Modelling the long-term post-employment liability in universities
The financial obligations that universities carry toward former staff extend well beyond the final payslip. Retiree superannuation, accumulated long service leave, and any continuing salary or health arrangements form a multi-decade liability that can rival the cost of a major capital program. For senior business officers, building a defensible model for this exposure is now a recurring requirement of annual reporting, of board strategy, and of conversations with rating agencies and state treasuries.
Australian universities operate within a layered environment. The Group of Eight and the regional networks in Queensland, Western Australia, and Tasmania each run a mix of defined benefit and accumulation vehicles, with UniSuper and AustralianSuper sitting alongside state public-sector schemes. The choice of vehicle shapes the on-balance-sheet exposure and the assumptions that feed the actuarial model.
Workforce ageing makes the work more pressing. Median staff ages in the sandstone universities of Sydney and Melbourne now sit comfortably above the mid-forties, while in Adelaide and Brisbane the share of academic staff within fifteen years of their preservation age is rising. Each retiring cohort converts an accruing service-cost line into a drawn-down cash obligation, and the curve steepens as the post-2009 superannuation guarantee increases compound.
The remainder of this article walks through the modelling decisions that matter most: the data foundations, the economic and demographic assumptions, the discount rate, scenario analysis, the disclosure pathway under Australian standards, and the risk tools that turn a balance-sheet number into a managed exposure.
Inventorying the obligations
Every modelling exercise begins with a complete map of the post-employment promises the institution has made. At most Australian universities this map includes three categories: defined benefit superannuation entitlements held in legacy funds and calculated on a final-salary or career-average formula; accumulation-phase superannuation, where the obligation is to contribute the mandated percentage of ordinary time earnings under the superannuation guarantee; and non-superannuation promises such as long service leave, salary continuations, and subsidised private health cover.
Long service leave deserves particular attention. Under the Fair Work Act and the relevant modern awards, an employee with seven or more years of continuous service is entitled to paid leave on termination, and many enterprise agreements negotiated at universities in Perth, Hobart, and the ACT extend this entitlement. The actuarial present value of this leave can be material and must be calculated separately from any defined benefit promise. Retiree health cover is less common in Australia than in North America, but where it exists, the model must distinguish between fully insured arrangements, where the insurer carries the technical provision, and self-funded schemes, where the university carries the obligation directly.
Building the data foundation
The quality of the model is constrained by the quality of the underlying data. Three streams matter. Payroll and human resources data feed the membership file: date of birth, gender, salary, service history, contribution rate, and benefit category. Investment data feeds the discount rate and the asset-liability matching analysis. Industry data, drawn from the Australian Bureau of Statistics and from sector-specific studies commissioned by Universities Australia, anchors the demographic assumptions.
A clean membership file is the single most useful investment a finance team can make. Records missing the date-of-birth field, that confuse commencement dates with re-engagement dates, or that double-count casual staff who move between faculties will quietly distort the liability. Reconciliation against single-touch payroll reports and against the contribution schedule lodged with UniSuper is a useful control. Local calibration on internal staff turnover data produces a more reliable model than any off-the-shelf default, because Australian universities operate across very different labour markets.
Economic assumptions and the discount rate
Three economic variables drive the present value of the liability: the discount rate, the rate of price inflation, and the rate of salary growth. The discount rate is the most contentious, because it sets the time value on cash flows that may extend forty years into the future.
Australian accounting standards require the discount rate to be determined by reference to high-quality corporate bond yields at the reporting date, denominated in the same currency as the obligation. The Commonwealth Government bond curve and the rated corporate bond curve in Australia have moved through a wide range over the past decade, and the liability figure moves with them. A model that produces a single point estimate without showing the sensitivity to a fifty or one hundred basis point shift in the discount rate is incomplete.
Salary growth assumptions should reflect the structure of the relevant enterprise agreement. The agreements in force at the University of Melbourne, the University of Queensland, and Curtin University contain different annual increment schedules, and any model that applies a single sector-wide rate will misrepresent the obligation. Where the defined benefit promise is indexed to prices rather than wages, the inflation assumption is the binding driver.
| Category | Typical scope | Key data inputs | Sensitivity drivers |
|---|---|---|---|
| Defined benefit superannuation | Pension promises under a trust deed, often legacy | Membership file, fund rules, payment history | Discount rate, salary growth, mortality improvements |
| Accumulation superannuation | Future contributions and member balances | Contribution rate, salary base, member investment choice | Salary growth, member investment returns |
| Long service leave | Paid leave on termination after qualifying service | Service history, leave balances, award provisions | Discount rate, salary growth, withdrawal rates |
| Retiree health and other | Subsidised cover, salary continuations | Claims history, eligibility rules, premium schedules | Claims inflation, take-up rates, mortality |
| Termination benefits | Redundancy and early retirement packages | Enterprise agreement provisions, project plans | Timing of decisions, discount rate |
Demographic assumptions
Mortality is the assumption that has changed most rapidly in recent years. Australian life tables published by the Australian Government Actuary continue to show gains in life expectancy at older ages, and defined benefit liabilities rise as a result. A model that uses a mortality table more than five years old will understate the obligation for a pension-paying population.
