Loading field map
Asset Liability Management (ALM) is the disciplined practice of coordinating an institution's assets and liabilities so that its financial commitments can be met over time without exposing the institution to unacceptable risk. Although the term is used across insurance, pensions, and corporate treasury, its most developed and demanding form is found in banking, where it functions as a continuous process of measurement, analysis, and decision-making that sits between the bank's front-office business lines and its senior management.
A bank's balance sheet is defined by a fundamental asymmetry. Its liabilities—deposits, borrowings, and other funding—are typically short-term, often repayable on demand or at short notice, and carry interest rates that can reset quickly. Its assets—loans, securities, and other investments—are typically longer-term, less liquid, and carry rates that are fixed for extended periods or that adjust according to different benchmarks. This mismatch is not an accident; it is the source of banking profit. A bank earns its spread by borrowing short and lending long, by transforming maturity, liquidity, and credit risk into interest income.
But the same mismatch creates the central problem of ALM: the bank's earnings and its solvency are exposed to movements in interest rates, to changes in the shape of the yield curve, to shifts in customer behavior, and to disruptions in funding markets. If short-term rates rise faster than the rates on the bank's long-term assets, its net interest margin compresses. If depositors withdraw funds en masse, the bank may be forced to sell assets at fire-sale prices. If the yield curve inverts, the bank's core business model becomes unprofitable. ALM is the set of techniques and governance structures that allow the bank to take these risks deliberately and in measured amounts, rather than accidentally and in uncontrolled quantities.
The discipline therefore addresses three interconnected questions. First, how much interest rate risk is the bank currently running, and how would its earnings and capital respond to various rate scenarios? Second, how much liquidity and funding risk does the bank face, and what buffers and contingency plans are needed to survive a stress event? Third, what is the appropriate transfer price between the business lines that originate assets and the treasury function that funds them, so that the profitability of each product is measured accurately and the bank's overall risk appetite is respected?
The practice of ALM emerged gradually from the banking crises and regulatory reforms of the late twentieth century. In the decades after World War II, many banking systems operated under interest rate controls, fixed exchange rates, and limited competition. Banks faced relatively stable funding costs, and the management of the balance sheet was largely administrative. The breakdown of the Bretton Woods system in the early 1970s, followed by the oil price shocks and the high and volatile inflation of the 1970s and early 1980s, changed this environment permanently. Interest rates in major economies became both higher and far more variable, and banks that had lent long at fixed rates while funding short found their margins destroyed.
The savings and loan crisis in the United States during the 1980s was the defining event. Hundreds of institutions failed because they had borrowed short-term deposits and made long-term fixed-rate mortgage loans, then watched short-term rates rise far above the yields on their loan portfolios. The crisis demonstrated that interest rate risk was not a secondary concern but a primary cause of insolvency, and it forced regulators and bankers alike to develop systematic methods for measuring and managing the mismatch. The response included the development of gap analysis, duration-based measures, and eventually the simulation models that became standard tools.
A second wave of development came in the 1990s and 2000s, driven by the growth of derivatives markets. Interest rate swaps, futures, options, and other instruments gave banks powerful tools to hedge their balance sheet risks without selling or repricing the underlying assets and liabilities. This transformed ALM from a purely analytical discipline into an active management function. Banks could now choose to run a matched book, hedging most of their rate exposure, or to run an intentional position, taking on risk in expectation of profit. The availability of hedging instruments also made it possible to separate the measurement of risk from its management, and to price internal transfers between business units with greater precision.
The global financial crisis of 2007–2009 added a third dimension. Before the crisis, ALM had focused heavily on interest rate risk, with liquidity treated as a secondary concern. The crisis revealed that funding risk could be sudden, severe, and fatal. Banks that were solvent on a mark-to-market basis failed because they could not roll over their short-term wholesale funding. The regulatory response—most notably the Basel III framework—introduced explicit quantitative requirements for liquidity coverage and stable funding, elevating liquidity risk to a co-equal pillar of ALM alongside interest rate risk.
The field is organized around two broad families of technique, corresponding to the two principal risks: interest rate risk and liquidity risk. Within each family, there are distinct methodological traditions that coexist and complement one another.
