Diamond Delegated Monitoring Model

Diamond Delegated Monitoring Model is the foundational microeconomic theory of commercial banking and financial intermediation in business and economics theory, developed by Nobel laureate financial economist Douglas W. Diamond in his landmark 1984 Review of Economic Studies article "Financial Intermediation and Delegated Monitoring." Addressing a central puzzle of financial economics—why commercial banks and financial intermediaries exist alongside direct bond and equity markets, and why banks simultaneously hold illiquid monitored loans on the asset side of their balance sheet while issuing fixed-payoff deposit contracts on the liability side—Diamond analyzed an economy with ex-post asymmetric information between borrowers (entrepreneurs) and small outside savers. When an entrepreneur requires capital from m > 1 small investors, direct lending via the bond market forces either costly duplication of monitoring (with all m bondholders paying the monitoring cost K, totaling mK) or a free-rider collapse of monitoring in which borrowers face deadweight bankruptcy / non-pecuniary default penalties. Delegating the task of monitoring to a single financial intermediary (bank) reduces physical monitoring costs from mK to K, but introduces a second-tier agency problem—delegation cost ("monitoring the monitor")—because the bank itself privately observes loan repayments and could under-report portfolio returns to its depositors. By issuing standard debt (fixed-claim deposit) contracts to depositors and diversifying its loan portfolio across N independent entrepreneurial projects, the bank exploits the Law of Large Numbers: as N → ∞, the bank's average portfolio return converges in probability to a constant exceeding the promised deposit rate, driving the probability of bank default and the delegation cost per borrower D(N) → 0 asymptotically.

Why Do Financial Intermediaries Exist?

In an Arrow–Debreu general equilibrium model or a frictionless Modigliani–Miller capital market with symmetric information and zero transaction costs, financial intermediaries are redundant: households with surplus savings can purchase bonds and equities issued directly by non-financial corporations without paying a bank spread. Yet in every modern economy, a large fraction of corporate and small-business investment is financed through financial intermediaries—commercial banks, savings institutions, and private-credit funds—that issue deposit or debt claims to households and originate and monitor non-traded loans.

Building on Hayne E. Leland and David H. Pyle (1977) and Robert M. Townsend's costly state verification (1979), Diamond (1984) identified the core informational service provided by a bank on the asset side of its balance sheet: monitoring borrower information, which encompasses screening creditworthiness before loan origination, enforcing loan covenants during the life of the project, and verifying the realized cash flow y of the borrower at maturity.

Setup of the Entrepreneur–Depositor Contracting Environment

Consider a one-period risk-neutral economy populated by N identical risk-neutral entrepreneurs (borrowers) and a large population of risk-neutral outside investors (households / depositors):

• Project Scale and Subdivision Across m Small Savers: Each entrepreneur i ∈ {1, …, N} has an indivisible investment project requiring 1 unit of capital at date 0. Each saver holds an endowment of 1/m units of capital (with m ≫ 1, e.g., m = 1,000 retail savers per loan), so financing a single project requires pooling funds from m small investors whose opportunity cost of funds (competitive gross risk-free rate) is R > 0.
• Stochastic Project Return (ỹ_i): At date 1, project i produces a random output ỹ_i ∈ [0, ȳ] drawn from cumulative distribution F(y) with expected value E[ỹ_i] = R + α > R (positive net present value α > 0).
• Ex-Post Asymmetric Information: The realized cash flow y_i ∈ [0, ȳ] is observed costlessly by entrepreneur i alone. Unless an outside lender pays a fixed monitoring cost K > 0, outside lenders cannot observe y_i and the entrepreneur could claim that y_i = 0 and retain the entire return.

