Does Industrial Policy Necessarily Lead to Soft Budget Constraints

There is indeed a strong intrinsic link between industrial policy and soft budget constraints, but saying "almost always" is too absolute—whether SBC arises depends more on the specific form of the policy and the institutional environment, rather than on the category of industrial policy itself.

The mechanism by which industrial policy tends to induce SBC is clear. The core issue Kornai focuses on is time inconsistency: the government can declare ex ante that it "will not bail out losers," but ex post, faced with unemployment, local fiscal pressure, sunk costs, and political lobbying, the commitment is hard to sustain. Once the government labels an industry as "strategic," beneficiary firms gain an implicit guarantee—both the market and the firms themselves make decisions based on the expectation that "they will be bailed out if things go wrong." Coupled with information asymmetry (the government finds it difficult to distinguish genuinely promising projects from rent-seeking ones disguised as such) and the formation of vested interest groups, softening is almost endogenous.

But it is important to distinguish different forms of industrial policy:

R&D subsidies, government procurement, infrastructure, talent training, and standard-setting do not directly link to firms' operating losses, so the SBC pressure is relatively small. What truly creates SBC are ongoing operational subsidies, directed credit, state-owned equity injections, and loss compensation—tools that directly tie a firm's financial status to the government's purse.

The selection mechanism is also crucial. If the policy is "wide net + market elimination" (subsidizing a large number of firms, letting them compete, and allowing losers to go bankrupt), SBC is weakened; if it is "picking champions" (betting on a few existing large firms), SBC is almost inevitable.

Whether the exit mechanism is credible is the third variable. Sunset clauses, performance thresholds, and clear graduation paths can make "hard" constraints more credible.

Empirical differences also support this differentiated judgment. During Park Chung-hee's era in South Korea, the government allowed large firms like Hanbo Steel to go bankrupt, showing that hard constraints can be maintained under a strong government. China's photovoltaic industry experienced large-scale bankruptcies of firms like Suntech and LDK, and the new energy vehicle industry also saw significant shakeouts—these are not typical SBC scenarios; but in the same period, local governments' implicit guarantees for local state-owned enterprises and "star firms" created a large number of zombie enterprises. Japan's MITI had partial success in the 1960s-70s, but the bank-firm-government triangle after the 1990s is a textbook case of SBC. The U.S. auto and financial bailouts in 2008 are a purer manifestation of SBC.

Thus, a more accurate statement is: industrial policy has a strong tendency toward SBC, especially when it takes the form of ongoing operational support for specific existing firms; but through competitive selection, time limits, performance thresholds, and credible exit commitments, this tendency can be significantly suppressed—it is not fate. Complete immunity requires quite strong fiscal discipline and institutional constraints, which are not easily met in most countries—so your intuition that "almost always" is actually close to reality in a statistical sense, but the mechanism should not be understood as inevitable.

The issue you raise is where this concept is most abused, and it is also the direction Kornai himself repeatedly revised in his later years. Let me state the conclusion directly: a purely hard budget constraint does not exist in any real economy. The binary usage of SBC as a "deviation" and HBC as a "benchmark" indeed implies an unrealistic idealized reference system.

Going back to Kornai's original definition makes it clearer. What he emphasized is not the ex post fact of "whether losses are bailed out," but the ex ante expectation structure: whether economic agents systematically expect to be bailed out when they incur losses, and accordingly change their investment, risk, and cost-control behavior. So SBC is essentially a concept about behavior and expectation, not about fiscal flows. Grasping this resolves much confusion.

Now, regarding the comparison you mentioned. In market economies, "softening" mechanisms are everywhere:

• Limited liability itself is a form of risk softening (shareholders' maximum loss = their investment)
• Bankruptcy procedures involve negotiated restructuring, not immediate liquidation
• Banks extend debt maturities and reduce interest rates for distressed borrowers
• Parent companies inject capital into subsidiaries and provide related-party transactions
• Insurance mechanisms socialize individual shocks
• Central banks' lender-of-last-resort (LLR) function systematically softens financial institutions' constraints
• The 2008 U.S. bailout was of private financial institutions

If the standard for "hard constraint" is "cash flow gap = immediate liquidation," then no modern economy meets this standard, not even the laissez-faire period of the 19th century. So mapping SBC vs HBC directly to "distortion vs. ideal market" is indeed a conceptual slippage.

So how should this concept be used? I think a more explanatory approach is to transform it from a binary judgment into a comparative diagnosis along several dimensions:

First, the source of softening. Is the bailout decision made by market participants who bear the risk (shareholders, creditors, insurers) at their own expense, or is it underwritten by a third party (government, parent organization) that does not bear direct economic consequences? The former is part of risk pricing; the latter is Kornai's "paternalism."

Second, the conditionality of softening. Market-based bailouts usually come with strong conditions: debt-to-equity swaps, management changes, asset sales, salary caps, and priority claims on future cash flows. Political bailouts are often unconditional or have weak conditions because the rescuer's goal is not to recover value but to avoid some political cost.

Third, systemic vs. occasional softening. Markets occasionally bail out, but each bailout is individually priced and involves bargaining; the real problem of SBC arises when a certain type of agent forms a stable expectation that "I belong to the category that will be rescued," thus relaxing constraints ex ante. In other words, SBC is not "being rescued" but "being expected to be rescued."

Fourth, who bears the cost. In market-based softening, losses are absorbed by parties who were informed at the time of contracting and received corresponding returns (shareholders, creditors, insurance recipients); in political softening, losses are often shifted to taxpayers, depositors, or future generations—third parties without ex ante pricing or consent.

Under this framework, the question is not "is there SBC?" but "in what form does softening exist, who pays for it, and under what conditions?" Some "softening" is socially desirable: LLR during financial crises, patient capital for innovation, bankruptcy reorganization rather than liquidation—these all deviate from pure HBC logic, but they address market failures (liquidity runs, externalities, insufficient long-termism) rather than creating them.

Thus, the truly useful application of this concept is to identify specific types of distortion: when certain firms (typically state-owned enterprises, local government financing platforms, politically sensitive "too big to fail" institutions, or entities designated as strategic) can systematically expect unconditional, third-party-funded bailouts, their ex ante behavior deviates from what market allocation would look like—excessive debt, overinvestment, insensitivity to costs, and insensitivity to risk. This is a concept used with a clear causal mechanism, not a vague claim that "government intervention = inefficiency."

Conversely, if an analyst uses SBC to criticize all non-pure-liquidation treatments—for example, calling bankruptcy reorganization itself SBC, central bank market intervention SBC, or social security SBC—that is the conceptual slippage you suspect, and it indeed hides an unrealistic ideal market benchmark.

In short: SBC should not be used as a label for "deviation from ideal competitive markets," but as a diagnostic tool to identify "systematically distorted expectations and unconditional externalization of losses." The former is ideology; the latter is analysis.