evaluating the impact of health card program on access to reproductive health services: an...

7/27/2019 Evaluating the Impact of Health Card Program on Access to Reproductive Health Services: An Indonesian Experience

1/26

3rd

International Conference on Reproductive Health and Social Sciences Research | 201

Evaluating the Impact of Health Card Program on Access to

Reproductive Health Services: An Indonesian Experience

Erlangga Agustino LandiyantoInstitute for Population and Social Research, Mahidol University, Thailand

Abstract

Health card program aims to protect the poor in Indonesia during the Asian

economic crisis. Health cards were targeted and allocated exclusively to the poor that

would provide free access to public health services. The impact of health card program

to reproductive health services was rarely discussed by previous studies that pay more

attention on health card utilization for both inpatient and outpatient. Using Indonesian

family life survey (IFLS) data 1997-2000 from RAND Corporation, this study aims to

evaluate the impact of health card program during Asian economic crisis on access to

reproductive health services and answer the question whether who had health cardreally have better access to reproductive health services. Discussion in this paper limit

on antenatal care, place of delivery and contraceptive use which are only reproductive

health components that covered by health card program. Using combination between

descriptive analysis and multivariate analysis, this study found that the health cards

were not well targeted and distributed. The study also found that, generally, there is no

significant effect of health card ownership to access to reproductive health services.

Keyword: Health Card, Antenatal Care, Contraceptive, Place of Delivery


2/26

202 | 3rd

International Conference on Reproductive Health and Social Sciences Research

Introduction

Background of Study

The Indonesian economic crisis began in 1997 when the rupiah depreciatedrapidly and brought Indonesia into the economic crisis of 1998. The devaluation ofthe rupiah increased levels of debt of private companies as well as increasedoperational cost resulting in bankruptcies. These conditions stimulated a reduction inlabor demand, rising unemployment, and, indirectly, a loss of social securitycoverage. Prices of goods and services increased greatly during the year, whichdecreased quality of life the lower income population as well as pushed lower middleincome population to be below population line (Strauss et al, 2002; Frankenberg et al,2002; Pritchett and Suryahadi, 2002; Sparrow, 2006).

The crisis negatively affected the health sector from both supply and demandsides. For the supply side, WHO (1998) and AUSAID (2002) reported that theIndonesian provinces and district health offices experienced a reduction in operating

budgets, which resulted in a cut in the budget for preventive programs. Accordingwater et al (2003) and AUSAID (2002), health services providers faced the increasing

in operation costs, extraordinary increasing in pharmaceutical and medical suppliescosts, and reduced supplies of modern health services. On the demand side, theseverity of the crisis affected households health care utilization and expenditures.Frankenberg et al. (2002) and AUSAID (2002) found that household consumptionwas going down in 1998, with decreasing investments in human capital (health andeducation) as well as decreasing utilization of modern health care services.

In response to the crisis, there were a series of publication about social safetynet programs that were initiated or reconfigured in Indonesia. Some of programs weredesigned to reach all population and some were targeted to reach the poor only(Strauss, 2002). The health component of the Indonesian Social Safety Net program,the health card program, was started in September 1998 and initiated to protect the

poor from the effects of the economic crisis through a targeted price subsidy and a

public spending component. The health cards entitled all household members to theprice subsidy at public health care providers. (Saadah et al, 2001; AUSAID, 2002;Sparrow, 2006; Somanathan 2008)

Most of previous studies on the impact health cards for protecting the poorduring Indonesian economic crisis focus on targeting of health cards distribution(Lanjauw et al 2001; Pritchett and Suryahadi, 2002; Sparrow, 2006; and Sparrow,2008) and utilization of health cards and its impact to outpatient (Saadah et al, 2001;Sparrow, 2006; Saadah et al, 2007). Other studies focus on the impact of health cardson childrens health care (Somanathan, 2008) and health care consumption (Johar,2007).

The issues were rarely discussed in previous Indonesia cases studies.Therefore, this thesis focus on the impact of the health card program on access to

reproductive health services like contraception, pre-natal care and assistance at birth.Base on discussion above, we can hypothesize the health cards were distributedaccurately to targeted beneficiaries and used it as purposes. We also expect that whoreceived health cards should have better access to reproductive health services.