Withdrawal and resignation patterns differ sharply between academic and professional staff, and between metropolitan and regional campuses. Academic staff in research-intensive roles tend to have longer tenures, lower resignation rates, and later retirement ages than colleagues in teaching-focused or administrative roles. A single withdrawal assumption applied across the whole population will distort the projection, particularly where the workforce includes a growing share of fractional appointments and parental-leave returnees.
Scenario analysis and stress testing
A single best-estimate figure is rarely enough for a board asked to approve a long-term funding plan. Scenario analysis lets the model speak in ranges, and stress testing lets it speak in tail outcomes. The most useful scenarios compare a high-inflation environment with a low-inflation environment, a high-discount environment with a low-discount environment, and a life-expectancy scenario in which current mortality improvements continue against one in which they stall.
Stochastic modelling takes this further by generating thousands of random paths for the economic variables and recording the distribution of the resulting liability. For a defined benefit superannuation fund, the stochastic output feeds directly into the asset-liability study and the investment strategy. For the university balance sheet, it provides a basis for setting a prudential margin above the best estimate. Climate-related transition risk is now a standard input to many institutional stress tests, with material from the Reserve Bank of Australia and the Australian Prudential Regulation Authority that applies directly to a long-horizon liability of this kind.
Reporting under Australian accounting standards
The disclosure framework is set by the Australian Accounting Standards, in particular the standard on employee benefits. The standard requires the present value of defined benefit obligations, the fair value of plan assets, and the net surplus or deficit to be recognised on the balance sheet, with movements in the year recognised in the income statement or in other comprehensive income.
Universities that participate in multi-employer defined benefit schemes, such as some state public-sector funds, are required to apply the defined contribution accounting treatment unless sufficient information is available to apply defined benefit accounting. The choice has a real effect on the reported result, and the assumptions supporting that choice must be documented. Audit and review committees are increasingly asking for an independent actuarial review at least every three years, with sensitivities disclosed to a one percentage point change in the major drivers.
Risk mitigation, funding strategy, and governance
Modelling shapes action through funding plans that smooth contributions over a multi-year horizon, salary packaging arrangements that contain the growth of the salary base, and benefit redesigns that close legacy defined benefit promises to new members while preserving the accrued entitlements of existing members.
A disciplined funding plan sets a target funding ratio and a recovery path that is reviewed annually. The review should consider the institution's own appetite for risk, the requirements of any bond covenants, and the expectations of the state or federal government in jurisdictions where the university is a statutory body. Several Australian universities have used the model output to renegotiate employer contribution rates with their industry fund, an option worth investigating wherever the trust deed allows.
Governance is the final piece. A standing committee that meets at least twice a year, with a documented assumption-setting protocol, will produce a more robust number than an annual exercise run between the chief financial officer and the actuary. The committee should include representatives from human resources, payroll, treasury, and the academic divisions, and its papers should be retained for review by internal audit.
Practical steps for a stronger modelling function
- Build a single, reconciled membership file as the foundation of every subsequent calculation, with clear ownership between payroll, human resources, and finance.
- Update the mortality table at least every three years, drawing on the most recent Australian Government Actuary publication and any sector-specific studies commissioned by Universities Australia.
- Disclose sensitivity to a one percentage point change in the discount rate, the inflation rate, and the salary growth rate, so the board can see the range rather than only the central estimate.
- Run a full stochastic asset-liability study at least every five years, with the results feeding into the investment strategy and the funding plan.
- Maintain a documented assumption-setting protocol that names the responsible officer, the data sources, and the approval pathway for each material assumption.
Use the next budgeting cycle to refresh the membership data, retest the key assumptions, and brief the audit and risk committee on the updated range. Commission the independent actuarial review in parallel with the year-end financial close so the new figures land in the financial statements without slippage. With a transparent, well-documented, and stress-tested number in hand, lead the conversation with UniSuper, the state government, the rating agency, and the bond trustees from a position of evidence rather than assertion.