Debt Covenants and the Macroeconomy: The Interest Coverage Channel

Save this PDF as:
 WORD  PNG  TXT  JPG

Size: px
Start display at page:

Download "Debt Covenants and the Macroeconomy: The Interest Coverage Channel"

Transcription

1 Debt Covenants and the Macroeconomy: The Interest Coverage Channel Daniel L. Greenwald MIT Sloan EFA Lunch, April 19 Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 1 / 6

2 Introduction Non-residential investment is a key driver of monetary policy response. - Natural link: $6T corporate debt market. - Large body of work on transmission through credit limits ( financial accelerator ). Firm credit limits typically modeled as limit on market leverage. - But actual covenants in debt contracts quite different from this stylized model. - Lian and Ma (17): importance of earnings based constraints. - But many covenants depend on more than earnings, firms often have several at once. Research question: how does firm credit limit structure influence macro dynamics? - Focus on Interest Coverage (IC) covenants that cap ratio of interest payments to earnings. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 / 6

3 This Paper Approach: combine structural model with firm-level empirical evidence. Stylized Facts: Interest Coverage covenants extremely common (seen in 84% of firms in DealScan sample with covenants), maximum ratios appear stable over time. Main Finding #1: Interest Coverage covenants amplify interest rate transmission. - Much stronger responses of debt, investment, output than under alternative covenant types. - Reason: directly shifted by interest rates. - Rates 1bp = extra 6.3% 8Q asset growth in data (4.4% in model). Main Finding #: Combination of interest coverage + other cov. = state dependence. - Whether interest coverage is tightest covenant determined by interest rate. - Stronger transmission when rates are already high (and IC covenants likely to bind). - High vs. low rates: 1bp = extra 4.9% 8Q asset growth in data (1.5% in model). Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 3 / 6

4 This Paper Approach: combine structural model with firm-level empirical evidence. Stylized Facts: Interest Coverage covenants extremely common (seen in 84% of firms in DealScan sample with covenants), maximum ratios appear stable over time. Main Finding #1: Interest Coverage covenants amplify interest rate transmission. - Much stronger responses of debt, investment, output than under alternative covenant types. - Reason: directly shifted by interest rates. - Rates 1bp = extra 6.3% 8Q asset growth in data (4.4% in model). Main Finding #: Combination of interest coverage + other cov. = state dependence. - Whether interest coverage is tightest covenant determined by interest rate. - Stronger transmission when rates are already high (and IC covenants likely to bind). - High vs. low rates: 1bp = extra 4.9% 8Q asset growth in data (1.5% in model). Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 3 / 6

5 This Paper Approach: combine structural model with firm-level empirical evidence. Stylized Facts: Interest Coverage covenants extremely common (seen in 84% of firms in DealScan sample with covenants), maximum ratios appear stable over time. Main Finding #1: Interest Coverage covenants amplify interest rate transmission. - Much stronger responses of debt, investment, output than under alternative covenant types. - Reason: directly shifted by interest rates. - Rates 1bp = extra 6.3% 8Q asset growth in data (4.4% in model). Main Finding #: Combination of interest coverage + other cov. = state dependence. - Whether interest coverage is tightest covenant determined by interest rate. - Stronger transmission when rates are already high (and IC covenants likely to bind). - High vs. low rates: 1bp = extra 4.9% 8Q asset growth in data (1.5% in model). Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 3 / 6

6 Background: Debt Covenants Covenants provide conditions that, if violated by the firm, allow lender to demand immediate repayment. - Often set thresholds for financial ratios = debt limits. - Applies to entire firm s statistics, not limited to individual loan. - Violation typically leads to (costly) renegotiation, but for today treat as hard caps. Three main types: 1. Interest Coverage (IC): restrict interest payments fraction θ IC of earnings (EBITDA).. Debt/Earnings (DE): restrict stock of debt fraction θ DE of earnings (EBITDA). 3. Leverage: restrict stock of debt fraction θ LEV of firm book value. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 4 / 6

