A UK Synthetic Household Panel
Published in , 2027
This project builds a synthetic panel of UK households that follows income, consumption, and wealth over time. The three variables are linked by an accounting identity: consumption is disposable income minus saving, and wealth changes with saving and valuation effects. A panel that respects this identity must be generated jointly and dynamically; stacking separately imputed cross-sections would not satisfy it.
We propose to use Understanding Society (UKHLS) and BHPS as longitudinal evidence for a synthetic population of people and households in a chosen geography, augmented with information on income, consumption and wealth from other sources. A candidate first stage synthesises a population at one reference date and then generates histories conditional on its attributes and the observed panel; a second candidate synthesises trajectories jointly. How best to connect the cross-sectional and longitudinal steps remains to be decided. The target is one synthetic person per member of the chosen population, not one record per survey respondent. Synthesis alone does not guarantee anonymity: release depends on licence conditions and disclosure assessment.
The panel serves (i) researchers who calibrate and populate large agent-based models, e.g. setting initial conditions, learning agents’ behaviour, and estimating microeconomic parameters, and (ii) social scientists studying consumption, income dynamics, and inequality.
