## 2.1 Data, identification, and sample ## Source details **Canonical URL:** [2.1 Data, identification, and sample](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023192-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023192-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023192-print-pdf.pdf.json) --- ### Firm-level data and sample - Quarterly balance sheet data from S&P Compustat North America and Compustat Global on nonfinancial listed firms for 49 countries (23 EMs and 26 AEs) over 2000q1-2019q4. - Financial firms excluded: banks, diversified financials, and insurance firms. - Nominal variables deflated with the respective country CPI deflator. - Final sample: an unbalanced panel of unique 24,371 nonfinancial firms over 2000q1-2019q4, resulting in 812,627 firm-quarter observations. - Main firm-level outcome variables: - Investment growth: log percentage change in the net capital stock. - Employment growth: year-on-year log percentage change in the total number of employees. - Debt growth: log percentage change in total debt (short and long). - Cost of debt: implicit interest rate = four-quarter rolling sum of interest expenses divided by total debt. ### Zombie firm definition and stylized facts - A firm is a zombie if all three conditions hold for at least two consecutive years: 1. Interest coverage ratio (ICR) is below one. 2. Leverage ratio is above the median firm in the same industry. 3. Real sales growth is negative. - Exit from zombie status allowed only once at least one indicator is reversed for two years in a row. - Stylized facts: - Typical listed zombie firm unconditional mean age: 33 years. - 8% of zombie firms are younger than ten years. - Less than 2% of zombie firms are younger than five years. ### Monetary policy shocks and instrument construction - Identification: US monetary policy surprises using high-frequency identification on three-month ahead Fed funds futures in a 30-minute window surrounding FOMC meetings (10 min before and 20 min after). - Aggregation to quarterly frequency: sum of all daily surprises within the respective quarter. - Monetary policy indicator used in estimation: country-specific one-year sovereign bond yield rate. - Robustness: alternative US shock constructions that control for central bank information shocks are used in Section 7. ### Country-level data and policy indicators - Bank indicators: Non-performing loans (NPLs) and capital ratios from IMF Financial Soundness Indicators; regulatory capital buffers computed as actual regulatory capital to risk-weighted assets minus minimum required ratio. - Macroprudential policies: IMF iMaPP database; instruments assigned +1 for tightening, −1 for loosening, 0 for no change; cumulative indices constructed from 1990 onward and grouped into seven categories. - Insolvency frameworks: crisis preparedness indicator from Araujo et al. (2022), time-invariant snapshot for 2020-21, matched to 39 countries. --- ### 3.2 First-stage regression and empirical strategy ### First-stage specification - First-stage equation: - R_{c,t} = α_c + δ_c S_t + Γ'_c X_{c,t−1} + u_{c,t} - R_{c,t} denotes the one-year government bond yield in country c at time t. - S_t denotes the US monetary policy surprises, aggregated to each quarter t by summing up all daily surprises within the respective quarter. - Controls X_{c,t−1} include four lags of: real GDP growth, consumer price inflation, current account balance as a percentage of GDP, the real effective exchange rate, and US GDP growth. - Country-by-country estimation: first-stage run separately for each country; predicted values R̂_{c,t} collected as identified country-specific monetary policy shocks. - Average F-statistic across countries in first-stage specifications: of around 12. ### Identification approach and LP-IV estimation - Use US monetary policy surprises as external instrument for country-specific one-year sovereign bond yields in a panel Local Projection Instrumental Variable (LP-IV) framework. - LP-IV advantage: allows impulse response estimation where surprises need only be contemporaneously exogenous to the monetary policy indicator and normalizes shocks in units of the endogenous monetary indicator (one-year yield). - Transmission channels emphasized: - Financial channel: exogenous impact of US policy on foreign interest rates drives local credit conditions, risk premia, and portfolio rebalancing. - Exchange rate and trade channels are acknowledged but financial channel is the focus (exchange rate channel more relevant for firms with foreign-currency debt; trade channel for exporters). --- ### Main empirical findings: monetary tightening, zombies, and credit ### Aggregate and heterogeneous effects calibrated to a 100 bps shock - A monetary policy shock calibrated to increase country-specific one-year sovereign bond yields by 100 bps is used for impulse-response analyses. - Differential effect on zombie versus nonzombie firm investment gap: - A 100 bps increase in the one-year yield leads the gap to shrink by roughly 25 percent. ### Differential credit and cost outcomes for zombies - Empirical regression results for zombie firms relative to nonzombies: - Debt growth falls by less for zombie firms by over 0.6 p.p. relative to nonzombies following the 100 bps tightening. - Cost of debt rises by less for zombie firms by roughly 0.25 p.p. relative to nonzombies. - Interpretation: credit standards tighten by less for zombies following a monetary policy shock, consistent with banks evergreening existing loans to zombies (zombie lending channel). ### Comparison with conventional financially constrained metrics - Financially constrained proxies: age, total assets, leverage, net liquid assets, 12-month PD. - Main findings: - Firms with limited net liquid assets, high probability of default, smaller, or younger tend to be hardest hit by higher interest rates. - High-leverage firms (debt-to-asset ratio proxy) seem to react less to monetary policy shocks, but estimates are imprecise. - Distressed nonzombies (ICR < 1 and leverage above median for two years, excluding profitability) show investment growth more responsive to monetary tightening, suggesting persistently unprofitable zombies are fundamentally different from conventional distressed firms. - Profitability (sales growth) is key: firms with negative sales growth for two consecutive years respond significantly less to monetary policy shocks. --- ### Theoretical framework: evergreening model (overview and implications) ### Environment and mechanism - Model environment builds on Faria-e-Castro et al. (2022) with Stackelberg loan contract: lender sets debt price Q, firm chooses borrowing b′ and investment k′. - Firm state variables: existing liabilities b and productivity z. - Lenders heterogeneous in capital a; competitive access to saving at fixed interest rate R with discount factor β_k = 1/R. - Firms default at t = 0 iff V(z,b;Q) < 0. - Under binding borrowing constraints, minimum debt price Qmin(z,b) is derived and is increasing in b and decreasing in z. ### Lender pricing, relationship