7 Simple Example of Interest Rate Transmission Consider firm with EBITDA $1M, max ratio of interest payments to EBITDA of 4%. - Max interest payment is $4M. - At 1% interest rate, firm can borrow up to $4M /.1 = $4M without violating. - If rates fall to 9%, firm can now borrow $4M /.9 = $44.4M, an increase of 11% This high sensitivity holds even if firm uses only fixed-rate debt. - In this case, relevant interest rate is rate on new fixed rate debt. If firm already holds floating rate debt, amount of new debt firm can take on without violating even more sensitive. - If firm already has $M at same floating rate (1%), can borrow addl. $M without violating. - After fall in rates, old debt only contributes $1.8M toward interest cap, can borrow $.M /.9 = $4.4M, increase of %. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 5 / 6

8 Simple Example of Interest Rate Transmission Consider firm with EBITDA $1M, max ratio of interest payments to EBITDA of 4%. - Max interest payment is $4M. - At 1% interest rate, firm can borrow up to $4M /.1 = $4M without violating. - If rates fall to 9%, firm can now borrow $4M /.9 = $44.4M, an increase of 11% This high sensitivity holds even if firm uses only fixed-rate debt. - In this case, relevant interest rate is rate on new fixed rate debt. If firm already holds floating rate debt, amount of new debt firm can take on without violating even more sensitive. - If firm already has $M at same floating rate (1%), can borrow addl. $M without violating. - After fall in rates, old debt only contributes $1.8M toward interest cap, can borrow $.M /.9 = $4.4M, increase of %. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 5 / 6

9 Simple Example of Interest Rate Transmission Consider firm with EBITDA $1M, max ratio of interest payments to EBITDA of 4%. - Max interest payment is $4M. - At 1% interest rate, firm can borrow up to $4M /.1 = $4M without violating. - If rates fall to 9%, firm can now borrow $4M /.9 = $44.4M, an increase of 11% This high sensitivity holds even if firm uses only fixed-rate debt. - In this case, relevant interest rate is rate on new fixed rate debt. If firm already holds floating rate debt, amount of new debt firm can take on without violating even more sensitive. - If firm already has $M at same floating rate (1%), can borrow addl. $M without violating. - After fall in rates, old debt only contributes $1.8M toward interest cap, can borrow $.M /.9 = $4.4M, increase of %. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 5 / 6

10 Covenant Prevalence by Type Plot: share with each covenant type for firms with at least one DealScan covenant. Share with Interest Coverage high and stable over time Interest Cov. Share Debt/Earnings Share Leverage Share Source: DealScan. Shares are equally weighted among DealScan firms with at least one covenant. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 6 / 6

11 Covenant Ratios Over Time Complication: covenant limits are endogenously set. Do lenders dynamically adjust simple covenants to achieve more complex debt policies? 5.5 Wtd. By Sales 5. Wtd. By Deal Amount (a) Min Interest Cov. Ratio 1 Wtd. By Sales Wtd. By Deal Amount (b) Max Debt/Earnings Ratio Source: DealScan, Compustat. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 7 / 6

12 Covenant Ratios Over Time Below: initial covenant ratios at origination in DealScan. Appear noisy but stable over time. - Note: Min Interest Cov. Ratio = EBITDA / payment. 5.5 Wtd. By Sales 5. Wtd. By Deal Amount (a) Min Interest Cov. Ratio 1 Wtd. By Sales Wtd. By Deal Amount (b) Max Debt/Earnings Ratio Source: DealScan, Compustat. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 7 / 6

13 Covenant Ratios Over Time Second check: maximum ratios on new loans stable even when underlying aggregate economic ratios move Min. Interest Cov. Corp NF Profit / Pay 7 Max. Debt/EBITDA Corp NF Debt/Profit (a) Min Interest Cov. Ratio (b) Max Debt/Earnings Ratio Source: DealScan, Compustat, NIPA, Flow of Funds. Covenant limits are weighted by deal amount. Debt payments assume 6bp spread over 3-Month Treasury. Min. Interest Cov. is the min. allowed Earnings / Interest ratio. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 8 / 6

14 Covenant Ratios Over Time Now look at all active covenants. Provide stable constraints even as variables move Min. Interest Cov. Corp NF Profit / Pay (a) Min Interest Cov. Ratio 7.5 Max. Debt/EBITDA Corp NF Debt/Profit (b) Max Debt/Earnings Ratio Source: DealScan, Compustat, NIPA, Flow of Funds. Covenant limits are weighted by deal amount. Debt payments assume 6bp spread over 3-Month Treasury. Min. Interest Cov. is the min. allowed Earnings / Interest ratio. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 9 / 6

