SimPaths: An open-source microsimulation model for life course analysis

The demographic transition currently unfolding across the world has profound implications for diverse aspects of societies, including the functioning and financing of the welfare state. As baby boomers (those born in the two decades after WWII) move out of work and into retirement and are replaced by smaller cohorts of working-age individuals, the shares of national populations in employment are projected to decline, which will reduce public tax receipts at the same time as needs in terms of health care and social assistance are projected to rise.

The old-age dependency ratio (the population aged 65 and over, relative to the population aged 20 to 64) of OECD countries doubled from 14% in 1950 to 30% in 2020 and is projected to double again to 59% by 2075.¹ Although increases in age dependency ratios are anticipated in all OECD countries, there is substantial cross-country variation. Korea is an outlier in this series, projected to rise from the lowest dependency ratio in 1950 (6%) to the highest in 2075 (79%). EU countries also feature prominently in the transition, accounting for eight of the ten countries with the highest projected age dependency ratios in the OECD by 2075.

The current rises in age dependency ratios are driven by unprecedented declines in fertility and rises in life expectancy, as well as the ageing of the baby boom generation. The OECD average total fertility rate fell by more than half from 3.3 children per woman in 1960 to 1.6 children in 2020.² During this same period, the total fertility rate in EU countries fell from 2.6 to 1.5, and from 6.0 to 0.8 in South Korea. Furthermore, average life expectancy at birth in OECD countries increased from 68.1 years in 1960 to 80.5 years in 2020, from 69.7 to 79.9 years in EU countries, and in Korea from 58.7 years in 1970 to 80.5 years in 2020.³

These remarkable shifts in fertility and life expectancy have a pervasive bearing on social and private organisation. From partner relations to education decisions, labour market participation to housing demand, changing gender roles, caring needs, and healthcare provisions; few aspects of modern life are left unaffected. With longer lives, inequalities in income, wealth and health also have more time to compound. In short, OECD countries are passing through a period of social revolution.⁴

Many current trends are now well established, displaying predictable patterns over time. The influence that these trends have on margins of concern are also often predictable. For example, an older population implies a greater prevalence of old-age pensions in payment and more demand for health care, both of which impose a burden on the public purse. Yet, to move beyond basic postulations, numerical analyses are required. This is particularly true when attempting to take into consideration multiple inter-related temporal trends.

Most numerical approaches used to anticipate the scale and scope of population ageing provide limited detail for exploring distributional effects at a given point in time, longitudinal effects over individual life courses, and implications for financing of the welfare state. The European Commission and OECD, for example, both adopt a cohort methodology to project the scale and effects of population ageing.⁵ These methods are based on assumptions concerning cohort-average effects for employment, fertility, health, and mortality. Such cohort averages, however, are ill-suited for exploring fiscal flows associated with the welfare state, which crucially depend upon distributional differences within (as well as between) cohorts.

Interest in within-cohort variation and heterogeneity in life course trajectories has motivated the development of dynamic microsimulation models, especially during the last three decades. In dynamic microsimulation models, the characteristics of each micro unit (individual people in our case) are projected through time from a starting point usually derived from cross-sectional survey (micro-)data. Temporal projections are based on biological, institutional, or behavioural rules. Examples of biological rules are ageing and death. Examples of institutional rules are tax and benefits systems. Examples of behavioural rules are any choices that the units can make, for instance related to education, household composition, fertility, labour supply, lifestyle and health behaviour, savings, and investments.

The output from a dynamic microsimulation can usefully be conceptualised as a database that reports evolving information for the population of interest.⁶ In a dynamic microsimulation, individuals can be linked, so that partner and household characteristics complement individual state variables. New individuals can enter the simulation at later periods, for instance as the result of immigration or fertility. The rules for updating the simulated population include parameters with values that are either exogenously assumed (e.g. tax-benefit parameters) or estimated from available survey data.

Use of dynamic microsimulation methods has grown substantially during the last four decades, benefitting from the increasing availability of high-quality microdata, analytical advances, and increases in computing power.⁷ Despite the emergence of generic software packages (GENESIS, JAS-mine, LIAM2, MODGEN, openM++),⁸ bespoke analytical frameworks continue to be (re-)implemented in the literature. Each independent research group has typically developed its own model code, which is often maintained as a proprietary asset. This imposes considerable developmental overhead on prospective entrants to the literature and limits external validation of reported results.

One way to mitigate developmental costs and facilitate external validation is to publish all research materials as open source. This approach is being actively promoted by the European Commission in its “open access” requirements for funded research, which extend to peer-reviewed publications and research metadata.

This paper describes a novel open-source framework for dynamic microsimulation modelling, which we refer to as SimPaths. All source code is freely available for download under a European Free/Open Source Software (F/OSS) EUPL-1.2 license, alongside evolving, increasingly detailed documentation.⁹ The framework incorporates many state-of-the-art features which are rarely combined in dynamic models.

First, SimPaths generates data for a diverse range of life course domains – education, work, family life and health – explicitly modelling the dynamic feedback effects between them.

Second, SimPaths is linked to an underlying tax-benefit model, which provides a realistic description of the impact of taxes and benefits at both the individual and population level. The detailed tax-benefit description that reflects prevailing public policy is important for any evaluation of the funding and distributional implications of population ageing for the welfare state.