lending, and thresholds - Lender maximizes relationship value W(z,b,a) and defines Qmax(z,b,a) where W(...) = 0. - Lender’s optimal debt price Q*(z,b,a) characterized by comparisons of Qmin, βk, and Qmax. - Thresholds: - b̄(z): Qmin(z, b̄(z)) = βk (competitive lending threshold). - b̂(z,a): Qmin(z, b̂(z,a)) = Qmax(z, b̂(z,a), a) (evergreening indifference threshold). - Verified properties hold: b̄(z) < b̂(z,a); Q* increases in b in the evergreening region; Q* decreases in z in that region. ### Regions and implications - Three regions for existing debt b (for given z): - b < b̄(z): competitive lending (Q* = βk). - b̄(z) ≤ b ≤ b̂(z,a): evergreening region (lender offers more favorable debt price to prevent default). - b > b̂(z,a): default region (lender forgoes extending credit). - Mapping to empirical zombies: intermediate b and low z firms receive favorable lending terms (consistent with zombie firms receiving rollovers and softer credit conditions). --- ### 5.4 Bank capital, monetary transmission, and policy effectiveness ### Role of bank capital in evergreening - Higher bank capital reduces lenders’ incentives to evergreen: the maximum acceptable debt price Q_max decreases as bank capital increases, making lenders less willing to support high outstanding debt. - Weakly capitalized banks are more inclined to extend relationship lending to highly indebted firms because they seek to avoid recognizing losses. - Evergreening motives stronger for weakly capitalized banks, but the model allows zombie lending irrespective of bank capital. ### Monetary tightening mechanics in the model - Contractionary monetary policy modeled as raising the risk-free interest rate R from R1 to R2 (β_k1 to β_k2). - Q_min unaffected by β_k change; Q_max and β_k horizontal line shift downward; thresholds b̄(z) and b̂(z,a) fall. - Implications: - More firms (including some with lower debt) rely on relationship lending to survive under higher rates. - More heavily indebted firms opt to default at the higher rates. - Firms already in relationship lending face limited passthrough of higher rates (Q* = Q_min remains unchanged), while competitively financed firms face higher borrowing costs. - Investment of firms in relationship lending remains relatively stable despite tightening; investment of competitively financed firms decreases. ### Empirical evidence on policy buffers and insolvency regimes - Empirical specification includes interactions of zombie status, policy indicators, and identified shocks to measure βh2 (additional differential effect). - Findings: - Countries with higher regulatory capital buffers or lower NPLs tend to experience a decline in zombies’ investment growth relative to nonzombies, consistent with higher bank capital reducing evergreening. - Overall macroprudential policy index shows weak evidence of mitigating zombie lending; disaggregation shows Loan-targeted measures and Supply-loan measures (limits to credit growth, loan loss provisions, loan restrictions, limits to loan-to-deposit ratio) associated with lower zombie lending but with greater uncertainty. - Well-prepared insolvency regimes are associated with declines in zombie firms’ investment growth relative to nonzombies over the medium term (effects visible after three years). - Caution: results are not strictly causal but control choices and fixed effects mitigate confounding. ### Extensions and robustness - Results robust to alternative US shock specifications that control for central bank information effects (Jarociński and Karadi (2020); Bauer and Swanson (2023)). - US sample: differential responses of employment, debt, and interest rates imprecise, but zombie firms’ investment growth falls more relative to nonzombies. - Euro area sample: zombie firms’ financial performance less responsive to ECB shocks, consistent with greater prevalence of zombie lending in regions with less creditor-friendly bankruptcy laws. - Alternative zombie definitions largely preserve the baseline qualitative result (zombie investment less impacted by contractionary shock), though some definitions (subsidized interest rate–based) yield different outcomes and may misclassify zombies in low-rate environments. --- ### Policy implications and recommendations - Empirical finding: zombie firms are less responsive to monetary policy shocks than other firms because contractionary policy can strengthen banks’ incentives to evergreen zombie loans. - Policy recommendations: - Strengthen banks’ balance sheets (higher capital buffers) to reduce evergreening incentives. - Limit banks’ incentives for risky evergreening behavior via stronger prudential supervision and macroprudential measures targeted at loans. - Improve insolvency regimes to allow efficient resolution and restructuring of weak firms; insolvency improvements have delayed but meaningful effects. - Rationale: reducing zombie lending when financial conditions tighten helps prevent misallocation of credit that harms more productive firms and aggregate productivity. _The Zombie Lending Channel of Monetary Policy — Working Paper No. WP/2023/192_ ### 2.1 Data ### 2.1 Data ### Firm-level - Quarterly balance sheet data from S&P Compustat North America and Compustat Global on nonfinancial listed firms for 49 countries, 23 EMs and 26 AEs, over 2000-2019. - Financial firms excluded: banks, diversified financials, and insurance firms. - Nominal variables deflated with the respective country CPI deflator. - Final sample: unbalanced panel of unique 24,371 nonfinancial firms over 2000q1-2019q4, resulting in 812,627 firm-quarter observations. - Data cleaning and variable definitions: see Appendix A (cleaning assumptions and variable definitions described there). - Main firm-level outcome variables: - Investment growth: log percentage change in the net capital stock (capital expenditures in physical capital, property, plant, and equipment). - Employment growth: year-on-year log percentage change in the total number of employees. - Debt growth: log percentage change in total debt (short and long). - Cost of debt: implicit interest rate = four-quarter rolling sum of interest expenses divided by total debt. ### Monetary policy shocks - Identification strategy: US monetary policy surprises using high-frequency identification of unexpected changes in the Fed policy rate (Gürkaynak et al. 2005, Gertler and Karadi 2015, Nakamura and Steinsson 2018). - Construction of surprises: unexpected changes in three-month ahead contracts on Fed funds futures in a 30-minute window surrounding FOMC meetings (10 min before and 20 min after). - Transformation to quarterly frequency: sum of all daily surprises within the respective quarter. - Rationale and instrument validity note: - The simple sum of surprises within each quarter makes sure that the constructed monetary policy shocks, as an external instrument, are uncorrelated with the same type of shocks in other periods. This meets the strict condition for instrument validity when employing local projection methods and instruments to identify the dynamic effects of an imperfectly observable