WP4 – Empirical Analyses
- Use SHARE data to collect and describe basic facts that give background and target for analyses.
- Fit both versions of the micro model to the SHARE data.
- Turn equations of micro model into projection model
Task 4.1: Stylized facts (Dauphine M1-M12): This task extracts from the SHARE data the country-specific trajectories
of all outcome and explanatory variables relevant for this project, serving as target against which the model outcomes
will be validated. Outcome trajectories include labour supply, income, and wealth; participation in social programs;
exchange of time and money between parents and children; location, living arrangements and mobility. The SHARE life
histories and the SHARE job episodes panel provide decade-long trajectories to fit the dynamics of the model. Extract
from SHARE data on health and cognitive impairment.
Task 4.2: Institutional features (Dauphine M1-M12): SHARE is augmented by SPLASH, the Social Policy Archive
for SHARE, which contains institutional and environmental variables. This task will add more detailed data on the
institutional features of LTC provision and LTC insurance in the SHARE countries, using the large international network
of SHARE country teams.
Task 4.3: Model estimation (MPG M13-M42): Fit both variants of the micro model to these trajectories. Where an
explicit solution of the first order conditions is possible, use OLS or instrumental variables, otherwise fit Taylor
approximations. For deep utility parameters, use calibration. Use three types of variation to identify: (a) policy variation
across SHARE countries; (b) macro variation across time and countries (e.g. dependency ratio, divorce rate, health; and
(c) predetermined individual variables (e.g. childhood conditions, education).
Task 4.4: Baseline projection (MEA M1-M42): Based on our preferred assumptions about demographic changes and
the evolution of family cohesion (e.g., prevalence of divorces, separations and patchwork families), we will derive the
potential supply of informal care. Based on assumptions about regional differences in productivity growth, we will
derive the distance between parents and children, labour supply of the children, and thus the actual supply of informal
care. Based on assumptions about the future development of health (e.g., increase in the prevalence of dementia or
another epidemic), we will derive the demand for LTC. Based on scenarios of LTC insurance coverage and generosity
and assumptions about future income distribution we will derive who can afford and who will choose formal care. This
will allow us to predict all main outcome variables (labor supply, saving, caregiving, unmet care needs).
Task 4.5: LTC policy design (Dauphine M1-M45): We then feed alternative policy scenarios from WP2 into the model
and proceed like in Task 4.4 to derive the probability distribution of the outcomes of interest under these alternative
policies. By varying policies, we will explore which policy mix maximizes welfare in our model.
Role of participants: Dauphine will lead WP4 and execute Tasks 1 and 2 jointly with MPG. Tasks 3 will be done in close
co-operation with MPG and MEA. Tasks 4 and 5 will be joint work by all participants.