15 Covenant Ratios Over Time Takeaway: covenants have structural meaning, reasonable to consider as fixed limits at business cycle frequency Min. Interest Cov. Corp NF Profit / Pay (a) Min Interest Cov. Ratio 7.5 Max. Debt/EBITDA Corp NF Debt/Profit (b) Max Debt/Earnings Ratio Source: DealScan, Compustat, NIPA, Flow of Funds. Covenant limits are weighted by deal amount. Debt payments assume 6bp spread over 3-Month Treasury. Min. Interest Cov. is the min. allowed Earnings / Interest ratio. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 9 / 6

16 Firm Characteristics by Covenant Firms with covenants larger, more levered than firms without covenants/syndicated loans. None IC DE Lev IC + DE IC + Lev DE + Lev Sales EBITDA Assets PPE Debt ST Debt LT Debt Cash Debt/EBITDA Debt/Assets Market-to-Book N 184,75 69,84 56,78 36,96 51,8 7,56 16,36 Source: Dealscan, Compustat. Additional Groupings Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 1 / 6

17 Firm Characteristics by Covenant Firms with IC covenants generally similar to firms with DE covenants. Firms with Leverage covenants a bit smaller. None IC DE Lev IC + DE IC + Lev DE + Lev Sales EBITDA Assets PPE Debt ST Debt LT Debt Cash Debt/EBITDA Debt/Assets Market-to-Book N 184,75 69,84 56,78 36,96 51,8 7,56 16,36 Source: Dealscan, Compustat. Additional Groupings Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 1 / 6

18 Model Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April / 6

19 Model Overview Demographics and preferences - Risk-neutral representative household consumes and provides labor. - Interest rate variation = time varying discount factor: log β t = (1 ρ β ) log β + ρβ t 1 + ε β,t. - Representative firm owns capital and pays dividends to household. Productive technology: f (K t 1, N t ) = Z t K α t 1 N1 α t Firm capital structure: - Risk-free floating rate debt at rate r t, interest is tax deductible (tax shield). - Dividend adjustment costs (financing frictions) following Jermann and Quadrini (1). - Combined: pathway from debt limits debt investment. Flexible prices and wages, monetary authority targets (and achieves) constant inflation. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 1 / 6

20 Model Overview Demographics and preferences - Risk-neutral representative household consumes and provides labor. - Interest rate variation = time varying discount factor: log β t = (1 ρ β ) log β + ρβ t 1 + ε β,t. - Representative firm owns capital and pays dividends to household. Productive technology: f (K t 1, N t ) = Z t K α t 1 N1 α t Firm capital structure: - Risk-free floating rate debt at rate r t, interest is tax deductible (tax shield). - Dividend adjustment costs (financing frictions) following Jermann and Quadrini (1). - Combined: pathway from debt limits debt investment. Flexible prices and wages, monetary authority targets (and achieves) constant inflation. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 1 / 6

21 Model Overview Demographics and preferences - Risk-neutral representative household consumes and provides labor. - Interest rate variation = time varying discount factor: log β t = (1 ρ β ) log β + ρβ t 1 + ε β,t. - Representative firm owns capital and pays dividends to household. Productive technology: f (K t 1, N t ) = Z t K α t 1 N1 α t Firm capital structure: - Risk-free floating rate debt at rate r t, interest is tax deductible (tax shield). - Dividend adjustment costs (financing frictions) following Jermann and Quadrini (1). - Combined: pathway from debt limits debt investment. Flexible prices and wages, monetary authority targets (and achieves) constant inflation. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 1 / 6

22 Model Overview Demographics and preferences - Risk-neutral representative household consumes and provides labor. - Interest rate variation = time varying discount factor: log β t = (1 ρ β ) log β + ρβ t 1 + ε β,t. - Representative firm owns capital and pays dividends to household. Productive technology: f (K t 1, N t ) = Z t K α t 1 N1 α t Firm capital structure: - Risk-free floating rate debt at rate r t, interest is tax deductible (tax shield). - Dividend adjustment costs (financing frictions) following Jermann and Quadrini (1). - Combined: pathway from debt limits debt investment. Flexible prices and wages, monetary authority targets (and achieves) constant inflation. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 1 / 6