Research question on this paper focus on to answer the question did the poorwho had health card really have better access to reproductive health services? If sowhy? If not, Why not? From the research questions, the research objective as follow:


3/26

3rd


y General Research objective: Evaluating the impact of the health card program

on access to reproductive health services like access to contraception, pre-natal

care and assistance at birth.

y Specific Research Objectives:

o Measure the performance of health cards targeting and distribution.

o Exploring the utilization of health card for reproductive healthservices.

o Evaluating whether the poor who have health card have better access to

reproductive health services or not.

Research Hypothesis

o Health cards were received by the poor only.

o Health cards utilized by the poor as intended.

o The poor who have health card have better access to reproductive

health service

Literature Review

Social Safety Net-Health Card Program In Indonesia

Program Design

The health component of the Indonesian Social Safety Net program (JPS-BK),health cards program, was designed to prevent the decline of health and nutritionalstatus as a result of the economic crisis. The community health centers (Puskesmas)and the village midwives are the key actors of the program. The health card programwas designed to allow poor households to obtain at least basic health care services. Asdemand side intervention, the health card provides access to health services to the

program beneficiaries by the use of a health card (Strauss et al, 2002; Sparrow, 2006).According Saadah et al (2001) Strauss et al (2002) Sparrow (2006), the typesof services covered by the health card include:

a) Basic health services, medical attention as first treatment or referrals, family

planning services, immunization and other basic health services.

b) Basic maternal health care and referrals for pregnant mother, delivery care,

post and neo-natal natal care.

c) Nutritional improvement through food supplementation to undernourished

poor families.

d) Eradication of communicable diseases such as malaria, tuberculosis and

diseases that could be prevented through immunization.

e) Revitalization of Posyandu (integrated health post), a health post improvementprogram to prevent negative effects of the crisis on the nutrition and health

status of mothers and young children.

Distribution and Implementation

The health card program followed a partly decentralized targeting process,involving both geographic targeting at district level and community based individual


4/26

204 | 3rd


targeting at village level. Households that were categorized as vulnerable to economicshocks were targeted to receive health cards. (Saadah et al, 2001; Sparrow, 2006)

The amount of subsidy for public health care providers to be distributed acrossdistricts and number of health cards to be issued base on National Family PlanningCoordinator Agency (BKKBN) headcounts per-district. The headcount wascalculated based on the survey data to investigated number of poor (Saadah et al,

2001; Sparrow, 2006)At the district level committees were formed to deal with the allocation of

funds to the health clinics, community health center (Puskesmas) and villagemidwives. The district committees were also responsible to allocate health cards andBKKBNs poverty measurement criteria guidelines to villages where the villageleaders headed village allocation committees (Saadah 2001; Sparrow, 2006).

The poverty measurement criterias of BKKBN to identify targeted householdsis called prosperity measurement status for identifies the poor based on who meetwith one of the following criteria (Strauss, 2002; Sparrow, 2006), such as: unable tohave 2 meals a day, unable to afford health services, the head of the household lost his

job due to retrenchment or households with school age children drops out due to thecrises

The village committees (consisting of village staff, family planning workers,village midwives, and community activists) distributed to the villagers base on theBKKBNs poverty criteria above. The identified poor households are given healthcards signed by the head of the community health post (Posyandu) and the head of thevillage. This card is valid for one year and can be extended as long as the householdsmeet those criteria (Strauss, 2002; Sparrow, 2006)

M&E Framework and Research Conceptual Framework

Because this thesis is a evaluation research, the monitoring and evaluationframework of the health card program (figure 2.2) is designed in order to developresearch conceptual framework for this thesis. There are four components ofmonitoring and evaluation frameworks. First is programs input and process that

provide summary of program activities (more detail information about health cardprogram can be found at chapter 2.2), second is programs output, third is programsoutcomes and fourth is programs goal.


5/26

3rd


Figure 2.2: Monitoring and Evaluation Framework

On the figure 2.2 can be seen that this thesis focus on evaluating whether thehealth card program achieve the goal for improving access to reproductive healthservices or not.

This thesis is also investigating the program output (health card targeting anddistribution) and the program outcomes (health card utilization) as well as the effect

of the program output and program impact for achieving the program goal controllingby some independent variables.

Programs Input

and Process

Programs

Output

Programs

Outcome

Programs Goal

1. Established

districts level

committee

and allocated

fund at

districts level.

2. District level

committee

allocated

fund to health

care provider

and distribute

health card to

village

committee.3. The

Health Cardarereceived/own

Health Cardare utilized bythe poor

The poor whohave healthcard have

better access tore roductive

BKKBNPovertyCriteria

Influencing Factors:x Economic Factors

x Socio-demographic

Factors

Unobservab


6/26

206 | 3rd


Figure 2.3: Research Conceptual Framework

Figure 2.3, research conceptual framework, shows the hypotheticalrelationship between dependent variables (program goal), program variables (programoutcomes and program output) and independent variables.

