How to perform the Asset Liability Modelling to determine Investment Strategy

 How to perform the Asset Liability modelling to determine the investment strategy.

 First, let’s understand Asset liability Modelling

Actuary try to project company’s assets and liabilities into future and then check whether insurance company has sufficient assets to meet there liabilities or not. Along with that Actuary also checks whether the solvency position as required by regulator is satisfactory or not.

Asset liability modelling is a tool to project assets proceeds and liability outgo into the future that can be used to help in setting investment strategy to control the risk of falling to meet an objective.

Now question in your mind would be what is objective?

For ex: Objective can be that assets should be more than liabilities for the next 10 years with 99.5% confidence interval. In Objective, we try to make it quantifiable along with time horizon and a defined probability interval.

Assets Projection

Now to project assets we have to see various types of asset classes such as:

·         Money market instruments (for that we need short term interest rates)

·         Bonds (for that we need inflation, gross redemption yield)

·         Equities (we need dividend yield and expected future growth)

·         Property (we need rental yield and expected rental growth)

·         Overseas assets (we need future exchange rate movements and yields in those countries)

First of all, bifurcate all the existing assets in major asset classes (govt. bonds, corporate bonds, equity and property etc) and assign expected future returns from each asset class. In case of bonds, we will also consider proceeds at maturity

While projecting asset values, we need to make following assumptions:

·         Future interest rates

·         Future inflation

·         Future real yields

·         Rental yields and future Rental growth rates (for property)

·         Dividend yields and future dividend growth rates (for equities)

The projection of assets includes both income (coupon/dividend/rent) and maturity proceeds, if any

 


Liability Projection

·         Liability here means Net liability i.e. Benefit payments to policyholders + Expenses and commissions incurred in servicing the contracts + Other insurance related liabilities – Premiums received.

·         Outgoes can be

o   Fixed such as Fixed sum assured in case of term insurance

o   Variable i.e. index linked

o   discretionary for example, bonuses on with-profit policy

·         Expenses are usually linked to price inflation or salary inflation. Estimate the expected expenses for the portfolio.

·         The term (i.e. duration) of outgoes depends on likely timing of the payments.

·         Choose appropriate model points (i.e. Model point is a set of data representing a single policy or a group of policies. It captures the most important characteristics of policies that it represents.) which reflects underlying portfolio.

·         Need to make assumptions based on liability. For ex, if product is annuity then level of longevity, level of annuity increase, taxes etc.

·         we can then estimate the expected outgoes in each year.

·         The ALM model can be deterministic or stochastic depending on the purpose of modelling.

 

Determine the Investment Strategy

·         The Actuary’s initial point for the model would be to specify the objective. For ex,  to determine an asset allocation / investment strategy such that the probability of the solvency level falling below x% over the next y years is less than z%. (i.e. Solvency ratio falling below 150% over the next 10 years is less than 0.5%)

·         Given the recent regulatory requirements, the solvency level might be defined by a regulator (such as in case of Solvency II), although an in-house valuation basis could also be used (such as in case of ECap i.e. economic capital)

·         The model would project the cashflows associated with the asset proceeds and liability outgo

·         A decision would have been needed as to which variables to model stochastically and which to model deterministically. For example,

o   Stochastic modelling for:

§  investment returns

§  inflation.

o   Deterministic modelling for:

§  Withdrawal rates

§  Mortality

§  Expenses

·         Model should exhibit joint relationship between variables. For ex,

o   Withdrawal rate may be correlated with economic conditions i.e. in case of economic downturn, withdrawal rates will increase

o   inflation rates and investment returns, bonus rates and investment returns

·         For the stochastically modelled variables, probability distributions will have been determined along with parameter values for ex, mean investment returns / inflation rates, variances and correlations between variables will be considered. Similarly parameter values for deterministic modelled variables will be assumed.

·         The actuary would have chosen a trial asset allocation and carried out a large number of simulations (say 1,00,000) based on different values generated from the random variables

·         The results of the model will have been compared with the objective

·         The Actuary will probably have tried many different asset allocations, leading to a range of potential asset allocation strategies that meet the objectives.

·         Sensitivity and scenario testing will have then been carried out on these strategies (sensitivity testing means checking the impact on output by changing one assumption or variable at a time. Scenario testing means by changing more than one variables at a time)

·         This will have led to a reduced number of acceptable strategies to be presented to the management. For example strategy would be

o   50% in Government bonds

o   25% in Domestic equities

o   25% in Corporate bonds

 Follow us on Linkedin: Actuary Sense

Follow us on Instagram: Actuary Sense

Comments

Popular posts from this blog

Do you know the difference between Excess and Deductibles ?

IFRS 17 - Actuarial Science - Part 1

Do you know about every Probability Distribution function with cool examples?