23 Representative Firm s Problem Rep. firm chooses dividends D t, labor demand N t, new debt B t and the investment rate i t to maximize V F (K t 1, B t 1 ) = Ψ(D t ) + E t [ Λt+1 V F (K t, B t ) ] where concave Ψ(D t ) represents adjustment costs for dividends, Λ t+1 is the household SDF, subject to the budget constraint ) D t = (1 τ) (f (K t 1, N t ) w t N t } {{ } after-tax profit (1 τ)r t πt 1 B t 1 }{{} interest payment and the borrowing constraint (debt covenants). Household s Problem + τδk }{{ t 1 } depreciation credit i t K }{{ t 1 } investment ) B t πt 1 B t 1 ( + } {{ } net principal Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April / 6

24 Covenant Implementations Denote EBITDA by X t = f (K t 1, N t ) w t N t. Covenant types: 1. Interest Coverage: B IC t = θic X t r t.. Debt/Earnings: B DE t = θ DE X t. 3. Leverage: B LEV t = θ LEV BV t 1 θ LEV K t 1. Only interest coverage directly shifted by interest rates. - Highly sensitive, elasticity of B IC to rates is 1. Overall debt limit is smoothed to allow for e.g., annual financial statistics: B t ρ B t + (1 ρ)πt 1 B t 1 Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April / 6

25 Collateralizability Debt limits are mechanically sensitive to interest rates under IC covenants. What about transmission into real investment? Optimality condition: [ q t = Ω t + M }{{}}{{} t E t (1 + r t ) ] B t+1 K t Tobin s q Value of CFs }{{} Collateral Benefit Key object is collateralizability of investment: B t+1 / K t : B IC t K t = θic f K,t+1 r t+1, B DE t K t = θ DE f K,t+1, B LEV t K t = θ LEV. All covenants are collateralizable, but only IC collateralizability varies with interest rate. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April / 6

26 Calibration For today, no capital adjustment costs, isoelastic (power) dividend smoothing. Parameter Name Value Internal Target/Source Discount factor mean β.986 N 6% real rate Discount factor persistence ρ β.99 N Autocorrelation of 3-Mo T-Bill Labor disutility scale η 1.7 Y N SS = 1 Tax rate τ.35 N Corporate tax rate Inflation rate π 1.5 N % inflation Capital share α.36 N Standard Depreciation δ.5 N Standard Dividend cost elasticity ζ D 1. N Moderate smoothing Borrowing limit smoothing ρ B.5 N Annualized ratios Max interest coverage ratio θ IC.18 Y Book leverage = 1/3 in IC economy Max debt-to-earnings ratio θ DE Y Book leverage = 1/3 in DE economy Max Leverage ratio θ LEV.338 Y Book leverage = 1/3 in Lev. economy Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April / 6

27 Results Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April / 6

28 Comparison: Covenant Types Main Result #1: Interest Coverage covenants amplify interest rate transmission. Compare linearized IRF to 1bp disc. rate shock in economies each with single constraint. Debt IRF to Discount Rate Capital 6 4 Interest Cov. Debt/EBITDA Leverage IRF to Discount Rate Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April / 6

29 Comparison: Covenant Types IC economy: large relaxation of debt limits = capital, EBITDA growth = feedback. Additional 8Q growth of debt (14.1%), capital (4.4%), output (4.4%) relative to DE economy. Debt IRF to Discount Rate Capital 6 4 Interest Cov. Debt/EBITDA Leverage IRF to Discount Rate Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April / 6

30 Empirical Evidence: Covenant Types Regression on merged Compustat (investment, debt) + DealScan (loan covenants) data: Percent of Assetst y i,t+h = α i + φ ind,t + I cov,t (β,cov + β 1,cov r t ) + γ X t 1 + δ (X t 1 r t ) + ε i,t cov Debt Interest Coverage Percent of Assetst Assets Interest Coverage Source: DealScan, Compustat. The sample spans 1997Q1 to 7Q4. Dark bands indicate 67% confidence bands, while light bands indicate 95% confidence bands. Standard errors are clustered at the firm level. MP Shocks Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April / 6

