Local Government Pension Scheme

Local Government Pension
Actuarial Mathematics Workshop 2013
Research opportunities
Local Government Pension Scheme
LGPS overview - some facts and figures
• Established by the Superannuation Act 1972
• Open to employees working in local government, those
working for participating employers and some councillors
• A multi-employer funded defined benefits pension scheme
• Comprises 101 regional funds in the UK
• Approx 10,000 employers and growing (e.g. academies)
• Approx 1.6 million contributing members
• Benefit payments in 2011-2012 were approx £7.5 billion
• Assets end of March 2012 were approx £148 billion
• 4th largest DB pension scheme in the world
LGPS overview – benefit structure
Pre April 2008 final salary (1/80th pension + 3/80th cash)
Post April 2008 final salary (1/60th pension + commutation)
Post April 2014 CARE (1/49th pension + commutation)
Complex legacy protections
Highly political
LGPS overview – what do actuaries do?
• Triennial funding valuations to recommend contribution rates
for existing employers
• Provision of strategic investment advice (stochastic modelling)
• Risk assessments and contribution rates for new employers
• Cessation valuations for exiting employers
• Financial reporting for individual employers under FRS17 and
IAS19 (now represents more than half of all activity)
LGPS overview – actuarial capacity issues
• Only 4 actuarial firms have any significant presence in the
LGPS marketplace
• Considered low margin business by most proprietary
consulting firms
• Regarded as highly specialist
• Significant capacity problems
• Currently low levels of investment in systems and staff
particularly with general decline in DB pension schemes
• Fund Actuaries also carry out work for employers – significant
potential for conflicts of interest
• Seven directors and two freelance consultants (actuarial,
LGPS, IT, legal, business skills)
• Innovative new service aimed at providing low-cost
independent actuarial advice directly to LGPS employers
• Core offering is a package comprising financial reporting and
related monitoring of funding measures, outsourcing advice
and general actuarial support
• Unique selling point is a web-based service delivery through
our NeXtStep platform
• Our objective is to significantly expand market capacity,
bringing affordable and independent actuarial services within
the reach of ordinary LGPS employers
Research opportunities
Selection criteria
Selection criteria
Suited to a mathematical or statistical treatment
Provides solutions needed by industry
Must be capable of practical application
Provides solutions that are of commercial value to consulting
actuarial firms, audit firms, standards setters and regulatory
• OR that serve the public interest (through cost reductions and
greater efficiency)
Research opportunities #1
Quantifying the error inherent in pension liability
roll-forward processes
Roll-forward – what is it?
• A simplified mathematical model designed as a substitute for
performing an accurate valuation of pension scheme liabilities
• A practical response to time and cost constraints
• Greatly simplifies data requirements
• Quicker to run
• Less skilled operators required
• Routinely used for the majority of financial reporting exercises
(FRS17 & IAS19) in both public and private sectors
Roll-forward – what are the issues?
• No-one has any real understanding of the magnitude of error
inherent in the process
• Auditors routinely ask questions about the size of the
potential error but place reliance on response from actuary
• Actuaries tend to be circumspect in their answers
• No real tools or techniques exist to identify the conditions
that give rise to errors and potential magnitude
• The risk is that there may be material misstatement of
pensions figures going into accounts
Roll-forward – research linked to the LGPS
• The LGPS presents a unique opportunity for a statistically
based study into roll forward methodology
• We estimate there are around 6,000 LGPS employers that
currently report under FRS17 & IAS19
• All FRS17 and IAS19 data exists in the public domain and may
also be obtained from Whole of Government Accounts
• Accurate results also exist as they are produced for the funds
every 3 years for reporting under IAS26
• Regression models could be developed that attempt to
explain the link between employer and data characteristics
and the observed level of reporting error
Roll-forward research – who benefits?
• Potentially of significant value to the major audit firms –
better placed to insist on full valuations – more accurate
financial reporting
• Actuarial firms unlikely to complain – more work!
• Investors and users of accounts can have greater confidence
in the reliability of pensions information
Research opportunities #2
Quantifying the error from using index returns in
place of actual investment returns
Index returns – the issue
• Index-based estimated returns are frequently used when
actual investment data is not available
• Common situations include bulk transfers, financial reporting,
interim valuations and in calculating new employer
contribution rates
• The approach has very simple data requirements (split
portfolio into major asset classes, multiply by index-based
returns, add up to get total estimated return)
• Almost invariably, index returns do not match the actual data
when it eventually becomes available
Index returns – potential research
• Statistical based studies or theoretical treatment possible
• Vast amounts of potential data available to generate actual vs
expected data
• Actuarial firms could be interviewed to understand
• Can a link be established between the data, fund or asset
characteristics that helps to quantify and explain the potential
• Can better models be constructed and tested in the light of
this research?
Index returns – who benefits?
• Likely to be of interest to actuarial and audit firms
• Better estimation models will improve accuracy in financial
• Auditors will be better positioned to know when estimation
techniques will be acceptable or will be better able to
understand the potential error
Research opportunities #3
Yield curves and the derivation of FRS17 &
IAS19 discount rates
FRS17 and IAS19 discount rates
• FRS17 requires use of a discount rate based on AA-rated
corporate bonds of equivalent term and currency to the
pension liabilities
• IAS19 requires use of high quality corporate bond yields
• In practice most firms use the same approach for both
• In the early days of the standards the use of a single index
figure was commonplace (e.g. iBoxx AA indices)
• Larger firms now moved away from this and most use
proprietary yield curves
• Audit firms appear to favour this approach
Yield curve research – the problem
• The principal problem is that the duration of liabilities for
most pension schemes is higher than the duration of the
bonds on which the curves are based
• Ignore issue and continue to use a simple index figure
• Curve fitting models developed using extrapolation
• Pragmatic solutions based on the gilts curve (does extend to
the required durations)
• Subjective individual “judgement” applied
Yield curves research – possibilities
• A research paper on various curve fitting methodologies, their
behaviour under different financial conditions and practical
implications for discount rates
• Research into the relationship between gilts and corporate
bonds (credit spreads) and the potential family of curves that
could be derived from using the gilts curve as a base
• Development of a high quality open source yield curve model
based on publicly available data for use as a reference source
for actuarial firms and auditors
Yield curve research – who benefits?
• Reduced cost to industry as numerous proprietary models
likely to become defunct if a free alternative exists
• Potential for greater consistency in financial reporting (one of
original purposes of the standards)
• If offered as open-source there is a significant opportunity for
the University to raise its profile within the industry
Research opportunities #4
The construction and design of
proxy mortality tables
Proxy mortality tables – the issue
• LGPS funds are big enough to generate their own statistically
significant mortality data
• Historically, scheme specific QX tables have been constructed
and expressed as adjustments to standard CMIB tables
(standard industry practice)
• In recent years we have seen the emergence of proprietary
mortality tables, usually postcode-based
• Very little information regarding these tables is available to
auditors and third-party providers of actuarial services
• Lack of transparency undermines confidence and leads to
higher costs
Proxy mortality tables – research
• Objective would be to develop an optimising algorithm that,
for a given set of data points (usually life expectancies at set
ages) leads to a best fit curve with desirable characteristics in
relation to a supplied set of input reference curves
• Output in the form of industry-standard adjustments to
standard tables
Proxy mortality tables – who benefits?
• Impact on auditors and third-party suppliers is minimised,
leading to reduced costs and time in preparing work
• Easier for third parties to manipulate valuation results for
other purposes if proxy tables can be identified quickly
Thank you

similar documents