Third, SimPaths features rich behavioural models over the principal economic margins of decision making (time-use and savings), where projected choices depend not only on individual characteristics, but also on the influence of fiscal incentives on future expectations.

Fourth, from an architectural perspective, SimPaths is built following a highly modular approach. This facilitates switching between alternative methods for projecting behaviour to allow for sensitivity and robustness analysis. The model is written in Java, using the JAS-mine suite of simulation libraries (Richiardi and Richardson, 2017).

Fifth, SimPaths is built with an eye to facilitate adaptation to different countries. This is achieved by decoupling the dynamic structure from the tax-benefit model, so that alternative tax-benefit systems can be easily interchanged within the model. Furthermore, care has been taken to describe model dynamics that can be estimated on a single standardised data source for European Union countries (Statistics on Income and Living Conditions, EU-SILC).

The remainder of the paper is structured as follows. Section 2 places SimPaths in the context of contemporary microsimulation models. Section 3 presents the architecture behind SimPaths. Section 4 discusses model estimation and validation. Section 5 describes existing applications of the framework, and planned extensions. Section 6 discusses the funding strategy and governance structure. Section 7 concludes.

Dynamic microsimulation models require significant resources to develop and maintain, and are consequently most commonly developed within policy institutions (e.g. government departments), or form part of the modelling infrastructure of research institutions (e.g. Statistics Canada, Urban Institute, CeMPA, GenIMPACT, NATSEM).¹⁰ This is a marked departure from the common academic practice framed upon ‘one model – one paper’.¹¹ It also presents challenges to assessment of prevailing best-practices in research, as models are often proprietary, and developers typically have few incentives to publish accompanying documentation.

This section reviews a selection of microsimulation models that satisfy three conditions: there is evidence that the model is in current active use; the model is publicly documented; and the model focuses on life-course dynamics of people. These filters identify seven examples for discussion. The condition on “active use” is a particularly important, as it excludes the majority of examples discussed in previous surveys (O’Donoghue and Dekkers, 2014; O’Donoghue and Dekkers, 2018; Li et al., 2014; Li and O’Donoghue, 2013; Harding, 2023; O’Donoghue, 2001; Klevmarken, 1997).¹²

DYNASIM (Favreault et al., 2015) projects a representative sample of the US population forward in time, simulating demographic events such as births, deaths, marriages, divorces, and health status, and economic events such as labour force participation, earnings, hours of work, and retirement. The model is developed at the Urban Institute and has evolved from the original work of Orcutt et al. (1976). The model simulates home and financial assets, living arrangements, and includes a detailed calculation of tax and benefit entitlements. In recent years the scope of the model has expanded considerably to cover health-related outcomes, including disability, chronic conditions, and projections of health insurance coverage, premium costs, and out-of-pocket medical spending.

MOSART (Andreassen et al., 2020) is a life course model based on administrative data for the entire Norwegian population, which projects birth, death, migration, marriage, divorce, educational activities, labour force participation, retirement, income and wealth based on estimated transition probabilities. The model first emerged in 1990 and is used by Statistics Norway and the Norwegian government for projections and policy analyses related to the pension system.

IrpetDin (Maitino et al., 2020) and T-DYMM (Conti et al., 2024) are two models calibrated to Italian data. IrpetDin is estimated on two different samples: the whole of Italy, and the Tuscany region. It simulates death, ageing, marriage, fertility, divorce, leaving parental home, migration, secondary school enrolment and graduation, university enrolment and graduation, labour force participation, employment status, income, health status, pensions, social assistance for old people and retirees, disability and long-term care. Education is a particular focus of IrpetDin, and (endogenous) projections of labour supply are matched with (exogenous) projections of labour demand derived from an auxiliary macroeconomic model.

T-DYMM, developed at the Italian treasury, is comprised of a demographic module (fertility, mortality, immigration and emigration, education, exit from parental home, marriages, divorces), a labour market module (employment), a pension module (public and private pensions), a wealth module (home ownership and income from other assets), and a tax-benefit module. Employment distinguishes between self-employment and dependent employment, contract type (open-ended vs. fixed term), part-time vs. full time, and public vs. private sector. All transitions are modelled as reduced-form probabilities.

A more limited focus on the labour market characterises SLAMM, a microsimulation model for Slovakia (Štefánik and Miklošovič, 2020). The microsimulation model projects labour supply, and is coupled with an external input-output model that projects sectoral employment levels, with wages endogenously adjusting to ensure market closure.

The LifeSim model by Skarda et al. (2021) projects developmental, economic, social and health outcomes from birth to death for each child in the Millennium Birth Cohort (MCS) in England. The model controls for a large number of individual characteristics and behaviour, including human capital development in childhood (social skills, cognitive skills, teenage smoking), and has a focus on mental and physical health, and well-being. All transitions are governed by reduced form probabilities, while life course income profiles are adjusted for individual shifters, such as disability. Taxes and benefits are modelled using stylised functions.

microWELT (Spielauer et al., 2020) reproduces demographic projections for Austria, Spain, Finland, and the UK, by simulating fertility, mortality, education, partnership formation and dissolution, and migration. These projections are then used to re-weight cross-sectional microdata generated by the EUROMOD tax-benefit model (Sutherland and Figari, 2013).

