Soft Budget Constraint Theory

Soft budget constraint theory is a foundational framework in institutional economics, banking theory, corporate governance, and business and economics theory introduced by Hungarian economist János Kornai (1979, 1980, 1986) to analyze chronic bailouts and shortage phenomena in socialist and mixed economies, and later formalized via dynamic contract theory and asymmetric information by Mathias Dewatripont and Eric Maskin in their landmark 1995 Review of Economic Studies article, "Credit and Efficiency in Centralized and Decentralized Economies." A soft budget constraint (SBC) arises whenever an economic organization (a state-owned enterprise, a systemically important bank, a subnational government, or a corporate division) expects that if it runs a financial deficit or cost overrun, a supporting institution (the state treasury, central bank, or parent conglomerate) will bail it out rather than enforce bankruptcy and liquidation. Dewatripont and Maskin (1995) demonstrated that soft budget constraints are not merely a product of political paternalism, but an endogenous dynamic time-inconsistency (credibility) failure driven by sunk costs: when a centralized creditor with deep capital reserves has already sunk an initial investment I₁ = 1 into a poor project, the creditor finds it ex-post rational to refinance the project with an additional outlay I₂ = 1 whenever the continuation return R_s exceeds the marginal refinancing cost I₂ = 1, even though total social return R_s < I₁ + I₂ = 2 is negative. Knowing that it will be bailed out at t = 1, an entrepreneur with a poor project is not deterred from applying for funding at t = 0. By contrast, decentralized credit markets—where small initial banks lack the liquidity to refinance cost overruns internally and must turn to outside co-financiers who face asymmetric information about project monitoring—harden the budget constraint and screen out unprofitable projects ex ante.

Paternalism, Bailouts, and the Economics of Shortage

In Economics of Shortage (1980) and "The Soft Budget Constraint" (Kyklos, 1986), János Kornai contrasted the classical Walrasian household or competitive firm—which faces a hard budget constraint where expenditures cannot exceed initial endowments plus market revenues on pain of legal bankruptcy—with enterprises in socialist and interventionist economies:

1. Four Channels of Budget Softening: Kornai identified four mechanisms through which a state or parent organization softens an enterprise's budget constraint: (a) Soft Subsidies (discretionary fiscal grants covering operating losses); (b) Soft Taxation (ad-hoc tax exemptions or tolerated tax arrears); (c) Soft Credit (automatic rollovers of non-performing loans without enforcing bankruptcy); and (d) Soft Administrative Prices (cost-plus pricing adjustments that pass cost overruns onto buyers).
2. Behavioral Consequences: When managers expect that losses will be absorbed by external bailouts, the firm's price elasticity of input demand drops toward zero, generating insatiable "investment hunger", chronic input shortages, weak cost control, and survival of negative-value-added firms.

While Kornai attributed soft budget constraints primarily to state paternalism, market economies repeatedly exhibited identical SBC pathologies during the U.S. Savings and Loan crisis, the Japanese keiretsu banking crisis of the 1990s, municipal bailouts, and the 2008 financial crisis. This universality prompted Mathias Dewatripont and Eric Maskin (1995) to ask why a rational, profit-maximizing creditor cannot simply sign an ex-ante contract committing to liquidate loss-making borrowers.

Projects, Entrepreneurs, and Adverse Selection

Consider a two-period economy (t = 1, 2) populated by risk-neutral entrepreneurs and risk-neutral banks, with zero discount rate:

• Two Types of Projects (Adverse Selection): Each entrepreneur has zero initial wealth and possesses an indivisible project requiring one unit of capital (I₁ = 1) at the start of Period 1. A fraction α ∈ (0, 1) of entrepreneurs have Good (Fast) Projects (θ = f), and a fraction 1 − α have Bad (Slow) Projects (θ = s). At t = 0, each entrepreneur privately knows θ ∈ {f, s}, whereas banks know the prior α.
• Period-1 Returns and Continuation Requirements:
  • Fast (Good) Project (θ = f): Requires I₁ = 1 at t = 0 and finishes at t = 1, yielding verifiable monetary return R_f > 1 and private entrepreneurial benefit B_f > 0.
  • Slow (Bad) Project (θ = s): Yields zero return at the end of Period 1 (R₁ = 0). At t = 1, if liquidated, the slow project yields 0 monetary return and inflicts a negative private payoff B_l < 0 on the entrepreneur; if refinanced (continued) into Period 2 with a second unit of capital (I₂ = 1), it completes at t = 2 with verifiable monetary return R_s and positive private benefit B_s > 0.

The Core Sunk-Cost Inequality (Ex-Ante Unprofitable vs. Ex-Post Profitable)

The central parameter condition of the Dewatripont–Maskin model is that slow projects are socially inefficient ex ante (at t = 0 before I₁ = 1 is spent), but profitable to complete ex post (at t = 1 once I₁ = 1 is sunk):

Measured from t = 0, a slow project requires 2 units of capital (I₁ + I₂ = 2) and generates total return R_s + B_s < 2 (negative net present value). If a slow entrepreneur knows the project will be liquidated at t = 1, the entrepreneur receives B_l < 0 and is deterred from applying for a loan at t = 0; if the entrepreneur anticipates refinancing at t = 1, the entrepreneur receives B_s > 0 and submits the bad project at t = 0.

