Holmström–Tirole Financial Intermediation and Liquidity Model

Holmström–Tirole Financial Intermediation and Liquidity Model is a unified contract-theoretic framework of corporate borrowing capacity, bank capital crunches, and macroeconomic liquidity supply in business and economics theory, developed by Nobel laureate economists Bengt Holmström and Jean Tirole in two landmark papers: "Financial Intermediation, Loanable Funds, and the Real Sector" (Quarterly Journal of Economics, 1997) and "Private and Public Supply of Liquidity" (Journal of Political Economy, 1998). Bridging the gap between microeconomic principal–agent models of moral hazard and macroeconomic models of credit transmission, Holmström and Tirole demonstrated that because an entrepreneur must retain a minimum inside equity stake in a project's upside to prevent shirking or private-benefit extraction, a firm's pledgeable income—the maximum expected return that can be credibly promised to outside investors without destroying managerial incentives—is always less than the project's total net present value (NPV). This wedge between total return and pledgeable income implies that firms with insufficient internal net worth (A) cannot finance positive-NPV projects directly in arm's-length bond markets. Financial intermediaries (commercial banks and venture monitors) alleviate this credit rationing by paying a monitoring cost c to curtail the entrepreneur's private benefit from B to b < B, segmenting the corporate sector by net worth into unfinanced firms, bank-dependent firms, and direct bond-market issuers, while making aggregate investment sensitive to the balance-sheet capital of both non-financial corporations (K_f) and banking intermediaries (K_m). Extended to interim liquidity shocks in Holmström and Tirole (1998), the same pledgeable-income friction explains why corporations purchase bank credit lines and why correlated macroeconomic shocks require publicly supplied outside liquidity (government Treasury bonds).

Moral Hazard with Limited Liability

Consider a single-period investment problem in the spirit of Innes's moral-hazard debt model (1990) and Tirole's (2006) Theory of Corporate Finance. A risk-neutral entrepreneur has access to an indivisible fixed-scale investment project requiring an upfront outlay of I > 0 at date 0 and possesses internal liquid assets (net worth) A ∈ [0, I). To undertake the project, the firm must raise I − A > 0 from competitive risk-neutral outside investors whose required gross rate of return is γ ≥ 1.

At date 1, the project yields a verifiable cash flow of either R > 0 (success) or 0 (failure). After financing is raised at date 0, the entrepreneur makes an unobservable effort choice:

• Working ("Behaving / Diligent Management"): The project succeeds with probability p_H ∈ (0, 1) and yields zero private benefit to the entrepreneur.
• Shirking ("Misbehaving / Private-Benefit Extraction"): The project succeeds with a lower probability p_L < p_H (denoting Δp = p_H − p_L > 0), while the entrepreneur enjoys a non-verifiable private benefit B > 0 (such as diverting resources to perks or pet projects).

The project has positive social NPV if and only if the entrepreneur works diligently:

The Incentive Compatibility Constraint and Pledgeable Income

By limited liability, neither party receives negative payouts in failure (when cash flow is 0, both receive 0). A financial contract splits the verifiable success return R between the entrepreneur (R_f ≥ 0) and outside investors (R_u = R − R_f ≥ 0). The entrepreneur chooses to work diligently (p_H) rather than shirk (p_L) if and only if the entrepreneur's incentive compatibility constraint (IC_f) holds:

Because the entrepreneur must retain at least R_f^min = B / Δp of the success payoff R, the maximum success payoff that can be promised to outside investors is R_u^max = R − B / Δp. Multiplying by p_H defines the project's pledgeable expected income (𝒫_direct):

For competitive uninformed bondholders (who require expected repayment γ(I − A)) to participate, pledgeable income must satisfy p_H(R − B/Δp) ≥ γ(I − A), which defines the minimum direct-finance net worth threshold Ā(B) (denoted A_market):

Any entrepreneur whose internal net worth falls below Ā(B) (A < Ā(B)) is credit-rationed out of the direct bond market: even though the project has positive social NPV (p_HR > γI), bondholders refuse to lend because promising R_u > R − B/Δp would induce the entrepreneur to shirk.

