A Working Model of Zombification: From Loan Rollovers to Domestic Disinflation

The evidence now supports a connected judgment about zombie firms rather than a collection of separate findings. This checkpoint links the financial definition, persistent bank relationships, firm and sector damage, and domestic price and growth effects. The useful test is whether the resulting model can explain both sides of the paradox—lower prices and lower productive performance—and generate a disciplined prediction about economies that import from zombie-intensive sectors.

One local state, two connected loops

The first loop begins with measurement. A firm enters the study’s zombie category when EBIT does not cover interest expense for at least three consecutive years. The rule identifies sustained financial weakness; it does not by itself determine legal insolvency or prove that the firm has no recoverable business. The transition evidence gives the measure economic significance: from 2011 onward, roughly 93–95% of firms already classified as zombies remained in that state in the following year, while only about 3% of healthy firms entered it in a given year.

The second step is the financing relationship. The evidence finds that zombies are more dependent on bank liabilities than comparable non-zombies, and that zombie lending is persistent at the bank level. Persistence is stronger when asset quality is weak or capital is thinner. At the firm level, the response to worsening relationship-bank NPL ratios differs by borrower type: never-zombie firms lose debt and leverage, while always-zombie firms retain higher recorded debt and leverage. Relationship exposure and the number of banks are also associated with a greater probability of remaining a zombie.

These results do not turn the interest-coverage rule into a causal explanation. They connect a measured state to a mechanism that can sustain it. The operating weakness is observed in the firm’s coverage ratio. The bank incentive is supported by the persistence and differential credit responses. The coordination interpretation is consistent with the relationship and multi-creditor results, but the data do not observe every restructuring negotiation. Keeping those levels separate is essential: a coherent chain is not permission to treat every link as equally directly estimated.

flowchart LR D["Sustained interest-coverage distress"] --> B["Bank accommodation and creditor coordination"] B --> Z["Zombie state persists"] Z --> L["Low productivity, lower markups, excess capacity"] L --> C["Healthy-firm congestion"] C --> O["Lower investment and productive growth"] L --> P["Downward pressure on prices"] P --> I["Domestic disinflation"] W["Weak bank asset quality or thinner capital"] -. reinforces persistence .-> B

The diagram contains a reinforcing relationship, not a claim that every arrow has been estimated as an independent causal coefficient. The firm-level regressions support lower ROA, sales growth, investment, employment growth, productivity, margins, and markups for zombies. The sector regressions support poorer performance among non-zombies when the country–sector zombie asset share is higher. The country regressions associate a higher lagged zombie share with lower inflation and GDP growth. Together, these results make the system-level interpretation more credible than any single coefficient would be.

The two loops are financial and real-economy loops.