The earliest and simplest approach to interest rate risk is gap analysis. The bank classifies its assets and liabilities into time buckets according to their repricing dates—the dates on which their interest rates reset. For each bucket, the bank calculates the gap, which is the difference between the volume of assets repricing and the volume of liabilities repricing. A positive gap means that more assets than liabilities will reprice in that period, so the bank's net interest income will rise if rates rise. A negative gap means the opposite. Gap analysis is intuitive and easy to compute, and it remains widely used as a first-level screening tool.
Its limitations are equally clear. Gap analysis treats all repricings within a bucket as if they occurred at the same time, ignoring the distribution within the bucket. It does not account for the fact that different rates move by different amounts, or that the yield curve can shift in non-parallel ways. It measures only the near-term impact on earnings, not the long-term impact on the economic value of the bank's equity. And it treats customer behavior as fixed, when in fact depositors and borrowers respond to rate changes in ways that can amplify or offset the mechanical gap.
Duration analysis addresses some of these limitations. Duration is a measure of the average time until an instrument's cash flows are received, weighted by their present values. It provides a single number that summarizes the price sensitivity of an asset or liability to a small parallel shift in interest rates. By comparing the duration of the asset portfolio with the duration of the liability portfolio, the bank can estimate the change in its economic value for a given rate move. Duration is more sophisticated than gap analysis because it accounts for the timing of all cash flows, not just the repricing date, and it can be extended to measure convexity, which captures the fact that price sensitivity changes as rates move.
But duration has its own assumptions. It is accurate only for small, parallel shifts in the yield curve. It assumes that cash flows are fixed and known, which is rarely true for banking products. Mortgages can be prepaid, deposits can be withdrawn early, and loan commitments can be drawn down. These embedded options mean that the actual cash flows of a bank's balance sheet are interest-rate dependent, and duration measures that assume fixed cash flows can be seriously misleading.
The recognition that banking cash flows are not fixed led to the development of dynamic simulation, which has become the dominant approach in modern ALM. The bank builds a detailed model of its balance sheet, including the contractual terms of its products, the behavioral assumptions about customer prepayment and deposit withdrawal, and the repricing rules for its administered-rate products such as savings accounts. The model then projects net interest income and the economic value of equity under a range of interest rate scenarios, typically including parallel shifts, steepening and flattening of the yield curve, and historical or hypothetical stress scenarios.
The strength of simulation is its flexibility. It can incorporate any assumption about customer behavior, any repricing rule, and any rate scenario. It can measure both near-term earnings at risk and long-term value at risk. It can be used to test the impact of new products, to evaluate hedging strategies, and to set limits on the amount of risk the bank is willing to accept.
The weakness of simulation is that its outputs are only as good as its assumptions. Behavioral assumptions—how quickly depositors will chase higher rates, how much prepayment will accelerate when rates fall—are inherently uncertain and can change with the economic environment. The models are complex, and their complexity can create a false sense of precision. A simulation that projects a specific dollar amount of earnings at risk is not a prediction; it is the output of a model that embodies a particular set of assumptions about the future. The discipline of ALM therefore requires not only the construction of models but also the governance to question their assumptions, to test their sensitivity, and to ensure that the people making decisions understand what the models can and cannot tell them.
Liquidity risk is managed through a complementary set of tools. The stock approach measures the bank's liquidity buffer—the pool of high-quality liquid assets that can be sold or pledged to raise cash in a stress event—and compares it with the potential cash outflows that could arise from deposit withdrawals, loan drawdowns, and the inability to roll over wholesale funding. The Basel III Liquidity Coverage Ratio formalized this approach, requiring banks to hold sufficient high-quality liquid assets to cover their net cash outflows over a thirty-day stress scenario.
The flow approach, embodied in the Net Stable Funding Ratio, takes a longer-term perspective. It requires banks to fund their longer-term assets with stable funding sources, reducing the reliance on short-term wholesale funding that proved so fragile during the crisis. The flow approach is structural rather than stress-based; it is designed to ensure that the bank's funding profile is sustainable over a one-year horizon, regardless of the specific stress scenario.