DYNASIM, MOSART, IrpetDin, and T-DYMM are all proprietary models. SLAMM code is available upon request, while LifeSim and microWELT are open-source, although with contributors limited to the original developers.

Many characteristics are common to the models discussed above. Most start with a representative population cross-section, which is evolved forward in time (LifeSim is cohort-based). Most simulate events at discrete (annual) intervals (microWelt is cast in continuous time). Most include demographic events related to family composition, health events, and economic events (SLAMM is limited to education and economic activity). Most give at least some consideration to tax and benefit policies (SLAMM is again an exception).

Differences between models mostly relate to the respective analytical focus, technical implementation, and econometric specification. In this regard, DYNASIM stands out for its comprehensiveness in both economic and health-related outcomes, while IrpetDin and SLAMM are noticeable for their interaction with macroeconomic projections.

Relative to the models discussed above, the main innovations of SimPaths are: (i) focus on facilitating new research in the field by maintaining open-source coding and associated documentation (in common with LifeSim and microWELT); (ii) externalisation of the tax-benefit component to a third-party dedicated tax-benefit model (see Section 3.9 below); and (ii) use of a structural model of individual decision-making, rather than simple transition probabilities.

The workhorse version of SimPaths employs a structural model of labour supply where households choose hours worked by each adult of each benefit unit in each period as though associated decisions are made to optimise the trade-off between leisure and disposable income. An advanced version extends this behavioural model to take into account intertemporal considerations along the income-leisure and consumption-savings margins (Section 3.8.2). The advantage of using structural approach to project behaviour, relative to the more commonly applied reduced-form statistical approach, is that the structural approach is theoretically designed to be invariant to the policy context. Use of utility theory in the case of SimPaths, allows simulated behaviour to respond in a coherent fashion to changes in incentives, including those described by tax and benefits policy (see, e.g. Blundell et al., 2016; Blundell et al., 2021).

In contrast to some of the models discussed above, SimPaths does not benefit from dedicated institutional funding. Rather, development is driven by modelling needs of specific research projects (and associated funding). Between 2014 and 2018, SimPaths (originally LABSim) was developed at the Institute for New Economic Thinking (INET), University of Oxford. The model now benefits from a growing community of researchers based at diverse institutions, with core development conducted at the Centre for Microsimulation and Policy Analysis (CeMPA), University of Essex, and the Social & Public Health Sciences Unit, University of Glasgow. See Section 7 for more details.

SimPaths models are currently estimated for the United Kingdom, Greece, Hungary, Italy and Poland, and are under development for Germany, Spain and Sweden. Two master versions are currently maintained: one for the UK, and one for EU countries. The UK version is the most comprehensive variant where developments are usually introduced and tested. For this reason, in this paper we focus on validation of the UK model parameterisation (Section 4).

SimPaths implements an hierarchical architecture where individuals are organised in benefit units (for fiscal purposes), and benefit units are organised in households.¹³ The model projects data at yearly intervals, reflecting the yearly frequency of the survey data used to estimate model parameters.¹⁴ The model is composed of eleven modules: (i) Ageing, (ii) Education, (iii) Health, (iv) Family composition, (v) Social care, (vi) Investment income, (vii) Labour income, (viii) Disposable income, (ix) Consumption, (x) Health (2), and (xi) Statistical display. Variables from different modules characterise (multi-dimensional) well-being. Each module is composed of one or more processes; for example, the ageing module contains ageing, mortality, child maturation, and population alignment processes. Empirical specification of dynamic processes makes extensive use of cross-module characteristics (state variables).¹⁵

The model described in this paper is the public release 2023.12.18 of SimPaths. Simulated modules and processes are organised as displayed in Figure 1 and Table 1.

Figure 1

Download asset Open asset

In each simulated year, agents are first subject to the ageing process, followed by population alignment. The alignment process adjusts the simulated population to match official population projections distinguished by gender, age (single-year brackets),¹⁶ and geographic region at NUTS1 level,¹⁷ which ensures that simulated output remains representative of the country’s population.

The education module determines transitions into and out of student status. Students are assumed not to work and therefore do not enter the labour supply module. Individuals who leave education have their level of education re-evaluated and can become employed.¹⁸

The health module projects an individual’s health status, comprising both self-rated general health and mental health metrics (based on a clinically validated measure of psychological distress using a Likert scale and a caseness indicator), and determines whether an individual is long-term sick or disabled (in which case, he/she is not at risk of work and may require social care).

The family composition module is the principal source of interactions between simulated agents in the model. The module projects the formation and dissolution of cohabiting relationships and fertility. Where a relationship forms, then spouses are selected via a matching process that is designed to reflect correlations between partners’ characteristics observed in survey data. The proportion of the population in a cohabiting relationship is, by default, aligned to the population aggregate in the years for which observational data is available, to account for changes in household structure introduced by the population alignment.

Females in couples can give birth to a (single) child in each simulated year, as determined by a process that depends on a range of characteristics including age and presence of children of different ages in the household. In case of divergence from the officially projected number of newborns, fertility rates are adapted by an alignment process to match population projections for new-born children distinguished by gender, region, and year.

The social care module projects provision and receipt of social care activities for people in need of help due to poor health or advanced age. The module is designed to distinguish between formal and informal social care, and the social relationships associated with informal care. The social care module accounts for the time cost incurred by care providers with respect to informal care, and the financial cost incurred by care receivers with respect to formal care.