31 Empirical Evidence: Covenant Types Industry-time fixed effects control for endogeneity of interest rate. Larger responses to rates 1bp for firms with Interest Coverage covenants. Percent of Assetst Debt Interest Coverage Percent of Assetst Assets Interest Coverage Source: DealScan, Compustat. The sample spans 1997Q1 to 7Q4. Dark bands indicate 67% confidence bands, while light bands indicate 95% confidence bands. Standard errors are clustered at the firm level. MP Shocks Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April / 6

32 Empirical Evidence: Covenant Types Challenge: firms with no covenants differ from IC firms on observables. Better comparison: firms with DE covenants. These show no increased response. Percent of Assetst Debt Interest Coverage Other Covenant Percent of Assetst Assets Interest Coverage Other Covenant Source: DealScan, Compustat. The sample spans 1997Q1 to 7Q4. Dark bands indicate 67% confidence bands, while light bands indicate 95% confidence bands. Standard errors are clustered at the firm level. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 / 6

33 Empirical Evidence: Covenant Types Formal comparison: estimate β 1,IC β 1,DE. Estimate: higher 8Q debt growth (.7%), asset growth (6.3%) for IC relative to DE covenant. Percent of Assetst Debt IC - Other Percent of Assetst Assets IC - Other Source: DealScan, Compustat. The sample spans 1997Q1 to 7Q4. Dark bands indicate 67% confidence bands, while light bands indicate 95% confidence bands. Standard errors are clustered at the firm level. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 1 / 6

34 Multiple Covenants Previous analysis considers economies with a single covenant at a time. Data: most firms with any covenants have both Interest Coverage + Debt/Earnings DE + IC Share DE + Lev Share IC + Lev Share Source: DealScan. Shares are equally weighted among DealScan firms with at least one covenant. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 / 6

35 Implementation: Debt/Earnings + Interest Coverage Covenant Assume common Debt/Earnings limit θ DE, but each firm i faces idiosyncratic IC limit: θ IC i,t = e i,t θ IC, e i,t iid Γ e Timing: - Firm re-draws e i,t each time it takes on new debt. - Must choose capital before it knows its draw of e i,t. ( ) Overall debt limit: B i,t = min B IC i,t, B DE i,t. Calibrate σ e to match IQR of θ DE i,t /θic i,t in DealScan data (σ e =.31). Calibrate θ IC, θ DE to match that 47% have tighter IC at steady state. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 3 / 6

36 State Dependence Whether Interest Coverage vs. Debt/Earnings is tighter uniquely determined by rates. - IC binds r t r i,t θic i,t / θ DE Share IC Tighter Date Source: DealScan, Compustat, equally weighted. Assumed interest rate is 6bp spread over the 3-Month T-Bill. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 4 / Mo Treasury 4 3 1

37 State Dependence DealScan data: substantial variation in implied fraction with IC as tighter covenant. Share IC Tighter Date Source: DealScan, Compustat, equally weighted. Assumed interest rate is 6bp spread over the 3-Month T-Bill. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 4 / Mo Treasury 4 3 1

38 State Dependence: DE + IC Covenants Main Result #: Combining IC + DE covs = state dependent interest rate transmission. Alternative regimes with SS interest (discount) rate high (+5bp) vs. low (-5bp). 7.5 IRF to Discount Rate 3 Low Rates Hybrid DE/IC High Rates IRF to Discount Rate Debt 5..5 Capital Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 5 / 6

39 State Dependence: DE + IC Covenants Stronger transmission when rates are high (79% IC binds) vs. low (8% IC binds). Additional 8Q growth in debt (6.%), capital (1.5%) in high vs. low rate regime. 7.5 IRF to Discount Rate 3 Low Rates Hybrid DE/IC High Rates IRF to Discount Rate Debt 5..5 Capital Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 5 / 6

