change in confidence, price, or funding into an immediate demand for liquidity.
That simultaneity is the central systemic complication. If a single holder sells an asset, other buyers may absorb the transaction without a visible change in the market. If many holders sell at the same time, the market's capacity to absorb those sales becomes relevant. The price may fall further, the cost of financing may rise, and the collateral value available to other borrowers may shrink. In that setting, a liquidity buffer is not independent of the behavior of other institutions. Its usefulness depends partly on whether cash and collateral can still be obtained at reasonable terms.
The initial losses can therefore differ while the liquidity response converges. One fund may face redemptions; one bank may face a funding withdrawal; a client may face a collateral call. All three may seek the same scarce forms of liquidity or sell claims that are valued by the same market. The resulting pressure is not captured by simply adding the direct loss recorded by each balance sheet. It is a question about timing, common exposures, and the capacity of markets and counterparties to absorb simultaneous actions.
This is why the IMF's systemwide subject includes banks, NBFIs, market-based finance, and their clients. NBFIs are not assumed to behave like banks, and market-based finance is not treated as a single institution. The point is to follow the relationships through which their different responses meet. A bank may lend to a client that also relies on market finance. A fund may hold an asset that appears on a bank's balance sheet. A collateral arrangement may connect a market price to a funding obligation. The same entity can be a source of liquidity in one relationship and a demander of liquidity in another.
Cross-border links make the timing problem harder to see from a single national ledger. The institution that needs cash may be domestic while the investor withdrawing funds, the lender providing financing, or the market holding the relevant asset is elsewhere. The IMF identifies cross-border amplification as one reason to build systemwide tools. That does not establish a universal direction of transmission. It means that the location of the visible institution cannot, by itself, identify the full perimeter of the stress.
A useful way to organize the process is to distinguish a trigger, a mediator, and an outcome. The trigger is the initial price, funding, or confidence shock. The mediator is the behavior that carries it—withdrawals, redemptions, collateral demands, or sales. The outcome is the change in prices, market liquidity, funding conditions, or other balance sheets. The outcome can then become a new trigger. This separation prevents a common analytical mistake: describing a final market disruption as if it were the initial shock, or treating a plausible response as an automatic one.
The feedback loop: from calls and redemptions to fire sales
The sequence becomes clearer if the events are placed in time. An asset price falls or a funding source becomes uncertain. Investors, lenders, or counterparties reassess their position. Some withdraw, redeem, or require more protection. The affected institution then chooses among using cash, borrowing, reducing risk, or selling assets. If the response is large relative to market capacity, prices and trading conditions change. Those changes affect the next round of funding and collateral decisions.
The first two responses in the sequence—runs and redemptions—are demands for immediate access to funds. A run is a withdrawal or repayment pressure that can make continued funding uncertain. A redemption is an investor's request to receive money from a fund. Neither event requires the institution to be insolvent. They become systemically important when the requested liquidity must be raised quickly, when many claims mature or are withdrawn together, or when the institution's usual funding channels are also under strain.
Margin and collateral calls add a market-price mechanism. When the value of an asset or position changes, a counterparty may require more cash or eligible collateral to protect the transaction. The call can be a direct consequence of a market move, not of a new failure inside the institution receiving it. Meeting the call may require selling assets or finding new funding. If many participants receive similar calls, the demand for cash and high-quality collateral can increase at the same moment that the value of the assets available to raise it is falling.
Fire sales are the point at which a private liquidity response becomes a market event. A sale is not necessarily a fire sale merely because it is forced. The term is useful when the need to sell quickly, the size of the sale, or the lack of willing buyers causes the seller to accept a price materially below the value that might be realized under calmer conditions. The lower price then affects every holder of the asset, including institutions that did not need to sell. It may also weaken collateral and create new margin demands.
Price dynamics can thus reverse the apparent direction of causality. At first, an institution sells because it needs cash. Later, another institution may need cash because the first sale lowered prices. A client may face tighter financing because collateral is worth less. A bank may reduce lending because a borrower has become harder to assess or because its own capital and liquidity conditions have deteriorated. A fund may face additional redemptions if investors interpret falling prices as evidence of further losses. The IMF's abstract names runs, redemptions, margin and collateral calls, fire sales, price dynamics, and disruptions in core markets as part of the systemwide liquidity problem; the exact sequence and strength of the feedback depend on the scenario being examined.
The reference to core markets is important. Some markets perform functions far beyond the balance sheet of one participant: they provide prices, collateral, funding, or a venue for transferring risk. Disruption there can make several other markets harder to use even if no single institution has failed. This is one reason market-based finance belongs in the same analytical picture as banks and NBFIs. Market conditions can mediate the path between an institution's action and another institution's constraint.
The feedback loop is not guaranteed to run to its most severe endpoint. A fund may meet redemptions with cash or highly liquid assets rather than selling into a falling market. A bank may replace withdrawn funding. A lender may accept additional collateral or extend the maturity of a loan. Buyers may enter the market, or the original shock may reverse. These possibilities are not proof that a system is safe. They show why a stress test should compare assumptions and scenarios instead of presenting a single chain as a mechanical forecast.
A model may also stop the chain at different points. One exercise may represent the initial loss and direct liquidity demand but hold prices fixed. Another may allow asset sales to affect prices and feed back into other balance sheets. A third may examine how cross-border counterparties or clients transmit the result. These are not merely technical variations. They answer different questions about where amplification might arise. The public description of the IMF paper does not specify a particular horizon, scenario design, or calibration, so it supports the mechanism and its policy relevance without establishing which branch is quantitatively dominant.
A disciplined interpretation asks five questions of any reported result. What was the initial shock? Which actors were required to respond? Which funding, collateral, or market constraints mediated their response? Did the model allow prices and market liquidity to change? Finally, which institutions or clients received the second-round effect? If those questions cannot be answered, a large loss estimate may be less informative than a smaller result whose transmission path is explicit.
The same discipline applies to cross-border stress. A sale in one jurisdiction can change prices used elsewhere, while a funding withdrawal abroad can force a domestic institution to raise cash. But the direction, size, and legal treatment of those flows are not supplied by the public abstract. They must be measured or assumed in the relevant application. Cross-border inclusion expands the question; it does not eliminate the need for evidence about the connections.
The strongest conclusion is therefore about mechanism, not inevitability. Liquidity stress travels from one balance sheet to many when simultaneous demands for cash or collateral prompt actions that alter prices, market functioning, or funding conditions for connected actors. Runs and redemptions create the demand; margin and collateral calls can intensify its timing; forced sales can transmit it through markets; and price effects can feed it back into banks, NBFIs, clients, and core markets. A systemwide stress test earns its wider perimeter by making that chain available for examination.
The next analytical task is to determine which tool can reveal which part of the chain. A broad map of claims and flows can show where the connections lie, while a focused liquidity analysis can examine how an investment fund's response becomes a market-impact problem. Keeping those roles distinct is essential: the map identifies possible routes, and the dynamic analysis tests how one route may amplify stress.