The Domestic Cost: Excess Capacity, Crowding Out, and Slower Growth
The bank–firm relationship explains why a zombie can remain active; this chapter follows what that survival does inside the economy. The focus is the interface between a zombie firm and the firms, workers, prices, and investment decisions around it. By separating the firm’s own performance from sector congestion and national outcomes, we can see why a low-price strategy may coexist with lower productivity and weaker growth rather than treating the zombie as an isolated bad borrower.
The zombie firm can keep selling while producing less value
The firm-level evidence begins with a comparison that needs careful wording. Conditional on firm and country–year fixed effects and lagged controls, zombie status is associated with a 2.77 percentage-point lower return on assets, about 5 percentage points lower sales growth, a roughly 1 percentage-point lower capital-expenditure ratio, and about 3 percentage points lower employment growth. These are not claims that every zombie has the same performance or that the classification mechanically causes each outcome. They are within-firm and controlled comparisons that point in the same direction: the operating weakness is visible in profits, expansion, investment, and hiring.
The productivity results reinforce the picture. A transition into zombie status is associated with about 8.5% lower sales per employee and about 14% lower value added per employee. The measures are revenue-based, so nominal prices and underpricing can affect them; they should not be read as a perfectly clean engineering measure of physical productivity. Even with that qualification, the combination of lower asset turnover, lower margins, and lower markups is inconsistent with a simple story of highly efficient firms choosing to compete on price.
The apparent contradiction is that the raw summary statistics show slightly higher average sales growth for zombies than for non-zombies—about 0.39 versus 0.35. Once leverage, size, assets, and fixed effects are taken into account, the conditional relationship is negative. That reversal is economically meaningful. A large, heavily leveraged company can continue to move a considerable volume of goods while generating too little return on each dollar of assets. Headline sales then measure activity, not necessarily value creation.
The financing condition makes that strategy sustainable for longer than the firm’s operating results would otherwise allow. Zombies pay an average interest rate of about 5.4%, compared with 6.0% for non-zombies. Because zombies also have higher leverage, a roughly 60-basis-point difference can materially reduce the cash flow needed to keep debt service going. The study describes this gap as a below-market borrowing cost associated with bank forbearance, not as a direct fiscal subsidy. Without the financing wedge, a volume-over-margin strategy would deepen losses and make exit more likely.
Excess capacity is the bridge from firm performance to prices. A firm that keeps low-return capacity running has an incentive to generate cash flow, even if doing so requires accepting a lower markup. If many firms in a sector behave this way, prices can be held below the level that would prevail after unproductive capacity exits. That is disinflationary, but it is not necessarily welfare-improving: the lower price can reflect resources that are being used inefficiently and competitors that are being prevented from expanding.
The diagram is a system rather than a list of symptoms. The financing wedge sustains the zombie state; the state changes the firm’s pricing and resource use; those changes cross an interface into healthy firms and into aggregate prices and growth. The data support the direction of these relationships, while the precise contribution of each interface—credit, labor, market share, or pricing—cannot be separated completely in the sector estimates.
Sector congestion harms firms that are not zombies
The next interface is between a zombie firm and its healthy competitors. The study measures zombie prevalence as a lagged share of total assets within a country–sector and asks how non-zombie firms perform when that share is higher. In the main specification, a higher zombie share is associated with lower ROA and lower sales growth among non-zombies. The coefficients are −1.072 for ROA and −0.168 for sales growth. When the sample is restricted to firms that were never zombies, the coefficients remain negative at −0.757 and −0.107.
The magnitudes are easiest to interpret as changes in the sector’s composition. A 10 percentage-point increase in the zombie asset share corresponds to approximately 0.11 percentage points lower non-zombie ROA and 1.68 percentage points lower sales growth in the main estimates. The calculation does not mean that the same number must appear in every sector. It translates a coefficient on a share measured from zero to one into a more intelligible change. The persistent negative sign in the never-zombie sample is the important result: the association is not explained only by firms that move back and forth between the two categories.
Several forms of congestion could produce this pattern. A zombie may retain a bank relationship that would otherwise finance a healthier borrower. Its continuing demand for workers can hold labor in a low-productivity use, even if employment growth at the zombie itself is weak. Its lower markup can pressure competitors’ prices and reduce the revenue available for investment. Its market presence can also make entry less attractive to younger firms. The study estimates the combined sector effect rather than assigning a separate coefficient to each route, so these should be understood as connected mechanisms, not independently measured contributions.
This is why “let the weak firm continue because it still produces something” is not a complete evaluation. Output from the zombie is visible, while the investment, hiring, and market share that a healthy competitor does not obtain are counterfactual and less visible. A policy that preserves a job at an existing firm may also delay the expansion of another firm or prevent assets from moving to a more productive use. The relevant comparison is not simply operating versus closed. It is the observed operating state against the reallocation that could occur after restructuring, sale, or exit.