structural shock, as pointed out by Stock and Watson (2018). - Monetary policy indicator used in estimation: country-specific one-year sovereign bond yield rate (chosen to assess heterogeneous transmission of US monetary policy surprises to country-specific borrowing costs). - Robustness: Section 7 uses other US monetary policy surprises from the literature that also control for central bank information shocks (Jarociński and Karadi 2020, Bauer and Swanson 2023). ### Country-level - Merged with aggregated country-level data (Section 6) to study how country characteristics mitigate/amplify zombie lending after monetary policy tightening. - Bank indicators: - Non-performing loans (NPLs) and capital ratios from IMF’s Financial Soundness Indicators. - Regulatory capital buffers computed as: banks’ actual regulatory capital to risk-weighted assets minus minimum required risk-based regulatory capital ratio (from World Bank’s Bank Regulation and Supervision Survey). - Motivation: capital buffers better capture banks’ resilience across countries than raw capital ratios. - Macroprudential policies: - IMF’s integrated Macroprudential Policy (iMaPP) database, monthly, covering 182 jurisdictions over 1990-2020 (Alam et al. 2019). - Database covers 17 macroprudential instruments; for each instrument assigns +1 for tightening, −1 for loosening, 0 for no change, based on effective date. - Cumulative indices constructed by summing dummy indices for each instrument-country pair from 1990 onward (following Akinci and Olmstead-Rumsey (2018)); instruments grouped into seven main categories as in Alam et al. (2019). - Insolvency frameworks: - Crisis preparedness indicator from Araujo et al. (2022) measures countries’ preparedness to handle large-scale corporate restructuring; time-invariant snapshot for 2020-21. - Indicator matched to 39 countries in the firm-level dataset. ### Zombie firms - Context: zombification documented since Japan’s late-1980s and 1990s and found in OECD and selected European evidence (literature citations summarized). - Trends: Recent research shows share of zombie firms has been increasing over the last two decades, especially since the GFC. After a temporary downward trend from 2016 to 2019, the share picked up again since the Covid-19 pandemic. - Heterogeneity: zombification more prevalent in countries with looser macroprudential stance, less-prepared insolvency frameworks, weaker GDP growth, and higher share of small firms; more prevalent in nontradable industries (real estate, energy sector, information technology, materials, consumer discretionary). - Definition used (following Albuquerque and Iyer (2023)): - A firm is a zombie if all three conditions hold for at least two consecutive years: 1. Interest coverage ratio (ICR) is below one. 2. Leverage ratio is above the median firm in the same industry. 3. Real sales growth is negative. - Exit from zombie status allowed only once at least one indicator is reversed for two years in a row (two-year horizon for entry and exit to minimize misclassification from cyclical fluctuations). - Rationale: combines financial distress (ICR, leverage) with persistent lack of growth (negative real sales growth); applicable to listed and private firms. - Rejected alternative definition: subsidized interest rate definition (Caballero et al. 2008, Acharya et al. 2019) not used due to potential identification problems in low-interest-rate environments and ambiguity about relative interest rates charged to zombies versus high-rated firms. - Stylized facts in dataset: - Typical listed zombie firm unconditional mean age: 33 years. - Only 8% of zombie firms are younger than ten years. - Less than 2% of zombie firms are younger than five years. - Comparative characteristics (summary of reproduced Table 1 findings from Albuquerque and Iyer (2023)): - Zombies exhibit substantially weaker balance sheets within the same industry and country: lower TFP, lower profitability, lower investment, lower liquid assets, smaller size, higher default risk, greater dependence on bank loans. - Regression panels in Table 1 include country-industry-quarter fixed effects; standard errors clustered by country-industry-quarter; statistical significance markers *, **, *** denote 10%, 5%, and 1% levels respectively. ### Empirical strategy: LP-IV approach - Objective: estimate dynamic causal effect of monetary policy shocks on non-US nonfinancial firms’ financial performance, focusing on corporate investment and differential effects on zombie vs nonzombie firms. - Identification challenges and solutions: - Reverse causality: interest rates respond to the economy; solution is to use US monetary policy shocks as exogenous drivers of global conditions, identified via high-frequency surprises in three-month Fed Funds Futures within 30-minute FOMC windows (10 min before, 20 min after). - Need for instrument: use surprises as instruments for country-specific one-year sovereign bond yield in a panel Local Projection Instrumental Variable (LP-IV) setup. - Estimation approach: - Panel data LP-IV following Jordà et al. (2020) and Cloyne et al. (2023) to flexibly estimate impulse response functions using monetary policy surprises as instruments for the monetary policy indicator. - Advantage: surprises need only be contemporaneously exogenous to the monetary policy indicator; LP-IV imposes unit-effect normalization so shock sizes interpretable in units of the endogenous monetary indicator (the one-year yield). - Focus on transmission channels: - Traditional Mundell-Fleming channels: demand-reducing channel (contractionary US policy reduces US domestic demand for home and foreign goods) and expenditure-switching channel (currency depreciation makes exports more competitive). - Empirical literature finds demand-reducing channel typically predominates, leading to negative spillovers. - Financial channel emphasized: cross-border financial linkages transmit US monetary policy via co-movement in interest rates, risk-taking, and portfolio rebalancing; contractionary US policy raises foreign interest rates and lowers risky asset prices. - This paper focuses on the financial channel by isolating the exogenous impact of US monetary policy on foreign interest rates, then studying how tighter foreign monetary conditions affect firms’ financial performance. - Channel relevance to firms: - Financial channel important for all nonfinancial firms as changes in sovereign borrowing costs determine prevailing local credit conditions. - Exchange rate channel primarily affects firms with foreign-currency debt. - Trade channel primarily affects exporting firms. *Source: wpiea2023192-print-pdf (https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023192-print-pdf.pdf)* ### 3.2 First-stage regression ### 3.2 First-stage regression ### First-stage specification and identification - First-stage equation: - R_{c,t} = α_c + δ_c S_t + Γ'_c X_{c,t−1} + u_{c,t} - Definitions and choices: - R_{c,t} denotes the one-year government bond yield in country c at time t. - S_t denotes the US monetary policy surprises, aggregated