Banque Nationale De Belgique Sa · EURONEXT:BNB

Banque Nationale De Belgique Sa — Technical Presentation (April 2004)

41 slides from Banque Nationale De Belgique Sa's technical presentation, Apr 27, 2004

1. Slide 1: The Magnitude and Cyclical Behavior of Financial Market Frictions

   The Magnitude and Cyclical Behavior of Financial Market Frictions Andrew T. Levin Fabio M. Natalucci Egon Zakrajšek Federal Reserve Board April 27, 2004

2. Slide 2: The Modigliani–Miller Theorem

   The Modigliani–Miller Theorem In a world of no financial market imperfections, capital structure is indeterminate and the aggregate mix of debt vs. equity is irrelevant for the evolution of the real economy. The MM theorem has provided an important justification for abstracting from financing decisi

3. Slide 3: Financial Market Frictions: Theory

   Financial Market Frictions: Theory The seminal work of Bernanke & Gertler (1989) shows that financial factors can play an important role in the propagation of aggregate shocks—the financial accelerator. • Debt-deflation during the Great Depression Fisher (1933) • Asymmetric information theory Akerlo

4. Slide 4: Financial Market Frictions: Evidence

   Financial Market Frictions: Evidence The quantitative significance of financial market frictions has remained elusive despite substantial empirical work: • Reduced-form evidence linking balance-sheet variables to investment and employment Gurley & Shaw (1960), etc. • Structural analysis using proxie

5. Slide 5: Financial Market Frictions: Evidence (contd.)

   Financial Market Frictions: Evidence (contd.) • Matching VAR-based impulse response functions Bernanke et al. (1999), Christiano et al. (2003), etc. Chari et al. (2003), Lee & Ohanian (2003) Missing: Direct estimates of the magnitude of financial frictions 4

6. Slide 6: Outline

   Outline • Revisit microfoundations of BGG (1999) financial accelerator model • Analyze firm-level panel data linking credit spreads, market-based measures of default risk, and balance sheet variables • Directly estimate the magnitude of financial frictions • Examine the behavior of model-implied ext

7. Slide 7: Summary of Results

   Summary of Results We obtain precise time-specific estimates of the structural parameter measuring the extent of financial market frictions. We clearly reject the null hypothesis of frictionless capital markets for the 1997–2003 period. The degree of financial frictions exhibits a strong cyclical pa

8. Slide 8: Overview of the BGG Framework

   Overview of the BGG Framework Bankruptcy Costs: lender incurs costs proportional to total value of the firm if borrower defaults, giving rise to the external finance premium Leverage-Spread Schedule: inverse relationship between the firm’s net worth and its borrowing costs Sources of Declining Net W

9. Slide 9: Microfoundations of the BGG Framework

   Microfoundations of the BGG Framework Total Return to Capital: ωiRk QKi, where Rk (return to capital) and Q (price of capital) are known to both lender and entrepreneur: Rk > R (R = risk-free rate) QKi = Bi + Ni (value of firm = debt + net worth) Idiosyncratic Productivity Shock: ωi ∼ f (ω), with E(

10. Slide 10: The Optimal Debt Contract

    The Optimal Debt Contract The optimal debt contract specifies loan amount Bi and default productivity threshold ω i: • No Default: if ωi ≥ ω i, entrepreneur pays RbBi to lender, where RbBi = ω iRk QKi and entrepreneur keeps (ωi − ω i)Rk QKi. • Default: if ωi < ω i, then entrepreneur gets 0, and lend

11. Slide 11: The Optimization Problem

    The Optimization Problem Choose loan amount Bi and default productivity threshold ω i to max ψ(ω i)Rk (Bi + Ni) s.t. ξ(ω i; µ)Rk (Bi + Ni) = RBi • ψ(ω i)Rk (Bi + Ni) = expected return to entrepreneur: Z ∞ ψ(ω i) = (ωi − ω i)f (ω)dω ωi • ξ(ω i; µ)Rk (Bi + Ni) = expected return to lender: Z ω Z ∞ i ξ(