Data and Methodology

Data

The major data source for this thesis is Indonesian Family Life Survey (IFLS)from RAND Corporation in Santa Monica, USA. The data is longitudinal survey dataat household and community level. In this thesis, data at the household level is used

because this thesis focuses on analysis from side. To investigate the impact of thehealth card, IFLS second wave (1997) is used as baseline data before intervention and

IFLS third wave (2000) is used as post-intervention year. The sample size of IFLSsecond wave is 7619 Households and IFLS third wave is 10435 Households. Formultivariate analysis, panel data are constructed from the longitudinal data.

For analysis, there are four versions of datasets. First is data set for descriptiveanalysis. The data is raw data set for univariate and bivariate analysis. Second is paneldata set for indentifying the effect of independent variables and control variables toantenatal care. Third is panel data set for indentifying the effect of independentvariables and control variables to place delivery, public health facility or the others.

Independent Variables

x Health Card Ownerships

x Year Dummy

x Interaction Variables between

health card ownership and year

dummy

Dependent Variables

x Antenatal Care

x Place for delivery

x Using modern contraceptives

Control VariablesA. Socio-demographic Variables

x Number of Household Members

x Highest Grade Completed by head of

households

x Highest Grade Completed by spouse of

head of households

B. Health facilities knowledge

C. Economic Variables

x Water sources

x Sanitation

x Economic Variables

x House ownership status

x Have Television


7/26

3rd


The fourth is panel data set for indentifying the effect of independent variables andcontrol variables to utilization modern contraceptives.

Methods

Two methods of analysis we employed in this study. Fist, descriptive statisticsdescribed the individual and households characteristics, the descriptive statistics

cover univariate and bivariate analysis. Second, Inference statistics cover bivariateand multivariate analaysis.

The multivariate analysis focuses on examining the effect of intervention andindependent variables on dependent variables. The major approach for multivariateanalysis in this thesis is identifying difference-in-difference estimator. To construct

panel dataset for identifying difference and difference estimators, the new entries istaken out from 2000 data and take the drop out cases out from 1997 data and merging

both dataset to create panel dataset for the multivariate analysis with consideration ofsample selection bias. The bias was tested using logistic regression to identify whetherthe dropping case will change the characteristics of observation or not.

Result and Discussion

This chapter presents the interpretation and discussion from data analysis. Partthree is divided as two parts: 4.1 provide descriptive analysis of health card ownershipand utilization and 4.2 provide multivariate analysis of the impact of health cardownership to antenatal care, place delivery and modern contraceptive.

Descriptive analysis of health card ownership and utilization

As discussed on chapter two, the health card program followed a partlydecentralized targeting process. Households that were categorized as vulnerable toeconomic shocks were targeted to receive health cards. It means health card allocated

for the poor household to protect them from the effect of crisis (Saadah et al, 2001;Sparrow, 2006).For evaluating the accuracy of targeting and allocation, wealth quintiles is use

to identify the poor and non poor. The poorest and second poorest quintiles arecategorized as poor, the rest quintiles are non poor or wealthier.

Wealth quintile in this study was constructed from economic variables such asusing electricity, have television, own house, access to improve water, access toimprove sanitation, asset and expenditure. Those variables were combined using

principal component analysis.


8/26

208 | 3rd


Table 4.1: Tabulation between wealth quintiles and health card ownerships

Wealth Index Quintileshhs have kartu sehat

0. No 1. Yes

Poorest N 1070 323

% 18.82 25.25Second N 1076 317% 18.93 24.78

Middle N 1135 257% 19.96 20.09

Fourth N 1160 233% 20.40 18.22

Richest N 1244 149% 21.88 11.65

Total N 5685 1279

% 100.00 100.00

Chi Square: Significant at 0.001

Table 4.1 shows that only 50% of all health cards were distributed to the poor(first and second poorest quintiles). The remaining health cards were miss-targetedand distributed to wealthier quintiles. The miss-targeting might be happen in locallevel when head of village that have rights to select who should receive health cards,gave health cards to head of villages relatives or friends.