to each quarter t by summing up all daily surprises within the respective quarter. - Controls X_{c,t−1} include four lags of: real GDP growth, consumer price inflation, current account balance as a percentage of GDP, the real effective exchange rate, and US GDP growth. - Country-by-country estimation: fully heterogeneous coefficients across countries by running the first-stage regression separately for each country. - Predicted values R̂_{c,t} are collected as the identified monetary policy shocks for each country. - Rationale for controls: - Current account balance captures the trade channel; real effective exchange rate captures the exchange rate channel. - Domestic and US GDP growth control for demand effects. - Instrument design and identification note: - Approach is similar to a Bartik-type instrument: common US shock with heterogeneous sensitivities across countries. Identification relies on the exogeneity of the monetary policy surprises. ### Choice of monetary policy indicator - Primary indicator: one-year government bond yield. - Motivations: - Fairly standard in literature studying monetary policy effects on the real economy (Gertler and Karadi 2015). - Captures forward rate guidance. - Longer-dated yields incorporate changes in risk premia (term premia and credit risk premia). - Limitations of shorter rates: - Shorter rates (central bank policy rates, one- or three-month rates) may not accurately capture how a country’s monetary conditions respond to US monetary policy. - Short-term rates in the rest of the world do not typically respond to US monetary policy news or may decline as central banks respond countercyclically. - Robustness: - Results are robust to using ten-year rates; one-year rates used due to larger sample size. - Using ten-year yields yields somewhat stronger effects; using three-month money market rates or treasury bill rates leads to weaker and short-lived effects and greater uncertainty. ### First-stage strength and related literature - Average F-statistic in the first-stage specifications across all countries: of around 12. - Literature context: - Heterogeneous transmission of US monetary policy to foreign interest rates (Kalemli-Özcan 2019, De Leo et al. 2023, Kearns et al. 2023). - Long-end yield responses reflect repricing of risk and portfolio rebalancing by investors. ### Notes and caveats - One-year rates capture forward guidance and longer-run risk premia but do not measure perfectly the price of new loans. - Implicit interest rate measure caveats (discussed later): interest expenses divided by stock of debt is slow-moving and backward-looking; may not accurately reflect marginal cost on new loans. ### 2.6 p.p.. A monetary policy shock calibrated to increase bond yields by 100 bps thus leads this ### 2.6 p.p.. A monetary policy shock calibrated to increase bond yields by 100 bps thus leads this ### Differential effects of monetary policy shocks on zombies versus nonzombies - A monetary policy shock that increases the country-specific one-year sovereign bond yields by 100 bps leads the gap to shrink by roughly 25 percent. - Results are not sensitive to using longer-dated yields (ten-year) as the monetary policy indicator. - Cumulative impulse responses show zombie firms’ investment and employment growth respond differently relative to nonzombies following the 100 bps tightening (confidence bands: 68 and 90 percent). ### Comparison with traditional financial-constraint metrics - Financially constrained firm proxies used: firm age, total assets (size), leverage (debt-to-asset ratio), net liquid assets, and 12-month probability of default (modified Merton distance-to-default from NUS-CRI). - Definition of financially constrained firms: time-varying dummy using median cut-offs of each proxy (e.g., net liquid asset ratio, total assets, age below median; 12-month probability of default or debt-to-asset ratio above median). - Main finding for investment response: - Firms with limited net liquid assets, high probability of default, smaller, or younger, tend to be hardest hit by higher interest rates (consistent with literature). - High-leverage firms (debt-to-asset ratio proxy) seem to react less to monetary policy shocks, but estimates are imprecise. - Leverage alone may not capture firm resilience and access to credit. - Distressed nonzombies (ICR below one and leverage above median for two years in a row, excluding the profitability indicator) have investment growth more responsive to monetary policy shocks, suggesting zombie behavior (persistently unprofitable) is fundamentally different from conventional financially constrained/distressed firms. - Profitability (proxied by sales growth) is key: firms with negative sales growth for two consecutive years respond significantly less to monetary policy shocks than other firms. ### Credit conditions, evergreening, and empirical patterns - Empirical regression (Equation (3)) finds for zombie firms relative to nonzombies: - Statistically significant and positive coefficient on debt growth. - Negative coefficient on the implicit interest rate. - Quantified relative responses: - Debt growth falls by less for zombie firms by over 0.6 p.p. relative to nonzombies. - Cost of debt rises by less for zombie firms by roughly 0.25 p.p. relative to nonzombies. - Interpretation: credit standards tighten by less for zombies following a monetary policy shock, consistent with banks engaging in evergreening—shifting lending to zombies so they receive more favorable credit conditions when financial conditions tighten (the zombie lending channel of monetary policy). - Robustness notes: - Similar results using ten-year rate as monetary policy indicator. - Different results with the three-month rate, underscoring the importance of local interest rates that properly account for transmission of US monetary policy. ### Theoretical framework: evergreening model (overview) - Model builds on Faria-e-Castro et al. (2022) with a Stackelberg loan contract: lender (leader) sets debt price Q, firm (follower) chooses borrowing b′ and investment k′. - Key mechanism: presence of existing debt b creates relationship lending incentives; lenders may roll over pre-existing debt of highly leveraged and low-productivity firms (evergreening) to avoid realization of losses from default. - Evergreening can arise as a normal feature of financial intermediation when relationship lending internalizes default externalities. ### 5.1 Environment - Agents: firms and lenders. - Firm state variables: existing liabilities b and productivity z. - Lenders: risk-neutral, heterogeneous in capital a, access to infinitely-elastic saving market at fixed interest rate R; discount factor βk = 1/R. - Two periods: t = 0 and t = 1. ### 5.2 Firm problem - Continuation value if firm repays and operates: V(z,b;Q) = max_{b′, k′ ≥ 0} [ −b − k′ + Q b′ + βf [ z (k′)^α − b′ ] ] s.t. b′ ≤ θ k′ - Firms default at t = 0 if and only if V(z,b;Q) < 0. - Under binding borrowing constraints (Q ≥ βf), firm borrows up to the constraint, yielding a minimum price of debt