The IMF's Two Lenses: Mapping Exposures and Modeling Fire Sales
The IMF's systemwide stress-testing approach uses two base tools to look at the same financial system from different angles. This chapter compares the Excel-based flow-of-funds framework with investment-fund liquidity analysis that includes fire-sale and market-impact dynamics. The practical judgment is not which tool “wins,” but which question each can answer: where are the relationships and exposure paths, and how might a stressed portfolio's response change prices and conditions for everyone else?
That distinction matters because a broad perimeter and a detailed feedback mechanism solve different analytical problems. A map can show that banks, NBFIs, clients, and markets are connected without showing how quickly a sale changes prices. A dynamic liquidity analysis can show how redemptions and sales amplify stress without, by itself, describing every balance-sheet relationship in the wider system. The IMF's public description supports this division of labor. It names both tools, places them within a systemwide program, and describes their application and enhancement in Financial Sector Assessment Programs (FSAPs). It does not provide enough detail to treat them as one common engine or to rank their predictive accuracy.
What a flow-of-funds framework reveals
At a conceptual level, a flow-of-funds framework is a structured map of financial claims, liabilities, and transactions among sectors. It asks who provides funding, who receives it, who holds a claim, and how financial resources move through the system. The IMF identifies its version as an Excel-based tool. That exact description is important, but limited: the available public record does not disclose the workbook's worksheets, sector definitions, data frequency, equations, or validation results.
The value of the map is that it turns an institutional list into a relationship problem. Suppose the price of an asset falls. A list of banks and funds may tell us which entities exist. A flow-of-funds view can pose more useful questions. Which sectors hold the asset? Which lenders finance those holdings? Which clients depend on the asset as collateral? Which liabilities would become harder to roll over if the asset lost value? Which claims connect domestic institutions to foreign counterparties? The answers identify possible routes through which the original shock could move.
This is a breadth advantage. Systemwide risk often hides in the relationships between categories that are reported separately. A bank may not look unusually exposed when its assets are viewed one by one, while several sectors may hold similar claims or depend on the same funding market. An NBFI may be outside the bank's regulatory perimeter but still be important because its portfolio overlaps with the bank's collateral or because its financing depends on a bank. A client may appear as a borrower in one table and as an investor or market participant in another. A map makes those overlaps available for inspection.
It also helps maintain the distinction between direct and indirect exposure. A bank can lose money because it owns a declining asset directly. It can also be affected indirectly because a client cannot refinance, a counterparty posts weaker collateral, or a market intermediary sells into the same price decline. Those effects should not be merged into one unexplained “systemic risk” number. They are different paths, with different assumptions and potentially different policy implications. The flow-of-funds lens is useful precisely because it makes the path an object of analysis.
A broad map can support several kinds of comparison without pretending to forecast behavior. An analyst can compare the concentration of claims across sectors, inspect which funding relationships are shared, or ask whether a national picture omits cross-border positions. The result may identify a connection worth stress testing more closely. It does not follow that every identified connection will transmit a shock, or that the largest recorded claim will generate the largest loss. Exposure is a condition for a pathway, not proof that the pathway will be activated.
That limitation becomes visible when time and behavior enter the picture. A relationship table may show that an investment fund holds an asset and that a bank has exposure to the same market. It may not tell us whether investors will redeem, whether the fund will use cash or sell securities, how quickly the sale will occur, or how many buyers are available. It can tell us where to look for a liquidity channel. It cannot, from the label alone, establish the size or speed of the feedback.
The same caution applies to the word “flow.” A flow-of-funds framework may organize transactions and positions, but the public description does not establish whether it represents a particular dynamic scenario engine. It is safer to treat the source-supported role as coverage and relationship mapping. Any stronger statement about forecasting, causal identification, or model performance would require the paper's detailed methods and validation evidence.
What investment-fund liquidity analysis adds
Investment-fund liquidity analysis begins with a narrower but more dynamic question: what happens when a fund must meet a redemption or another immediate liquidity need? The IMF's description explicitly connects this analysis to fire sales and market-impact dynamics. That connection is the crucial feature. The fund is not only a holder of assets exposed to a shock; its response can alter the market prices and conditions faced by other participants.
A stylized sequence makes the difference clear. Investors request redemptions. The fund first considers available cash, maturing instruments, borrowing, or the sale of portfolio assets. If sales are large, urgent, or concentrated in less liquid assets, the realized prices may be lower than they would be under orderly conditions. Those lower prices affect other holders. They may reduce the value of collateral, increase funding pressure, or prompt additional investors to reassess the fund. The process can turn a fund-level liquidity need into a market-level event.
The analysis is not claiming that every redemption produces a fire sale. It is designed to examine the conditions under which that outcome becomes plausible and consequential. A fund with sufficient cash and liquid assets may meet withdrawals without materially disturbing prices. A fund whose assets are difficult to sell, or whose investors redeem at the same time as other market participants, may face a different path. The point of including market impact is to make those branches visible rather than assuming that an asset can always be sold at an unchanged price.