Table 4.2: Tabulation between wealth quintiles and health card utilization forinpatients

Wealth Index QuintilesUse HC for Inpatient

0. No 1. Yes

Poorest N 25 10% 5.33 50.00

Second N 56 6% 11.94 30.00

Middle N 72 3% 15.35 15.00

Fourth N 128 1% 27.29 5.00

Richest N 188 0% 40.09 0.00

Total N 469 20% 100.00 100.00

Chi Square: Significant at 0.05

Table 4.2 shows that only 80% from who use health cards for inpatients werethe poor (first and second poorest quintiles). The remaining health cards were miss-utilized by who were in wealthier quintiles and did not have rights to use health cards.Miss-utilization shows that there is no good verification system on health services


9/26

3rd


providers, when the providers allow who have health card get the free inpatientservices even they are not the poor.

Table 4.3: Tabulation between wealth quintiles and health card utilization foroutpatients

Wealth Index QuintilesUse HC for Outpatient

0. No 1. Yes

Poorest N 288 24% 13.17 26.37

Second N 368 28% 16.83 30.77

Middle N 442 16% 20.22 17.58

Fourth N 494 20% 22.60 21.98

Richest N 594 3

% 27.17 3.30Total N 2186 91

% 100.00 100.00

Chi square: significant at 0.01

Table 4.3 shows that only 57% from who use health cards for outpatients werethe poor (first and second poorest quintiles). The remaining health cards were miss-utilized by who were in wealthier quintiles and did not have rights to use health cards.Miss-utilization shows that there is no good verification system on health services

providers, when the providers allow who have health card get the free outpatientservices even they are in a wealthier economic status.

Multivariate analysis

The Impact of health card ownership on antenatal care

To explore the impact of health card ownership on antenatal care, sevenmodels of multiple regressions are used. Each model has different purposes that can

be seen at the following explanation. The result of logistic regression of the followingmodels can be seen at table 4.4.


10/26

210|3rdI

nternationalConferenceonReproductiveHealthandSocialSciencesResearch

Table4.4:Regression

coefficientsandstandarderrorsfrom

multipleregressionanalysisofTheimpactofhealthcardowne

rshipon

antenatalcare.

IndependentVariables

Model

1

Model2

Model3

Model4

Model5

Model6

Model7

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

C

oefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Stand

ard

Erro

r)

Coefficient

(Standard

Error)

healthcardownership

-0.23

24

-0.4

496

-0.6

261

-0.6

010

-0.4

128

-0.5927

-0.5

497

(-0.2

587)

(0.4

441)

0.4

346

(0.4

547)

(0.4

594)

(0.4

788)

(0.4

617)

yeardummy

-0.0

190

-0.0

992

-0.5

205*

-0.1

990

-0.6

494*

-0.5

157*

(0.2

236)

(0.2

335)

(0.2

617)

(0.2

445)

(0.2

869)

(0.2

598)

Interactionvariable

0.3

214

0.4

585

0.6

937

0.3

445

0.6180

0.6

286

(0.5

515)

(0.5

735)

(0.5

721)

(0.5

712)

(0.5

964)

(0.5

803)

knowwhereispublic

hospital

0.2

895

0.1565

(0.2

508)

(0.2

598)

knowwhereisprivat

ehospital

0.3

395

0.0443

(0.2

669)

(0.2

758)

knowwhereispublic

/auxiliaryhealth

center

-0.7

587

-0.8675

-0.7

374

(0.4

712)

(0.4

817)

(0.4

666)

knowwhereisprivat

eclinic

-0.2

258

-0.3486

(0.3

621)

(0.3

812)

knowwhereisprivat

ephysician

0.0

622

-0.0130

(0.2

456)

(0.2

557)

knowwhereismidw

ife

0.5

431*

0.5888*

0.7

458**

(0.2

515)

(0.2

645)

(0.2

467)


11/26

3rdI

nternationalConferenceon

ReproductiveHealthandSocialSciencesResearch

|211

Table4.4:Regression



rshipon

antenatalcare.(Continued)

Independent

Variables

Model1

Model2

Model3

Mo

del4

Model5

Model6

Model7

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coef

ficient

(Standard

Er

ror)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

knowwhereisnurse

0.1

100

0.0

799

(0.2

339)

0.2

436

knowwhereistraditionalbirth

attendant

-0.0

039

0.1

634

(0.2

537)

0.2

683

knowwhereistraditionalpracticioner

-0.1

673

0.0

519

(0.2

356)

0.2

489

knowwhereispharm

acy

0.4

012

0.2

594

(0.2

555)

