Informal, formal, or both? Assessing the drivers of home care utilization in Austria using a simultaneous decision framework

ABSTRACT

Understanding the relationship between different modes of home care for the elderly and the determinants of mode choice is fundamental for an efficient care policy in ageing societies. However, empirical research on this issue has revealed that policy conclusions will depend on both national and methodological factors. Using data for Austria from the Survey of Health, Ageing and Retirement in Europe, the purpose of the present paper is twofold: First, at least to our knowledge, it is the first comprehensive assessment of this kind for Austria. Second, it adds to the literature explicitly focusing on the combined use of informal and formal care in addition to the exclusive use of these services based on an econometric framework accounting for the simultaneity and interdependence between these modes. Our results provide strong evidence for a task-specific and complementary relation of formal and informal home care in Austria, with the health status and functional limitations as the main determinants of home care choice.

KEYWORDS:

• Home care
• formal care
• informal care
• bivariate probit
• SHARE

JEL CLASSIFICATION:

• C35
• H42
• I11
• I12
• J14

Acknowledgments

The authors thankfully acknowledge helpful comments by Christoph Badelt, Ludovico Carrino, Ulrike Famira-Mühlberger, David Roodman, Birgit Trukeschitz, Hannes Winner, and an anonymous reviewer.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Multivariate analyses on Austria are limited to the role of the socioeconomic status on long-term care utilization in the capital Vienna (Schmidt 2017). This metropolitan area, however, is characterized by sociodemographic and socioeconomic conditions that are very different from the rest of the country.

² The Pflegeregress system was abolished in 2018 and thus after the data used in this paper were surveyed (2013–2017).

³ See Stabile, Laporte, and Coyte (2006) and Byrne et al. (2009) for theoretic models of household decision-making with respect to informal care-giving and the use of formal home care services.

⁴ Balia and Brau (2014), Bruni and Ugolini (2016) and Carrino, Orso, and Pasini (2018) also estimated simultaneous models but did not model the determinants for cases of FIHC.

⁵ Due to limited space we restrict the discussion to papers focused on home care.

⁶ Additionally, McMaughan Moudouni et al. (2012) conclude for the U.S. that IHC and FHC are neither substitutes nor complements.

⁷ Previous waves of the SHARE do not include the question related to formal home care (HC127) used in our empirical analysis. Waves 1–2 contain a related question (HC032) that, however, has a different wording and is more restrictive in its scope.

⁸ See Appendix B.2 for a discussion of the consequences of this age restriction.

⁹ For details on the questions in the questionnaire defining FHC and IHC see online Appendix A.

¹⁰ NUTS (Nomenclature des unités territoriales statistiques) is a geocode standard for referencing the subdivisions of European Union countries. The NUTS-3 level corresponds to the smallest spatial unit above the local administrative units and is the most disaggregated regional level for which residence information is available in the SHARE. For 136 respondents information about the NUTS-3 region of residence is not available. To keep these observations in the sample they are subsumed in the NUTS-3 category ‘Missing’.

¹¹ While the majority of our observations were surveyed in SHARE wave 5 (2,358), 195 of the interviews in our sample were conducted in wave 6 and 8 in wave 7. Because the low number of observations we use from wave 7 may lead to multicollinearity problems (there is, for example, no respondent receiving both $IHC$ and $FHC$ in the wave 7 data we use) non-wave-5 respondents are subsumed in a single category.

¹² Alternatively, the choice situation could also be modelled as a multinomial logit where the dependent variable has four levels: (i) no home care, (ii) FHC only, (iii) IHC only and (iv) both FHC and IHC. However, the multinomial logit presumes independence of irrelevant alternatives (IIA), which is questionable given that the choice represented by level (iv) is a combination of the choices represented by levels (ii) and (iii) and it is thus highly likely that, for example, choosing between FHC only and a combination of FHC and IHC is not independent of whether IHC only is available or not. Indeed, when empirically testing for violations of IIA we find that the null hypothesis of IIA can be rejected at the 5% significance level if IHC is excluded from the choice set. Results are available from the authors upon request.

¹³ In an alternative specification, we clustered standard errors at the household rather than at the regional level. The significance levels of almost all variables are unaffected by this modification. Results are available from the authors upon request.

¹⁴ An exception is Table 3 reporting marginal effects rather than the regression coefficients. As wild cluster bootstrap $p$-values cannot be calculated for marginal effects, we stick to $p$-values from the cluster-robust variance estimation in Table 3. However, given the negligible differences we found between the two options for calculating $p$-values for the regression coefficients of the bivariate model of Table 2 in additional robustness checks (available upon request), we are confident in the conclusions we draw from our model are also valid for the marginal effects illustrated.

¹⁵ To preserve space, the estimated coefficients of the NUTS-3 dummies are not shown in the regression results. Detailed results are available from the authors upon request.

¹⁶ We also estimated a model that included the distance to the nearest child in the model. The proximity variable was, however, not statistically significant for either HC mode. The results are available from the authors upon request.

¹⁷ Although we do not find a significant effect on the extensive margin, we cannot exclude that there may be effects on the intensive margins of formal and informal care provision. For example, Balia and Brau (2014) – -although the effects on the extensive margins are insignificant–-the number of hours of IHC provided does have a positive effect (significant at the 10% level) on the number of hours of FHC provided, and that the number of hours of FHC has a significantly negative effect on the number of hours of IHC provided.

Additional information

Funding

Research was conducted within the research programme “Austria 2025“, which has received funding from the Austrian Federal Ministry for Transport, Innovation and Technology, the Austrian Federal Ministry of Science, Research and Economy, the Oesterreichische Nationalbank, the Climate and Energy Fund, the Austrian Federal Ministry of Labour, Social Affairs and Consumer Protection, the Hannes Androsch Foundation at the Austrian Academy of Sciences. The Austrian Chamber of Labour, the Austrian Federal Ministry of Agriculture Forestry, Environment and Water Management, the Austrian Chamber of Agriculture and the Austrian Economic Chambers funded each a project to be included into the research program.