12. Slide 12: The Optimization Problem (contd.)

    The Optimization Problem (contd.) From definitions and FOCs, we obtain: Rk ψ 0(ω i) = (1) R ψ(ω i)ξ 0(ω i; µ) − ψ 0(ω i)ξ(ω i; µ) Bi ψ 0(ω i)ξ(ω i; µ) = − (2) Ni ψ(ω i)ξ 0(ω i; µ) Rb Rk ! Bi = ωi 1 + (3) R Ni R Analytical solutions to 1–3 obtained by assuming log-normality: σ2 ! ln ω ∼ N − , σ2 2 11

13. Slide 13: Figure 1A: Implications of Monitoring Costs

    Figure 1A: Implications of Monitoring Costs (annual rates in percent) Default Threshhold Default Probability 0.8 60 0.75 50 0.7 0.65 40 0.6 30 0.55 0.5 20 0.45 10 0.4 0.35 0 −0.5 0 0.5 1 −0.5 0 0.5 1 Log(Leverage) Log(Leverage) External Finance Premium Credit Spread over Risk−Free Rate 30 30 25 25 µ

14. Slide 14: Figure 1B: Implications of Idiosyncratic Risk

    Figure 1B: Implications of Idiosyncratic Risk (annual rates in percent) Default Threshhold Default Probability 1.6 100 90 1.4 80 1.2 70 60 1 50 0.8 40 0.6 30 20 0.4 10 0.2 0 −1 0 1 2 3 −1 0 1 2 3 Log(Leverage) Log(Leverage) External Finance Premium Credit Spread over Risk−Free Rate 30 30 25 25 20 20

15. Slide 15: Data Description

    Data Description We linked firm-level data from the following sources: • Compustat: income and balance sheet variables (quarterly) • Merrill Lynch: yields on outstanding publicly-traded corporate securities (daily) • Moody’s/KMV: expected default probabilities (monthly) 12

16. Slide 16: Data Description (contd.)

    Data Description (contd.) Panel dimensions: • 918 nonfinancial publicly-traded corporations • sample period: 1997Q1–2003Q3 (27 quarters) • 14,124 total observations • min. tenure = 4; median tenure = 14; max. tenure = 27 13

17. Slide 17: Key Variables

    Key Variables Leverage Ratio: book value of long-term debt relative to market capitalization Credit Spreads: corporate yield less estimated Treasury yield • Treasury yield curve estimated daily (Svensson, 1994) • tax adjustment (Cooper & Davydenko, 2002) Default Probabilities: expected default frequ

18. Slide 18: Summary Statistics

    Summary Statistics Variable Min Median Max Sales ($ bil.) 0.002 0.7 57 Mkt. Capitalization ($ bil.) 0.006 2.2 309 Leverage Ratioa 0.02 0.50 15.9 Credit Spread (%) 0.3 2.3 29 No. of Issues Traded 1 2 59 Avg. Portfolio Maturity (years) 1 8 30 Share of Traded Debtb (%) 2 51 100 S&P Credit Rating C2 BBB

19. Slide 19: Corporate Balance Sheets: Aggregate Growth Rate of Real Sales

    Figure 2A Corporate Balance Sheets Aggregate Growth Rate of Real Sales Percent 4 Our Dataset Compustat¹ Corr. = 0.95 2 0 −2 −4 −6 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 Note. Data are seasonally adjusted. ¹ All nonfinancial corporations.

20. Slide 20: Corporate Balance Sheets: Leverage Ratio

    Figure 2B Corporate Balance Sheets Leverage Ratio Percent 60 Our Dataset¹ Compustat¹ Flow of Funds² 50 40 30 20 10 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 Note. Data are seasonally adjusted. ¹ Median book value of long−term debt relative to market value of equity, weighted by firm sales. ²

21. Slide 21: Corporate Credit Spreads

    Figure 3A Corporate Credit Spreads Median BBB 7−10 Years Credit Spread Percentage Points 3.5 Our Dataset Merrill Lynch 3.0 2.5 2.0 1.5 1.0 1997 1998 1999 2000 2001 2002 2003 Note. Medians are weighted by the market value of bonds outstanding. Merrill Lynch data are for all nonfinancial issues.

22. Slide 22: Corporate Credit Quality

    Figure 3B Corporate Credit Quality Median Expected Year−Ahead Default Frequency Percent 1.0 Our Dataset Moody’s / KMV 0.8 0.6 0.4 0.2 1997 1998 1999 2000 2001 2002 2003 Note. Firm−level estimates of default weighted by firm liabilities.

23. Slide 23: Leverage and Credit Spreads

    Figure 4A Leverage and Credit Spreads Percentage Points 30 • • • • 25 Annualized Tax-Adjusted Credit Spread at t+1 • • • • • • • • • • • • ••• 20 • • • • • • • • • • •••• • •• • • • • • • •• • • • • • • • • •• • •• • • • • • • •• • • • • • ••• • • • • • • • • • • • •• • • • • • • • • • •• • • • • •

24. Slide 24: Leverage and Expected Probabilities of Default

    Figure 4B Leverage and Expected Probabilities of Default Percent • • •• • • •• • 20 • • • • • • • • • • • • • ••• • •• • ••• • • • • • • • • • • • • • • • •• • • • • • • • ••• • • • • • • • • •••• • • • • • • • • •• • • • • • • •• • • •• ••• • • •• • • • • •• • • • Expected Year-Ahead Default Freque

25. Slide 25: Estimation Methodology

    Estimation Methodology Assumptions: 2 , σ2 ) • Productivity shock: ln ωit ∼ N(−0.5σit it • Bankruptcy costs: µt common across firms but can vary over time. In period t, given µt, solve for ω it and σit B ψ 0(ω it, σit)ξ(ω it, σit; µt) = − N it−1 ψ(ω it, σit)ξ 0(ω it, σit; µt) 2 ! ln ω it − 0.5σit ED

26. Slide 26: Estimation Methodology (contd.)

    Estimation Methodology (contd.) Solutions ω ∗it and σit ∗ imply a contractual credit spread #∗ b " ! R B = ω ∗it 1 + R it N it−1 ψ 0(ω ∗it, σit ∗) ! × ψ(ω ∗it, σit ∗ )ξ 0 (ω ∗ , σ ∗ ; µ ) − ψ 0 (ω ∗ , σ ∗ )ξ(ω ∗ , σ ∗ ; µ ) it it t it it it it t Difference between actual and model-implied credit spr

27. Slide 27: Estimation Methodology (contd.)

    Estimation Methodology (contd.) RATINGit−1 is a fixed effect for firm i’s credit rating at the end of period t − 1 (i.e., liquidity premiums) uit is a mean-zero error term PN Objective: Choose µt to minimize i=1 u2 it Robustness check: • trimmed data (2.5th Pctl. ≤ leverage ≤ 97.5th Pctl. and spread

28. Slide 28: Estimation Methodology

    Estimation Methodology (annual rates in percent) Default Probability 100 90 80 70 60 50 40 σ = 0.5 σ = 0.25 σ = 0.125 30 20 10 0 −1 −0.5 0 0.5 1 1.5 2 2.5 3 Log(Leverage) Credit Spread 30 25 20 15 σ = 0.25 10 5 0 −1 −0.5 0 0.5 1 1.5 2 2.5 3 Log(Leverage)

29. Slide 29: NLLS Estimates of the Bankruptcy Cost Parameter µ

    Figure 5A The Magnitude of Financial Market Frictions NLLS Estimates of the Bankruptcy Cost Parameter µ 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1997 1998 1999 2000 2001 2002 2003 Note. The shaded region represents +/− 2 standard error bands computed using a heteroscedasticity−consistent asymptotic covarianc

30. Slide 30: The Magnitude of Financial Market Frictions: Recovery Rates

    Figure 5B The Magnitude of Financial Market Frictions Recovery Rates on Defaulted Bonds Percent 100 80 60 40 20 Model−implied (µ = 0)¹ Model−implied (estimated µ )¹ Actual² 0 1997 1998 1999 2000 2001 2002 2003 ¹ Average model−implied recovery rate weighted by the book value of bonds outstanding. ² A

31. Slide 31: Evolution of Model-Implied External Finance Premium

    Figure 6 Evolution of Model-Implied External Finance Premium 75th Percentile Percentage Points 4 Estimated µ Constant µ = 0.12 3 2 1 0 1997 1998 1999 2000 2001 2002 2003 50th Percentile Percentage Points 2.0 Estimated µ Constant µ = 0.12 1.5 1.0 0.5 0.0 1997 1998 1999 2000 2001 2002 2003 25th Percen

