Home Care

European Conference on
Long-Term Care
Department of Economics
University of Bologna
Italy
The role of paid home care in
elderly living arrangements
decisions
Matteo Lippi Bruni, Dept. of Economics, University of Bologna
Cristina Ugolini, Dept. of Economics, University. of Bologna
Zew, Mannheim, 22 October 2005
1
With the expected growth in the demand for LTC, there is a mounting
interest in understanding the determinants of family decisions
regarding elderly living arrangements (ELA).
Italy has the oldest population in the world:
• fertility rate decline
• increased life expectancy.
Most LTC is still ensured by unpaid informal caregivers, but
households are less and less likely to be in the condition to provide
care directly.
•reduction in households’ size
•decline in family ties
•increased women labour-force participation.
2
BACKGROUND
The theoretical literature varies along several dimensions:
common preferences [Kotlikoff and Morris, NBER1988; Hoerger et al., Rev Econ
Stat 1996] or family bargaining [Stern, J Hum Resour 1995; Pezzin and Schone
Am Econ Rev 1997; Engers and Stern, Int Econ Rev 2002];
the type of care (formal or informal) or living arrangements considered
(shared housing, live independently, nursing home), role of children in the
decision process, etc.
The empirical literature is as varied as the theoretical one.
It has been conducted with a variety of econometric methods, but displays
remarkably consistent results.
The great majority of existing works relates to the United States.
3
This paper builds on the existing body of research in two ways.
Very little is known of the Italian situation:
•Results of the international literature on ELA not entirely
generalisable.
•The very few studies about Italy: Wolf and Pinelli, 1980;
Tomassini and Wolf, 1999, suffer of poor information on
disability conditions.
•We examine ELA choices by Italian households using for the
first time detailed information on the health conditions of the
elderly person.
Main focus : the role of formal caregiving when elderly
people are assisted at home.
Formal vs. Informal care typically grounds on the assumption that
formal care provided by paid helpers occurs exclusively when the
elderly person resides in a living assisted facility.
i.e. in the traditional approach, informal and home care tend to coincide.
Claim: when a dependent person is kept at home, the choice
between informal and paid care is an increasingly relevant
issue.
Since the late 1980s Italy experienced significant migration flows from
ex-socialist and African countries, mostly undocumented and illegal,
with a high female component that joined the informal labour market.
Two third of these women are engaged in domestic or personal care.
Social norms about filial responsibility still tend to attach a consistent
amount of social stigma to the institutionalisation of the elderly.
For many Italian families, the possibility to delegate caregiving by buying
services (although in most of cases exploiting the black market)
ensured a substantial reduction in out of pocket expenditures with
respect to professional services
contributed significantly to reduce nursing home admission rates.
5
THE SURVEY
Interviews drawn from a cross-sectional survey carried out on a sample
of 1405 families of the Italian region Emilia Romagna (around 4 millions
inhabitants) in year 2002.
The main purpose was to elicit WTP for covering LTC expenditure risk.
A specific section of the questionnaire is devoted to register the
existence of a disabled person aged 50 or more inside the family unit.
We record a total of 279 households with an elderly dependent,
231 individuals live at home,
48 are institutionalised.
179 receive informal home care,
52 receive paid home care
Given the available data set, we assume common preferences among
family members and consider that the living arrangement decision is
taken once-and-for-all.
6
THE DECISION TREE
The decision process is
articulated in two steps.
Household i
Home care
n=231
Formal Home Care
n=52
Residential care
n=48
Informal Care
n=179
The household decides
whether to institutionalise the
elderly in a living assisted
facility (Residential Care, RC)
or to provide care at home
(Home Care, HC).
For those who stay at home,
the household decides
whether to provide care
directly (informal home care,
IHC) or to hire a external
person as primary caregiver
(paid home care, PHC).
THE DATA
DE characteristics
Age,
LTC spell
Num ADL
Heavy help
Sex,
Lived alone
Spell of disability in years
Number of ADLs and IADLs in which DE is not self-sufficient
Public support for > 40 days (identification variable)
Family characteristics
House tenure
family owner of the house
Family income
Family income in Euro (HH+ HH spouse, if present)
Family size
Residence
Household residence influenced by the will to live close to
other relatives (family ties)
Universal access “everything to everybody”
Need-based access “fundamental services to everybody”
(experience based) Family opinions
Opinion_LTC bad quality of existing LTC services
