Skip to Content

You are in the Finance archive | Archive Home Page | Return to the Finance homepage | Contact Us

Finance archive > Superannuation > Pension Indexation Costings > Cumpston Sarjeant - Review of Actuarial Advice

The Department of Finance and Deregulation Archive

The content on this page and other Finance archive pages is provided to assist research and may contain references to activities or policies that have no current application. See the full archive disclaimer.

Cumpston Sarjeant - Review of Actuarial Advice

Assistant Secretary
Superannuation Branch
Department of Finance and Deregulation
John Gorton Building, King Edward Terrace
PARKES ACT 2600
23 December 2010

 

Dear Sir

Review of Actuarial Costings of Effects of Changes to Indexation Arrangements for the Australian Government’s Civilian and Military Superannuation Schemes

This report provides a review of calculations originally performed by Mercer and the Australian Government Actuary, estimating the financial impact of potential changes to pension indexation arrangements in the Australian Government’s civilian and military superannuation schemes.

In our review we have found:

We conclude that the estimates of financial impacts of changes to indexation arrangements within the Australian Government’s civilian and military superannuation schemes are reasonable. 

Yours sincerely

John Rawsthorne
Fellow of the Institute of Actuaries of Australia


Glossary

AGA Australian Government Actuary
CPI Consumer Price Index
CSS Commonwealth Superannuation Scheme
DFRB Defence Forces Retirement Benefits Scheme
DFRDB Defence Force Retirement and Death Benefits Scheme
Mercer Mercer (Australia) Proprietary Limited
MSBS Military Superannuation and Benefits Scheme
MTAWE Male Total Average Weekly Earnings
NECR notional employer contribution rate
PBLCI Pensioner and Beneficiary Living Cost Index
PNG Scheme the scheme under the Papua New Guinea (Staffing Assistance) (Superannuation) Regulations 1973
PSS Public Sector Superannuation Scheme
OSS he scheme under the Superannuation Act 1922

Introduction

We have been engaged by the Superannuation Branch, Department of Finance and Deregulation, to provide an independent review report on the actuarial advice in respect of pension indexation arrangements for Commonwealth superannuation schemes provided by AGA and Mercer.  The purpose of the review is to determine the validity of the methodology and assumptions and an assessment of the reasonableness of the results. The costing involves a comparison of the results under the policy change with the results from a base case of the 2009 Unfunded Liability Reports (which applies the same methodology and assumptions as the 2008 Long Term Cost Reports).

As the base calculation uses the methodology and assumptions from the 2008 Long Term Cost Reports (which has previously been audited), a review of the methodology and assumptions for the base case is not required. Likewise, the data used in the calculations will be as per used in the 2009 Unfunded Liability Reports, and does not require review.

In determining the reasonableness of the results, a detailed review of the programming of actuarial models and spreadsheets used in the calculation will not be required.

The review should specifically address the key assumptions regarding the pension indexation, being the rate of indexation and the behavioural change in Public Sector Superannuation Scheme members and Military Superannuation and Benefits Scheme members taking up pensions.

This review has been prepared at the request of the Assistant Secretary of the Superannuation Branch of the Department of Finance and Deregulation.

We understand that this review report will be made publicly available.

Actuarial Standards

This report is intended to comply with the requirements of Institute of Actuaries of Australia’s Code of Professional Conduct, November 2009.

Acknowledgements

We are grateful for assistance provided by Darren Wickham and Roger Bohlsen at Mercer, and Michael Burt, Susan Antcliff and Limin Wang at AGA.

Background

The Australian Government provides superannuation benefits to civilian public servants and military personnel via several superannuation schemes.  Some of those superannuation schemes have pension entitlements as a benefit option, and are subjects of this review.

Civilian schemes included in this review are:

Military schemes included in this review are:

Pension Indexation Arrangements

Indexed pensions in the above schemes are currently increased in line with changes in the CPI every July and January.  Some benefits are available as unindexed pensions – there is no proposed change to indexation arrangements for those benefits.

The two proposed new indexation scenarios are:

As part of this review we received a copy of a previous review of pension indexation arrangements for the schemes in question:

Data received

Civilian schemes

In respect of civilian schemes we received a letter from Darren Wickham of Mercer to the Director, Policy and Legislation Section, Superannuation Branch, dated 10 November 2010 with subject “Indexation Arrangements of Pension Entitlements” (the Mercer report) on 10 November.  We subsequently received an amended version of this advice, with a change in the split of results between CSS and OSS, on 1 December. 

John Rawsthorne met with Darren Wickham in Sydney on 12 November to discuss the calculations performed by Mercer, and to determine what additional information might be available to assist with checking.  As a result of these and other discussions and emails, Mercer also provided:

Military schemes

In respect of military schemes we received on 16 November three letters from AGA addressed to the Assistant Secretary, Superannuation Branch, as follows:

John Rawsthorne met with Michael Burt and Susan Antcliff of AGA in Canberra on 17 November to discuss the calculations performed by AGA, and to determine what additional information might be available to assist with checking.  As a result of these and other discussions and emails, AGA also provided:

Review of Assumptions

Economic Assumptions

The base economic assumptions used by Mercers and AGA in assessment of costs are:

We have not been asked to review the appropriateness of these base assumptions.