The result also clarifies the role of banks in the domestic system. Bank forbearance does not merely keep a single borrower’s debt on the books. It changes the selection environment for an entire country–sector. If lenders reduce credit to never-zombie firms as their own NPL ratios rise but continue exposure to always-zombie firms, the sector receives a signal that is opposite to productive opportunity. The healthy firm is asked to prove its creditworthiness in a tightening market, while the zombie receives time because its lender is reluctant to recognize the past loss.
From sector congestion to national disinflation and weaker growth
The sector-level mechanisms aggregate through two related paths. First, lower profitability, weaker sales growth, reduced investment, and slower hiring at both zombie and healthy firms reduce the expansion of productive capacity. Second, the excess capacity that remains in operation puts downward pressure on selling prices as firms seek cash flow and market share. The first path lowers output growth; the second lowers inflation. They can occur together because lower prices are not the same as higher productivity.
The country-level estimates are consistent with this prediction. A one-percentage-point increase in the lagged zombie share measured as a proportion of total assets is associated with approximately 0.04 percentage points lower CPI inflation and approximately 0.04 percentage points lower real GDP growth. In plain terms, moving a country’s asset-weighted zombie share from, for example, 10% to 11% is associated with a four-hundredths-of-a-percentage-point reduction in each outcome in the reported specification. The abstract’s estimate of roughly 0.2 percentage points per year reflects the larger changes observed across several economies and the specification used; it is not a replacement for the coefficient’s denominator.
The country results are meaningful but should not be treated as a universal structural formula. The inflation regressions contain 108 observations and the baseline GDP-growth regressions 124 observations. They use lagged zombie shares, country and year fixed effects, and controls such as unemployment, commodity prices, investment, credit, population growth, and lagged outcomes. That design helps compare changes within countries while accounting for common time conditions, but it cannot give the same level of detail as the firm panel or establish that every national price movement comes from zombie capacity.
The direction is also economically coherent with the firm and sector evidence. If zombie firms have lower returns and markups, and healthy firms in zombie-congested sectors perform worse, then a higher aggregate zombie share should weaken the supply of productive growth. If the distressed capacity remains active and prices its output aggressively, the economy can experience disinflation at the same time. The key is that the two effects have different welfare meanings: a lower price caused by an efficient process can accompany higher real income, while a lower price caused by excess capacity can coexist with lost investment and lower output.
Employment makes the trade-off more concrete. Zombie firms accounted for roughly 6–9% of reported employment in the ten Asian economies in 2015–2020, with about 2.5–3.2 million workers in zombie firms in 2016–2017. That employment share was below the asset share, which was around 11–13% in the same broad period. Zombies were therefore more capital-intensive than labor-intensive, but their workforce was still large enough to make sudden exit socially costly. The case for restructuring is not the case for ignoring those workers. It is the case for distinguishing temporary support and going-concern preservation from indefinite maintenance of a low-productivity operating state.
The same distinction applies to capital. A firm can preserve machines, buildings, and supplier relationships while destroying the returns those assets could generate elsewhere. An orderly process may keep a viable business together or transfer a business as a going concern; a poorly managed process may liquidate valuable capacity too early. Conversely, delay can allow assets to remain tied to a firm whose operating model is not recovering. The domestic problem is therefore an allocation problem with distributional consequences, not a simple contest between compassion and discipline.
What the domestic evidence establishes
Taken together, the firm, sector, and country results describe a connected pattern. Zombie firms underperform after controls and operate with lower margins and productivity measures. Their presence is associated with poorer outcomes for non-zombie firms in the same country–sector. At the country level, a higher asset-weighted zombie share is associated with lower inflation and lower growth. The financing wedge identified in the bank evidence gives the pattern a persistence mechanism: low-return capacity can remain active long enough to affect prices and competitors.
The pattern is strong enough to motivate the next question, but not so complete that it eliminates uncertainty. The country panel is small, sector coefficients do not decompose the congestion channels, and revenue-based productivity measures can incorporate price effects. The evidence also does not observe a universal counterfactual in which every zombie exits and every displaced healthy firm expands. What it does establish is why a zombie firm’s survival has system-level consequences: the firm’s low-return activity crosses interfaces into credit allocation, competitive conditions, employment, prices, and investment.
That local chain now provides a testable prediction beyond national borders. If excess capacity lowers the prices of goods that move through global value chains, an importing economy may receive lower input prices. But if those prices reflect distortion rather than productivity, the importer may also lose production and investment. The next step is to combine the domestic mechanism with trade exposure and ask where the international effect should appear first.