Why Centralized Credit Cannot Commit to Liquidation

Suppose credit is centralized: a single large bank (or state treasury / conglomerate headquarters) has at least 2 units of capital per borrower and finances projects at t = 0. Suppose a slow project is funded with I₁ = 1 at t = 0 and reaches t = 1 with zero Period-1 output (R₁ = 0). Even if the initial contract signed at t = 0 contained a clause stating that any project yielding 0 at t = 1 will be liquidated, what happens when t = 1 arrives?

• At t = 1, the initial loan I₁ = 1 is a sunk cost—whether the bank liquidates or refinances, I₁ = 1 cannot be recovered.
• If the centralized bank liquidates the project at t = 1, its incremental Period-2 payoff is 0 (for a cumulative two-period loss of −1).
• If instead the centralized bank renegotiates the contract and refinances the project with an extra I₂ = 1 at t = 1 (taking 100% of the verifiable Period-2 return R_s, which leaves the entrepreneur with B_s > 0 > B_l so both parties prefer continuation), the bank's incremental Period-2 net profit is:

(reducing the bank's cumulative two-period loss from −1 to R_s − 2 > −1). Because R_s > 1, refinancing at t = 1 is a Pareto improvement ex post for the bank and the entrepreneur. Consequently, in any subgame-perfect equilibrium (where incomplete contracts can be renegotiated), the centralized bank cannot credibly commit to liquidating slow projects at t = 1.

Equilibrium Under Centralized Credit (The SBC Trap)

Because bad entrepreneurs (θ = s) rationally anticipate at t = 0 that the centralized bank will bail them out at t = 1 (giving them B_s > 0), all bad entrepreneurs apply for loans alongside good entrepreneurs at t = 0. The centralized bank's expected profit per borrower across the pooled population is:

Here R_s − 2 < 0 is a deadweight social loss on every bad project. Moreover, if the share of good projects α is below the threshold α^SBC ≡ (2 − R_s) / (R_f − R_s + 1), then Π_centralized < 0: the soft budget constraint problem causes the centralized banking system either to suffer aggregate losses or to shut down lending altogether.

Period-1 Bank Monitoring and Asymmetric Information Across Banks

Dewatripont and Maskin (1995) showed that credit decentralization—dividing capital across small, liquidity-constrained banks so that refinancing a slow project at t = 1 requires bringing in a new outside creditor—naturally creates an asymmetric information friction at t = 1 that blocks refinancing and hardens the budget constraint. Extend the slow-project specification to incorporate unobserved Period-1 bank monitoring effort e ∈ [0, 1]:

• During Period 1, the initial bank (Bank 1) that funded the project at t = 0 can exert non-verifiable monitoring effort e ∈ [0, 1] at private cost ψ(e) (with ψ(0) = 0, ψ′(0) = 0, ψ′(e) > 0, ψ″(e) > 0).
• If the project turns out to be a slow project (θ = s) and is refinanced at t = 1 with a second unit of capital I₂ = 1, its expected Period-2 verifiable monetary return is an increasing function of Bank 1's Period-1 monitoring effort, R_s(e):
  • With high monitoring (e = e^*), expected Period-2 return exceeds the cost of monitoring plus Period-2 capital: 1 < R_s(e^*) − ψ(e^*) < 2.
  • With zero or low monitoring (e = 0), expected Period-2 return falls below the Period-2 capital requirement: R_s(0) < 1.

Now compare the centralized and decentralized credit regimes:

1. Under Centralized Credit (Single Bank Provides Both I₁ = 1 and I₂ = 1): Because the single large bank finances both Period 1 and Period 2 internally and captures 100% of the Period-2 return R_s(e), the centralized bank has full incentives to exert the optimal monitoring effort e^* in Period 1. Because R_s(e^*) > 1, the centralized bank always refinances slow projects at t = 1, locking the economy into a soft budget constraint.
2. Under Decentralized Credit (Bank 1 Has Only I₁ = 1 and Must Recruit External Bank 2 at t = 1): Suppose each bank holds only 1 unit of capital. When a slow project fails to pay off at t = 1, Bank 1 has exhausted its capital and must ask an outside creditor (Bank 2) to supply the second unit of capital I₂ = 1.
   • Bank 2 cannot observe the monitoring effort e that Bank 1 exerted during Period 1.
   • To break even on its I₂ = 1 outlay at t = 1, Bank 2 must be promised a senior repayment share s₂ = 1 / R_s(e) of the Period-2 return, leaving Bank 1 with only the residual fraction (1 − s₂)R_s(e) = R_s(e) − 1.
   • Because Bank 1 must surrender a large share s₂ of the Period-2 return to outside Bank 2 (and because Bank 2's contract offer at t = 1 cannot condition on Bank 1's unobserved sunk effort e), Bank 1's marginal return to monitoring effort in Period 1 is diluted (the classic Holmström moral hazard in teams / Myers debt overhang effect).
   • When this dilution is strong enough, Bank 1 chooses under-monitoring e_{decentralized} < e^* such that R_s(e_{decentralized}) < 1. Rationally anticipating that Bank 1 under-monitored the project in Period 1, outside Bank 2 refuses to provide refinancing I₂ = 1, forcing the slow project into liquidation at t = 1.
   • Because bad entrepreneurs (θ = s) know at t = 0 that a decentralized banking system will liquidate slow projects at t = 1 (inflicting private loss B_l < 0), bad entrepreneurs never apply for loans at t = 0—achieving the Hard Budget Constraint equilibrium.

