23
W W W . W A T S O N W Y A T T . C O Real life ALM Russell Beaumont 16 April 2004

Russell Beaumont, Watson Wyatt

Embed Size (px)

DESCRIPTION

 

Citation preview

Page 1: Russell Beaumont, Watson Wyatt

W W W . W A T S O N W Y A T T . C O M

Real life ALMRussell Beaumont

16 April 2004

Page 2: Russell Beaumont, Watson Wyatt

2Copyright © Watson Wyatt Worldwide. All rights reserved.

Agenda

Why life companies use ALM

Different approaches used

Practical considerations

Example for CEE markets

Two further examples

Page 3: Russell Beaumont, Watson Wyatt

3Copyright © Watson Wyatt Worldwide. All rights reserved.

Why companies use ALM

Understand financial dynamics of business

Helps identify and quantify risks

Improved – capital allocation– investment strategy– liability management (products, bonuses, guarantees)

Better informed decision-making

Enhanced business performance

For (future) regulatory requirements

Page 4: Russell Beaumont, Watson Wyatt

4Copyright © Watson Wyatt Worldwide. All rights reserved.

Reasons for growth in ALM

Economic conditions– poor equity performance / low interest rates– cost of guarantees– impact on solvency levels

Competitive pressure– product innovation– maturity value out-performance– if others are, why don’t we?

Increased systems power / capacity

Page 5: Russell Beaumont, Watson Wyatt

5Copyright © Watson Wyatt Worldwide. All rights reserved.

ALM applications

Investment strategy

Product development

Statutory reserving

Business planning

Risk management

Realistic reporting

Capital allocation

Performance measurement

Page 6: Russell Beaumont, Watson Wyatt

6Copyright © Watson Wyatt Worldwide. All rights reserved.

Deterministic approach

Deterministic approach involves choosing individual economic scenarios to investigate, eg– best estimate– market crash– high / low interest rates

+ Advantages+ consider scenarios specific to circumstances+ useful first step to gain basic understanding+ quicker to generate results

– Disadvantages– scenarios chosen are limited and subjective– no indication of weight for each scenario– may overlook key scenarios

Page 7: Russell Beaumont, Watson Wyatt

7Copyright © Watson Wyatt Worldwide. All rights reserved.

Stochastic approach

Model determines future economic parameters using mathematical relationships between them

+ Advantages+ many scenarios can be randomly generated+ mathematical relationships based on empirical evidence+ can be calibrated to current market conditions

- Disadvantages- practical considerations may limit number of scenarios

that can be investigated- harder to analyse results to check accuracy

Page 8: Russell Beaumont, Watson Wyatt

8Copyright © Watson Wyatt Worldwide. All rights reserved.

What approach in practice?

Start with current programs e.g. valuation, embedded value

Investigate a number of deterministic scenarios, including extreme scenarios to increase understanding of model

Develop management decision rules based on above

Add stochastic asset model

Page 9: Russell Beaumont, Watson Wyatt

9Copyright © Watson Wyatt Worldwide. All rights reserved.

Some practical considerations

What are the objectives of the ALM study?

Choice of asset model

Model point creation

Decision rules

Projection period

Page 10: Russell Beaumont, Watson Wyatt

10Copyright © Watson Wyatt Worldwide. All rights reserved.

Objectives of the ALM study

Clear and agreed objectives for the study

Understanding the objectives will help determine the answers to the other practical issues

Don't try to answer too many questions with one study - confuses the message and the modelling!

Choice of asset model depends on the objectives– econometric– stochastic

Page 11: Russell Beaumont, Watson Wyatt

11Copyright © Watson Wyatt Worldwide. All rights reserved.

Choice of asset model

Econometric– long and short term

models– uses historical data– aims for best estimate

‘realistic’ probability distribution

– used for investigating extreme scenarios

– used for– business planning– asset allocation– risk management

Market consistent– uses current stock market

and derivative data– aims to replicate market

prices– gives average values,

based on likelihood of each scenario

– used for– pricing guarantees– liability hedges e.g.

guaranteed annuities– realistic balance sheets

The choice of asset model depends on the objective

Page 12: Russell Beaumont, Watson Wyatt

12Copyright © Watson Wyatt Worldwide. All rights reserved.

Long-term econometric models Assess risk and reward for different asset allocations over

the long term

Theory, history and judgement combined

Expected returns relate to views of the future

Uncertainty (volatility) judged with an eye to the past

Investment manager views also included

Example structure– Inflation drives bond yields, which drive equity total returns– Inflation & bond yields autoregressive– Equities random walk– Returns are log-normally distributed

Page 13: Russell Beaumont, Watson Wyatt

13Copyright © Watson Wyatt Worldwide. All rights reserved.