0.2

687

knowwhereisposya

ndu

0.5

494

0.4

888

(0.3

003)

0.3

109

householdsize

0.1

328**

0.1

429**

0.1

404**

(0

.0455)

(0.0

488)

(0.0

462)

highesteducationhhh

0.6

898

0.4

374

0.8

667**

(0

.3614)

(0.3

827)

(0.2

963)

highesteducationshh

0.2

672

0.3

801

(0

.3991)

(0.4

259)

houseownership

0.2

929

0.2

207

0.2

585

(0.2

758)


12/26

212|3rdI


Table4.4:Regression



rshipon

antenatalcare.(Continued)

Independent

Variables

Model1

Model2

Model3

Mo

del4

Model5

Model6

Model7

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coef

ficient

(Standard

Er

ror)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

usingelectricity

0.5

194

0.3

793

0.3

034

(0.3

294)

havetelevision

0.2

620

0.1

177

0.2

338

(0.2

572)

improvewatersource

0.3

223

0.2

873

0.2

208

(0.2

427)

improvesanitation

-0.2

191

-0.1

609

(0.2

311)

(0.2

468)

constant

0.5

279***

0.5

366***

0.0

497

-0.3

383

-0.2

385

-1.3

268

-0.2

671

0.1

114

0.1

515

0.5

366

0.3

222

0.3

755

0.7

053

0.5

632

LogLikelihood

-278.7

017

-278.5

184

-267.1

242

-269.4

201

-272.1

993

-256.7

999

-263.9

404

N

420

420

420

420

420

420

420

LRChi2

0.8

0

1.1

7

23.9

6

19.3

6

13.8

1

44.60

30.3

2

Prob>Chi2

0.3

709

0.7

609

0.0

464

0.0

036

0.0

870

0.0

030

0.0

001

PseudoR2

0.0

014

0.0

021

0.0

429

0.0

347

0.0

247

0.0

799

0.0

543

*=Significantat0.05;**=Significantat0.01;***=Sig

nificantat0.001


13/26

3rd


Model 1 on table 4.4 shows that there is no direct effect of health cardownership on antenatal care without controlling for other factors. The model is alsonot significant that shown by prob>chi2 more than 0.05.

Model 2 shows that there is no effect of health card ownership and dummy ofprogram interventions periods as well as there is no combination effect betweenhealth card ownership and program duration on antenatal care without controlling for

other factors. The model is also not significant that shown by prob>chi2 more than0.05.

Model 3 show that there is no effect of health card ownership and dummy ofprogram interventions periods as well as there is no combination effect betweenhealth card ownership and program duration on antenatal care controlling forknowledge of health facilities. In this model, can be seen that knowledge of where ismidwife have significant effect on adequate antenatal care.

Model 4 show that there is no effect of health card ownership and dummy ofprogram interventions periods as well as there is no combination effect betweenhealth card ownership and program duration on antenatal care controlling for socio-demographic variables. In this model can be seen household size have significantfactor of antenatal care.

Model 5 show that there is no effect of health card ownership and dummy ofprogram interventions periods as well as there is no combination effect betweenhealth card ownership and program duration on antenatal care controlling foreconomic variables. Model 5 also shows that there is no effect of economic factor onantenatal care.

Model 6 show that there is no effect of health card ownership as well as thereis no combination effect between health card ownership and program duration onantenatal care controlling for all variables including knowledge of health facilities,socio-demographic and economic variables. But there is significant effect of dummyof program interventions periods (before and after). Consistent with model 3 andmodel 4, there is positive effect of knowledge where is midwife and household size.

Model 7 show that there is no effect of health card ownership as well as thereis no combination effect between health card ownership and program duration onantenatal care controlling for selected knowledge of health facilities variables, socio-demographic variables and economic variables. Consistent with model 6, there issignificant effect of dummy of program interventions periods (before and after).Consistent with model 3, model 4 and model 6, there is positive effect of knowledgewhere is midwife and household size.

Impact of health card ownership to place delivery

To explore the impact of health card ownership on place of delivery, sevenmodels of multiple regressions are used. Each model has different purposes that can

be seen at the following explanation. The result of logistic regression of the following

models can be seen at table 4.5Model 1 on table 4.5 show that there is no direct effect of health cardownership on place of delivery without controlling for other factors. The model is alsonot significant that shown by prob>chi2 more than 0.05.