The investment income module projects income from investment returns and (private) pensions. The approach taken to project these measures of income depends upon the model variant considered for analysis. Where consumption/savings decisions are simulated using a structural behavioural framework, then asset income is projected based on accrued asset values and exogenously projected rates of return. Alternatively, computational burden of model projections can be economised by proxying non-labour income without explicitly projecting asset holdings.

The labour income module projects potential (hourly) wage rates for each simulated adult in each year and their associated labour activity. Given potential wage rates, hours of paid employment by all adult members of a benefit unit are generated. Labour (gross) income is then determined by multiplying hours worked by the wage rate.

The disposable income module uses information concerning disability, relationship status and fertility, social care, investment income and labour income to evaluate taxes and benefits and disposable income for each projected benefit unit in each year. The model includes alternative methods for projecting employment status, some of which involve interactions between the labour income and disposable income modules to identify preferred combinations of labour supply and disposable income. An alignment routine is used to match projected rates of employment against population aggregates, to correct for biases in the labour supply model.¹⁹

Given disposable income and household demographics, the consumption module projects measures of benefit unit expenditure. Where the model projects wealth, then a simple accounting identity is used to track the evolution of benefit unit assets through time. A regression-based homeownership process predicts if the primary residence is owned by either of the responsible adults in a benefit unit, in which case the benefit unit is considered to own its home.

A secondary mental health process adjusts estimates obtained by the primary process to account for the effect of exposure to labour market transitions, such as moving in and out of employment and/or poverty.

At the end of each simulated year, SimPaths generates a series of year specific summary statistics. All of these statistics are saved for post-simulation analysis, with a subset of results also reported graphically as the simulation proceeds.

Early applications of the SimPaths framework focussed on the short-to-medium term impact of social policies on mental health outcomes. Kopasker et al. (2024) consider the UK policy response to COVID-19, comparing baseline simulations with the policies that the UK government enacted in 2020-2021 to sustain incomes during the pandemic with counterfactual simulations where pre-crisis policies remained in place. Their period of analysis is 2017 to 2025. Results show that the policy response prevented a further 3.4 percentage points (pp) increase in the prevalence of common mental disorders (CMDs) with respect to pre-covid levels, on top of the more than 10pp increase observed in 2020-2021 with covid-19 legislation in place. This amounts to approximately 1.2 million additional cases of CMDs prevented by the covid-19 policy response. Beyond 2021, as employment levels rapidly recovered, psychological distress returned to the pre-pandemic trend.

Thomson et al. (2024) consider the effects of hypothetical basic income schemes over the period 2022-2026. They show that the policy has potential to improve short-term population mental health by reducing poverty, particularly among women, but impacts are highly dependent on whether individuals choose to remain in employment following its introduction. Sensitivity analysis concerning employment alternatives is conducted by replacing the labour supply module of SimPaths with alternative assumptions on the behavioural responses to the policy change.

Ongoing work investigates the employment effects and associated impact on psychological distress of the job search requirements of Universal Credit (UC), a major social protection scheme in the UK. The analysis distinguishes two channels by which conditionality affects employment outcomes. On the one hand, being on UC forces individuals to search more. On the other, the same search requirements impact mental health, reducing search effectiveness. Two more effects are identified. Individuals who are successful in their job search get a boost to their mental health associated with both entering employment and exiting UC. However, some individuals might be deterred by the search requirements from taking-up UC, resulting in reduced coverage and effectiveness of the policy. All of these effects are reinforced over time, hence requiring a dynamic analysis.

A separate analysis explores the impact of maintenance payments on mothers, fathers, and children. Both the provision and the receipt of maintenance payments impact labour supply decisions and income dynamics. These in turn affect the life course trajectories of children.

Other applications focus less on the impact of policies, and more on life course dynamics. van de Ven et al. (2024) study the individual and societal implications of increased care needs associated with population ageing over the next five decades in the UK. The analysis projects a sharp rise in the incidence of need for social care, from just under 4.0 million in 2020 to 7.6 million by 2070, with 70% of this increase driven by people aged 80 and over. The coincident costs of social care are projected to rise from £35 billion in 2020 (1.6% of GDP) to 152 billion by 2070 (2.8% of GDP), with a large fraction of this cost being sustained by depleted savings.

Informal social care provided by partners is projected to account for over half of all social care throughout the simulated time period, with the formal sector being the next largest provider accounting for approximately 15% of provisions. van de Ven et al. (2024) also compare simulations where individuals make decisions based on future expectations (the ‘expectations explicit’ approach described in Section 3) with simulations without forward looking behaviour. They show that the anticipation of the need to provide informal care leads to reduced employment and earnings. Yet the model projections suggest that precautionary savings set aside against the risk of requiring formal social care more than off-set the above factors, so that the net impact of care on savings to age 55 is positive when averaged over the population cohorts born between 1990 and 1999.