How Financial Intermediaries Expand Pledgeable Income

Holmström and Tirole (1997) introduce financial intermediaries (commercial banks or lead lenders) that inspect the firm's operations and enforce loan covenants. By incurring a private monitoring cost c > 0, a bank prevents opportunistic diversion and reduces the entrepreneur's private benefit from shirking from B down to b ∈ (0, B):

However, because monitoring effort c is privately borne by the bank, the bank itself faces a moral-hazard problem ("monitoring the monitor"). For the bank (intermediary m) to have an incentive to pay the monitoring cost c, the bank must retain a payoff stake R_m in the project's success satisfying its intermediary incentive compatibility constraint (IC_m):

Delegated Monitoring with Limited Bank Capital (K_m)

In a monitored loan, the success payoff R = R_f + R_m + R_u is shared among the entrepreneur (R_f ≥ b/Δp), the monitoring bank (R_m ≥ c/Δp), and uninformed depositors (R_u ≤ R − (b + c)/Δp).

Suppose the banking sector has limited inside capital (K_m), so that banks earn an equilibrium scarcity return β ≥ γ per dollar of bank capital I_m co-invested in a loan. Since the bank receives expected payoff p_HR_m ≥ p_H(c/Δp), the maximum own-capital I_m that the bank contributes per project is I_m(β) = (p_H/β)(c/Δp), while uninformed depositors contribute at most I_u = (p_H/γ)[R − (b + c)/Δp]. The firm can obtain monitored bank financing (A + I_m + I_u ≥ I) if and only if its net worth exceeds A_bank(β):

Whenever b + (1 − γ/β)c < B, the bank threshold lies below the direct bond-market threshold (A_bank(β) < Ā(B) = A_market), partitioning firms into three tiers:

1. Well-Capitalized Firms (A ≥ A_market = Ā(B)) — Direct Bond / Commercial Paper Market: Firms with large net worth A borrow I − A directly from uninformed bondholders, bypassing banks to save the monitoring cost c.
2. Intermediate-Capitalized Firms (A_bank(β) ≤ A < A_market) — Monitored Bank-Dependent Mixed Financing: Firms with moderate net worth hire a bank monitor to reduce B to b, raising co-financing from both the bank (I_m) and uninformed depositors (I_u).
3. Under-Capitalized Firms (A < A_bank(β)) — Credit-Rationed / Unfinanced: Firms with very low net worth cannot pledge enough income even under bank monitoring; their projects go unfunded.

Credit Crunches vs. Collateral Squeezes in General Equilibrium

Aggregating across a continuum of firms with net worth distribution G(A) and intermediary capital K_m, Holmström and Tirole (1997) distinguish two financial contractions with opposite implications for bank return β:

• Bank Capital Crunch (ΔK_m < 0): When loan losses erode bank equity K_m, monitoring capital becomes scarce, driving up the equilibrium return on bank equity β↑ and pushing up the minimum net worth cutoff A_bank(β)↑. Small and medium bank-dependent firms are rationed out of credit while large bond-market issuers (A ≥ A_market) are insulated.
• Corporate Balance-Sheet Squeeze (ΔK_f < 0): When a recession erodes non-financial corporate net worth A, marginal bank-dependent firms fall below A_bank, reducing overall demand for bank monitoring and lowering β↓.

The Three-Date Model with Interim Liquidity Shocks (ρ)

In their 1998 Journal of Political Economy paper "Private and Public Supply of Liquidity," Holmström and Tirole extended the pledgeable-income framework to three dates (t = 0, 1, 2):

• Date 0 (Initial Investment I): A risk-neutral firm with net worth A chooses project scale I and arranges a financing and liquidity plan.
• Date 1 (Interim Liquidity Shock ρ): Before the project matures, it is hit by a verifiable continuation cost overrun ρ ≥ 0 per unit of scale I, drawn from distribution F(ρ). If the firm injects ρ I at date 1, the project continues to date 2; otherwise it is liquidated at date 1 for 0.
• Date 2 (Moral Hazard and Payoff R): Conditional on continuation, the entrepreneur must retain at least R_f ≥ (B/Δp)I at date 2 to work diligently, leaving a maximum date-2 pledgeable income per unit of scale of:

where ρ₁ = p_HR is the full social continuation value per unit of scale and ρ₀ < ρ₁ is the maximum pledgeable continuation value that can be promised to spot investors at date 1.

Spot Market Failure at Date 1 and Inside vs. Outside Public Liquidity

If a firm waits until date 1 to raise ρI on the spot capital market, new spot investors will advance at most ρ₀ I. Consequently, whenever the realized liquidity shock falls in the intermediate wedge ρ₀ < ρ ≤ ρ₁, continuing the project is socially efficient (p_HR ≥ ρ), yet spot financing fails! To survive shocks ρ ∈ (ρ₀, ρ], the firm must pre-arrange liquidity at date 0 by holding liquid reserves or purchasing an irrevocable committed bank line of credit that cross-subsidizes high-shock states ρ ∈ (ρ₀, ρ] using the surplus pledgeable capacity (ρ₀ − ρ)I generated in low-shock states ρ < ρ₀.