Qmin(z,b): Qmin(z,b) = βf + 1/θ − (βf α z)^{1/α} θ [ (1−α)/α b ]^{(1−α)/α} - Qmin(z,b) is increasing in b and decreasing in z: high-debt and low-productivity firms have higher Qmin and thus tend to default more. ### 5.3 Lender problem - Competitive lending: - If no prior relationship, lender offers Q = βk (competitive price) if firm will not default at that price. - Firm marginal product of capital under competitive lending: MPK = z α (k′)^{α−1} ≡ (1 − θ (βk − βf)) / βf - Relationship lending: - Lender internalizes default risk from existing debt and solves: W(z,b,a) = max_{Q ≥ βk} 1[V(z,b;Q) ≥ 0] × [ u(a+b) − Q b′(z;Q) + βk b′(z;Q) − u(a) ] - Define Qmax(z,b,a) s.t. W(z,b,a; Qmax) = 0: highest debt price the lender is willing to offer in a relationship. - Lender’s optimal debt price Q*(z,b,a): - Q*(z,b,a) = βk if Qmin(z,b) ≤ βk ≤ Qmax(z,b,a) - Q*(z,b,a) = Qmin(z,b) if βk ≤ Qmin(z,b) ≤ Qmax(z,b,a) - Q*(z,b,a) = 0 otherwise - Define thresholds: - b̄(z): threshold where Qmin(z, b̄(z)) = βk (lender willing to offer competitive price) - b̂(z,a): threshold where Qmin(z, b̂(z,a)) = Qmax(z, b̂(z,a), a) (indifference to evergreen) - Verified properties: 1. b̄(z) < b̂(z,a), ∀ z; 2. Q*(z,b,a) is increasing in b, strictly if b ∈ [ b̄(z), b̂(z,a) ]; 3. Q*(z,b,a) is decreasing in z, strictly if b ∈ [ b̄(z), b̂(z,a) ]. ### Model implications and intuition - Three regions by existing debt b (for given z): - b < b̄(z): competitive lending region (Q* = βk). - b̄(z) ≤ b ≤ b̂(z,a): evergreening region — lender offers more favorable debt price (Q* > βk) to prevent default and protect existing exposure. - b > b̂(z,a): default region — lender forgoes extending credit as Qmax < Qmin. - By productivity z and debt b: - High b and low z: default region (no lending relationship). - Low b or high z: competitive pricing region (Q = βk). - Intermediate b and z: relationship lending / evergreening region—these firms (relatively more indebted and less productive) receive favorable lending terms, aligning with the empirical definition of zombie firms. *Source: wpiea2023192-print-pdf* ### 5.4 The role of bank capital ### 5.4 The role of bank capital ### Bank capital and lending incentives - Bank capital determines whether a lender is inclined to maintain a relationship lending arrangement with a firm. - The level of bank capital directly influences the marginal benefit derived from recovering previous debt, implying a decrease in the maximum debt price Q_max as bank capital increases. - Lenders with higher levels of bank capital are less willing to support firms with high levels of outstanding debt. - Lenders with lower bank capital may be more inclined to extend relationship lending to firms with higher indebtedness because they may not have enough capital to absorb losses from firms defaulting. - Evergreening (zombie lending) motives may be stronger for weakly capitalized banks, but zombie lending occurs in the model irrespective of bank capital concerns: - Banks may gamble on zombie firms’ resurrection; zombies’ reputation grows with the length of their lending relationship (Hu and Varas 2021). - Rolling over bad loans allows zombie firms to remain in business and potentially access market financing in the future (Hu and Varas 2021). ### Relationship lending and monetary policy transmission - Simulation setup: - A contractionary monetary policy is modeled by raising the risk-free interest rate R from R1 to R2, equivalent to a decrease in lenders’ effective discount factor, shifting β_k1 to β_k2. - Mechanics and price thresholds: - The minimum debt price Q_min remains unaffected by a change in β_k. - Contractionary monetary policy causes a downward movement of the Q_max curve and the β_k horizontal line. - Thresholds fall: b̄(z) (threshold below which lender does not offer higher debt prices) and b̂(z,a) (threshold beyond which lender chooses not to evergreen) decrease. - Implications for firm outcomes: - More firms, particularly those with lower existing debt levels, rely on relationship lending to survive under higher interest rates. - More heavily indebted firms opt to default at the higher interest rate. - Firms already in relationship lending see their effective debt price Q* = Q_min remain unchanged (no passthrough of higher rates), provided the relationship existed prior to the policy change. - Firms relying on competitive lending rates experience higher interest rates because their Q_min is lower than the competitive loan price β_k2; at prevailing market interest rates these lenders offer the competitive loan price and firms do not default. - Effect on zombie firms: - Relationship lending tends to dampen the transmission of monetary policy to zombie firms’ investment. - Mapping higher interest rates to investment growth shows: - Investment of firms in relationship lending remains relatively stable despite tightening. - Firms relying on competitive lending rates exhibit decreased investment. - Model illustrates extreme cases of passthrough (no passthrough or full passthrough) to explain why high-debt and low-productivity firms (akin to empirical zombies) are less responsive to monetary policy shocks. ### Role of policies in mitigating the zombie lending channel of monetary policy - Empirical and theoretical background: - Higher interest rates can create stronger incentives for banks to offer better credit conditions to zombie firms to prevent default, lowering passthrough of higher rates to zombies and reducing credit supply declines to zombies relative to other firms. - Zombie firm survival can create negative spillovers (congestion effects) to nonzombies along intensive and extensive margins, reducing productivity growth (Caballero et al. 2008; McGowan et al. 2018; Acharya et al. 2019; Banerjee and Hofmann 2022; Albuquerque and Iyer 2023). - Policy channels examined: - Banks’ regulatory capital buffers (actual regulatory capital to risk-weighted assets minus the minimum required risk-based regulatory capital ratio). - Banks’ NPLs as a ratio of total loans. - Stance of macroprudential policies (Alam et al. 2019 index). - Quality of insolvency regimes (Araujo et al. 2022), time-invariant. - Empirical specification (investment growth as outcome): - ∆h Inv i,t+h = αh i + αh c,s,t + βh1 (Zom i,t−1 × R̂ c,t) + βh2 (Zom i,t−1 × Policy c,t−1 × R̂ c,t) + βh3 (Zom i,t−1 × Policy c,t−1) + Γ′h Z i,t−1 + e h i,t - Main coefficient of interest: βh2 measures additional differential effect of tighter monetary policy on zombies’ investment growth relative to nonzombies in countries with Policy above the median versus below. - Policy indicators are converted into binary variables (above/below median for each quarter). - Given positive βh1 from Section 4.2, a positive (negative) βh2 implies stronger (weaker) evergreening motives in countries with Policy above the median. - Empirical findings on policy effectiveness: - Countries with higher regulatory capital buffers or lower NPLs tend to experience a decline in zombies’ investment growth relative to nonzombies (Figure 10). - Similar results obtain when using continuous measures of