Unfunded Pension Liabilities

1
Unfunded Pension Liabilities The Corporate CDS
MarketRonan GallagherFinance PhD
CandidateQueens University Belfast

• ATRC 13th August 2009

2
Introduction

• PhD Essays on Unfunded Pension Liabilities
  Financial Markets
• 2nd of 3 Essays
• Co-authored with Professor Donal McKillop
• Focus on Defined Benefit schemes and the Credit
  Default Swap market
• Does a firms pension deficit play a significant
  contributing role in its overall default risk as
  measured by the Credit Default Swap market?

3
Context

• Perfect Storm (2001 - 2002)
• Falling equity markets
• Falling discount rates
• Prevalence of significantly underfunded schemes
• Triggers regulatory perfect storm
• Enhanced role of pension regulator
• Widespread reform to pension accounting
• FRS17, IAS19, FAS87/158
• Deficits brought on balance sheet

4
Context

• Secondary wave of Perfect Storm
• Credit crunch
• Onset of global recession
• Equity market declines
• Falls in Central Bank Interest Rates
• Accounting Quirk
• Standards advocate using the AA (/ high
  quality) corporate bond yield as discount rate
  for pension liabilities
• Credit crunch has exposed a frailty in this
  approach

5
Context

• Yield spread widened significantly
• AA rated yield at 31/12/2008 was 111 basis points
  higher than its 5 year average
• Historically high discount rates serve to lower
  reported liabilities below their true economic
  value
• Yield Risk
• Improving conditions in credit markets
• Effect of Quantitative Easing
• Reform to accounting standards
• Longevity Risk

6
Context
Source DataStream Authors Calculations
7
Some Striking Statistics

• Estimated Aggregate Deficit (07/08/2009)

8
Some Striking Statistics

• Company Level Data (2008 Fin Year)

9
The Pension Plan The Firm

• Theoretical Literature
• Two schools of thought
• Traditional balance sheet separation
• Integrated balance sheet augmentation
• Our study is an implicit test of the integrated
  perspective
• Empirical Literature
• Corporate Equity Mixed results
• Tests conducted on raw returns, betas and
  portfolios
• See Feldstein Morck (1983), Jin et al (2006)
  and Franzoni Marin (2006)
• Pension accounting veil particularly in US
• Corporate Debt More support for integrated
  school
• Tests conducted on credit ratings and bond
  spreads
• See Carroll Niehaus (1998) and Cardinale (2007)
• Asymmetry between surplus and deficit effects

10
The CDS Market

• Credit Default Swaps (CDS) are credit derivatives
• Swap contracts in which
• buyer makes series of payments to seller
• in exchange receives payoff if credit instrument
  goes into default
• Created in 1997 by JP Morgan Chase
• Designed to shift risk of default to a
  third-party
• Exponential growth until onset of credit crunch
• Spread (premium) expressed in basis points of
  notional principal per annum
• Ceteris paribus, spread is a direct measure of
  the default risk of the reference entity

11
The CDS Market

• Why focus on this measure of default risk?
• Competing with credit ratings and bond spreads
• Free from conflicts of interest associated with
  ratings
• Continuous variable avoids discrete
  categorisation bias
• CDS market is much more liquid than the bond
  market
• Price discovery is faster (Blanco et al 2003 )
• Unburdened by choice of a reference risk free
  yield curve in order to extract a spread

12
Modelling Default Risk
Traditional Models
Structural Models

• Based on fundamental analysis
• Examine factors such as cash flow adequacy, asset
  quality, earning performance and capital adequacy
• Draw a small set of accounting variables,
  financial ratios and other information into a
  quantitative score.
• In some cases this score can literally be
  interpreted as the probability of default where
  in others it is used as some sort of
  classification system (e.g. credit ratings).
• e.g. Beaver 1966, Altman 1968,1975

• Contingent claims analysis
• If at time T the firms asset value exceeds the
  promised payment, D, the lenders are paid the
  promised amount and the shareholders receive the
  residual asset value
• If the asset value is less than the promised
  payment the firm defaults, the lenders receive a
  payment equal to the asset value, and the
  shareholders get nothing
• ET max(AT D, 0)
• Similar to option payoff
• Merton (1974) model emerged from early work
  utilising Black Scholes framework
• Continual revision Moodys KMV model