Richiardi et al. (2024) use SimPaths to explore the feedback loops between health, family and labour market outcomes, and the associated implications for income and health inequalities over alternative time horizons. They devise a computational procedure to isolate direct and indirect (mediated) effects in a microsimulation framework, consisting in artificially shocking one domain (e.g. by forcing a partnership dissolution on the simulated individuals) and comparing simulations where the impact of the shock is allowed to affect other domains (e.g. because labour market outcomes depend on, amongst other things, partnership status) with simulations where the evolution of the other domains remains unaffected and follow baseline simulations – the shock is only allowed to influence the future evolution of the shocked variable itself. Results indicate that partnership status has significant effects on other life domains, and highlight attenuation mechanisms that facilitate bouncing back to a partnered status following a union dissolution. On the other hand, health is found to have fewer connections to other life domains, with limited feedback that attenuate or exacerbate the effects of an adverse shock.

A number of ongoing funded projects provide further applications, together with the need for extending and updating the model. Major areas of development are the inclusion of measures of well-being, interaction with macro models, adoption of a spatially disaggregated synthetic population as a starting point of the simulations, and improved modelling of wealth dynamics (and associated impacts on health). The latter will distinguish between housing, pensions, other assets and unsecured debt. Private transfers between households (e.g. parents helping with the home purchase of their children) will also be included, with implications for intergenerational fairness and the intergenerational transmission of inequality. Self-rated health and psychological distress are being replaced by the physical score component (PSC) and mental score component (MCS) respectively of the SF-12 health assessment. More health outcomes are also being added, including loneliness, nutrition, and tobacco consumption. Measures of neighbourhood safety and housing quality are being introduced as pathways between socio-economic position and health outcomes. A separate workstream looks at improving modelling of external migration, which is currently treated as a residual (see Section 3.1.1). Estimates are being updated to include more recent waves of the input data, account for migrant status and ethnicity, and improve the modelling of health outcomes (including mortality). Finally, as mentioned, some of these models have already been estimated (e.g., Germany, Greece, Hungary, Italy, Poland, Spain and Sweden).

While the open source nature of SimPaths imposes no limitation on potential users, the framework is primarily used by academic researchers. It has no institutional backing nor benefits from institutional funding. Rather, it is developed mostly through research grants. A first version of the model, then called LABSim, was developed with funding from the National Institute for the Analysis of Public Policies (INAPP), an Italian government institution (2016-2017; 2019-2022),²⁷ building on a simpler model developed for Eurofound, the EU Agency for the improvement of living and working conditions (2015).²⁸ The modelling framework received further funding from the UK Health Foundation (2021-2022),²⁹ the UK National Institute for Health Research (NIHR, 2021-2028),³⁰ the Joint Programming Initiative More Years, Better Lives (JPI MYBL, 2021-2025),³¹ the European Research Council (ERC, 2021-2025),³² the European Commission Horizon Europe programme (HE, 2023-2027),³³ the European Spatial Planning Observation Network (ESPON, 2023-2025),³⁴ and the Collaboration of Humanities and Social Sciences in Europe – NORFACE network (CHANSE, 2025-2028).³⁵ The two main institutions involved in the modelling work around the SimPaths framework for the above projects are the Centre for Microsimulation and Policy Analysis (CeMPA) at the University of Essex, and the MRC/CSO Social and Public Health Sciences Unit at the University of Glasgow.

As with all open-source projects, governance presents challenges and opportunities. In terms of the six open-source governance models identified by Neary et al. (2020),³⁶ SimPaths initially followed a Founder-leader configuration, and is now migrating towards a Self-appointing council or board structure.³⁷ Active developers meet with a weekly frequency to coordinate project developments, and maintain single versions of the master code for the UK and the EU models respectively. “SimPaths community meetings” are held with a monthly frequency, to support adopters and users around the globe (these are mainly PhD students and policy analysts). Procedural rules are being drafted to guide contributions and ensure, as far as possible, a shared and unitary management of the project. Ultimately, the project rests on trust between developers, and on the incentives provided by the benefits of an enlarged team of developers in the context of a highly competitive funding environment.

In this paper we have introduced SimPaths, a novel dynamic microsimulation model of individual life course trajectories covering demographic, family, health and work-related events, and discussed parameterisation to the United Kingdom. The innovative features of the model lie in the flexible integration with an external tax-benefit model, and in the structural behavioural modelling at the core of individual decision making over important margins of interest. Extensive validation over the period 2011-2019 shows that the model is able to replicate well the joint evolution of individual characteristics.

SimPaths is fully open source with increasingly detailed documentation, which sets it apart from most comparable models in the existing literature. The objective of the developmental team in this regard is to facilitate, and thereby encourage researcher entry into a field that we believe presents extensive opportunities to further understanding of a wide range of practically important phenomena. From theoretical descriptions of individual decision making, sustainability of personal and public finances, and the nature of burdens and opportunities to which people are and will be subject during the prospective half century, there is much of interest to explore.

The open-source nature and governance structure of SimPaths is a radical departure from the major past and current dynamic microsimulation endeavours, which typically developed as proprietary, institutional models. While some of them passed the test of time, few have produced a consistent stream of academic outputs over the years.³⁸ History will tell whether SimPaths will remain an ephemeral convergence of interests by a relatively small group of researchers, or will evolve into a broader research infrastructure freely available to the scientific and policy communities.

