Superannuation models

RIM’s early work (and much of its current work) involves the use of superannuation models.

The parameter and superannuation models must be sufficiently disaggregated to provide insight into a wide range of public policy questions including: retirement benefits, the age pension and social security systems, workforce participation and retirement patterns, private sector and national saving, superannuation tax concessions and fiscal balance.

Other issues of interest include patterns of retirement benefit commutation and lump sum dissipation, contribution/ earnings patterns over the life cycle, key demographic, macroeconomic and microeconomic variables, the retirement age decision and relevant tax, superannuation and social security parameters.

RIM has developed three types of superannuation micro models to address its terms of reference.

Hypothetical tax-benefit models cover one individual, a couple (or income unit). The RIM Unit's RIMHYPO model takes an individual or couple from work force entry to death. All relevant combinations of life events, government policies and retirement income sources can be modelled. This model captures in detail the legislative structure defining the interactions between superannuation, taxation and social security legislation. The complexity of this structure is illustrated by the fact that RIMHYPO contains some 16 thousand lines of code.

Cohort models track the lifecycle of groups in the population. They are also referred to as group projection models, actuarial models or cell based models. Cohort models concentrate on population changes and accumulation processes over time, and on incrementing or decrementing processes within a time period.

Because the whole population is covered, cohort models can be used for costings to the extent that the group structure is sensitive to the parameters of the costing. When group models are used for projections, insufficient or inappropriate group disaggregation can lead to inappropriate pooling of accumulations. For example, a coarse income distribution can lead to a poor costing of a new income test but give a reasonable approximation to aggregate superannuation contributions and earnings. Because of these effects, most costing spreadsheets can only be regarded as short period group models.

RIMGROUP is the main policy model of the RIM Unit. It is a comprehensive cohort projection model of the Australian population which starts with population and labour force models, tracks the accumulation of superannuation in a range of account types, accumulates non superannuation savings, and calculates tax payments and expenditures, social security payments including pensions and the generation of other retirement incomes. Again the full richness of the underlying legislation is captured.

RIMGROUP projections are done for each year of the projection period separately for each birthyear, gender, and career earnings decile cohort. The model projections begin in July 1992 and can run out to June 2060. There are over 4300 cohorts in the model covering the Australian population, which includes relevant demographic detail for each cohort. Each cohort is tracked through its lifecycle (between age 18 and 100+ depending on birth year) with superannuation and asset modelling through all of working life and retirement.

Static microsimulation models, in contrast, concentrate on the distributional detail of single year flows and statuses and can only be used for short-run projections. For example, each year in RIM's microsimulation model based on personal income taxation data is represented by 800,000 records. RIMGROUP provides demographic profiles by which the static microsimulation model base populations are reweighted for later years. Treasury economic forecasts provide indices for uprating incomes.

RIMHYPO - Retirement Income Modelling Hypothetical Model

RIMHYPO - which is a very detailed lifecycle projection model of working life incomes, superannuation, other savings and retirement incomes for hypothetical individuals and couples written in SAS.

RIMGROUP - Retirement Income Modelling Group Superannuation Model

RIMGROUP is a aggregate projection model. RIMGROUP projects the superannuation, other savings and retirement incomes of age, gender, career income decile groups of the population by tracking mortality, labour force status, sector of employment, income and type of superannuation fund across every year of a group's working life. Calculations are done at the average for the group and accumulated assets are pooled. The approach is hence at a level of aggregation above unit records but below age-gender cohorts. The model gives projections on both the 'quantum and distribution' of taxation, saving, social security payments and tax concessions.

Demographic models

Treasury uses ABS data to model a number of demographic variables, including in the context of the InterGenerational Report and to provide data for successful modelling of aggregate superannuation in RIMGROUP. Demographic variables of interest include total population, fertility, mortality, migration, labour force, and so on. The models used are non-parametric models, and include a population model (POPMOD), a life expectancy model (LIFE), a labour force status model (LFSMOD) and a fertility model (FERTILITY).