Short-term econometric models Derives scenarios to stress test the free capital Concentrates on events that occur at the tail over short term

horizons Covers a wide range of economic and non-economic risks to

which a life company is exposed, including– asset and interest rate levels and volatility– credit spread levels– persistency levels– mortality levels– operational risk losses

Some risk factors need estimating for the model. Once calibrated, the model is run to produce 10,000

scenarios of monthly risk factor information for 1 year

Page 14: Russell Beaumont, Watson Wyatt

14Copyright © Watson Wyatt Worldwide. All rights reserved.

Key requirements for market consistent model

Ability to replicate today’s market prices Simple calibration process Efficiency (convergence to market prices) Transparency Replicating portfolios an alternative approach What happens when the market is not complete?

– length of options– away from strike prices– options not available

Page 15: Russell Beaumont, Watson Wyatt

15Copyright © Watson Wyatt Worldwide. All rights reserved.

Model point creation

Compromise between number of model points and model run times

What are your objectives, eg are you: – investigating future cashflows; or – investigating future solvency position

Reconciliation to full data runs

Appropriate criteria for grouping business– criteria for being 'in the money'

Page 16: Russell Beaumont, Watson Wyatt

16Copyright © Watson Wyatt Worldwide. All rights reserved.

Decision rules Need to reflect what would happen in practice

– management decisions– policyholders' decisions

Need to be objective, not subjective Should be realistic and achievable, e.g. in line with

policy documentation, projections, etc. Examples for asset mix:

– equity proportion = 40% if solvency level > 10%– else equity proportion = 20%

Withdrawal rates– surrender rates 5%– rates increase to 50% if guarantee 'in the money'

Page 17: Russell Beaumont, Watson Wyatt

17Copyright © Watson Wyatt Worldwide. All rights reserved.

Projection period

What are the objectives, e.g.– long term asset allocation– short term solvency cover

Need to cover significant future events, e.g. maturity of large block of business

The longer the projection period the bigger the ‘funnel of results’ (econometric model)

Availability of financial instruments for calibration of market consistent model

More scenarios needed for market consistency

Impact on run times

Page 18: Russell Beaumont, Watson Wyatt

18Copyright © Watson Wyatt Worldwide. All rights reserved.

Current issues for CEE markets

Decreasing interest rate environment

Some relatively high technical interest rates

EU accession speeds the process

Multinational head offices already developing techniques

Asset models for CEE markets?

Page 19: Russell Beaumont, Watson Wyatt

19Copyright © Watson Wyatt Worldwide. All rights reserved.

Example stress test: Hungary

Yield curve falls immediately to Euro-zone rates

Simple deterministic run on representative policy

What additional reserves required?

0%

2%

4%

6%

8%

10%

12%

0 5 10 15

Term

Yie

ld

Yield curves

Eurozone

Hungary

Page 20: Russell Beaumont, Watson Wyatt

20Copyright © Watson Wyatt Worldwide. All rights reserved.

Stress test results

0

20

40

60

80

100

120

140

160

180

Res

erve

s(H

UF

'000

s)

3.5 4.5 5.5

Techical interest rate (%)

Base reserve Cost of guarantee

Page 21: Russell Beaumont, Watson Wyatt

21Copyright © Watson Wyatt Worldwide. All rights reserved.

Example 1: Group wide study

Study in mid-2001 for UK insurer with bank, life and non-life subsidiaries, and final salary pension scheme

Long-term econometric model to help asset allocation

Key findings– highlighted conflict between short term statutory solvency

and long term policyholder performance– asset model does not assume market recovery– buy protection from equity falls and monitor closely– tax and risk/reward moves equities out of pension fund– proposed asset allocations and close matching principles

for different parts of business

Page 22: Russell Beaumont, Watson Wyatt

22Copyright © Watson Wyatt Worldwide. All rights reserved.

Example 2: FSA calibration

Help FSA in calibration of enhanced capital requirement Method

– Develop an "average" realistic balance sheet– Develop a stochastic model of multiple risk factors – Project the RBS over one year stochastically– Derive required capital at a range of confidence levels– Generate risk capital margin stress tests that require this

level of capital– Carry out sensitivities to the assumptions made– Comment on the appropriate structure of the RCM stress

tests Results expected to be published in Spring 2004

Page 23: Russell Beaumont, Watson Wyatt

W W W . W A T S O N W Y A T T . C O M

Real life ALMDiscussion