Model 2 show that there is no effect of health card ownership and dummy ofprogram interventions periods as well as there is no combination effect betweenhealth card ownership and program duration on place of delivery without controllingfor other factors. The model is also not significant that shown by prob>chi2 more than0.05.


14/26

214 | 3rd


Model 3 show that there is no effect of health card ownership and dummy ofprogram interventions periods as well as there is no combination effect betweenhealth card ownership and program duration on place of delivery controlling forknowledge of health facilities. In this model, can be seen that knowledge of where is

public hospital have significant positive effect on delivery at public health facilities.In this model also can be seen that knowledge of where traditional birth attendants are

has significant negative effect on delivery at public facilities. Interestingly, knowledgewhere traditional practitioners are has significant effect to delivery at public healthfacilities.

Model 4 show that there is no effect of health card ownership and dummy ofprogram interventions periods as well as there is no combination effect betweenhealth card ownership and program duration on place of delivery controlling forsocio-demographic variables. In this model can be seen that there is no socio-demographic variables have significant effect to delivery in public facilities.


15/26

3rdI



|215

Table4.5:Regression

coefficientsandstandarderrorsf

rom

multipleregressionanalysisoftheimpactofhealthcardowner

shiponplace

delivery.

IndependentVariables

Model

1

Model2

Model3

Model4

Model5

Model6

Model7

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

C

oefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Stand

ard

Erro

r)

Coefficient

(Standard

Error)

healthcardownership

0.04

24

0.0

232

-0.8

342

0.0

440

-0.3

945

-0.7806

-0.6

049

(0.4

213)

(0.7

954)

(0.8

902)

(0.8

026)

(0.8

442)

(0.9

351)

(0.8

531)

yeardummy

0.3

486

0.3

896

-0.0

236

0.2

484

0.3213

0.4

831

(0.3

678)

(0.4

173)

(0.4

267)

(0.4

345)

(0.5

734)

(0.4

451)

Interactionvariable

-0.0

883

0.5

042

0.0

406

0.5

282

0.3812

0.5

562

(0.9

442)

(1.0

641)

(0.9

716)

(1.0

082)

(1.1

445)

(1.0

491)

knowwhereispublic

hospital

2.2

773**

2.13

45**

2.3

300**

(0.7

720)

(0.7

944)

(0.7

537)

Knowwhereispriva

tehospital

0.3

339

0.3576

(0.4

166)

(0.4

616)

knowwhereispublic

/auxiliaryhealth

center

-0.1

157

-0.2828

(0.7

791)

(0.8

799)

knowwhereisprivat

eclinic

-0.5

605

-1.0420

(0.6

183)

(0.6

697)

knowwhereisprivat

ephysician

-0.3

669

-0.5995

(0.4

087)

(0.4

509)

knowwhereismidw

ife

-0.4

558

-0.2954

(0.4

569)

(0.5

076)


16/26

216|3rdI


Table4.5:Regression


rom


shiponplace

delivery.(Continued)

Independen

tVariables

Model1

Model2

Model3

Model4

Model5

Model6

Model7

Coefficien

t

(Standard

Error)

Coefficien

t

(Standard

Error)

Coefficien

t

(Standard

Error)

Coe

fficien

t

(Sta

ndard

Error)

Coefficien

t

(Standard

Error)

Coefficient

(Standard

Error)

Coefficien

t

(Standard

Error)

knowwhereisnurse

-0.4

844

-0.4

415

(0.4

528)

(0.4

738

)

knowwhereistraditionalbirthattendant

-1.3

987**

-1.3

506**

-

1.2

259***

(0.4

070)

(0.4

450

)

(0.3

844)


0.9

254*

1.2

074**

0.7

666*

(0.4

174)

(0.4

614

)

(0.3

877)

knowwhereispharm

acy

0.5

207

0.2

895

(0.4

351)

(0.4

893

)

knowwhereisposya

ndu

1.1

049

1.2

418

(0.6

522)

(0.6

988

)

householdsize

0.0

678

-0.0

081

(0.0

650)

(0.0

746

)

highesteducationhhh

-0.5

524

-1.0

147

-0.5

232

(0.5

611)

(0.6

286

)

(0.4

913)

highesteducationshh

1

.4414*

1.3

266*

(0.5

623)

(0.6

426

)

houseownership

0.7

112

1.0

392

(0.5

236)

(0.5

985

)


17/26

3rdI



|217

Table4.5:Regression


rom


shiponplace

delivery.(Continued)

Independen

tVariables

Model1

Model2

Model3

Model4

Model5

Model6

Model7

Coefficien

t

(Standard

Error)