1. 1
   The dynamic cross-sectional microsimulation model MOSART

   1. L Andreassen
   2. D Fredriksen
   3. HM Gjefsen
   4. E Halvorsen
   5. NM Stølen

   (2020)

   International Journal of Microsimulation 13:92–113.

   https://doi.org/10.34196/ijm.00214

   • Google Scholar
2. 2
   Accounting for methodological, structural, and parameter uncertainty in decision-analytic models: a practical guide

   1. J Bilcke
   2. P Beutels
   3. M Brisson
   4. M Jit

   (2011)

   Medical Decision Making 31:675–692.

   https://doi.org/10.1177/0272989X11409240

   • PubMed
   • Google Scholar
3. 3
   Female Labor Supply, Human Capital, and Welfare Reform

   1. R Blundell
   2. M Costa Dias
   3. C Meghir
   4. J Shaw

   (2016)

   Econometrica 84:1705–1753.

   https://doi.org/10.3982/ECTA11576

   • Google Scholar
4. 4
   Wages, Experience, and Training of Women over the Life Cycle

   1. R Blundell
   2. M Costa-Dias
   3. D Goll
   4. C Meghir

   (2021)

   Journal of Labor Economics 39:S275–S315.

   https://doi.org/10.1086/711400

   • Google Scholar
5. 5
   SimPaths: an open-source microsimulation model for life course analysis

   1. P Bronka
   2. J van de Ven
   3. D Kopasker
   4. SV Katikireddi
   5. MG Richiardi

   (2023)

   CeMPA WP 6/23, University of Essex.

   • Google Scholar
6. 6
   Coping with ageing: a dynamic approach to quantify the impact of alternative policy options on future labour supply in OECD countries

   1. JM Burniaux
   2. R Duval
   3. F Jaumotte

   (2003)

   Coping with ageing: a dynamic approach to quantify the impact of alternative policy options on future labour supply in OECD countries, OECD Economic Department WP 371.

   • Google Scholar
7. 7
   Long-Term Labour Force Projections for the 25 EU Member States: A Set of Data for Assessing the Economic Impact of Ageing

   1. G Carone

   (2005)

   SSRN Electronic Journal 235.

   https://doi.org/10.2139/ssrn.871086

   • Google Scholar
8. 8
   Guy H. Orcutt’s Engineering Microsimulation to Reengineer Society

   1. C-T Cheng

   (2020)

   History of Political Economy 52:191–217.

   https://doi.org/10.1215/00182702-8718000

   • Google Scholar
9. 9
   The Italian Treasury Dynamic Microsimulation Model (T-DYMM): Data, Structure and Baseline Results

   1. R Conti
   2. M Bavaro
   3. S Boscolo
   4. E Fabrizi
   5. C Puccioni
   6. O Ricchi
   7. S Tedeschi

   (2024)

   SSRN Electronic Journal 17:23–68.

   https://doi.org/10.2139/ssrn.4546893

   • Google Scholar
10. 10
    Confidence intervals for policy reforms in behavioural tax microsimulation modelling

    1. J Creedy
    2. G Kalb
    3. H Kew

    (2007)

    Bulletin of Economic Research 59:37–65.

    https://doi.org/10.1111/j.0307-3378.2007.00250.x

    • Google Scholar
11. 11
    LIAM2: a New Open Source Development Tool for Discrete-Time Dynamic Microsimulation Models

    1. G de Menten
    2. G Dekkers
    3. G Bryon
    4. P Liégeois
    5. C O’Donoghue

    (2014)

    Journal of Artificial Societies and Social Simulation 17:2574.

    https://doi.org/10.18564/jasss.2574

    • Google Scholar
12. 12
    Adding immigrants to microsimulation models

    1. HO Duleep
    2. DJ Dowhan

    (2008)

    Social Security Bulletin 68:51–65.

    • Google Scholar
13. 13
    The 2021 Ageing Report. Underlying Assumptions and Projection Methodology”, DG ECFIN

    1. European Commission

    (2020)

    The 2021 Ageing Report. Underlying Assumptions and Projection Methodology”, DG ECFIN, European Economy, Institutional Papers No 142.

    • Google Scholar
14. 14
    The Dynamic Simulation of Income Model (DYNASIM

    1. M Favreault
    2. KE Smith
    3. RW Johnson

    (2015)

    The Dynamic Simulation of Income Model (DYNASIM, Urban Institute research report.

    • Google Scholar
15. 15
    Perspectives on Free and Open Source Software

    1. J Feller
    2. B Fitzgerald
    3. SA Hissam
    4. KR Lakhani

    (2005)

    The MIT Press, Cambridge MA.

    • Google Scholar
16. 16
    Which moments to match?

    1. AR Gallant
    2. GE Tauchen

    (1996)

    Econometric Theory 12:657–681.

    https://doi.org/10.1017/S0266466600006976

    • Google Scholar
17. 17
    GENESIS - The GENEric simulation system for modelling state transitions

    1. MS Gillman

    (2017)

    Journal of Open Research Software 5:24.

    https://doi.org/10.5334/jors.179

    • PubMed
    • Google Scholar
18. 18
    Testing the statistical significance of microsimulation results: a plea

    1. T Goedemé
    2. KV den Bosch
    3. L Salanauskaite
    4. G Verbist

    (2013)

    International Journal of Microsimulation 6:50–77.

    https://doi.org/10.34196/ijm.00088

    • Google Scholar
19. 19
    Indirect inference

    1. C Gourieroux
    2. A Monfort
    3. E Renault

    (1993)

    Journal of Applied Econometrics 8:S85–S118.

    https://doi.org/10.1002/jae.3950080507

    • Google Scholar
20. 20
    Challenges and opportunities of dynamic microsimulation modelling

    1. A Harding

    (2023)

    International Journal of Microsimulation 16:1–13.

    https://doi.org/10.34196/IJM.00280

    • Google Scholar
21. 21
    Behavioral Modeling in Micro Simulation Models. A Survey”, Working Paper 1997:31, Department of Economics, Uppsala University, republished as Klevmarken, N.A

    1. NA Klevmarken

    (1997)

    International Journal of Microsimulation 15:78–88.

