Background and Motivation Regional microsimulation models face critical data challenges: survey data lack local representativeness, administrative data access is often restricted, and commercial datasets are costly. Increasingly, researchers turn to synthetic data generated through statistical models, but questions remain about their validity for policy analysis compared to established commercial alternatives like Experian.
This study evaluates synthetic population data against Experian commercial data for Essex microsimulation applications. Both datasets represent different forms of modelled data: Experian combines administrative records, commercial sources, and modelled estimates, while synthetic data is generated through statistical algorithms. Although both are typically validated against official statistics (ONS Census, household size, income benchmarks, age structure), validation against marginal distributions does not guarantee equivalence for policy analysis, where joint distributions and correlation structures matter critically. Our existing work developed UKMOD-aligned weights for Essex by reweighting FRS survey data to match Experians joint distribution of household characteristics. This provides a validated benchmark against which to assess synthetic data alternatives.
Methodology The comparison evaluates synthetic data against our established reweighted UKMOD variant by applying identical tax-benefit policy scenarios through UKMOD using each dataset. Policy simulation outputs - including income distribution changes, gains and losses by household type, and demographic impacts, are compared to assess whether synthetic data replicates the distributional patterns produced by our machine learning-based reweighted data. This direct comparison provides practical evidence on whether synthetic data offers comparable analytical utility to established reweighting methodologies for regional microsimulation applications.
Expected Contribution This research provides practical evidence on whether synthetic data produces comparable policy analysis results to established reweighting methodologies for regional microsimulation. The study addresses a fundamental question: can synthetic data adequately substitute for more resource-intensive data alignment approaches while maintaining analytical reliability? Findings will inform data strategy decisions for researchers and local authorities conducting sub-national distributional analysis, particularly where resource constraints, data access limitations, or timeliness considerations favor synthetic data solutions.
Most studies that report distributional comparisons of income focus on income evaluated over periods that vary between one week and one year. Distributional studies of weekly income recognise the importance of short-term constraints, particularly in relation to material deprivation and poverty. Distributional studies of annual income recognise the capacity of many people to save the proceeds of temporary income peaks to carry them through temporary income troughs. Income measured over longer periods is rarely analysed due to the relative (in)availability of survey data, rather than any more fundamental motivation. Unfortunately, analysis of lifetime incomes for contemporary population cross-sections is complicated in part by the limited historical context captured by existing panel studies, and in part because future incomes are unobservable. Microsimulation is one method to fill gaps in the available statistical record. This study describes how microsimulation methods were used to project lifetime incomes for a contemporary population cross-section of the UK.
Modelling France’s Agirc-Arrco supplementary pension scheme. The Agirc-Arrco federation runs a pension scheme complementary to France’s first pillar CNAV Old-Age pension scheme. This compulsory pay-as-you-go point system is second only to the CNAV by its size. It covers most private sector wage-earners, with 27 million contributors over a given year, and represents on average a third of pension income paid to 14 million pensioners. It is run together by trade unions and employer organisations. They determine most regulatory aspects of the scheme, such as contribution rates or the point’s buying price. Most notably, they set the annual revision of the point value to guarantee Agirc-Arrco’s reserve funds are at least equivalent to 6 months of benefits for the next 15 years. To cover this and other projection needs, the Agirc-Arrco technical department developed a microsimulation model in native Python to project the system’s future income and expenditure at an individual level until 2070 through a range of economic, demographic and regulatory hypotheses. Using a random sample of more than 2 million contributors, retirees and survivor pensioners, it models affiliation to the Agirc-Arrco scheme, labour-market transitions, wages, mortality and retirement decisions. It computes individual contributions, points bought during individual careers, and retirement and survivor pensions consequently paid by Agirc-Arrcos pension system. It is thus a key input for Agirc-Arrco’s decision makers, to ensure the scheme’s sustainability goals are met. It also provides periodic insights regarding the system’s joint management by social partners and is used for aggregate projections of the French pension system coordinated by the Retirement Orientation Council (COR). This presentation provides an overview of Agirc-Arrcos microsimulation model main features and graphical results, shedding light on France’s second most important pension scheme and sharing methodological choices of interest to the microsimulation community,
Nearly half of the unemployed persons are excluded from the unemployment benefit system by the re-form which has been gradually implemented since 1 January 2026 in Belgium, particularly those with a long unemployment history. The main feature of the reform is to limit the duration of unemployment benefit payments to a maximum of two years, depending on work experience, instead of unlimited payments in time as in the old system. Further, the progressivity of the benefit scheme has been strengthened by an increase of 10% of the ceiling amount during the six first months of unemployment, and a re-duction to a lump sum for unemployed persons who are entitled to payments after one year of unemployment.
