The role of a quantitative tool in Mtds analysis

1. The role of a quantitative tool in Mtds analysis UNCTAD, World Bank and IMF Workshop Geneva, February 06-10 2017

2. Outline • What is a risk model? Why is it useful? • The structure of a simple scenario analysis model • Model input, engine and output • Description of the simulation process • Implementation issues • How are scenario models used in practice? • Off-the-shelf or in-house developed model? • Scenario analysis vs. stochastic modeling • Summary

3. What is a Model? • A “model” is a representation of a system for the purpose of studying it • A specific representation of something more general and usually more complex • We need a model to • understand the relationships between the components of a system • to predict how system will operate under a new policy or in a new environment

4. What MAY WE Need a Model: PROBLEM DEFINITION Public Debt Management (PDM) uses bonds/bills/loans to meet financing needs of the government considering • the costs • the risks associated • and other macroeconomic policies The objective is to find an “optimum” combination of debt (debt management strategy) in terms of • maturity (short vs. long term) • coupon structure (zero-coupon or coupon bonds) • interest type (fixed vs. floating rate) • currency (local vs. foreign)

5. What MAY WE Need a Model: PROBLEM DEFINITION Complexities : • Objectives are conflicting • Borrowing short term is generally less costly, • but more risky • Uncertainty in • macroeconomic variables • market conditions

6. What MAY WE Need a Model? In the area of public debt management, models canbe used • to evaluate different strategies under alternative scenarios for future interest rates and exchange rates in a systematic analysis • Simplified representation of the debt process • to provide an understanding of the trade-offbetween cost and risk in the portfolio • requires clear definition of cost and risk • to help make decisions about the composition of public debt • Supports the identification and choice of indicators/targets

7. Models are widely used by debt managers to provide input to decision-making, and to better understand the cost and risk trade-offs Provides supplement to qualitative analysis A model should only contain elements that are needed to answer specific questions Additional details => additional complexity Risk Models in Debt Management

8. Is a Model Needed to Develop a Debt Strategy? • No, recent examples include • Indonesia, Peru, Colombia • In the above cases, the strategy was initially formulated as broad guidelines based on “intuition” • More domestic debt • Longer maturity etc. • A natural next step is providing more precision in the form of targets for specific risk indicators – this requires a model

9. The Structure of a Scenario Analysis Model • INPUT • Existing debt cash flows • Macro Variables • Primary fiscal balance • Structure of new debt • Borrowing strategy • Financial variables • - Exchange rates • - Interest rates OUTPUT Cost Risk ENGINE Cash-flow

10. The Structure of a Scenario Analysis Model • How much to finance? • Funding needs • How to finance with which instruments? (Strategies) • Which instruments: maturity, currency type, interest rate type etc. • At what price to finance? (Scenarios) • Interest rates and exchange rates • Time Horizon and unit for cost measurement • Medium to long term time horizon • Local currency?

11. Basic Budget Arithmetic – the Foundation for the Scenario Analysis Model Primary Balance - Interest payments = Fiscal Balance - Principal payments = Funding need • Future debt charges and redemptions – and therefore future funding needs – will depend on the borrowing actions as well as market rates • A higher than expected funding need can be the result of loose fiscal policy and/or higher market rates

12. INPUT: EXISTING DEBT CASH FLOWS • Cash flows based on outstanding balance as of a specific date (often end of previous fiscal year) • By currency • Domestic debt • Foreign currency debt (USD, EUR, JPY, etc) • By interest rate • Fixed • Floating • Information should be available from debt recording system

13. Input: Macroeconomic Variables • In • What are the future paths for the macro economy? • Primary balance projections • Expenditure plans • Projected revenues • This will determine the new borrowing requirement • Projections for GDP growth and revenues if used for cost and risk indicators • These variables are usually exogenous (though some countries have tried to use structural models to link interest rates and GDP)

14. Input: Structure of New Debt • What is amount of new debt that needs to be issued? • Primary balance • Maturing existing debt • Interest cost of existing debt • Maturing ‘new’ debt • Interest cost of ‘new’ debt • Assumption: Funding need fully covered by borrowing • Select a borrowing strategy: For example, all domestic, 50% 1 year / 50% 5 year

15. Input: Financial Variables • Scenarios for future market rates • Exchange rates • E.g. existing rates, Interest Rate Parity etc. • Interest rate • E.g. existing rates, forward rates etc. • Among potential scenarios for future market rates a baseline scenario is chosen – this will function as the basis for measuring cost and risk • A sound design of the base scenario is vital for the risk analysis

16. Model Output: Cost and Risk Cost Risk Scenario 1 Risk1,X Baseline Scenario Cost1,X Time

17. Cash Flow Simulation • Decide on the time frame of analysis, e.g. 5 years • The debt service flows generated by the baseline scenario for a given new debt issuance strategy will be defined as the expected cost • The sensitivity of a borrowing strategy to market rates can be analyzed by comparing cost and risk under alternative scenarios for market rates • Different borrowing strategies can be analyzed by comparing cost and risk for one or more risk scenarios for market rates

18. Example of Model Output Interest to GDP, end of period Debt to GDP, end of period • The above charts are based on the future cash flows generated by a model • Identifying the preferred strategy is typically difficult – different cost indicators will give different ranking

19. Using Debt Portfolio Modeling in Practice • Pre-requisites for modeling • High quality and timely data on the outstanding debt portfolio • Dedicated staff with good knowledge of spreadsheets and finance • Issues for modeling • Selection of market variable scenarios, or period of history for parameterizing a simulation, may be difficult when the economy has been through periods of instability • The process of developing a model represents a considerable investment

20. Typical Experiences from Working with Risk Modeling • Not the main basis for decision-making, rather a supplement to experience, sound judgment etc. – provide additional information for making better choices • Increase knowledge of the cost/risk trade offs • Requires dedicated resources, time-consuming • Clarifies framework for decision-making

21. Model development requires Adequate staff and software Time – often trial and error Focus on key person risk Buying an off-the-shelf model is tempting, but supply is very limited – and will often imply acquiring a black box The MTDS Analytical Tool Off-the-Shelf or In-House-Developed Model?

22. A simple deterministic scenario analysis model provides a basis for more advanced stochastic models In stochastic models the number of market scenarios are increased from a few to several thousand Allows quantification of cost and risk Cost-at-Risk models are related to the VaR concept “What is the maximum cost of the debt in a given year with a probability of 95%” Deterministic vs. Stochastic Scenario Analysis

23. A scenario model provides input on the direction and magnitude of risks – requires clear definition of cost and risk A simple scenario analysis model can provide input on The costs and risks of the existing borrowing strategy The choice between alternative borrowing strategies Strategic targets can be derived from the cost/risk analysis Trial and error process that is very time consuming A simple scenario model provide the basis for more advanced Cost-at-Risk models Summary

24. In LieU of Conclusion • “Remember that all models are wrong; the practical question is how wrong do they have to be to not be useful. • …all models are approximations. Essentially, all models are wrong, but some are useful.” • George E.P. Box

25. “A theory has only the alternative of being right or wrong. A model has a third possibility: it may be right, but irrelevant.” • Manfred Eigen

26. “I can never satisfy myself until I can make a mechanical model of a thing. If I can make a mechanical model, I can understand it. As long as I cannot make a mechanical model all the way through I cannot understand.” • William Thomson Kelvin

27. “Everything Should Be Made as Simple as Possible, But Not Simpler’ • Attributed to Albert Einstein