capital buffers or NPLs (Figure B.8). - Interpretation: banks with higher bank capital (or lower NPLs) are more bound to write off zombie loans and absorb losses. - Overall macroprudential policy index does not show strong evidence of mitigating zombie lending (Figure B.9), possibly because the index aggregates diverse measures. - Disaggregating, macroprudential measures targeting loans (Loan – targeted) are associated with lower zombie lending. - Measures imposing limits to credit growth, loan loss provisions, loan restrictions, and limits to the loan-to-deposit ratio (Supply – loans) show some evidence of reducing zombie lending, though with larger parameter uncertainty. - Insolvency regimes: - Well-prepared insolvency regimes, including laws enhancing out-of-court restructuring, are associated with declines in zombie firms’ investment growth relative to nonzombies over the medium term (statistical evidence appears after three years) (Figure B.10). - Interpretation: insolvency frameworks matter but effects materialize with delay as restructuring takes time. - Caution on causality: - Results do not necessarily imply causal links from tighter policies to lower zombies’ investment growth, but inclusion of rich controls and country-industry-time fixed effects mitigates confounding factors. - Policy takeaway: - Policies that strengthen banks’ balance sheets, limit banks’ incentives to engage in risky behavior, and enable efficient resolution of weak firms can help mitigate zombie lending when financial conditions tighten. - This aligns with literature advocating stronger bank supervision and regulation to break evergreening incentives (Giannetti and Simonov 2013; Acharya et al. 2021; Schivardi et al. 2022; Blattner et al. 2023; Bonfim et al. 2023). ### Extensions and robustness checks (high-level results) - Alternative monetary shock identification: - Results robust to shocks that control for central bank information effects using Jarociński and Karadi (2020) (“poor man’s sign restrictions”) and Bauer and Swanson (2023) orthogonalized shocks (Figures B.11, B.12). - US sample: - Using US one-year government bond yield in first stage and US domestic controls, differential responses of employment, debt, and interest rates are imprecise, but zombie firms’ investment growth falls more relative to nonzombies (Figure B.13). - Interpretation: stronger effect in US possibly reflects higher regulatory capital buffers, tighter macroprudential measures on loans, and well-developed insolvency regimes in the US; zombie lending less prominent in US due to efficient bankruptcy resolution (Favara et al. 2022). - Euro area sample: - Using ECB high-frequency shocks (Jarociński and Karadi 2020), zombie firms’ financial performance in the euro area is less responsive to ECB monetary policy shocks (Figure B.14), consistent with more prevalent zombie lending in the euro area due to less creditor-friendly bankruptcy laws and greater bank reliance for firm financing. - Alternative zombie definitions: - Banerjee and Hofmann (2022) definition (ICR < 1 and Tobin’s q below industry median over two years): baseline result (zombie investment less impacted by contractionary shock) remains, with less precise estimates (Figure B.15). - Acharya et al. (2019) subsidized interest rate-based definition yields opposite result; authors caution that subsidized interest rate concept may misclassify zombies due to capitalization of interest and low-rate environment issues. ### Conclusion (policy implications summarized) - Empirical finding: zombie firms are less responsive to monetary policy shocks than other firms because contractionary policy can strengthen banks’ incentives to evergreen zombie loans. - Evergreening by banks, especially those with lower capital or higher NPLs, mitigates passthrough of monetary tightening to zombies but can harm more productive firms and aggregate productivity. - Policy recommendations: - Strengthen banks’ balance sheets (higher capital buffers). - Limit banks’ incentives for risky evergreening behavior via stronger prudential supervision and macroprudential measures targeted at loans. - Improve insolvency regimes to allow efficient resolution and restructuring of weak firms. *Source: wpiea2023192-print-pdf - 5.4 The role of bank capital.* ### Appendix A: Sample selection and zombie firms ### Appendix A: Sample selection and zombie firms ### Sample overview - Data source: S&P Compustat North America and Compustat Global. - Coverage: nonfinancial listed corporations for 49 countries (23 EMs and 26 AEs). - Final sample: an unbalanced panel of 24,371 nonfinancial firms over 2000q1-2019q4, totaling 812,627 firm-quarter observations. - Excluded: financial firms (banks, diversified financials, and insurance firms) defined by GICS codes ranging from 4010 to 4030. ### Sample adjustments and cleaning rules (following Albuquerque and Iyer (2023)) - Convert non-USD to USD for key variables in levels using: - end-of-period exchange rates for stock balance sheet data; and - quarterly average exchange rates for income statement and cash flow data, and for financial market data. - Drop observations for missing assets and liabilities. - Replace negative values for assets and liabilities with zeros. - Drop observations if capital stock or total debt are missing. - Drop observations when acquisitions are larger than five percent of total assets to exclude potential mergers and acquisitions. - Drop firms with total debt larger than 100 percent of total assets. - Drop firms with fewer than three years of data on the leverage ratio, the capital stock, the ICR, and sales. - Drop observations for countries with fewer than five firms for each quarter. - Winsorize key variables at the 2.5/97.5 percentiles at the country level. - Drop countries with fewer than eight years of data on the leverage ratio, the capital stock, and the ICR from 2000 onwards. - Compute zombie firms for industries with at least three firms per country-quarter pair. - Drop countries with fewer than 12 years of data on zombie shares. - Take four-quarter rolling sums of flow variables—EBIT, sales, interest expenses—before computing ratios when the denominator is a stock variable (e.g., the implicit interest rate). - Deflate nominal variables with the respective country CPI deflator. ### Variable definitions (selected) - Net capital stock: PPENTQ — Compustat - Employment: EMP — Compustat - Note: Annual data interpolated to quarterly. - Total debt (book value): DLCQ + DLTTQ — Compustat - Long-term debt (book value): DLTTQ — Compustat - Short-term debt (book value): DLCQ — Compustat - Total assets (book value): ATQ — Compustat - Current assets: ACTQ — Compustat - Current liabilities: LCTQ — Compustat - Net current assets: ACTQ - LCTQ — Compustat - Cash + short-term investments: CHEQ — Compustat - Net income: NIQ — Compustat - Interest payments: XINTQ — Compustat - Depreciation & amortization: DPQ — Compustat - Stock prices: PRCQQ — Compustat - EBITDA: DASALEQ - COSGQ - XSGAQ — Compustat - EBIT: SALEQ - COSGQ - XSGAQ - DPQ — Compustat - Debt ratio: (DLCQ + DLTTQ) / ATQ — Compustat - ICR: EBIT / XINTQ — Compustat - ROA: EBIT / ATQ — Compustat - Equity (book value): SEQQ + TXDITCQ - Preferred stock — Compustat - Tobin’s Q: (ATQ + PRCCQ×CSHOQ - Equity) / ATQ — Compustat - Implicit interest rate: XINTQ / (DLCQ + DLTTQ) — Compustat - Sales: SALEQ — Compustat - Acquisitions: AQCYa — Compustat - a. Transformation from year-to-date to quarterly. - Loan share: (NP + DLTO) / (DLCQ + DLTTQ) — Compustat - Age: Foundation year — Capital IQ - PD in 12 months: PRCQQNUS-CRI ### Figures and analyses referenced - Number of distinct firms over 2000-19 (Figure A.1). - Average zombie shares for listed firms over the 2000-2021 period by country (Figure A.2). - Share of listed zombie firms in 2021 by country quartiles (Figure A.3). - Average share of zombie firms by industry at two-digit and four-digit GICS levels (Figure A.4). - Multiple impulse-response figures (Appendix B) analyze cumulative responses to monetary policy shocks that increase country-specific sovereign bond yields or short-term interest rates by 100 bps, including: - Average effects on investment growth, employment growth, debt growth, and interest rate change (Figures B.1, B.2). - Regional effects and AEs vs EMs comparisons (Figure B.3). - Differential effects on zombies versus nonzombies using various zombie definitions and indicators (Figures B.4–B.7, B.15). - Marginal effects of bank indicators (capital buffers, NPLs) and macroprudential policies on zombies versus nonzombies (Figures B.8, B.9). - Marginal effects in countries with well-prepared insolvency regimes (Figure B.10). - Effects using alternative US monetary policy shock specifications and regional policy shocks (Figures B.11–B.14). - Notes on figures: all impulse responses report cumulative responses to a monetary policy shock that increases the relevant country-specific sovereign bond yield or interest rate by 100 bps. Confidence bands reported are 68 (dark grey) and 90 (light grey) percent. *Source: Appendix A and Appendix B, wpiea2023192-print-pdf.* ### References ### wpiea2023192-print-pdf - References ### Monetary policy spillovers and transmission - Ahmed, S., Akinci, O. and Queraltó, A. (2021), U.S. Monetary Policy Spillovers to Emerging Markets: Both Shocks and Vulnerabilities Matter, International Finance Discussion Papers 1321, Board of Governors of the Federal Reserve System (U.S.). - Arbatli-Saxegaard, E., Firat, M., Furceri, D. and Verrier, J. (2022), U.S. Monetary Policy Shock Spillovers: Evidence from Firm-Level Data, IMF Working Papers 22/191, International Monetary Fund. - Bauer, M. D. and Swanson, E. T. (2023), A Reassessment of Monetary Policy Surprises and High-Frequency Identification, in ‘NBER Macroeconomics Annual, volume 37’, NBER Chapters, National Bureau of Economic Research, Inc. - Bräuning, F. and Ivashina, V. (2020), ‘U.S. monetary policy and emerging market credit cycles’, Journal of Monetary Economics112, 57–76. - Bräuning, F. and Sheremirov, V. (2023), ‘The Transmission Mechanisms of International Business Cycles: International Trade and the Foreign Effects of US Monetary Policy’, IMF Economic Review71, 300–325. - Cesa-Bianchi, A., Ferrero, A. and Rebucci, A. (2018), ‘International credit supply shocks’, Journal of International Economics112, 219–237. - Cesa-Bianchi, A. and Sokol, A. (2022), ‘Financial shocks, credit spreads, and the international credit channel’, Journal of International Economics135, 103543. - Dedola, L., Rivolta, G. and Stracca, L. (2017), ‘If the Fed sneezes, who catches a cold?’, Journal of International Economics108, S23–S41. - Degasperi, R., Hong, S. and Ricco, G. (2020), The Global Transmission of U.S. Monetary Policy, CEPR Discussion Papers 14533, C.E.P.R. Discussion Papers. - Di Giovanni, J. and Rogers, J. H. (2023), ‘The Impact of U.S. Monetary Policy on Foreign Firms’, IMF Economic Review (forthcoming). - Dornouni? [Note: Only entries from the source were used. — omitted if this is not in source] - Jarociński, M. (2022), ‘Central bank information effects and transatlantic spillovers’, Journal of International Economics139, 103683. - Jarociński, M. and Karadi, P. (2020), ‘Deconstructing Monetary Policy Surprises – The Role of Information Shocks’, American Economic Journal: Macroeconomics12(2), 1–43. - Miranda-Agrippino, S. and Nenova, T. (2022), ‘A tale of two global monetary policies’, Journal of International Economics136, 103606. - Miranda-Agrippino, S. and Rey, H. (2020a), ‘The Global Financial Cycle after Lehman’, AEA Papers and Proceedings110, 523–28. - Miranda-Agrippino, S. and Rey, H. (2020b), ‘U.S. Monetary Policy and the Global Financial Cycle’, The Review of Economic Studies87(6), 2754–2776. - Miranda-Agrippino, S. and Ricco, G. (2021), ‘The Transmission of Monetary Policy Shocks’, American Economic Journal: Macroeconomics13(3), 74–107. - Pinchetti, M. and Szczepaniak, A. (2023), ‘Global spillovers of the Fed information effect’, IMF Economic Review (forthcoming). - Rey, H. (2013), ‘Dilemma not trilemma: the global cycle and monetary policy independence’, Proceedings - Economic Policy Symposium - Jackson Hole. - Stavrakeva, V. and Tang, J. (2019), The Dollar During the Great Recession: US Monetary Policy Signaling and The Flight To Safety, CEPR Discussion Papers 14034, C.E.P.R. Discussion Papers. - Stock, J. H. and Watson, M. W. (2018), ‘Identification and Estimation of Dynamic Causal Effects in Macroeconomics Using External Instruments’, The Economic Journal128(May), 917–948. - Dumont? [Note: Only entries from the source were used. — omitted if this is not in source] ### Zombie firms, zombie lending, and corporate distress - Andrews, D. and Petroulakis, F. (2019), Breaking the shackles: Zombie firms, weak banks and depressed restructuring in Europe, Working Paper Series 2240, European Central Bank. - Banerjee, R. and Hofmann, B. (2022), ‘Corporate zombies: anatomy and life cycle’, Economic Policy37(112), 757–803. - Caballero, R. J., Hoshi, T. and Kashyap, A. K. (2008), ‘Zombie Lending and Depressed Restructuring in Japan’, American Economic Review98(5), 1943–77. - Hu, Y. and Varas, F. (2021), ‘A Theory of Zombie Lending’, The Journal of Finance76(4), 1813–1867. - Kulkarni, N., Ritadhi, S., Mukherjee, S. and Waldock, K. (2021), Unearthing Zombies, Available at https://ssrn.com/abstract=3495660, SSRN. - McGowan, M. A., Andrews, D. and Millot, V. (2018), ‘The walking dead? Zombie firms and productivity performance in OECD countries’, Economic Policy33(96), 685–736. - Storz, M., Koetter, M., Setzer, R. and Westphal, A. (2017), Do we want these two to tango? On zombie firms and stressed banks in Europe, Working Paper Series 2104, European Central Bank. - Albuquerque, B. and Iyer, R. (2023), The Rise of the Walking Dead: Zombie Firms Around the World, IMF Working Papers 23/125, International Monetary Fund. - Altman, E. I., Dai, R. and Wang, W. (2022), Global zombies, Available at https://ssrn.com/abstract=3970332, SSRN. - Bonfim, D., Cerqueiro, G., Degryse, H. and Ongena, S. (2023), ‘On-site inspecting zombie lending’, Management Science69(5), 