13
The Merton (1974) Model

• Equity limited liability
• ET max(Residual Asset Value, 0)
• Equity Call option on firm assets with strike
  price equal to the face value of debt
• Iteratively solve system of simultaneous
  equations such that the Black Scholes option
  price equals the market value of the firms equity
  

• Solver routine jointly implies an asset value and
  volatility which is used to generate a default
  probability with respect to a default point
• CONVENTION Default Boundary Point ST Debt
  (0.5 LT Debt)
• The distance to default is the number of
  standard deviations by which the expected asset
  value exceeds the default point (and is a
  function of the implied asset value and
  volatility).

14
Adjusting The Merton Model

• Traditional default point is ST Debt (0.5 LT
  Debt)
• Does not include pension liabilities which as a
  result of FRS17 / IAS19 / FAS87-158 are on
  balance sheet
• Adj Default Point ST Debt (0.5 LT
  DebtPDEF)
• Pension deficits drive firm closer to default
• Re-solve for asset value and implied volatility
• Newton-Raphson iterative simulation
• Generate PADTD
• DTD PADTD Pension Risk
• New pension risk measure

15
Dataset Estimation

• Matched CDS data from Datastream and accounting
  data from ThomsonReuters One Banker by hand
• 1, 5 and 10 year CDS instruments
• Constrained to 2003 due to pension disclosure
• 5 years data, roughly 500 pension sponsors,
  unbalanced panel
• US, UK and European coverage
• Employed a hybrid structural and traditional
  credit risk model
• Modelled in a fixed effects panel due to
  unobserved heterogeneity

16
The Model

• Dependent Variable
• Log(CDS Premium)
• Structural Variables
• Distance to Default, Pension Risk
• Traditional Variables
• Firm Size, Return on Assets, Income growth
• Interest Coverage, Quick Ratio, Sales Growth,
• Book Leverage, Retained Earnings

17
Principal Findings

• See table 1
• Traditional and Structural variables are
  complements
• Defined benefit pension risk is a highly
  significant determinant of 1 and 5 year CDS
  premia
• CDS market participants price pension risk such
  that it is a legitimate contributing factor in
  default risk
• A unit increase in the distance by which the
  pension plan pushes the firm closer to its
  default point increases 1and 5 year CDS premia by
  6 and 2 respectively
• This supports the integrated balance sheet
  perspective of the pension plan
• Does the relationship apply equally to surplus
  and deficits?
• This and other studies show that the company is
  indeed responsible for a pension deficit but can
  it reap the rewards of an overfunded scheme?

18
Testing For Asymmetry

• Empirical Expansion
• Create subdivided pension risk variables
  (PRISKSUR PRISKDEF)
• If surplus, PRISKSUR PRISK, otherwise 0
• If deficit, PRISKDEF PRISK, otherwise 0
• If there is zero asymmetry the coefficients on
  each pension risk subdivision should be
  significant and equal in magnitude
• See table 2
• Companies running a pension surplus do not attain
  a reduction in their default risk as measured in
  the CDS market
• Deficits increase 1 and 5 year CDS premia to a
  similar extent as in the initial specification
• Why the asymmetry?
• Mandated sharing of excess pension assets with
  scheme members
• Benefit Improvements, contribution holidays etc
• Enshrined in prevailing legislation

19
The Models Explanatory Power

• Distance to default is a key driver of models
  performance
• Pension risk and traditional variables such as
  firm size, return on assets and book leverage are
  incrementally informative
• Performance Metrics

20
Cumulative Accuracy Profiles

• Formed by creating cumulative percentile bins of
  the actual and predicted log-premia and charting
  the overlap between bins
• Examine rank order predictive ability of default
  risk models
• Accuracy Ratio 2(Area between CAP and 45
  degree line)

21
Conclusions

• Perfect storm in defined benefit pension plans
• Further pain on the horizon
• Sponsors responsible for funding
• Deficits on balance sheet
• CDS premia is a market measure of default risk
• Pension risk significantly increases sponsoring
  firms CDS premium
• Defined benefit pension plan underfunding
  contributes significantly to default risk
• Implications for investors and sponsoring firms
• Contributions
• New structural PR measure
• First study on the CDS market