Against this base, Mercer and AGA have considered the likely change to pension indexation rates on the two different pension indexation scenarios.  Broadly, Mercer and AGA have assumed, in the absence of a long historical record, that PBLCI will be at the same rate as CPI on average over the long term, i.e. 2.5% pa, while MTAWE will be at the same rate as general salary inflation, i.e. 4.0% pa.

Age Pension Indexation Basis

Mercer and AGA were requested to consider an indexation scenario where superannuation pensions were increased at a rate equivalent to the Age Pension Indexation Basis.  Age pensions are indexed at the higher of CPI and PBLCI, but with an overall base level dollar amount of 27.7% of MTAWE.

Graph 6.1 of the Matthews Report shows that age pensions have been paid at a level broadly in line with 25% of MTAWE from the mid-1970s through to 1997, during which time the age pension was indexed to CPI.  After 1999 age pensions have been paid at the level of 25% of MTAWE, although that percentage recently increased to 27.7%. 

Mercer and AGA have both concluded that, because the CPI/PBLCI-based pension level is now significantly below the level of 27.7% of MTAWE and because MTAWE is expected to grow at a higher rate than CPI or PBLCI in the long term, it is highly likely that age pensions will now only grow in line with MTAWE.  Thus Mercer and AGA have assumed that the Age Pension Indexation Basis will result in pension indexation at a rate of 4.0% in the long term.

We have reviewed the logic of this derivation, as well as the practical likelihood of the CPI/PBLCI-based pension exceeding 27.7% of MTAWE for significant periods, and we consider the 4.0% per annum pension growth rate adopted by Mercer and AGA for this scenario to be reasonable and consistent with the base economic assumptions set out above.

Higher of MTAWE, CPI and PBLCI

Mercer and AGA were requested to consider a second indexation scenario, where superannuation pensions were indexed each half-year at the highest 6-monthly increase amongst the indices for MTAWE, CPI and PBLCI.  Because of fluctuations in the rate of increase in each index over time, the “higher of” indexation rate will in the long term be higher than each of the individual indices.

Mercer and AGA independently arrived at the same conclusion, based on different modelling techniques, that the “higher of” rate would average around 4.6% per annum in the long term.  In arriving at their conclusion of 4.6% pa, Mercer and AGA both considered the actual outcome using historical data, while Mercer also considered the average simulated future outcome using best estimate assumptions in a stochastic economic scenario generator. 

To check the conclusion of Mercer’s and AGA’s analysis, we performed our own economic modelling taking account available historical information about averages, standard deviations, cross-correlations and serial correlations in each index.  We repeatedly simulated the indexation outcome by randomly selecting from the distribution of possible rates for the three indices, and then we averaged the “higher of” rate derived in these 1000 simulations.  There was very little information available for the PBLCI index, and so we assumed that it had the same properties as CPI (i.e. the same average, spread and correlations), but further we assumed that CPI and PBLCI were 80% correlated with each other.  For simplicity, our model simulated annual results, rather than 6-monthly results, and so our result should be slightly lower than the result when allowance is made for 6-monthly indexation.  Our projection suggested that the “higher of” indexation rate should be about 0.5% higher than the wage indexation rate, or in this case about 4.5% per annum.  The Mercer and AGA rate of 4.6% per annum is slightly higher than our rough approximation, as expected when 6-monthly indexation is allowed for.

We are satisfied, based on our review of the data and methods used by Mercer and AGA, and also based on our own modelling, that 4.6% per annum is an appropriate rate to use for the long term “higher of MTAWE, CPI, PBLCI” indexation rate, when compared to the base economic assumptions that have been made.

Behavioural Assumptions

Any change to pension indexation rates may have flow-on changes to other aspects of scheme experience.  In particular if pensions become more valuable by way of higher indexation then members may be more likely to choose pensions rather than lump sums.

PSS pension take-up

Mercer examined the potential impact of higher pension indexation on the rate of choosing pensions for PSS members.  There is no data available on which to base an assumption for change in behaviour.  Mercer assumed that the proportion opting to take up pensions would increase from 60% to 70%, noting that not all members make decisions which would appear financially rational to an external observer.  Overall we consider that the assumed change is reasonable but highly uncertain.

We asked Mercer about the relative costs of the behavioural change and the underlying changes in indexation rates.  Mercer estimate that the change in pension take-up rates increases the unfunded liability at 2011 by about $2.0B, while the total changes are $9.8B and $13.9B on the 4.0% and 4.6% indexation scenarios respectively.  Thus the behavioural change is a relatively minor component of the overall cost of changes, and the uncertainty in this part of the costing is acceptable.