Beyond these regulatory minimums, banks conduct their own internal liquidity stress tests. These typically include idiosyncratic scenarios (a loss of confidence in the bank itself), market-wide scenarios (a general freeze in funding markets), and combined scenarios. The bank must also maintain a contingency funding plan that specifies the actions it would take in a crisis: which assets to sell, which funding sources to activate, and how to communicate with counterparties and regulators.
A third major component of ALM is funds transfer pricing (FTP), which is the internal mechanism by which the cost of funding is allocated to business lines. When a loan officer originates a fixed-rate five-year loan, the bank's treasury must fund that loan with a combination of deposits and borrowings. The FTP system assigns a funding cost to the loan, typically based on the bank's marginal cost of raising funds at the loan's maturity. The business line earns the loan's interest rate minus the FTP charge, and the treasury earns the difference between the FTP charge and its actual funding cost.
FTP serves two purposes. First, it measures profitability accurately. Without FTP, a business line that originates long-term fixed-rate loans would appear profitable when short-term rates are low, even if the bank would lose money when rates rose. FTP transfers the interest rate risk to the treasury, where it can be managed centrally, and leaves the business line with a clean measure of its credit and customer relationship profitability. Second, FTP aligns incentives. If the FTP curve is set correctly, the bank's business lines will naturally originate the products that the bank can fund profitably, and the bank's overall risk position will reflect its deliberate risk appetite rather than the accidental aggregation of business line decisions.
FTP is not without controversy. The choice of the funding curve—whether to use the bank's own credit spread, a risk-free rate, or a benchmark such as LIBOR or its successors—can have a significant impact on measured profitability. The treatment of liquidity costs, optionality, and capital charges within FTP is a matter of ongoing debate. And the internal transfer price can create conflicts between business lines and treasury, particularly when market conditions change rapidly.
The practice of ALM today is shaped by three enduring features. The first is regulation. The Basel III framework, and its regional implementations, have made liquidity and funding requirements explicit and quantitative. Banks must hold minimum levels of high-quality liquid assets, must maintain stable funding profiles, and must conduct regular stress tests. The regulatory framework also includes the Net Interest Income simulation requirements of the Basel interest rate risk in the banking book standards, which require banks to measure and disclose their exposure to interest rate changes. These requirements have made ALM a compliance function as much as a risk management function, and they have driven significant investment in data infrastructure and modeling capability.
The second feature is the persistence of behavioral uncertainty. The models used in ALM depend on assumptions about how customers will behave—when they will prepay mortgages, when they will withdraw deposits, how quickly they will move their savings to higher-yielding alternatives. These behaviors are not stable; they change with the economic environment, with technology, and with competition. The rise of online banking and fintech has made deposit funding more rate-sensitive, as customers can move their money with a few clicks. The low-interest-rate environment that persisted in many economies after the global financial crisis created different behavioral patterns than the high-rate environment of the 1980s. ALM practitioners must therefore treat their behavioral models as hypotheses to be tested and updated, not as fixed truths.
The third feature is the integration of ALM with the broader risk management framework. ALM is no longer a standalone function that manages interest rate and liquidity risk in isolation. It is connected to credit risk, because the credit quality of the loan portfolio affects the cash flows that the ALM model projects. It is connected to capital management, because the economic value of equity is a measure of the bank's capital adequacy. It is connected to strategic planning, because the bank's choice of business lines and products determines its structural risk position. The modern ALM function is therefore less a specialized technical unit and more a central nervous system for the balance sheet, providing the information and analysis that senior management needs to make strategic decisions.
The field also continues to evolve in response to new challenges. The transition away from LIBOR to alternative risk-free rates has required banks to reprice their products and re-engineer their models. The possibility of prolonged low or negative interest rates has created new challenges for models that assume rates cannot go below zero. The growth of non-maturity deposits and the changing composition of bank funding have required new approaches to behavioral modeling. And the increasing availability of granular data and advanced analytics has opened the possibility of more sophisticated, more dynamic approaches to ALM, even as the fundamental questions remain the same: how much risk is the bank running, how much risk should it run, and how should it manage the difference?