    • Google Scholar
22. 22
    Evaluating the influence of taxation and social security policies on psychological distress: A microsimulation study of the UK during the COVID-19 economic crisis

    1. D Kopasker
    2. P Bronka
    3. RM Thomson
    4. V Khodygo
    5. T Kromydas
    6. P Meier
    7. A Heppenstall
    8. C Bambra
    9. N Lomax
    10. P Craig
    11. M Richiardi
    12. SV Katikireddi

    (2024)

    Social Science & Medicine (1982) 351:116953.

    https://doi.org/10.1016/j.socscimed.2024.116953

    • PubMed
    • Google Scholar
23. 23
    Simulation estimation of time-series models

    1. B-S Lee
    2. BF Ingram

    (1991)

    Journal of Econometrics 47:197–205.

    https://doi.org/10.1016/0304-4076(91)90098-X

    • Google Scholar
24. 24
    A survey of dynamic microsimulation models: uses, model structure and methodology

    1. Jinjing Li
    2. C O’Donoghue

    (2013)

    International Journal of Microsimulation 6:3–55.

    https://doi.org/10.34196/ijm.00082

    • Google Scholar
25. 25
    Handbook of Microsimulation Modelling

    1. J Li
    2. C O’Donoghue
    3. G Dekkers

    (2014)

    Dynamic Models, Handbook of Microsimulation Modelling, Emerald Group Publishing.

    • Google Scholar
26. 26
    IrpetDin. A Dynamic Microsimulation Model for Italy and the Region of Tuscany

    1. ML Maitino
    2. L Ravagli
    3. N Sciclone

    (2020)

    International Journal of Microsimulation 13:27–53.

    https://doi.org/10.34196/IJM.00224

    • Google Scholar
27. 27
    Microsimulation Modelling for Policy Analysis

    1. L Mitton
    2. H Sutherland
    3. M Weeks

    (2000)

    Cambridge: Challenges and Innovations, Cambridge University Press.

    • Google Scholar
28. 28
    Assessing the Reliability of Complex Models: Mathematical and Statistical Foundations of Verification, Validation, and Uncertainty Quantification

    1. National Research Council

    (2012)

    The National Academies Press.

    • Google Scholar
29. 29
    Understanding open source governance models, Red Hat Blog July 17

    1. D Neary
    2. J Berkus
    3. K Novakovic
    4. B Behrenshausen

    (2020)

    https://www.redhat.com/en/blog/understanding-open-source-governance-models, Accessed, 20 December 2024.

    • Google Scholar
30. 30
    Dynamic Microsimulation: A Methodological Survey

    1. C O’Donoghue

    (2001)

    Brazilian Electronic Journal of Economics, 4.

    • Google Scholar
31. 31
    Dynamic microsimulation

    1. C. O’Donoghue
    2. G. Dekkers

    (2014)

    In: C. O’Donoghue, editors. Handbook of Microsimulation Modelling, Contributions to Economic Analysis, 293. Leeds: Emerald Group Publishing Limited. pp. 305–343.

    • Google Scholar
32. 32
    Increasing the Impact of Dynamic Microsimulation Modelling

    1. C O’Donoghue
    2. G Dekkers

    (2018)

    International Journal of Microsimulation 11:61–96.

    https://doi.org/10.34196/ijm.00174

    • Google Scholar
33. 33
    Practical Microsimulation Modelling

    1. C. O’Donoghue

    (2021)

    Oxford: Oxford University Press.

    • Google Scholar
34. 34
    Old-age dependency ratio (indicator)

    1. OECD

    (2023)

    https://www.oecd.org/en/data/indicators/old-age-dependency-ratio.html, Accessed, 29 March 2023.

    • Google Scholar
35. 35
    Fertility rates (indicator)

    1. OECD

    (2023)

    https://www.oecd.org/en/data/indicators/fertility-rates.html, Accessed, 29 March 2023.

    • Google Scholar
36. 36
    Life expectancy at birth (indicator)

    1. OECD

    (2023)

    https://www.oecd.org/en/data/indicators/life-expectancy-at-birth.html, Accessed, 29 March 2023.

    • Google Scholar
37. 37
    Policy Exploration through Microanalytic Simulation

    1. GH Orcutt
    2. S Caldwell
    3. R Wertheimer

    (1976)

    Washington DC: Urban Institute.