Whether private credit lines can supply this liquidity without government intervention depends on the correlation of shocks across firms:

1. Idiosyncratic Liquidity Shocks (Private Inside Liquidity Suffices): If shocks ρ_i are independent across firms i ∈ [0, 1], a diversified intermediary pools all firms' claims at date 0 and uses the excess pledgeable income (ρ₀ − ρ_i)I of low-shock firms to fund the liquidity injections (ρ_i − ρ₀)I of high-shock firms. Inside corporate claims are self-sufficient and Ricardian equivalence holds.
2. Aggregate Liquidity Shocks and Government Debt as Outside Liquidity: Under a common macroeconomic shock ρ ∈ (ρ₀, ρ^*], all firms simultaneously need a net cash injection (ρ − ρ₀)I > 0 at date 1, and no firm in the private corporate sector has spare pledgeable capacity to sell. Because the government can tax future household labor endowments, issuing sovereign Treasury bonds (outside liquidity) backed by future tax capacity supplies the missing store of value needed for banks and firms to weather aggregate liquidity crises, improving social welfare and generating a liquidity premium on government debt.

Why can't a firm with a positive-NPV project always borrow the full investment cost I by promising investors a higher interest rate?

Promising outside investors a higher repayment R_u in the success state reduces the entrepreneur's residual equity stake R_f = R − R_u. Once R_f drops below B / Δp, the entrepreneur prefers to shirk and take the private benefit B, lowering the probability of success from p_H to p_L and reducing the investors' expected payoff from p_H(R − B/Δp) to p_LR_u. Thus, pledgeable income has a ceiling p_H(R − B/Δp), and any firm whose required borrowing γ(I − A) exceeds this ceiling is credit-rationed.

How does the Holmström–Tirole (1997) model distinguish a "bank capital crunch" from a general recession in loan demand?

In a general corporate recession (a drop in firm net worth K_f), fewer firms qualify for loans, reducing the demand for bank monitoring and driving down the return on bank capital β. In a bank capital crunch (a drop in bank equity K_m), bank monitoring capacity becomes the scarce bottleneck: the equilibrium return on bank equity β rises, the bank-to-bond spread widens, and intermediate bank-dependent firms (A_bank ≤ A < A_market) suffer sharp credit cuts relative to large bond-market issuers (A ≥ A_market).

Why do corporations purchase committed bank lines of credit rather than relying on spot bond markets when liquidity shocks hit?

At date 1 when a cost overrun ρI occurs, a firm can raise at most its future pledgeable income ρ₀I = p_H(R − B/Δp)I in the spot market, which is less than the social continuation value ρ₁I = p_HR I. A committed bank credit line arranged at date 0 commits the bank to advance ρI > ρ₀I in high-shock states ρ ∈ (ρ₀, ρ^*] in exchange for collecting commitment fees and excess pledgeable income (ρ₀ − ρ)I in low-shock states ρ < ρ₀.

Why does Ricardian equivalence fail in Holmström and Tirole (1998) when liquidity shocks are correlated across firms?

Under an aggregate macroeconomic shock ρ ∈ (ρ₀, ρ^*], all firms simultaneously suffer a pledgeable-income shortfall (ρ − ρ₀)I > 0 at date 1, so no private corporation has surplus capacity to lend to another. Because the government can levy taxes on unpledgeable future household labor income, sovereign Treasury bonds provide net outside liquidity that private corporate claims cannot replicate, breaking Ricardian equivalence.

1. Holmström, B., & Tirole, J. (1997). "Financial Intermediation, Loanable Funds, and the Real Sector." Quarterly Journal of Economics, 112(3), 663–691. https://doi.org/10.1162/003355397555316
2. Holmström, B., & Tirole, J. (1998). "Private and Public Supply of Liquidity." Journal of Political Economy, 106(1), 1–40. https://doi.org/10.1086/250001
3. Diamond, D. W. (1991). "Monitoring and Reputation: The Choice between Bank Loans and Directly Placed Debt." Journal of Political Economy, 99(4), 689–721. https://doi.org/10.1086/261775
4. Innes, R. D. (1990). "Limited Liability and Incentive Contracting with Ex-Ante Action Choices." Journal of Economic Theory, 52(1), 45–67. https://doi.org/10.1016/0022-0531(90)90066-S90066-S)
5. Tirole, J. (2006). The Theory of Corporate Finance. Princeton, NJ: Princeton University Press. https://doi.org/10.1515/9781400830183