Transcript C3_PS03_11_pres04_Willingness to pay for Voluntary Health

Willingness to pay for Voluntary Health
Insurance in Tanzania
August J. Kuwawenaruwa
2nd Conference of the African Health Economics and Policy Association (AfHEA)
Saly – Senegal, 15th - 17th March 2011
BACKGROUND

Increasingly moves are being made to expand
health insurance cover in Africa as a means of
reducing out of pocket payments as well as
improving access to formal health care

However, fragmentation of insurance schemes
in many settings, along with limited regulation
of the health insurance sector, has hampered
expansion efforts in many countries.
CHALLENGES

How to expand coverage among the informal
sector which constitute a large proportion of the
population

How to finance and sustain the expansion of
health insurance, and

To
what
extent
mandatory
insurance
contributions from the formal sector can be used
to cross-subsidise contributions from informal
sector groups.
3
CONTEXT OF TANZANIA

Highly fragmented system of health insurance.

National coverage is low around 13% (Humba September,
2010). The aim being to reach 45 % by 2015 (Humba
September, 2010)

The National Health Insurance (NHIF) is financed through
mandatory payroll contributions amounting to 3% of salaries
from the employee which is matched by the employer.

CHF is based on voluntary contribution per annum per
household
TIKA recently introduced in urban areas

4
BENEFIT PACKAGE
Insurance Scheme
Who is eligible
Contribution rate
Benefit Package
National
Health Mandatory for public 6% of gross salary, split Inpatient & outpatient
Insurance Fund (NHIF)
servants and up to 5 between employer and care from public and
dependents.
employee
accredited faith based &
private
facilities
&
pharmacies.
Community Health Fund Rural – voluntary, for a Between Tsh
(CHF)
couple and children 15,000
under 18 years.
year/household
5,000- Primary level public
per facilities.
Limited
referral care in some
districts
5
HEALTH INSURANCE SUBSIDIZATION

There is a huge discrepancy between the benefit
package offered to NHIF members compared with CHF
members, as well as the amount of revenue
generated by each scheme.

While there is cross-subsidization across NHIF
members, there is no cross-subsidization across the
schemes. Nor is there cross-subsidisation across
districts/councils for the CHF.
6
OBJECTIVE OF THE STUDY

To elicit household’s willingness to join (WTJ) and
willingness to pay (WTP) for voluntary health insurance

To assess how WTJ/WTP varies according to benefit
package offered, and

To examine households willingness to cross-subsidise
poorer groups in the community
7
METHODS




1,163 uninsured and 1,061 insured household heads were
interviewed in 2008 from 3 urban councils and 4 rural
districts.
Rural districts were selected such that they had a minimal
level of CHF coverage (at least 10%), and to offer some
geographical variation.
Uninsured household heads were asked about their WTJ
and WTP for health insurance
Scenarios: First reflecting the current design of the CHF
with premium Tsh 5,000, the second offering expanded
benefits
8
ELICITING WILLINGNESS TO CROSS-SUBSIDIZE
 Willingness to cross-subsidize questions were addressed to
insured household heads.

They were asked:
“Would you be willing to contribute to any health insurance
scheme or to the council any amount of money so that the very
poorest in your community can benefit from free care when
they are sick?”.

Those who said “Yes”, were asked to state how much they
would be willing to pay per annum to protect the poor
9
DATA ANALYSIS

Bivariate analysis was done to assess the level of
association between WTJ, WTP for insurance and
willingness to cross-subsidise and a range of
individual variables.

Pearson chi-square and t-test test statistics were used
for binary explanatory variables and mann-whitney
test for continuous explanatory variables to test the
significance of the results.
10
REGRESSION ANALYSIS

A logit model was constructed to assess the determinants
of WTJ community health insurance and willingness to
cross-subsidize the poor.

An OLS log linear model was constructed to assess the
determinants of the amounts people were WTP as well as
willing to cross-subsidise.