Coefficien

t

(Standard

Error)

Coefficien

t

(Standard

Error)

Coe

fficien

t

(Sta

ndard

Error)

Coefficien

t

(Standard

Error)

Coefficient

(Standard

Error)

Coefficien

t

(Standard

Error)

usingelectricity

1.6

348

1.0

987

(1.0

599)

(1.1

177

)

havetelevision

0.7

474

0.7

025

0.7

400

(0.4

117)

(0.5

035

)

(0.4

171)

improvewatersource

0.8

083*

0.5

224

(0.3

648)

(0.4

480

)

improvesanitation

0.3

706

0.2

403

(0.4

109)

(0.5

075

)

constant

-

1.9

332***

-

2.1

026***

-

3.9

553***

-

2.6038***

-

5.1

959***

-

6.4

359***

-

4.0

059***

0.1

835

0.2

648

1.1

710

0.5

347

1.1

707

1.7

018

0.8

412

LogLikelihood

-125.7

808

-125.2

846

-99.3

196

-12

1.3

882

-112.9

377

-90.9

328

-99.6

007

N

330

330

330

330

330

330

330

LRChi2

0.01

1.0

0

52.9

3

8.8

0

25.7

0

69.71

52.3

7

Prob>Chi2

0.9

202

0.8

007

0.0

000

0.1

854

0.0

012

0.0

000

0.0

000

PseudoR2

0.0

000

0.0

040

0.2

104

0.0

350

0.1

021

0.2

771

0.2

082

*=Significantat0.05;**=Significantat0.01;***=Sign

ificantat0.001


18/26

218 | 3rd


Model 5 show that there is no effect of health card ownership and dummy ofprogram interventions periods as well as there is no combination effect betweenhealth card ownership and program duration on place of delivery controlling foreconomic variables. Model 5 also shows that there is no effect of economic factor ondelivery in public facilities.

Model 6 show that there is no effect of health card ownership and dummy of

program interventions periods as well as there is no combination effect betweenhealth card ownership and program duration on place of delivery controlling for allvariables including knowledge of health facilities, socio-demographic and economicvariables. Consistent with model 3, there is positive effect of knowledge of where is

public hospital and knowledge where traditional practitioners are as well as significantnegative effect of knowledge of where traditional birth attendant were on delivery in

public facility.Model 7 show that there is no effect of health card ownership and dummy of

program interventions periods as well as there is no combination effect betweenhealth card ownership and program duration on place of delivery controlling forselected knowledge of health facilities variables, socio-demographic variables andeconomic variables. Consistent with model 3 and model 6, model 7 shows that there is

positive effect of knowledge of where is public hospital and knowledge wheretraditional practitioners are as well as significant negative effect of knowledge ofwhere traditional birth attendant were on delivery in public facility.

Impact of health card ownership to utilization of modern contraceptive

To explore the impact of health card ownership on place of delivery, sevenmodels of multiple regressions are used. Each model has different purposes that can

be seen at the following explanation. The result of logistic regression of the followingmodels can be seen at table 4.6.

Model 1 show direct effect of having health cards on using moderncontraceptives. People who have health card are more likely use modern contraceptivewithout controlling for other factors.

Model 2 show that there is positive effect of health card ownership andcombination effect between health card ownership and program duration onutilization of modern contraceptives without controlling for other factors. Consistentwith model 1, household who have health cards are more likely have use moderncontraceptive.


19/26

3rdI



|219

Table4.6:Regression


rom


shipon

utilization

ofmoderncontraceptive.

Independen

tVariables

Model1

Model2

Model3

Model4

Model5

Model6

Model7

Coefficien

t

(Standard

Error)

Coefficien

t

(Standard

Error)

Coefficien

t

(Standard

Error)

Coe

fficien

t

(Sta

ndard

Error)

Coefficien

t

(Standard

Error)

Coefficient

(Standard

Error)

Coefficien

t

(Standard

Error)

healthcardownership

0.1

947**

0.4

325***

0.3

399**

0.4250***

0.3

461**

0.3

128**

0.3

174**

(0.0

683)

(0.1

114)

(0.1

132)

(0.1

114)

(0.1

127)

(0.1

139

)

(0.1

133)

yeardummy

-0.1

042

-0.1

756**

-0.1697**

-0.1

538*

-0.1

793**

-

0.1

907***

(0.0

551)

(0.0

567)

(0.0

605)

(0.0

606)