MSBS pension take-up

AGA examined the potential impact of higher pension indexation on the rate of choosing pensions for MSBS members.  As with PSS, there is no data available on which to base an assumption for change in behaviour.  AGA assumed that the proportion opting to take up pensions would increase from 75% to 90% for officers and from 60% to 80% for other ranks.  We consider that the assumed change is reasonable but highly uncertain.

We asked AGA about the relative impact of the change in pension take-up and other changes.  In discussions, Michael Burt indicated that the change in pension take-up represented perhaps 20-30% of the initial increase in liability, although he had not directly quantified it.  Our own estimate is that the change in pension take-up increases liability at 2011 by about $1.2B out of total increases in MSBS liability of $6.4B at the 4.0% indexation rate or $9.1B at the 4.6% indexation rate.  Thus the behavioural change is a relatively minor component of the overall cost of changes, and the uncertainty in this part of the costing is acceptable.

Other potential behavioural changes

AGA and Mercer made only one estimate of behavioural change, despite there being two different indexation changes.  It would be possible to make a different estimate for the two different indexation scenarios.  We asked about this, and both AGA and Mercer indication that the estimated behavioural change was intended to be a coarse estimate of likely outcomes, and any attempt to finesse the costing in this respect was spurious.  We accept this explanation, and consider that it is reasonable to only consider one degree of behavioural change for the two changes in indexation arrangements.

We note that, while CSS and DFRDB both have remaining contributors, the scheme designs mean that there is no effective option about whether to take a pension with the employer component of benefit.  Therefore in both of these schemes there is no potential change due to change in take-up of pensions.  For DFRB and OSS there are no remaining contributors, hence no potential change in take-up of pensions.

We asked Mercer and AGA about any other potential changes in behaviour that might influence costs.  Potential changes include:

In each case after discussion with AGA and Mercer we concluded that the potential increases or decreases in employer cost were likely to be much smaller than the change in pension take-up considered above.  We consider that the behavioural changes allowed for (i.e. higher take up of pensions in MSBS and PSS) are the only changes that are likely to have significant associated employer costs.

Review of Results

We have not been required to perform detailed checks of all calculations, nor to check the precise details of the actuarial modelling performed.  Specifically, we have not checked the derivation of the assumptions used in the base case (which are from the 2008 Long Term Cost Reports), and we have not checked the machinations of the model used to arrive at the base case results (which are from the 2009 Unfunded Liability Reports).

Base case results

We have checked that the base case results presented in the Mercer and AGA reports do correspond to the 2009 Unfunded Liability Reports.  We note that for each scheme the liability at 2009 on the base case matches the unfunded liability set out in the 2009 Unfunded Liability Reports.

For civilian schemes the projected liabilities and cash flows set out by Mercers for all years match the base case liabilities in the Unfunded Liability Report.  For military schemes the projected base case liabilities and cash flows are not shown.  AGA only showed the changes in liabilities and cash flows relative to the base case, rather than setting out the projected liabilities themselves. 

Checking of alternative scenarios

Because our checking is focussed on the relativities of the alternative scenarios to the base case, we have been able to perform a series of approximate checks on liability results, cash flows and employer contribution rates which take into account:

Our checking is designed to be coarse, and to establish only the broad appropriateness of the results while focussing on the appropriate relativities between results on different bases.  Appendix A gives a demonstration of the checking on numerical results performed, by stepping through a hypothetical checking example.

Civilian Schemes

For CSS, OSS and PSS schemes we established, for each alternative indexation scenario, that:

For PSS we established that the change in projected payments in the short and longer term was consistent with the changed pension take-up assumption, i.e. a decrease in lump sum payments in the short term, with correspondingly higher pension payments in the longer term.

For PSS we established for each alternative scenario that the change in NECR was consistent with the change in indexation.

Mercer did not include any results in their advice in respect of the PNG Scheme.  We understand that this scheme is small, and any costs in respect of this scheme are likely to be entirely immaterial in the overall context of this exercise.

Based on our checking of civilian scheme projections we are satisfied that the projected liabilities and cash flows on each of the alternative indexation scenarios are materially correct, and the indications of change due to change in indexation arrangements are materially correct.

Military schemes

For DFRB and DFRDB schemes we established, for each alternative indexation scenario, that:

For MSBS we established, for each alternative indexation scenario, that:

Based on our checking of military scheme projections we are satisfied that the projected liabilities and cash flows on each of the alternative indexation scenarios are materially correct, and the indications of change due to change in indexation arrangements are materially correct.

Conclusion

In our review we have found:

We conclude that the estimates of financial impacts of changes to indexation arrangements within the Australian Government’s civilian and military superannuation schemes determined by Mercer and AGA are reasonable. 

John Rawsthorne
Fellow of the Institute of Actuaries of Australia


Appendix A


Contact for information on this page: superbranch@finance.gov.au


16 September, 2013