POPMOD - Population Model

POPMOD provides annual projections of Australia's population by year for males and females by single year of age. The model is driven by parameter matrices for fertility, mortality and overseas migration. The overseas migration sub-model accepts projection of permanent and long-term arrivals and departures and measures of category jumping.

LIFE - Life expectancy Model

The life expectancy model calculates survival rates, survivors to age x, deaths at age x to x+n, life table populations and life expectancy for males and females by single year of age. The estimates are constructed from the mortality parameters used in POPMOD. Life expectancies are used in both demographic and policy models. Factors for annuity and allocated pension pricing are provided in policy models.

LFSMOD - Labour Force Status Model

This is long-run annual model of the Australian labour force to capture structural (trend) behaviour at fine detail. The model projects persons by labour force status, age, gender and income decile. . There is no short-run behavioural response in LFSMOD, as the model simply runs off the observed underlying long-run movements of key, and hopefully stable, parameters, which are estimated as non-linear trends with consistent asymptotic values. Apart from these time-varying parameter matrices, the model's only exogenous inputs are population projections from a population model, such as POPMOD, and aggregate unemployment rates for males and females.FERTILITY – Fertility ModelThe fertility model uses fertility history of the Australian population to project future fertility by year of age. The fertility projections are then used to calculate births for inclusion in the population model to produce annual projections of total Australian population.

Household Income Models

Personal Tax Microsimulation Model Based On Sample Tax File

The Personal Tax Microsimulation Model is based on a sample file provided by the ATO each year. It is a 1 in 12.5 unit record file, sampling records taken from those of all individuals who submitted a personal tax return in the relevant financial year. Details of age, incomes, deductions and rebates are provided.

RIM Modified Version of STINMOD OUTYEARS is an extension of STINMOD developed by NATSEM with some assistance from the Treasury. The static microsimulation model of household incomes is used for family payment and family taxation costings up to 4 to 5 years into the future.