    • Google Scholar
38. 38
    JAS-mine: A new platform for microsimulation and agent-based modelling

    1. MG Richiardi
    2. RE Richardson

    (2017)

    International Journal of Microsimulation 10:106–134.

    https://doi.org/10.34196/ijm.00151

    • Google Scholar
39. 39
    UKMOD – A new tax-benefit model for the four nations of the UK

    1. M Richiardi
    2. D Collado
    3. D Popova

    (2021)

    International Journal of Microsimulation 14:92–101.

    https://doi.org/10.34196/IJM.00231

    • Google Scholar
40. 40
    Attenuation and Reinforcement Mechanisms Over the Life Cycle

    1. M Richiardi
    2. P Bronka
    3. J van de Ven

    (2024)

    CeMPA WP 2/24.

    https://doi.org/10.2139/ssrn.4735597

    • Google Scholar
41. 41
    Age of withdrawal from the labour force in the OECD countries

    1. P Scherer

    (2002)

    Labour Market and Social Policy Occasional Papers.

    • Google Scholar
42. 42
    A Brief, Global History of Microsimulation Models in Health: Past Applications, Lessons Learned and Future Directions

    1. DJ Schofield
    2. MJB Zeppel
    3. O Tan
    4. S Lymer
    5. MM Cunich
    6. RN Shrestha

    (2018)

    International Journal of Microsimulation 11:97–142.

    https://doi.org/10.34196/ijm.00175

    • Google Scholar
43. 43
    Editorial: Special Issue in honour of Vale Emerita Professor Ann Harding AO

    1. DJ Schofield

    (2023)

    International Journal of Microsimulation 16:i–iii.

    https://doi.org/10.34196/IJM.00278

    • Google Scholar
44. 44
    LifeSim: A Lifecourse Dynamic Microsimulation Model of the Millennium Birth Cohort in England

    1. I Skarda
    2. M Asaria
    3. R Cookson

    (2021)

    International Journal of Microsimulation 14:2–42.

    https://doi.org/10.34196/IJM.00228

    • Google Scholar
45. 45
    microWELT: A Dynamic Microsimulation Model for the Study of Welfare Transfer Flows in Ageing Societies from A Comparative Welfare State Perspective

    1. M Spielauer
    2. T Horvath
    3. M Fink

    (2020)

    WIFO Working Papers.

    • Google Scholar
46. 46
    Approximate Solutions to Stochastic Dynamic Programs

    1. S Stern

    (1997)

    Econometric Theory 13:392–405.

    https://doi.org/10.1017/S0266466600005867

    • Google Scholar
47. 47
    EUROMOD: the European Union tax-benefit microsimulation model

    1. H Sutherland
    2. F Figari

    (2013)

    International Journal of Microsimulation 6:4–26.

    https://doi.org/10.34196/ijm.00075

    • Google Scholar
48. 48
    Quality Assessment of Microsimulation Models The Case of EUROMOD

    1. H Sutherland

    (2018)

    International Journal of Microsimulation 11:198–223.

    https://doi.org/10.34196/ijm.00178

    • Google Scholar
49. 49
    Short-term impacts of Universal Basic Income on population mental health inequalities in the UK: A microsimulation modelling study

    1. RM Thomson
    2. D Kopasker
    3. P Bronka
    4. M Richiardi
    5. V Khodygo
    6. AJ Baxter
    7. E Igelström
    8. A Pearce
    9. AH Leyland
    10. SV Katikireddi

    (2024)

    PLOS Medicine 21:e1004358.

    https://doi.org/10.1371/journal.pmed.1004358

    • PubMed
    • Google Scholar
50. 50
    SIDD: An adaptable framework for analysing the distributional implications of policy alternatives where savings and employment decisions matter

    1. J van de Ven

    (2017)

    Economic Modelling 63:161–174.

    https://doi.org/10.1016/j.econmod.2017.02.007

    • Google Scholar
51. 51
    Dynamic simulation of taxes and welfare benefits by database imputation

    1. J van de Ven
    2. P Bronka
    3. M Richiardi

    (2022)

    CeMPA Working Paper.

    • Google Scholar
52. 52
    CeMPA WP 7/24

    1. J van de Ven
    2. P Bronka
    3. M Richiardi

    (2024)

    CeMPA WP 7/24.

    • Google Scholar
53. 53
    Modelling foreign labour inflows using a dynamic microsimulation model of an ageing country - Slovakia

    1. M Štefánik
    2. T Miklošovič

    (2020)

    International Journal of Microsimulation 13:102–113.

    https://doi.org/10.34196/IJM.00220

    • Google Scholar

Author details

1. Patryk Bronka

   Centre for Microsimulation and Policy Analysis, Colchester, United Kingdom

   For correspondence

   pbronk@essex.ac.uk

   "This ORCID iD identifies the author of this article:" 0000-0003-1068-3186

2. Justin van de Ven

   Centre for Microsimulation and Policy Analysis, Colchester, United Kingdom

   "This ORCID iD identifies the author of this article:" 0000-0003-4965-0218

3. Daniel Kopasker

   MRC/CSO Social & Public Health Sciences Unit, Glasgow, United Kingdom

   "This ORCID iD identifies the author of this article:" 0000-0002-2636-8431

4. Srinivasa Vittal Katikireddi

   MRC/CSO Social & Public Health Sciences Unit, Glasgow, United Kingdom

   "This ORCID iD identifies the author of this article:" 0000-0001-6593-9092

5. Matteo Richiardi

   Centre for Microsimulation and Policy Analysis, Colchester, United Kingdom

   "This ORCID iD identifies the author of this article:" 0000-0002-3749-7386