Cash Care 1 cash transfer even if the principal caregiver is a family member
Cash Care 2 cash transfer only if there is an for caregiver external to the family
THE ECONOMETRIC SPECIFICATION
Bivariate probit model with sample selection (e.g. van de Ven , van Pragg,
1981 Journal of Econometrics)
U ijt   jt X it   ijt
i= 1….N t=1, 2
j = RC vs HC; PHC vs.IHC
yi*1  (  HC1   RC1 ) X i1  ( iHC1   iRC1 )  1 X i1  vi1
yi*2  ( P HC2   IHC 2 ) X i 2  ( iPHC 2   iIHC 2 )   2 X i 2  vi 2
1st Stage
y1=1 if Home Care ;
y1=0 if Residential Care
2nd Stage
y2=1 if Paid H C;
y2=0 if Informal H C
missing information if y1=0
 log   x , x ,    log   x , x
y1i 1, y2 i 1
2
1 1i
2 2i
y1i 1, y2 i 0
2
1 1i
2 2i
,     log  1 x1i 
y1i 0
Probit model with sample selection /1
Coef. Std. Err.
Coef. Std. Err.
CARE HOME
AGE DE
Sex DE
LTC spell
Lived alone
Heavy help
Num ADL
Family size
Residence
choice
Universal
access
Need-based
access
House ownership
Family income
urb5_25
urb_inf5
Opinion_LTC
Cash care 1
Cash care 2
Constant
-.0285
.0468
-.0248
-.5009
-1.1483
-.1704
.0679
.7215
.0089***
.2292
.0084***
.2444**
.2667***
.0456***
.0872
.2752***
-.0269
.0214
-.0227
-.5905
-1.0250
-.1803
.0761
.7826
.0093***
.2286
.0081***
.2523**
.2790***
.0455***
.0888
.2638***
-.2986 .3063
-.3313 .3075
-.0755 .3188
-.0886 .3287
.3159
-.0002
-.3296
-.4190
.3299
-.0002
-.2743
-.4518
.2887
.0343
-.0434
4.2525
.2722
.0001**
.2184
.4867
4.447 .9226***
.2716
.0001***
.2271
.4585
.2430
.2303
.3161
.9236***
Probit model with sample selection /2
PAID HOME CARE
AGE DE
Sex DE
LTC spell
Lived alone
Num ADL
Family size
Residence choice
Universal access
Need-based ac
House ownership
Family income
urb5_25
urb_inf5
Opinion_LTC
Cash care 1
Cash care 2
Constant
rho
Log pseudo
likelihood
Sample size
279
.0224
.3595
.0065
.5140
-.0137
-.1132
-.2063
-.4643
-.5614
.1448
.0004
-.3282
-.8385
.0089**
.2337
.0099
.2751*
.0552
.0934
.2608
.2663*
.2883**
.2817
.0001***
.2257
.6181
-3.1204 .8593***
.3724 .5740
-196.5156
.0254
.2798
.0024
.4959
.0042
-.1727
-.3526
-.6332
-.7513
-.0200
.0004
-.5288
-.9830
-.3207
-.3556
.9306
-2.6541
.3229884
-186.5766
.0097***
.2458
.0094
.2816*
.0555
.1028*
.2816
.2956**
.3144**
.2870
.0001***
.2470**
.4922**
.2336
.2536
.3296***
.9400***
.5085
EMPIRICAL RESULTS
 Severity related variables increase the probability of
institutionalisation but have no effect on the decision of hiring an external
caregiver
 Age partial exception displays similar effects in the two stages
 Income increase both the probability of choosing residential care and
formal home care (much larger marginal effect)
 Household size apparently no effect (checked for spouse/housewife
etc..)
 Very frequent public support captures extremely severe cases
(highly institutionalised)
 Strong family ties reduce the probability of residential care
 Econometric Issue
 H0: 0 not rejected
Separate estimations generate unbiased coefficients
probit RC vs HC on the whole sample
probit PHC vs IHC on the subsample where y1=1
Probit model /1
Coef.
CARE HOME
AGE DE
Sex DE
LTC spell
Lived alone
Heavy help
Num ADL
Family size
Residence choice
Universal access
Need-based access
House ownership
Family income
urb5_25
urb_inf5
Constant
Wald chi2(14)
Prob >chi2
Pseudo R2
Log pseudo
likelihood
Sample size
-.0298
.0327
-.0246
-.5063
-1.1353
-.1699
.0690
.6787
-.2726
-.0628
.3152
-.0002
-.3343
-.4334
4.5217
56.80
0.0000
0.2381
-97.5970
279
DF/dx Std. Err.
-.00543
.00601
-.00449
-.10837
-.31912
-.03098
.01259
.09725
-.05024
-.01161
.06463
-.00004
-.06406
-.10440
.0087***
.2323
.0084***
.2404**
.2717***
.0449***
.0872
.2717***
.2933
.3143
.2722
.0001*
.2168
.4659
.9306***
Probit model /2
Coef.
PAID HOME CARE
AGE DE
Sex_DE
LTC spell
Lived alone
Num ADL
Family size
Residence choice
Universal access
Need-based access
House ownership
Family income
urb5_25
urb_inf5
constant
Wald chi2(14)
Prob > chi2
Pseudo R2
Log pseudo
likelihood
Sample size
.0252
.3607
.0093
.5767
-.0077
-.1255
-.2697
-.4648
-.5794
.0797
.0005
-.3123
-.8116
-3.3054
54.94
0.0000
0.1858
-99.0769
231
DF/dx Std. Err.
.00632
.08455
.00233
.16646
.00192
-.03138
-.06282
-.11422
-.13186
.01945
.00011
-.07507
-.13628
.0082**
.2344
.0091
.2562**
.0458
.0934
.2591
.2679*
.2864**
.2793
.0001***
.2274
.6275
.8050***
CONCLUSIONS
Assisting elderly dependent people by means of formal care is an
increasingly followed strategy also when families opt for a home care
solution.
 The determinants of formal care differ substantially if one considers
PHC vs IHC instead of the more traditional choice between RC vs HC

specific investigation of the issue is needed, in particular today that public
policies strongly encourage home care and that the opportunity cost of informal
care rise

Residential care more and more likely when health deteriorates ,
similar trend does not hold for formal home care.
 Economic barriers to the access of formal care seem much more
relevant in case of home care

Open
Issues
Endogeneity problems (income, opinion variables)
 Further investigate the role of family composition
 Improve the quality of the data with ad hoc surveys