32. Slide 32: Evolution of Corporate Leverage

    Estimated µ Figure 7A Evolution of Corporate Leverage Constant µ = .12 75th Percentile Ratio 1.2 1.0 0.8 0.6 1997 1998 1999 2000 2001 2002 2003 50th Percentile Ratio 0.50 0.45 0.40 0.35 0.30 0.25 0.20 1997 1998 1999 2000 2001 2002 2003 25th Percentile Ratio 0.18 0.16 0.14 0.12 0.10 0.08 1997 1998 19

33. Slide 33: Evolution of Expected Default Probabilities

    Estimated µ Figure 7B Evolution Constant µ = .12 of Expected Default Probabilities 75th Percentile Percent 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 1997 1998 1999 2000 2001 2002 2003 50th Percentile Percent 0.6 0.5 0.4 0.3 Estimated µ 0.2 Constant µ = .12 0.1 0.0 1997 1998 1999 2000 2001 2002 2003 25th Perce

34. Slide 34: Evolution of Model-Implied σ

    Figure 7C Evolution of Model-Implied σ 75th Percentile 0.68 0.66 0.64 0.62 0.60 0.58 0.56 0.54 Estimated µ 1997 1998 1999 2000 2001 2002 2003 Constant µ = .12 50th Percentile 0.50 0.48 0.46 0.44 0.42 0.40 0.38 0.36 1997 1998 1999 2000 2001 2002 2003 25th Percentile 0.34 0.32 0.30 0.28 0.26 0.24 1997

35. Slide 35: Implications of Excluding Credit Ratings

    Figure 8A Implications of Excluding Credit Ratings NLLS Estimates of the Bankruptcy Cost Parameter µ 1.0 0.8 0.6 0.4 Baseline Estimate¹ 0.2 0.0 1997 1998 1999 2000 2001 2002 2003 Note. The shaded region represents +/− 2 standard error bands computed using a heteroscedasticity−consistent asymptotic c

36. Slide 36: Implications of Excluding Credit Ratings: Recovery Rates

    Figure 8B Implications of Excluding Credit Ratings Recovery Rates on Defaulted Bonds Percent 100 80 Baseline Estimate³ 60 40 20 Model−implied (estimated µ )¹ Actual² 0 1997 1998 1999 2000 2001 2002 2003 ¹ Average model−implied recovery rate weighted by the book value of bonds outstanding. ² Average

37. Slide 37: Leverage and the Volatility of Idiosyncratic Risk

    Figure 9 Leverage and the Volatility of Idiosyncratic Risk 1.4 · · 1.2 · · · · · · · · ·· · ··· · · · 1.0 ···· · ·· · · ·· · ··········· · ····· · ·· · · ···· ··· · · ······ ·· ······ · · Model−Implied σ at t+1 ········· ············· ················ · ·· · · · ····· · ······ ······················

38. Slide 38: Directions for Future Research

    Directions for Future Research Address limitations of existing framework: • allow for cross-sectional heterogeneity in µ (e.g., firm size, credit quality, etc.) • allow for non-Gaussian distribution of idiosyncratic productivity shock ω • allow for multi-period debt contracts 19

39. Slide 39: Directions for Future Research (contd.)

    Directions for Future Research (contd.) Investigate the link between model-implied external finance premium and capital expenditures. Examine macroeconomic implications of financial market frictions in light of micro evidence (e.g., the relationship between corporate leverage and financial fragility

40. Slide 40: Financial Fragility

    Financial Fragility Health of corporate balance sheets may be an important state variable in the transmission of shocks to the economy. (Bernanke & Gertler 1990, Bernanke & Campbell 1998, etc.) Examine the link between firms’ financial condition and model-implied external finance premium over the bu

41. Slide 41: Time-Varying Leverage External Finance Premium Schedule

    Time-Varying Leverage External Finance Premium Schedule Semiparametric Specification Percentage Points 35 30 2003:Q3 2003:Q2 2003:Q1 Annualized Model-Implied External Finance Premium at t 2002:Q4 2002:Q3 25 2002:Q2 2002:Q1 2001:Q4 2001:Q3 2001:Q2 20 2001:Q1 2000:Q4 2000:Q3 2000:Q2 2000:Q1 15 1999:Q4

Banque Nationale De Belgique — Technical Presentation (April 2004) on 6ix