2547–3155. - Favara, G., Minoiu, C. and Perez, A. (2022), Zombie Lending to U.S. firms, Available at https://ssrn.com/abstract=4065886, SSRN. - Faria-e-Castro, M., Paul, P. and Sanchez, J. M. (2022), Evergreening, Working Papers 2022-024, Federal Reserve Bank of St. Louis. - Chari, A., Jain, L. and Kulkarni, N. (2022), The Unholy Trinity: Regulatory Forbearance, Government-Owned Banks and Zombie Firms. mimeo. - Hu, Y. and Varas, F. (2021), ‘A Theory of Zombie Lending’, The Journal of Finance76(4), 1813–1867. ### Bank lending, credit supply, and firm-level effects - Acharya, V. V., Borchert, L., Jager, M. and Steffen, S. (2021), ‘Kicking the Can Down the Road: Government Interventions in the European Banking Sector’, The Review of Financial Studies34(9), 4090–4131. - Becker, B. and Ivashina, V. (2014), ‘Cyclicality of credit supply: Firm level evidence’, Journal of Monetary Economics62(C), 76–93. - Becker, B. and Ivashina, V. (2022), ‘Weak Corporate Insolvency Rules: The Missing Driver of Zombie Lending’, AEA Papers and Proceedings112, 516–20. - Berlin, M. and Mester, L. J. (1992), ‘Debt covenants and renegotiation’, Journal of Financial Intermediation2(2), 95–133. - Bernanke, B. S. and Gertler, M. (1995), ‘Inside the Black Box: The Credit Channel of Monetary Policy Transmission’, Journal of Economic Perspectives9(4), 27–48. - Bernanke, B. S., Gertler, M. and Gilchrist, S. (1999), The Financial Accelerator in a Quantitative Business Cycle Framework, in J. B. Taylor and M. Woodford, eds, ‘Handbook of Macroeconomics’, Handbook of Macroeconomics, Elsevier, chapter 21, pp. 1341–1393. - Blattner, L., Farínha, L. and Rebelo, F. (2023), ‘When losses turn into loans: The cost of weak banks’, American Economic Review113(6), 1600–1641. - Gertler, M. (1988), ‘Financial Structure and Aggregate Economic Activity: An Overview’, Journal of Money, Credit and Banking20(3), 559–588. - Gertler, M. and Gilchrist, S. (1994), ‘Monetary Policy, Business Cycles, and the Behavior of Small Manufacturing Firms’, The Quarterly Journal of Economics109(2), 309–340. - Gertler, M. and Karadi, P. (2015), ‘Monetary Policy Surprises, Credit Costs, and Economic Activity’, American Economic Journal: Macroeconomics7(1), 44–76. - Giannetti, M. and Simonov, A. (2013), ‘On the Real Effects of Bank Bailouts: Micro Evidence from Japan’, American Economic Journal: Macroeconomics5(1), 135–67. - Gilchrist, S. and Zakrajšek, E. (2012), ‘Credit Spreads and Business Cycle Fluctuations’, American Economic Review102(4), 1692–1720. - Hadlock, C. J. and James, C. M. (2002), ‘Do banks provide financial slack?’, The Journal of Finance57(3), 1383–1419. - Ippolito, F., Ozdagli, A. K. and Perez-Orive, A. (2018), ‘The transmission of monetary policy through bank lending: The floating rate channel’, Journal of Monetary Economics95(C), 49–71. - Jiménez, G., Ongena, S., Peydró, J.-L. and Saurina, J. (2012), ‘Credit Supply and Monetary Policy: Identifying the Bank Balance-Sheet Channel with Loan Applications’, American Economic Review102(5), 2301–26. - Kahle, K. M. and Stulz, R. M. (2013), ‘Access to capital, investment, and the financial crisis’, Journal of Financial Economics110(2), 280–299. - Kashyap, A. and Stein, J. (2000), ‘What Do a Million Observations on Banks Say about the Transmission of Monetary Policy?’, American Economic Review90(3), 407–428. - Peek, J. and Rosengren, E. S. (2005), ‘Unnatural Selection: Perverse Incentives and the Misallocation of Credit in Japan’, American Economic Review95(4), 1144–1166. - Ottonello, P. and Winberry, T. (2020), ‘Financial Heterogeneity and the Investment Channel of Monetary Policy’, Econometrica88(6), 2473–2502. - Holm-Hadulla, F. and Thürwüchter, C. (2021), ‘Heterogeneity in corporate debt structures and the transmission of monetary policy’, European Economic Review136, 103743. - Müller, K. and Verner, E. (2023), ‘Credit Allocation and Macroeconomic Fluctuations’, Review of Economic Studies (forthcoming). - Li, D., Magud, N. E. and Valencia, F. (2020), ‘Financial Shocks and Corporate Investment in Emerging Markets’, Journal of Money, Credit and Banking52(2-3), 613–644. ### Macroprudential policy, regulation, and crises - Akinci, O. and Olmstead-Rumsey, J. (2018), ‘How effective are macroprudential policies? an empirical investigation’, Journal of Financial Intermediation33, 33–57. - Alam, Z., Alter, A., Eiseman, J., Gelos, G., Kang, H., Narita, M., Nier, E. and Wang, N. (2019), Digging Deeper–Evidence on the Effects of Macroprudential Policies from a New Database, IMF Working Papers 19/066, International Monetary Fund. - Borio, C. and Zhu, H. (2012), ‘Capital regulation, risk-taking and monetary policy: a missing link in the transmission mechanism?’, Journal of Financial Stability8(4), 236–251. - Giannetti, M. and Simonov, A. (2013), ‘On the Real Effects of Bank Bailouts: Micro Evidence from Japan’, American Economic Journal: Macroeconomics5(1), 135–67. - Storz, M., Koetter, M., Setzer, R. and Westphal, A. (2017), Do we want these two to tango? On zombie firms and stressed banks in Europe, Working Paper Series 2104, European Central Bank. - Schivardi, F., Sette, E. and Tabellini, G. (2022), ‘Credit Misallocation During the European Financial Crisis’, The Economic Journal132(641), 391–423. - Araujo, J. D., Garrido, J., Kopp, E., Varghese, R. and Yao, W. (2022), Policy Options for Supporting and Restructuring Firms Hit by the COVID-19 Crisis, Departmental Paper 22/002, International Monetary Fund. ### Empirical methods, identification, and measurement - Baltagi, B. H. (2011), Econometrics, Springer Texts in Business and Economics, Springer. - Montiel Olea, J. L. and Plagborg-Møller, M. (2021), ‘Local projection inference is simpler and more robust than you think’, Econometrica89(4), 1789–1823. - Nakamura, E. and Steinsson, J. (2018), ‘High Frequency Identification of Monetary Non-Neutrality: The Information Effect’, Quarterly Journal of Economics3(133), 1283–1330. - Mundell, A. R. (1963), ‘Capital mobility and stabilization policy under fixed and flexible exchange rates’, Canadian Journal of Economics and Political Science29(4), 475–485. - Fleming, J. M. (1962), ‘Domestic financial policies under fixed and under floating exchange rate’, IMF Staff Papers9, 369–380. - Gürkaynak, R. S., Sack, B. and Swanson, E. (2005), ‘Do Actions Speak Louder Than Words? The Response of Asset Prices to Monetary Policy Actions and Statements’, International Journal of Central Banking1(1), 55–93. - Jordà, Ò., Schularick, M. and Taylor, A. M. (2020), ‘The effects of quasi-random monetary experiments’, Journal of Monetary Economics112, 22–40. - Stock, J. H. and Watson, M. W. (2018), ‘Identification and Estimation of Dynamic Causal Effects in Macroeconomics Using External Instruments’, The Economic Journal128(May), 917–948. - Montiel Olea, J. L. and Plagborg-Møller, M. (2021), ‘Local projection inference is simpler and more robust than you think’, Econometrica89(4), 1789–1823. _The Zombie Lending Channel of Monetary Policy — Working Paper No. WP/2023/192_ --- _Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023192-print-pdf.pdf_