This paper gives the ex-ante evaluation results on income and labour supply of this reform. The simulation exercise concludes that approximately one third of the excluded persons are expected to return to work, approximately 40% are predicted to receive the social minimum income. The remaining fourth withdraw from the labour market without replacement income, resulting in a significant loss of disposable revenue. Nonetheless the impact on the poverty rate is moderate. Older unemployed, aged 55 years and over, are particularly affected by the reform as more than three fourth of them lose their benefit, representing nearly 22% of the excluded sample. Regional differences are substantial, with Flanders being the least impacted and Brussels-Capital region the most.
To fulfil the evaluation, we use:
The data from the Crossroad Bank which is a representative sample of the Belgian population with more than 300,000 individuals. It includes a wealth of economic, social and demographic information at both individual and household level, The microsimulation model EXPEDITION which is a highly detailed transcription of the Belgian fiscal and social laws into a microsimulation program, which enables to compute a.o. the dis-posable income and its components for each household based on gross salary, labour supply and all the individual and household characteristics. For this exercise, we are particularly interested in the net income from work, replacement income from unemployment or social minimum allowance, depending on eligibility, and additional benefits (children benefit e.g.). EXPEDITION offers also summary statistics and distribution analysis, such as inequality measures depending on income decile, family type, age group, etc. The labour supply model LASER which is a Van Soest-style microeconomic model, enhanced by non-parametric random coefficients on the household preference parameters. Furthermore, we introduce unemployment as an additional labour supply ‘’choice’’. Nonetheless, unlike most similar models, we explicitly define equations capturing the unemployment probability, and the potential unemployment duration. In addition, the replacement income from unemployment is de-fined as the average benefit over the expected unemployment period to account for the degressive scheme of the Belgian unemployment benefit (and the limitation in time after reform), when calculating the income associated with the unemployment ‘’choice’’ alternative. Subsequently, each individual (not only those directly affected by the reform) is impacted when choosing the best employment alternative.
We present a continuous-time longitudinal microsimulation module for labour activity careers in microWELT, a comparative microsimulation model for socio-demographic projections and policy scenario analysis. Individual careers evolve through transitions among employment, unemployment, non-participation, retirement, and family-related leave using an event-driven waiting-time approach. Transition processes are modelled with piecewise-constant hazard regressions featuring competing risks and explicit duration dependence, enabling realistic spell dynamics - especially for unemployment - that are often difficult to reproduce in discrete-time transition models. A central design feature is alignment to scenario targets for aggregate unemployment and labour force participation. Because microWELT, like most policy microsimulations, do not model market mechanisms, these targets provide macro-level closure for projections and a transparent instrument for counterfactual scenario design. Estimating transition hazards in microWELT is challenging because the underlying comparative survey data typically have limited longitudinal depth and small effective sample sizes, implying higher parameter uncertainty than administrative-history models such as Austria’s microDEMS. To address this, we adapt the microDEMS hazard-based framework and implement two alignment mechanisms. First, simulated aggregates are calibrated to scenario targets for overall unemployment and participation. Second, an optional but pivotal mechanism for microWELT constructs group-specific targets using cross-sectional imputation models defined by categorical characteristics (age group, gender, education, health, and family status). These group targets are reconciled with aggregate constraints via monthly calibration of the cross-sectional model during the simulation. Operationally, alignment exploits simulated waiting times for selected transitions - entries into unemployment and exits from the labour force - to rank individuals by implied event timing and then induces the required number of events by selecting those with the shortest waiting times to meet target margins. All other transitions remain fully continuous-time; only alignment-induced events are implemented on monthly steps to enforce cross-sectional constraints. We evaluate the framework through benchmarking and retrospective validation, comparing careers generated from hazards estimated on administrative versus survey data and assessing fit to recent observed outcomes. Validation focuses on the distribution of unemployment spell durations and heterogeneity in unemployment risk across population groups, quantifying how alignment delivers target consistency while preserving plausible continuous-time dynamics and empirically observed group differences.
The distribution of income differs in Ireland and Northern Ireland. Differences in demographics, working patterns, wage levels and the tax-benefit system all contribute to these differences and could prove a barrier to increased co-operation on the Shared Island of Ireland. Using harmonised microsimulation models for Ireland (SWITCH) and Northern Ireland (UKMOD), we identify the drivers of the differences in income distribution between Ireland and Northern Ireland. Using a decomposition technique, we isolate the relative contributions of market income differences - attributable to demographics, labour market participation and wage levels - and the tax-benefit system to differences in income distribution in the two jurisdictions. In a final step, we simulate the implementation of the Irish tax-benefit system in Northern Ireland and vice versa. This exercise indicates the potential costs and distributional effect of harmonising the direct tax and welfare system across the two jurisdictions.
