Interest Rate Risk Management for a Government Debt Portfolio

1
Interest Rate Risk Management for a Government
Debt Portfolio

• Paul Ferris
• paul.w.ferris_at_hotmail.com
• 7 November 2007

2
Government Borrowers- who are we talking about ?

• Government Borrowers
• Sovereign Commonwealth of Australia
• State Government
• Local Government
• Government Owned Authorities e.g. water, rail
• Focus is on government borrowings funding real
  assets
• However,
• Commonwealth no net debt
• States are in a range of different circumstances
• In some cases debt is partly funding financial
  assets
• Similar principals applicable to range of other
  borrowers

3
Interest Rate Risk is one of a number of
financial risks

• Interest Rate Risk (including yield curve risk)
• Unexpected increase in (a) interest costs, or (b)
  market vale of debt from unexpected interest rate
  changes
• Spread/Basis Risk (form of interest rate risk)
• Increase due to relative rate changes
• Currency Risk
• Due to currency fluctuations
• Liquidity/Funding Risk
• Liquidity inabilty to meet cash obligations as
  they fall due
• Funding inability to rollover debt
• Credit Counterparty Risk
• Counterparty risk Failure of counterparty to
  met obligations e.g. on swaps or repurchase
  agreements
• Operational Risk
• Loss due to operational failure e.g. tender
  process, settlement process, legal contract
• Focus is on Interest Rate Risk

4
Clarifying Objectives is a key issue for Interest
Rate Risk Management for Government borrowers

• Management requires objectives.
• Lowering the expected cost of debt
• Lowering risk
• How to measure the outcome e.g. accounting
  interest cost or market value of debt
• Example government borrower has just rolled
  over 10 billion of fixed rate debt for five
  years at 7 and then 2 days later rates increase
  to 8
• Should the debt manager be pleased that mv has
  fallen
• Should the debt manager be concerned they should
  have locked in for longer
• What outcomes are we interested in?
• What are the objectives ? ( in terms of risk to
  those outcomes and expected outcomes)
• given those objectives, what should be the debt
  benchmark ?

5
Potentially Relevant Outcomes

• Measure Accounting Outcomes
• (Accrual rather than mark-to-market
  accounting)
• Interest Accrual in current year
• Average Interest Accrual over budget forecast
  period, e.g. 3 years
• Average Interest Accrual over a longer forecast
  period, e.g. 10 years
• Impact of interest costs on budget
• Measure Economic Outcomes
• Accumulated value of debt after 3 or 10 years
• Accumulated real value of debt after 3 or 10
  years
• Accumulated value of debt as proportion of
  GDP/GSP
• Interest expense as a proportion of government
  revenues
• Other Outcomes
• Diversity of funding base
• Maintaining market lines of stocks

6
Potential Objectives

• Expected Outcome
• Risk
• Impact of immediate or gradual 1 rate increase
• Volatility of Outcome (statistical)
• This multiples the potential measures, therefore
  need to cull.
• Focus in order on -
• Medium/long term risk measures - e.g.
  Volatility of Accumulated debt/GDP
• Short term risk measures - e.g.
  Volatility of current year interest cost
• Expected outcome measure - e.g.
  Expected accumulated debt in 10 years

7
Rationale for Outcomes and Objectives

• Accounting objectives essentially what is
  reported in budget
• Especially the current and next year
• Economic objectives look at what the economy
  (i.e. tax payers) capacity to service
• Projected and actual accumulated debt after x
  years
• Real dollars rather than nominal dollars
• Proportion of GDP/GSP
• Resolution (Potential Objectives)
• Two objectives
• Economic risk objective such as minimising
  accumulated real debt after 10 years
• Minimising expected cost
• One constraint
• Subject to acceptable (e.g. x dollars ) level of
  volatility in current year and next year s
  accounting outcomes

8
Passive vs Active Management

• Benchmark is likely to be the same for active or
  passive management
• Active Management
• Accounting outcomes are accrual
• Performance to benchmark has to be
  marked-to-market accounted
• Buying back debt - crystallising gains or losses
• Active Management thru a derivative overlay

9
Role of Central Borrowing Authorities

• Intermediary between government and market place
• Balanced balance sheet
• Mark-to-market accounting
• Can immunize their own interest rate risk
• No benchmark required, objectives much clearer

10
Why a typical government borrower cant immunise
its risk

• Assets significantly exceed liabilities
• Assets are generally non-financial or where
  financial are equity rather than debt
• Off balance sheet revenues and expenses can
  dominate
• E.g. tax receipts and cost of providing services
• These revenues and expenses are highly correlated
  to inflation as are interest rates
• Therefore require benchmark
• No way of eliminating all risk
• No obvious risk neutral position
• No other means of measuring debt management
  performance
• Our aim is to get the government debt manager as
  close as we can to the same clarity of a typical
  fund manager with a benchmark by addressing those
  issues

11
Funding Financial Assets

• Government Balance Sheet often has financial
  assets, in particular loan assets
• Should be segregating these and segregate
  specific funding for them
• Therefore, a debt portfolio in two portions
• That funding financial assets
• No benchmark is required as you can immunise
  assets and liabilities
• Other funding which is managed to benchmark

12
Typical Government Debt Portfolio Benchmark

• Typical government debt benchmark
• 80 fixed rate debt
• Rolling 10 year i.e. 8 maturing in 12 months,
  8 in 24 months,.. 8 in 10 years
• 20 floating rate debt
• Benchmark can be described in terms of parameters
• Modified Duration of 4 years
• 20 floating rate debt
• Benchmark specified in terms of parameters
• Simpler and more practical but
• precludes performance measurement unless you use
  an overlay