Examination of theoretical validity was done

It was hypothesized that willingness to pay would be
affected by socio-demographic variables
11
SELECTED CHARACTERISTICS OF THE HOUSEHOLDS
Variables
Insured
N=1,162 %
73.1
69.7
4.8
Uninsured
N=1,060 %
76.5
4.7
8.0
Health Average (SAH)
Health Good (SAH)
25.6
69.15
26.5
64.82
0.393
0.956
Education=Completed
primary and above
Region =urban
93.7
79.2
0.001
19.2
25.5
0.001
Gender = Male
Occupation=Formal
Health Poor (SAH)
p-value
0.275
0.001
0.130
Exemption eligibility
7.9
18.7
0.001
Outpatient visit to formal
providers in previous month
23.5
13.2
0.001
Age
41 [12.2]
44.38 (14.4)
0.325
Household size
5.22 (2.9)
5.18 [2.6]
0.001
12
WILLINGNESS TO PAY JOIN

Respondents were more willing to join health
insurance in urban than rural areas at the proposed
rate of Tsh 5,000 (93% compared to 74%)

Proposed benefit package had a significant effect on
people’s willingness to join in rural areas,

78% were willing to join in Mbulu and Singida districts
where inpatient care was covered compared to 72% in
Kigoma and Kilosa districts where only primary care
was covered
13
ASSESSMENT OF THE BENEFIT PACKAGE

Expansion of the benefit package in Kigoma and
Kilosa, will increase the probability of joining from
72% to 79%.

In urban areas, the benefit package had little effect on
people’s WTJ.

Further, there was very limited willingness to pay more
than Tsh 5,000, even with an expanded benefit
package
14
DETERMINANTS OF WTJ/WTP
Variable
Logit model
Coefficient (SE)
Logit model
Marginal effect (SE)
Coefficient (SE)
Gender
0.844 (0.27)**
0.139 (0.05)**
0.152 (0.09)*
Occupation
-0.553 (0.53)
-0.092 (0.10)
-0.086 (0.08)
Education
0.438 (0.28)*
0.068 (0.05)*
0.098 (0.11)
Exemption Eligibility
-0.400 (0.48)
-0.062 (0.08)
-0.070 (0.12)
Outpatient visit to formal
providers in previous
month
Age
0.270 (0.32)
0.036 (0.04)
-0.058 (0.08)
-0.004 (0.01)
-0.001 (0.00)
0.006 (0.00)*
Income
0.083 (0.05) *
0.012 (0.01) *
0.054 (0.01)***
Household Size
-0.046 (0.04)
-0.007 (0.01)
-0.010 (0.01)
Above 59 years
0.124 (0.23)
0.017 (0.03)
-0.042 (0.07)
15
WILLINGNESS TO CROSS-SUBSIDIZE

There was a greater willingness to cross-subsidize the poor
among rural compared to urban households (46.0% vs
41.2%)

The actual average amounts stated were lower in rural
compared to urban areas (mean Tsh 6,620 vs Tsh 13,940).

39% of NHIF members were willing to cross-subsidize
compared to 53% CHF member,

However, NHIF members stated higher average amounts
than CHF members (mean Tsh 13,690 vs Tsh 4,790).
16
DETERMINANTS OF WILLINGNESS TO CROSS-SUBSIDISE

Results from logit showed that household being male headed it
increased the probability of willingness to cross-subsidise by
10.5 percentage points.

Having outpatient visit to formal providers increases the
probability of willing to cross-subsidise by 7.4 percentage
points.

Richer households, and those working in the formal sector were
willing to pay more for the poor and those who had recently
sought care were less willing to pay for the poor.
17
POLICY RECOMMENDATIONS

People are willing to join health insurance if they are made
aware of the principles of insurance and properly understand
the concept of risk pooling. However, willingness to pay
remains limited.

The greater willingness to join insurance in urban compared to
rural areas suggests that cross subsidisation should also be
promoted between urban and rural districts for the CHF.

At present, funds are pooled at the district level, but there is no
pooling of funds across districts
18
POLICY RECOMMENDATIONS CONT.

NHIF members are willing to cross-subsidise the poor, and
would potentially be willing to cross-subsidise CHF

The amounts that could be generated by NHIF members in
cross-subsidises would be Tsh 3, 765, 874,000 per annum
(=mean amount x 316,460 [NHIF principal members in
2008]).

This means that additional 753,175 CHF members will be
enrolled per annum at a premium of Tsh 5,000.

Its equivalent to 11% of the households in Tanzania using
2000 census data. This could have a dramatic effect on
national insurance coverage
19
CONCLUSION

Households are willing to pay for voluntary health insurance and
are willing to cross-subsidise poorer groups within society

In setting the premium policy makers need to consider the
variation in the household's socio-economic characteristics.

To achieve the targeted 45% insurance coverage in Tanzania
fragmentation of health insurance schemes should be
addressed and the size of the risk poor must be maximized.

Maintenance of membership goes parallel with improvement in
health care and availability of drugs within the accredited
facilities.
20
Health Insurance for the benefit of all (Financial protection)
21