The current German system of means-tested social benefits, which include citizen’s benefit, housing benefit, and supplementary child benefit, is characterized by high effective marginal tax rates, which are often higher than 90 percent across wide income brackets. As a result, even substantial increases in working hours typically yield small gains in disposable household income. These high effective marginal tax rates apply not only to citizen’s benefit, but also to the housing benefit and supplementary child benefit. Additionally, the coexistence of competing benefits – citizen’s benefit on the one hand, housing benefit and supplementary child benefit on the other – makes the social benefits system complex to navigate for those affected. For these reasons, and against the backdrop of considerable fiscal pressures, there is currently intense political and academic debate about reforming the system.
This study evaluates reform proposals currently under political consideration, categorized into two distinct approaches. The first approach comprises variants of a far-reaching reform in which housing benefits and child benefits are combined with citizen’s benefit and basic income support for the elderly to form a single means-tested benefit that is conceptually modeled on the existing citizen’s benefit framework. Variants within this group differ primarily in earned income exemption design. The second approach represents a more moderate intervention in the existing system: housing and supplementary child benefit are not abolished, but are modified and merged, while adjustments to citizen’s benefit exemptions are introduced. Both reform approaches have in common that they aim to reduce the high effective marginal tax rates in the existing system while simplifying the system for recipients and reducing administrative costs.
We analyze the reform proposals with the behavioral tax and transfer microsimulation model of the Institute for Employment Research (IAB-MSM), which uses household microdata from the German Socio-Economic Panel (GSOEP). The IAB-MSM consists of two components: First, a static tax-and-transfer module that simulates the effects of a tax-benefit reform on the disposable income of individual households, including taxes on income, social security contributions, and public transfers. Second, a discrete choice labor supply model that also endogenously models benefit take-up decisions. We use the IAB-MSM to simulate labor supply effects, fiscal effects, the change in the number of claiming households, and distributional impacts of the reform proposals.
We highlight conflicting goals that must be weighed against each other in political decisions. In particular, “generous” reform scenarios, which attempt to reduce marginal tax rates while ensuring that households are not worse off financially than under the current framework, are generally associated with relatively high fiscal costs. “Restrictive” scenarios, on the other hand, which reduce incentives to work in marginal employment or part-time and strengthen incentives to work full-time, sometimes involve significant budget savings, but are accompanied by income losses for households with low earned income, at least in the short term.
The recognition of health as an effective facilitator of productivity growth provides an important avenue for promoting good health and securing funding for health conditions with the greatest impact. Despite health being recognised as a critical policy lever for increasing productivity, there is no cohesive measure of the productivity impacts of health, nor clarity on which conditions result in the greatest productivity loss. While several studies have analysed the impact of health on labour force participation and income, they have used different methods; reported outcomes for a single disease; or combined all health conditions together, making it difficult to rank conditions based on the productivity loss. There has also been relatively little work extending the financial costs of lost productivity to increased welfare dependence or reduced income taxation, or examining the impacts of mortality. We have developed a new Australian microsimulation model, ProductiveLifeMOD, to quantify the national productivity impacts of chronic illness and mortality and the associated financial costs of productivity loss to individuals, families and government. This model integrates morbidity, mortality, fiscal impacts, and productivity effects across the full working-age population.
ProductiveLifeMOD is an extension of our microsimulation model, Health&WealthMOD, the first Australian microsimulation model of the economic impacts of early retirement due to chronic illness, which focused on the impacts on mature-aged workers aged 45 to 64 years. The new model analyses impacts across the whole working-age population, as there is growing evidence that health is also a major factor limiting labour force participation among younger people of workforce age, although the relevant health conditions may differ from those affecting mature-aged workers. The model analyses the productivity impacts of both morbidity and mortality together, as well as the related financial costs to individuals, families and government.
The base population of ProductiveLifeMOD is built from three Surveys of Disability, Ageing and Carers (SDAC) conducted in 2012, 2015 and 2018. These are nationally representative cross-sectional household surveys conducted by the Australian Bureau of Statistics (ABS) and provide detailed data on socio-demographic characteristics, chronic illness, disability and labour force participation. Mortality estimates are obtained from the Person Level Integrated Data Asset (PLIDA), a linked administrative dataset of the Australian population developed by the ABS. The three SDAC datasets are statically aged to form the base population. Statistical matching of the base data with STINMOD output is undertaken using a set of matching variables to create a synthetic dataset for modelling. This paper describes the development of ProductiveLifeMOD and its application for simulating policy impacts.
ProductiveLifeMOD enables direct comparisons of productivity losses and associated financial costs across specific chronic conditions. The model will provide a mechanism to determine where investment in health would generate the greatest productivity gains and economic benefits.