Applications: Banking Architecture, Conglomerates, and Fiscal Federalism

• Arm's-Length Markets vs. Bank-Centered Systems: Dispersed Anglo-American capital markets impose hard budget constraints that rapidly terminate unprofitable startups, but can exhibit short-termism ("myopia") when slow projects actually have high long-term value (R_s(e^*) − ψ(e^*) > 2, such as long-gestation basic R&D). Bank-centered systems excel at patient long-term investment when R_s > 2, but are vulnerable to zombie lending and soft budget constraints when R_s ∈ (1, 2).
• Corporate Conglomerates vs. Standalone Firms: Within a diversified corporate conglomerate, headquarters operates an internal capital market (centralized credit), pooling cash flows across divisions and cross-subsidizing weak divisions (R_s > 1) at the expense of strong divisions (Scharfstein and Stein, 2000), whereas spinning off a division as a standalone firm hardens its budget constraint.
• Fiscal Federalism (Qian and Roland, 1998): Yingyi Qian and Gérard Roland (1998) showed that when subnational governments compete for mobile capital and lack access to central bank money creation, their fiscal budget constraints harden, forcing them to liquidate loss-making local state enterprises.

Why is a soft budget constraint a dynamic time-inconsistency problem rather than just a poorly written contract?

At t = 0 (before the initial loan I₁ = 1 is sunk), the creditor prefers to commit to liquidating any slow project at t = 1, because doing so deters bad entrepreneurs from applying and saves 2 − R_s > 0 per bad borrower. However, once t = 1 arrives and I₁ = 1 has already been spent, I₁ is a sunk cost. Comparing the forward-looking payoff of refinancing (R_s − 1 > 0) against the forward-looking payoff of liquidation (0), both the creditor and the entrepreneur benefit from renegotiating the liquidation clause and refinancing the project. Because courts allow mutually consenting parties to renegotiate their contracts, any ex-ante promise by a deep-pocketed creditor to liquidate is time-inconsistent.

How does decentralizing credit among small banks solve the soft budget constraint?

When the initial bank (Bank 1) has only 1 unit of capital, it cannot refinance the second-period cost I₂ = 1 on its own and must recruit an external co-financier (Bank 2) at t = 1. Because Bank 2 must be promised a senior slice of Period-2 returns R_s(e) to cover I₂ = 1, and because Bank 2 cannot observe how hard Bank 1 monitored the borrower during Period 1, Bank 1's incentive to monitor in Period 1 collapses. Knowing that Bank 1 under-monitored the borrower in Period 1, Bank 2 refuses to lend at t = 1. Thus, decentralization uses an ex-post inter-bank information and free-rider friction to make refinancing unprofitable at t = 1, which credibly commits the banking system to a hard budget constraint at t = 0.

Is a hard budget constraint always socially superior to a soft budget constraint in the Dewatripont–Maskin model?

No. Decentralized credit carries a mirror-image pathology called short-termism (myopia): if an economy has valuable long-horizon projects whose Period-2 return exceeds total two-period capital costs (R_s(e^*) − ψ(e^*) > 2) despite yielding zero cash flow in Period 1, a decentralized credit market may fail to refinance them at t = 1 due to inter-bank asymmetric information.

1. Dewatripont, M., & Maskin, E. (1995). "Credit and Efficiency in Centralized and Decentralized Economies." Review of Economic Studies, 62(4), 541–555. https://doi.org/10.2307/2298076
2. Kornai, J. (1986). "The Soft Budget Constraint." Kyklos, 39(1), 3–30. https://doi.org/10.1111/j.1467-6435.1986.tb01252.x
3. Kornai, J., Maskin, E., & Roland, G. (2003). "Understanding the Soft Budget Constraint." Journal of Economic Literature, 41(4), 1095–1136. https://doi.org/10.1257/002205103771799999
4. Qian, Y., & Roland, G. (1998). "Federalism and the Soft Budget Constraint." American Economic Review, 88(5), 1143–1162. https://www.jstor.org/stable/116864
5. Scharfstein, D. S., & Stein, J. C. (2000). "The Dark Side of Internal Capital Markets: Divisional Rent-Seeking and Inefficient Investment." Journal of Finance, 55(6), 2537–2564. https://doi.org/10.1111/0022-1082.00299