Technical Report: Toward comparative elderly care modeling

Escalating demand for long-term care (LTC) against the backdrop of rapidly aging populations in Asia and Europe poses significant challenges to the sustainability of social protection systems. However, ageing is only one among several dimensions in which considerable transformations are co-occuring, many of which, e.g. fertility decline, the education expansion or changing family structures, are simultaneously affecting care supply and demand. This technical report is part of an effort to design comprehensive analytical tools that keep track of multiple dimensions while providing comparative and forward-looking assessments of LTC-systems and policy responses in the face of upcoming challenges. In particular, we develop a method to obtain parameters on care demand and supply in hours from the harmonized Survey of Health, Ageing and Retirement in Europe (SHARE) to be used as inputs in the dynamic microsimulation model microWELT (www.microwelt.eu). We proceed along the following steps. First, we adjust SHARE weights to correct for the undercoverage of the nursing home population. Second, we develop a care need assessment algorithm based on an administrative approach from the Austrian universal care allowance system to map characteristics observed in SHARE (ADLs, etc.) to hours. We calibrate this approach against Austrian statistics. Third, we model care needs (any hours) by age, sex and education. Fourth, we model the distribution of hours (those with care needs) by age, sex and education. Fifth, we model institutionalization, the probability of being in a nursing home by age, sex, care needs in hours, having a partner and number of children. Sixth, we model care received, the probability of receiving care given that there is a need by care need (hours), presence of a (caring) partner and having children. Seventh, we model the care mix received, share of total hours of care received from different sources by care need (hours), presence of a (caring) partner and having children. Eighth, we model care giving, the average hours of care provided by age and sex. The report is structured as follows. Section 1.1 introduces the data source and data adjustments. Section 1.2 discusses our care need assessment algorithm. Finally, Section 1.3 presents five parameterization steps.