Dynamic microsimulation requires generating coherent multivariate micro-trajectories over time across multiple outcomes (e.g., employment, income, hours), while handling panel gaps and propagating uncertainty into downstream indicators. Common approaches—transition models, chained regressions, and hot-deck imputation—often yield a single deterministic completion and can struggle to preserve high-dimensional joint structure, especially under block nonresponse.
We propose a conditional diffusion approach for dynamic microsimulation in which each unit is represented as a multivariate monthly trajectory, conditioned on static covariates and an observation mask. The method borrows the same core mechanism that made diffusion models widely known through text-to-image systems such as Stable Diffusion: a sample is generated by starting from noise and repeatedly denoising until realistic structure emerges—images are simply a particularly dramatic, high-dimensional domain where this iterative reversal is easy to appreciate. Here we apply the same denoising principle to structured longitudinal microdata: a one-dimensional residual convolutional denoiser with timestep and month positional embeddings learns to reverse a gradual Gaussian corruption process on trajectories, so observed months are pinned while missing months (or, via one-sided masking, future months) are generated as multiple plausible, jointly coherent completions consistent with the observed history and covariates.
We demonstrate the method on a large household panel dataset as a case study (SIPP). We evaluate the model under both random missingness and contiguous block gaps, and assess not only point error but dynamic and joint realism, including unemployment spell-length distributions, employment transition matrices, and income distributions conditional on employment status. A key benefit is multiple imputation: repeated sampling yields a distribution over plausible completions and uncertainty bands for downstream statistics, allowing the model to correctly express uncertainty instead of returning a single deterministic trajectory. The width of these bands reflects how identifiable missing months are given observed history and covariates; poor empirical coverage provides a diagnostic for missing predictors or miscalibration. We further demonstrate one-sided masking as a forecasting/nowcasting use case, and scenario-style constrained sampling for stress-testing counterfactual assumptions (e.g., income floors or top-ups), while noting that causal policy inference requires additional identification assumptions.
Overall, conditional diffusion offers a flexible, uncertainty-aware generative layer for microsimulation that can preserve multivariate temporal structure and support robust uncertainty propagation.
INFORM2 is a dynamic microsimulation model developed within the Department for Work and Pensions (DWP) for forecasting claim volumes that underpin the benefit expenditure forecast for Universal Credit. Development began in 2018 and builds on earlier iterations of the INFORM framework. The model simulates independent benefit units and individuals on a monthly time step, using Universal Credit (UC) administrative data as its core input.
INFORM2 outputs provide a detailed, benefit-unit level simulation of the UC caseload and its composition for Great Britain over the medium term. Its design allows results to be broken down by UC eligibility rules including age, health and carer status, family composition, housing costs, and work and earnings patterns. This detailed compositional information is crucial not only for accurate and consistent estimation of caseloads but also estimating the average benefit amounts that complete the expenditure forecast.
The model integrates onflows synthesised from historical data across both UC and the six “Legacy” working age benefits that UC replaces, simulating new claims, transitions from the Legacy system, and the complex flow dynamics at the margin of benefit entitlement and take-up. Internal transition- and off-flow-probabilities are handled using a combination of discrete probability matrices and logistic regressions in addition to deterministic ageing rules, for example when claimants move to the pension-age Benefits.
A range of structural and data constraints shaped the development of the model. Although 2019 offered the first full year where UC was fully rolled out to new claims, the system was still far from steady state: the legacy stock remained substantial, transitions were immature; and these behavioural patterns risked biasing estimated probabilities. COVID19 added further complexity by disrupting labour market dynamics to such an extent that the modelling data could not be robustly updated until 2022-23, when operational and claimant behaviour appeared to be stabilising. Even then, major increasing pace of Legacy-to-UC transition, and economic and policy changes continued to produce new discontinuities that required substantial development work. For example, the earnings modelling was refined separately up to 2024-25 to account for substantial changes in earnings distributions and conditionality rules.
Several research and development strands are in progress. A major one, investigated through a co-sponsored PhD study with the Centre for Microsimulation and Policy Analysis the Centre for Microsimulation and Policy Analysis at The University of Essex is the incorporation of economic forecasts into INFORM2 modelling, via a new onflows model estimated on the link between UC onflow volumes to unemployment rates, levels of benefit and working-age population changes. Also, machine‑learning techniques such as neural networks are being explored as alternatives to the increasingly large DPM structures used in the current off‑flows module.
INFORM2 is a highly significant model in Government, and is receiving increased scrutiny at all levels of Government, placing high value on explainability of its outputs, which is challenging to balance alongside the appetite for accuracy and detail that a microsimulation model affords.