13
Rational for Benchmark

• What if we lock in fixed rates for 10 years or
  more, i.e duration of around 8 years
• Positive
• Stable interest cost for 10 years or more
• Negative
• Mark-to-market risk
• Early repayment risk
• Positive yield curve could make it more expensive
• Inflation rates fall
• Debt locked in in nominal terms but has increased
  in real terms
• Should consider analysis in real terms as well
  as nominal dollars

14
Real and Nominal Interest Rates

• To analyse or model real debt outcomes
• Volatility of real rates,
• Volatility of nominal rates,
• Volatility of inflation rates and
• Correlation between the two
• Implications of considering real rates
• Modelling real outcomes tends to discourage
  locking in debt for longer terms
• Longer debt
• protects against real rate increases
• but
• Creates a risk to inflation rates falling
• Therefore, for example, might choose a duration
  of 4 years rather than 6 years

15
Indexed Debt

• Indexed debt has been issued by both the
  Commonwealth and most States
• Indexed debt can be considered a hybrid between
  fixed and floating debt
• Interest Rate Real rate Inflation rate
• Fixed Rate debt has real rate and inflation rate
  fixed
• Floating rate debt has real rate floating and
  inflation rate floating
• Indexed debt has a fixed real rate and floating
  inflation rate
• Introducing Indexed debt can act to reduce risk
  in real terms under a wide range of measures and
  assumptions
• Compared with a portfolio of fixed and floating
  rate debt
• Resolution
• Introduce 20 or more indexed debt on risk
  reduction grounds for most government borrowers
• Expected Cost - Currently real rate on indexed
  debt significantly lower than implied real rate
  on nominal debt (unless you are not confident
  about RBA in the medium term)
• Consider switching 10 year fixed nominal debt for
  20 year indexed debt

16
Purpose of the benchmark

• Describes a debt portfolio that is deemed
  consistent with risk and expected cost objectives
• Describes a risk neutral position (rate view
  neutral position)
• Basis for performance measurement
• Question is how to determine the benchmark

17
Determining the Debt Benchmark

• Identify key relevant Outcomes
• Accounting, Economic Other
• Identify Objectives
• Expected Outcomes, Volatility of Outcomes, Stress
  Outcomes
• Universe of Potential Debt Portfolios
• E.g. Debt of various maturities with and without
  swaps, currency restrictions, maturity
  restrictions, options
• Forecast Period
• For how long are you going to forecast interest
  costs, the budget and accumulated debt
• Expected Path of Interest Rates for forecast
  period
• How much more expensive (if at all) are you going
  to assume 10 year fixed rates are in the longer
  term
• Methodology how to model interest rate
  volatility
• Scenarios or Monte Carlo analysis
• Holding Period analysis
• Model interest rate volatility over maybe 3 years
• Potential Interest Rate volatility

18
Universe of Potential Interest Rate Risk Profiles

• Liquidity/Funding risk neutral
• Modelling and decisions based on swap rates
• Decide the Interest Rate profile not funding risk
• Endless number of potential combinations of
  different maturities
• Duration parameter (Economic Outcomes)
• Any two portfolios with same duration will have
  similar cost/risk trade-off
• Floating rate proportion (Current year accounting
  outcomes)
• Proportion of debt maturing in current year
  determines current year accounting volatility
• Only consider efficient debt/swap combinations
• No currency risk ?
• No options risk ?
• Nominal and Indexed Debt acceptable ?

19
Forecast period

• Project interest costs and other budget items
  over forecast period, e.g. 10 year period
• Baseline projection based on expected path for
  interest rates
• Alternate projections based on alternate interest
  rate risk scenarios
• Forecast period needs to be
• At least as long as longest debt in portfolio
• As long as life of assets being funded
• Need to model more than one forecast period

20
Expected Path of Interest Rates

• Need to determine what debt portfolios have the
  lowest expected cost
• Historically market yield curve has been positive
  sloped
• If, market forward swap rates rate
  expectations
• All portfolios have same expected cost/outcome
• Debt benchmark can then be chosen solely on risk
  grounds
• Only need the relativities between short term and
  long term rates
• Resolution
• Term premium between bank bill rate and ten year
  swap rate
• Implications of zero
• Implications of 0.80

21
Modelling interest rate volatility

• Modelling
• Scenario modelling
• Monte Carlo modelling
• Monte Carlo modelling
• Mathematically generate 10000 interest rate
  scenarios from statistical process
• Forecast all of the relevant outcomes under each
  scenario
• Identify mean outcomes and standard deviation or
  volatility of outcome
• More sophisticated but usual dependence of
  assumptions
• Scenario modelling
• Identify a range of specific interest rate
  assumptions
• E.g. immediate 1 increase in rates
• Simpler and more transparent but potentially
  arbitrary choice
• Resolution
• Scenario modelling is generally sufficient
• One key risk factor, parallel shock to rates
• Outcomes are either linear or close to linear

22
Holding Period

• 1 year, 3 year or 10 year holding period analysis
• Holding period can be and commonly is shorter
  than forecast period

23
Determining the Debt Benchmark

• Identify key relevant Outcomes
• Accounting, Economic Other
• Identify Objectives
• Expected Outcomes, Volatility of Outcomes, Stress
  Outcomes
• Universe of Potential Debt Portfolios
• E.g. Debt of various maturities with and without
  swaps, currency restrictions, maturity
  restrictions, options
• Forecast Period
• For how long are you going to forecast interest
  costs, the budget and accumulated debt
• Expected Path of Interest Rates for forecast
  period
• How much more expensive (if at all) are you going
  to assume 10 year fixed rates are in the longer
  term
• Methodology
• Scenarios or Monte Carlo analysis
• Holding Period analysis
• Model interest rate volatility over maybe 3 years
• Potential Interest Rate volatility