Regime 1: Direct Unmonitored Lending with Non-Pecuniary Bankruptcy Penalties (ϕ^*)

Suppose the m small investors lend directly to entrepreneur i without monitoring (K is not incurred). Diamond (1984, Section 1) shows that without monitoring, incentive compatibility requires an optimal standard debt contract with non-pecuniary bankruptcy penalties (representing time spent in bankruptcy court, loss of managerial reputation, or loss of control rights). If the contract specifies a promised face value h and imposes a non-pecuniary deadweight utility penalty ϕ(z) = max(h − z, 0) whenever the entrepreneur pays z < h, then the entrepreneur chooses payment z(y_i) = min(y_i, h). However, whenever y_i < h (project default), the non-pecuniary penalty ϕ(y_i) = h − y_i > 0 is a pure deadweight social loss.

To minimize the expected deadweight bankruptcy penalty E[ϕ^*(ỹ_i)], the competitive face value h⁰ is set at the smallest root satisfying the investors' break-even participation constraint:

which yields the minimum expected bankruptcy cost of unmonitored direct lending:

Regime 2: Direct Monitored Lending (mK) and the Free-Rider Problem

Alternatively, suppose lenders pay the monitoring cost K > 0 to observe y_i directly, eliminating deadweight bankruptcy penalties (E[ϕ^*] = 0). When m small bondholders each hold 1/m of the loan in direct capital markets, two inefficiencies arise:

1. Duplication of Monitoring Costs (mK): If each investor must verify y_i independently, aggregate physical monitoring costs equal m K per borrower, which exceeds E[ϕ^*(ỹ_i)] when m is large.
2. The Free-Rider Problem: Even if covenant enforcement is a public good among bondholders, each small bondholder receives only 1/m of the recovery while bearing 100% of their own monitoring cost K, inducing every bondholder to free-ride when K > (1/m)E[ϕ^*].

The Three-Way Comparison and Why N = 1 Fails

To avoid duplicating K across m investors, the m investors can form a financial intermediary (a commercial bank) that collects deposits, pays the single monitoring cost K per borrower (instead of mK), and observes y_i. Because the bank's monitoring results y_i are not observed by the m depositors, depositors must write an incentive-compatible deposit contract with the bank to ensure the bank turns over the loan proceeds. Let D(N) ≥ 0 denote the delegation cost per borrower—the expected deadweight bankruptcy penalty incurred at the bank level when the bank monitors N borrowers. Delegated monitoring via a bank is socially optimal if and only if:

What is the delegation cost D(1) if a bank monitors only a single entrepreneur (N = 1)? The bank collects y₁ from the entrepreneur at cost K, but the contracting problem between the m depositors (who require expected return R + K) and the bank (which privately observes y₁) is identical to the unmonitored contracting problem with the entrepreneur, except that the bank must cover R + K instead of R. Thus, the required bank-deposit face value H₁ > h⁰ yields:

Proposition 1 of Diamond (1984): With risk-neutral agents and N = 1 borrower, a financial intermediary never improves upon direct unmonitored debt!

How Diversifying Across N Independent Projects Eliminates Delegation Costs

Now consider a bank that contracts with N > 1 entrepreneurs whose project returns ỹ₁, ỹ₂, …, ỹ_N are statistically independent with mean E[ỹ_i] = R + α > R + K and variance σ_y² < ∞. The bank monitors all N projects at cost N K, pools the aggregate cash inflow G_N = ∑_i=1^N y_i, and issues fixed-claim deposit contracts with per-borrower face value H_N satisfying the depositors' break-even constraint:

where Ȳ_N ≡ (1/N) ∑_i=1^N y_i is the sample average portfolio return. The resulting delegation cost per borrower D(N) is:

Asymptotic Convergence of D(N) → 0 (Proposition 2 of Diamond 1984)

By the Weak Law of Large Numbers, as N → ∞, the sample mean Ȳ_N converges in probability to the deterministic constant R + α > R + K with variance Var(Ȳ_N) = σ_y² / N → 0:

Because the probability that the bank's diversified portfolio falls short of R + K vanishes at rate e^{−N I} (by Cramér's theorem), the required deposit face value converges to the risk-free cost (lim_N→∞ H_N = R + K) and the delegation cost per borrower converges to zero:

Theorem (Diamond's Asymptotic Viability of Delegated Monitoring, Proposition 2): If project returns are bounded and independent across borrowers with E[ỹ_i] > R + K and K < min(E[ϕ^*(ỹ_i)], mK), then D(N) decreases with N and lim_N→∞ [ K + D(N) ] = K.