	Superannuation models RIM’s early work (and much of its current work) involves the use of superannuation models. The parameter and superannuation models must be sufficiently disaggregated to provide insight into a wide range of public policy questions including: retirement benefits, the age pension and social security systems, workforce participation and retirement patterns, private sector and national saving, superannuation tax concessions and fiscal balance. Other issues of interest include patterns of retirement benefit commutation and lump sum dissipation, contribution/ earnings patterns over the life cycle, key demographic, macroeconomic and microeconomic variables, the retirement age decision and relevant tax, superannuation and social security parameters. RIM has developed three types of superannuation micro models to address its terms of reference. Hypothetical tax-benefit models cover one individual, a couple (or income unit). The RIM Unit's RIMHYPO model takes an individual or couple from work force entry to death. All relevant combinations of life events, government policies and retirement income sources can be modelled. This model captures in detail the legislative structure defining the interactions between superannuation, taxation and social security legislation. The complexity of this structure is illustrated by the fact that RIMHYPO contains some 16 thousand lines of code. Cohort models track the lifecycle of groups in the population. They are also referred to as group projection models, actuarial models or cell based models. Cohort models concentrate on population changes and accumulation processes over time, and on incrementing or decrementing processes within a time period. Because the whole population is covered, cohort models can be used for costings to the extent that the group structure is sensitive to the parameters of the costing. When group models are used for projections, insufficient or inappropriate group disaggregation can lead to inappropriate pooling of accumulations. For example, a coarse income distribution can lead to a poor costing of a new income test but give a reasonable approximation to aggregate superannuation contributions and earnings. Because of these effects, most costing spreadsheets can only be regarded as short period group models. RIMGROUP is the main policy model of the RIM Unit. It is a comprehensive cohort projection model of the Australian population which starts with population and labour force models, tracks the accumulation of superannuation in a range of account types, accumulates non superannuation savings, and calculates tax payments and expenditures, social security payments including pensions and the generation of other retirement incomes. Again the full richness of the underlying legislation is captured. RIMGROUP projections are done for each year of the projection period separately for each birthyear, gender, and career earnings decile cohort. The model projections begin in July 1992 and can run out to June 2060. There are over 4300 cohorts in the model covering the Australian population, which includes relevant demographic detail for each cohort. Each cohort is tracked through its lifecycle (between age 18 and 100+ depending on birth year) with superannuation and asset modelling through all of working life and retirement. Static microsimulation models, in contrast, concentrate on the distributional detail of single year flows and statuses and can only be used for short-run projections. For example, each year in RIM's microsimulation model based on personal income taxation data is represented by 800,000 records. RIMGROUP provides demographic profiles by which the static microsimulation model base populations are reweighted for later years. Treasury economic forecasts provide indices for uprating incomes. RIMHYPO - Retirement Income Modelling Hypothetical Model RIMHYPO - which is a very detailed lifecycle projection model of working life incomes, superannuation, other savings and retirement incomes for hypothetical individuals and couples written in SAS. RIMGROUP - Retirement Income Modelling Group Superannuation Model RIMGROUP is a aggregate projection model. RIMGROUP projects the superannuation, other savings and retirement incomes of age, gender, career income decile groups of the population by tracking mortality, labour force status, sector of employment, income and type of superannuation fund across every year of a group's working life. Calculations are done at the average for the group and accumulated assets are pooled. The approach is hence at a level of aggregation above unit records but below age-gender cohorts. The model gives projections on both the 'quantum and distribution' of taxation, saving, social security payments and tax concessions. Demographic models Treasury uses ABS data to model a number of demographic variables, including in the context of the InterGenerational Report and to provide data for successful modelling of aggregate superannuation in RIMGROUP. Demographic variables of interest include total population, fertility, mortality, migration, labour force, and so on. The models used are non-parametric models, and include a population model (POPMOD), a life expectancy model (LIFE), a labour force status model (LFSMOD) and a fertility model (FERTILITY). POPMOD - Population Model POPMOD provides annual projections of Australia's population by year for males and females by single year of age. The model is driven by parameter matrices for fertility, mortality and overseas migration. The overseas migration sub-model accepts projection of permanent and long-term arrivals and departures and measures of category jumping. LIFE - Life expectancy Model The life expectancy model calculates survival rates, survivors to age x, deaths at age x to x+n, life table populations and life expectancy for males and females by single year of age. The estimates are constructed from the mortality parameters used in POPMOD. Life expectancies are used in both demographic and policy models. Factors for annuity and allocated pension pricing are provided in policy models. LFSMOD - Labour Force Status Model This is long-run annual model of the Australian labour force to capture structural (trend) behaviour at fine detail. The model projects persons by labour force status, age, gender and income decile. . There is no short-run behavioural response in LFSMOD, as the model simply runs off the observed underlying long-run movements of key, and hopefully stable, parameters, which are estimated as non-linear trends with consistent asymptotic values. Apart from these time-varying parameter matrices, the model's only exogenous inputs are population projections from a population model, such as POPMOD, and aggregate unemployment rates for males and females.FERTILITY – Fertility ModelThe fertility model uses fertility history of the Australian population to project future fertility by year of age. The fertility projections are then used to calculate births for inclusion in the population model to produce annual projections of total Australian population. Household Income Models Personal Tax Microsimulation Model Based On Sample Tax File The Personal Tax Microsimulation Model is based on a sample file provided by the ATO each year. It is a 1 in 12.5 unit record file, sampling records taken from those of all individuals who submitted a personal tax return in the relevant financial year. Details of age, incomes, deductions and rebates are provided. RIM Modified Version of STINMOD OUTYEARS is an extension of STINMOD developed by NATSEM with some assistance from the Treasury. The static microsimulation model of household incomes is used for family payment and family taxation costings up to 4 to 5 years into the future.