40 Empirics: State Dependence Augment original regression so coefficients depend on interest rate regime (cutoff = 3.5%): y i,t+h = α i + φ ind,t + s {hi, low} } ( ) I s,t { I cov,t β s,cov + βs 1,cov r t + γ sx t 1 + δ s(x t 1 r t ) + ε i,t cov Percent of Assetst Debt Hi - Low (Both) Percent of Assetst Assets Hi - Low (Both) Source: DealScan, Compustat. Dark bands indicate 67% confidence bands, while light bands indicate 95% confidence bands. Standard errors are clustered at the firm level. The sample spans 1997Q1 to 7Q4. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 6 / 6

41 Empirics: State Dependence Increased 8Q growth in debt (4.4%), assets (4.9%) in high vs. low rate environment. Stronger response in high vs. low rate regime (despite smaller proportional change). Percent of Assetst Debt Hi - Low (Both) Percent of Assetst Assets Hi - Low (Both) Source: DealScan, Compustat. Dark bands indicate 67% confidence bands, while light bands indicate 95% confidence bands. Standard errors are clustered at the firm level. The sample spans 1997Q1 to 7Q4. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 6 / 6

42 Empirics: State Dependence No state dependent response for firms with IC covenant only. Percent of Assetst Debt Hi - Low (IC) Percent of Assetst Assets Hi - Low (IC) Source: DealScan, Compustat. Dark bands indicate 67% confidence bands, while light bands indicate 95% confidence bands. Standard errors are clustered at the firm level. The sample spans 1997Q1 to 7Q4. D/E Diff-in-Diff Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 7 / 6

43 Conclusion Novel model capturing key facts about corporate debt limits. - Interest Coverage limits are extremely common, caps stable over time. - Typical firm has multiple covenants. Main results: - Interest Coverage covenants amplify interest rate transmission. - State dependent transmission: stronger when rates are high. - Findings supported by firm-level data. Next steps: - Improved empirics, breakdowns by firm/debt characteristics. - More realistic firm profile, violation risk instead of hard caps. - Scraping EDGAR data. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 8 / 6

44 Monetary Policy Shocks Replace r t with identified MP shocks following Gertler and Karadi (1) Percent of Assetst Debt IC - Other Percent of Assetst Assets IC - Other Source: DealScan, Compustat. Dark bands indicate 67% confidence bands, while light bands indicate 95% confidence bands. Standard errors are clustered at the firm level. The sample spans 1997Q1 to 7Q4. Back Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 1 / 6

45 Empirics: State Dependence No state dependent response for firms with IC covenant only. Percent of Assetst Debt Hi - Lo (DE) Percent of Assetst Assets Hi - Lo (DE) Source: DealScan, Compustat. The sample spans 1997Q1 to 7Q4. Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 / 6

46 Empirics: State Dependence Difference in difference (high vs. low, both covenants vs. IC only). Percent of Assetst Debt Diff-in-Diff (Both vs. IC) Percent of Assetst Assets Diff-in-Diff (Both vs. IC) Source: DealScan, Compustat. The sample spans 1997Q1 to 7Q4. Back Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 3 / 6

47 Empirics: State Dependence Difference in difference (high vs. low, both covenants vs. DE only). Percent of Assetst Debt Diff-in-Diff (vs. DE) Percent of Assetst Assets Diff-in-Diff (vs. DE) Source: DealScan, Compustat. The sample spans 1997Q1 to 7Q4. Back Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 4 / 6

48 Representative Household s Problem Rep. household chooses consumption C t, labor supply N t and new debt B t to maximize subject to the budget constraint C t = Ψ(D t ) }{{} dividends V H (B t 1 ) = u(c t ) v(n t ) + βe t [ V H (B t ) ] + (1 τ)w t N t + r t πt 1 }{{} labor income B t 1 }{{} interest payment ( ) Bt πt 1 B t 1 }{{} net debt issuance + T S t }{{} transfer Back Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 5 / 6

49 Firm Characteristics by Covenant: Additional Groupings None IC Non-IC IC Only DE Only Lev Only Sales EBITDA Assets PPE Debt ST Debt LT Debt Cash Debt/EBITDA Debt/Assets Market-to-Book N 184,75 69,84 13,86 5,76 3,83 7,53 Source: Dealscan, Compustat. Back Daniel L. Greenwald Debt Covenants and the Macroeconomy EFA Lunch, April 19 6 / 6