Hedging Systematic Factors and Townsend Auditing Equivalence

If loan returns follow a factor structure ỹ_i = β_iF̃ + ε̃_i where the systematic factor F̃ (such as Treasury interest rates or exchange rates) is publicly observable while the idiosyncratic shock ε̃_i requires bank monitoring, Diamond (1984, Section 5) proved that the bank should hedge its systematic exposure β̄ F̃ in futures or swap markets (or index deposit rates to F̃), leaving only the idiosyncratic shocks (1/N)∑ε̃_i on its balance sheet so that D(N) → 0. In Section 6, Diamond showed that the same diversification theorem holds under Townsend's costly state verification, where diversification drives the probability of bank default and expected depositor auditing costs to zero. Together with the Diamond–Dybvig model (1983) of deposit liquidity on the liability side and the Holmström–Tirole model (1997) of bank capital crunches, Diamond (1984) forms the cornerstone of modern banking theory.

Why does a bank that lends to only a single borrower (N = 1) fail to reduce agency costs in the Diamond (1984) model?

When an intermediary lends to only one entrepreneur (N = 1), paying the monitoring cost K eliminates asymmetric information between the entrepreneur and the bank, but leaves intact the asymmetric information between the bank and its m outside depositors. To induce the undiversified bank to turn over the loan proceeds, depositors must impose a debt contract with bankruptcy penalties on the bank that exceed those of direct unmonitored debt (K + D(1) > E[ϕ^*]).

How does portfolio diversification across N independent loans solve the "monitoring the monitor" problem?

When the bank holds N independent loans, the Law of Large Numbers causes the sample average cash inflow per loan Ȳ_N = (1/N)∑_i=1^N y_i to concentrate tightly around E[ỹ_i] = R + α > R + K. Good realizations on some loans cross-subsidize bad realizations on other loans within the bank's portfolio, so the probability that total loan collections fall below total promised deposits N H_N vanishes as N → ∞, driving D(N) → 0 without depositors ever needing to inspect individual loans.

Why must the claims issued by the bank to its depositors take the form of fixed-payoff debt (deposits) rather than outside equity?

Because depositors do not observe the bank's realized portfolio return G_N = ∑y_i, any contract whose payout in unverified states varied with G_N would give the bank an incentive to report the lowest possible portfolio return. By the Costly State Verification revelation principle (Townsend, 1979), the optimal contract is a standard debt (fixed-claim deposit) contract that pays a constant amount H_N in solvent states and triggers bankruptcy penalties or auditing only when Ȳ_N < H_N.

Why does Diamond (1984) imply that banks should hedge interest-rate and macroeconomic risks using derivatives?

Any systematic risk (F̃) that cannot be diversified away across the bank's N loans prevents portfolio variance Var(Ȳ_N) from shrinking to zero, raising the probability of bank default and increasing D(N). Because aggregate interest-rate and exchange-rate shocks are publicly observable without monitoring, hedging them in derivative markets removes non-diversifiable volatility from the bank's balance sheet without weakening borrower monitoring incentives.

1. Diamond, D. W. (1984). "Financial Intermediation and Delegated Monitoring." Review of Economic Studies, 51(3), 393–414. https://doi.org/10.2307/2297430
2. Townsend, R. M. (1979). "Optimal Contracts and Competitive Markets with Costly State Verification." Journal of Economic Theory, 21(2), 265–293. https://doi.org/10.1016/0022-0531(79)90031-090031-0)
3. Leland, H. E., & Pyle, D. H. (1977). "Informational Asymmetries, Financial Structure, and Financial Intermediation." Journal of Finance, 32(2), 371–387. https://doi.org/10.2307/2326770
4. Williamson, S. D. (1986). "Costly Monitoring, Financial Intermediation, and Equilibrium Credit Rationing." Journal of Monetary Economics, 18(2), 159–179. https://doi.org/10.1016/0304-3932(86)90074-790074-7)
5. 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