This lens adds behavior to the exposure map. The relevant question is no longer only “who holds the asset?” It becomes “who may sell, in what circumstances, and what does the sale do to the market?” The answer depends on assumptions about redemptions, portfolio liquidity, timing, buyer capacity, and price response. Those assumptions are not minor implementation details. They determine whether a private response remains private or becomes a source of second-round stress.
The market effect can reach beyond investment funds. If the same asset is held by banks, other NBFIs, or clients, a price decline can weaken their positions even if they did not sell. If the asset is used as collateral, a lower price can lead to another call for cash or eligible assets. If clients rely on the market to refinance, trading conditions can affect their access to credit. In this way, focused investment-fund analysis connects back to the systemwide relationships identified by the broader map.
There are limits on what this focused lens can establish. A model may represent a particular fund behavior and a particular market-impact relationship while leaving other sectors simplified. The public description does not state the redemption rule, asset-liquidity assumptions, price-impact function, fund sample, or measured amplification. The tool's named dynamics are therefore evidence that this feedback is within scope, not evidence that a specific magnitude or threshold applies everywhere.
A useful model should make the sensitivity visible. If the conclusion changes when redemptions are slightly faster, when a portfolio contains fewer liquid assets, or when market impact is stronger, that sensitivity is part of the result. It tells the policymaker what information matters. If the conclusion is stable across plausible assumptions, confidence may be more justified, though never absolute. Without the underlying methods, one cannot claim which of these properties the IMF applications demonstrate. One can say why they are the questions a reader should ask.
The diagram's comparison uses the same criteria for both tools. The flow-of-funds framework is broader in the relationships it is intended to organize. Investment-fund liquidity analysis is more focused on a particular response and its market consequences. The first can identify where a dynamic question should be asked; the second can examine one important way that question may unfold. Calling them complementary does not imply that they share data, run in sequence, or yield one integrated number. Those facts are not established by the available publication record.
Matching the lens to the decision
The choice between the two lenses should begin with the decision, not with a preference for a more elaborate model. If the policy concern is an unseen concentration of claims, a shared funding dependency, or a cross-border exposure path, the flow-of-funds framework is the natural starting point. It can reveal which institutions, sectors, clients, and markets belong in the next question. Its output is most useful when the analyst needs to widen the perimeter or identify connections that a single-balance-sheet test would omit.
If the concern is whether a particular kind of liquidity pressure can destabilize markets, investment-fund analysis adds the necessary behavioral step. It can examine how redemptions or other demands for cash translate into sales, how sales affect prices, and how those prices feed back into other positions. Its output is most useful when the central uncertainty is not where the exposure sits, but how a stressed holder may behave and how much market capacity can absorb that behavior.
Neither lens is sufficient for every question. A map without behavior can overstate the significance of a connection that is unlikely to activate and understate the speed of a connection that is highly sensitive to liquidity. A focused fire-sale model without a wider map can miss who else holds the asset, who finances the fund, or which clients receive the price shock. The best interpretation keeps the two levels separate: first identify the relevant relationships, then examine the feedback mechanism that could make them consequential.
This is a breadth-versus-detail trade-off, but not a choice between good and bad analysis. Broader coverage requires more data reconciliation and more decisions about how sectors and cross-border claims are represented. Greater behavioral detail requires assumptions about investor actions, asset liquidity, buyers, prices, and timing. More detail can make one pathway clearer while leaving other pathways outside the model. More breadth can expose more possible channels while making each response more stylized. The relevant question is whether the chosen balance fits the financial-stability decision at hand.
The public description of the IMF paper supports a disciplined conclusion. Its two base tools address different parts of the systemwide problem, and the paper describes their application and enhancement within FSAPs across various countries. That is enough to understand the intended architecture and its policy relevance. It is not enough to name a best tool, assert a common computational pipeline, or report a country-level result. Those claims require direct evidence from the paper's methods and applications.
A reader evaluating either tool should therefore ask: What relationship or behavior is represented? What is measured, and over what period? Which assumptions determine the transmission path? What happens if those assumptions change? Which sectors, clients, markets, or borders remain outside the lens? And what decision would change if the result were different? These questions prevent apparent precision from substituting for a clear mechanism.
The essential distinction is now in place. The flow-of-funds framework helps make the system legible by organizing claims, liabilities, and flows across sectors. Investment-fund liquidity analysis helps make one feedback loop consequential by linking redemption pressure, asset sales, fire sales, and market impact. Together they support a more credible systemwide question than either a list of institutions or a single stylized fund response could provide. Their value lies in connecting coverage to mechanism while keeping the boundary of each tool visible.
That distinction also prepares the practical judgment that follows. A systemwide stress test is useful when it shows which connections matter under a stated scenario, which responses amplify the stress, and which assumptions or missing data could change the conclusion. The final question is how those findings should be used in an FSAP-oriented assessment without turning scenario evidence into an unwarranted forecast.