Loading field map
Why do banks, insurance companies, and other financial intermediaries exist? A stock market or a bond market can, in principle, channel savings directly from households to firms. Yet in every developed economy, intermediaries handle the bulk of this flow. The central puzzle of financial intermediation theory is to explain what these institutions do that arm's-length markets cannot. Over the past half-century, six major frameworks have offered competing and complementary answers, each focusing on a different kind of market imperfection: transaction costs, information asymmetries, coordination failures, and systemic externalities.
The earliest systematic answer, emerging around 1970, was that intermediaries exist to reduce transaction costs. Buying and selling securities involves search, negotiation, and legal expenses; a small saver who wants to lend to a distant borrower faces prohibitive per-unit costs. By pooling funds from many depositors, a bank can spread these fixed costs over a large portfolio. This framework treated the intermediary as a kind of cost-minimizing black box: it did not ask why the borrower and lender could not simply write a contract directly, but it provided a clear rationale for the intermediary's existence. The transaction cost approach never disappeared; it became an embedded background assumption in later models. Every subsequent framework implicitly accepts that intermediaries economize on some friction, even if the friction of interest shifts from physical costs to information or coordination.
By the mid-1970s, theorists began to ask what happens after the loan is made. A borrower may have an incentive to take on more risk than the lender intended, or to shirk effort. This is the problem of moral hazard, a species of agency conflict. Agency theory reframed the intermediary not merely as a cost-reducer but as a delegated monitor that mitigates incentive conflicts between savers and borrowers. The distinctive contribution of this framework was to show that the intermediary's capital structure—its mix of debt and equity—could itself be a device for aligning incentives. A bank that puts its own equity at risk has a stronger incentive to screen and monitor borrowers than a passive bondholder would. Agency theory thus moved the field from asking "why pool?" to asking "who watches the watcher?" and how the intermediary's own governance solves that second-order problem.
Agency theory identified the incentive problem; the credit rationing framework, formalized around 1981, showed that the same information asymmetry could cause markets to break down in a distinctive way. In a classic loan market, a lender might respond to the risk of adverse selection not by raising the interest rate but by limiting the quantity of credit—rationing some borrowers out of the market entirely. This is a non-price equilibrium: raising the rate would attract riskier borrowers (adverse selection) or induce safer borrowers to take on more risk (moral hazard), so the lender keeps the rate below the market-clearing level and rations credit. Credit rationing is not a separate framework from agency theory so much as a deeper application of the same information logic. It formalized a specific market failure that earlier agency models had gestured at but not fully characterized: the possibility that some borrowers are denied loans even when they are willing to pay a higher interest rate. This insight became central to macroeconomics, where credit rationing helps explain why monetary policy affects output through a "credit channel" distinct from the interest rate channel.
If information asymmetries are the core problem, why should a single intermediary monitor borrowers on behalf of many savers? The delegated monitoring model, introduced in 1984, provided a precise efficiency argument. Without an intermediary, each saver would have to monitor each borrower, duplicating effort. A bank that monitors once and certifies the loan to many depositors eliminates this duplication. The model formalized the intuition that the intermediary's comparative advantage lies in avoiding the "costly state verification" that each depositor would otherwise undertake. This framework explicitly synthesized the transaction cost and agency traditions: it kept the cost-minimization logic of the earlier approach but grounded those costs in the information asymmetry that agency theory had identified. Delegated monitoring remains the workhorse model for understanding why banks, rather than bond markets, dominate lending to opaque small and medium-sized firms.
All of the information-based frameworks share a common assumption: the intermediary's main function is to solve problems that arise because borrowers know more than lenders. The liquidity insurance model, published in 1983, offered a fundamentally different rationale. In this framework, the intermediary exists because it provides maturity transformation—converting short-term demand deposits into long-term loans. The key mechanism is a coordination problem among depositors. If all depositors withdraw their funds at once, even a solvent bank must liquidate long-term assets at a loss, triggering a bank run. The demand deposit contract, which promises first-come-first-served redemption at par, is both the source of liquidity insurance and the source of fragility. This framework contrasted sharply with the information-centric models: it showed that intermediaries could fail even when all parties are fully informed and honest. The run equilibrium is a pure coordination failure, not a consequence of hidden information. The liquidity insurance model thus carved out a distinct intellectual space, explaining both the social value of banks (smoothing consumption for depositors who face random liquidity needs) and their vulnerability to self-fulfilling panics.
The 1990s and early 2000s saw a broadening of scope from the individual intermediary to the financial system as a whole. The financial frictions and systemic risk paradigm, crystallizing around 1999, integrated insights from earlier frameworks into a macro-financial setting. From the liquidity insurance model, it took the idea that runs can be self-fulfilling and amplified by network connections. From agency theory and delegated monitoring, it took the recognition that leverage and moral hazard create feedback loops: a drop in asset prices erodes bank capital, forcing deleveraging, which depresses prices further. The systemic risk paradigm added new mechanisms—contagion through interbank exposures, fire sales, and the procyclicality of collateral constraints—that earlier models had treated as secondary. This framework did not replace its predecessors; it absorbed them into a richer picture in which the financial system is not a neutral conduit but an amplifier of macroeconomic shocks. The 2007–2009 global financial crisis gave this paradigm enormous practical urgency, and it now dominates research on macroprudential regulation, stress testing, and the design of central bank liquidity facilities.
Today, the six frameworks coexist as specialized tools for different questions. The transaction cost approach remains the default explanation for why intermediaries exist at all, though it is rarely the focus of active research. Agency theory and delegated monitoring are the standard lenses for understanding corporate lending, venture capital, and the governance of financial institutions. Credit rationing is central to monetary economics and development finance. The liquidity insurance model is the foundation for studying bank runs, deposit insurance, and the lender-of-last-resort function. The systemic risk paradigm has become the leading framework for macro-financial analysis, especially in policy-oriented work.
The major persistent disagreement in the field is between information-based explanations and coordination-based explanations. Information models treat financial fragility as a consequence of hidden actions or hidden types; the liquidity insurance model treats it as a consequence of the maturity mismatch inherent in banking. The systemic risk paradigm attempts to bridge this divide by showing that information problems and coordination problems interact—for example, a run may be triggered by a signal about asset quality, but its severity depends on the coordination game among depositors. What the leading frameworks agree on is that financial intermediaries are not redundant; they perform functions that arm's-length markets cannot replicate. The debate is over which function is fundamental and under what conditions the intermediary's solution becomes a source of instability.