(0.0

657

)

(0.0

567)

Interactionvariable

-0.3

530*

-0.2

798

-0

.3246*

-0.1

961

-0.2

068

-0.2

036

(0.1

420)

(0.1

441)

(0.1

426)

(0.1

440)

(0.1

453

)

(0.1

447)

knowwhereispublic

hospital

0.1

237*

0.0

843

(0.0

626)

(0.0

636

)

Knowwhereispriva

tehospital

0.0

314

-0.0

067

(0.0

625)

(0.0

634

)

knowwhereispublic

/auxiliaryhealth

center

0.1

867

0.1

850

(0.1

098)

(0.1

105

)

knowwhereisprivat

eclinic

0.0

266

-0.0

139

(0.0

780)

(0.0

794

)

knowwhereisprivat

ephysician

0.0

959

0.0

534

(0.0

597)

(0.0

608

)

knowwhereismidw

ife

0.1

942**

0.1

992**

0.2

130***

(0.0

625)

(0.0

629

)

(0.0

620)

knowwhereisnurse

0.0

832

0.0

755

(0.0

551)

(0.0

554

)


20/26

220|3rdI


Table4.6:Regression


rom


shipon

utilization

ofmoderncontraceptive.(Continu

ed)

Independent

Variables

Model1

Model2

Model3

Mo

del4

Model5

Model6

Model7

Coefficien

t

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Er

ror)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

knowwhereistraditionalbirth

attendant

-0.0

866

-0.0

133

(0.0

595)

(0.0

617)


-0.2

168***

-0.1

956***

-0.1

878***

(0.0

570)

(0.0

574)

(0.0

566)

knowwhereispharm

acy

0.0

940

0.0

241

0.0

786

(0.0

639)

(0.0

654)

(0.0

562)

knowwhereisposya

ndu

0.4

369

0.4

366

0.4

676***

(0.0

710)

(0.0

714)

(0.0

683)

householdsize

-0.0

049

-0.0

097

(0

.0117)

(0.0

119)

highesteducationhhh

0.0

938

-0.0

669

(0

.0795)

(0.0

832)

highesteducationshh

0.1

404

0.0

239

(0

.0923)

(0.0

953)

houseownership

-0.1

263

-0.1

159

(0.0

734)

(0.0

758)

usingelectricity

0.0

263

-0.0

534

(0.0

851)

(0.0

870)

havetelevision

0.3

906***

0.3

512***

0.3

531***

(0.0

588)

(0.0

626)

(0.0

583)


21/26

3rdI



|221

Table4.6:Regression


rom


shipon

utilization

ofmoderncontraceptive.(Continu

ed)

Independent

Variables

Model1

Model2

Model3

Mo

del4

Model5

Model6

Model7

Coefficien

t

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Er

ror)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

Coefficient

(Standard

Error)

improvewatersource

0.1

410**

0.1

406

0.1

417**

(0.0

548)

(0.0

575)

(0.0

549)

improvesanitation

0.0

074

-0.0

141

(0.0

565)

(0.0

574)

constant

0.1

258***

0.1

756***

-0.5

620***

0.1

959**

-0.0

420

-0.5

882***

-0.6

045***

0.0

275

0.0

381

0.1

202

0.0

765

0.1

086

0.1

634

0.0

825

LogLikelihood

-4377.6

903

-4369.7

141

-4305.5

943

-4365.5

847

-4328.3

262

-4282.4

358

-4288.5

722

N

6350

6350

6350

6350

6350

6350

6350

LRChi2

8.1

8

24.1

3

152.3

7

32.3

9

106.9

1

198.69

186.4

1

Prob>Chi2

0.0

042

0.0

000

0.0

000

0.0

000

0.0

000

0.0

000

0.0

000

PseudoR2

0.0

009

0.0

028

0.0

174

0.0

037

0.0

122

0.0

227

0.0

213

*=Significantat0.05;**=Significantat0.01;***=Sign

ificantat0.001


22/26

222 | 3rd


Model 3 show that there is positive effect of health card ownership and butthere is no combination effect between health card ownership and program durationon utilization of modern contraceptives controlling for knowledge of health facilities.Consistent with model 1 and 2, household who have health cards are more likely haveuse modern contraceptive. But the periods variable show that household in 2000 areless likely use modern contraceptive. The model also show positive effect of

knowledge of where is public hospital, midwife, traditional practitioners andposyandu are on utilization modern contraceptives

Model 4 show that there is positive effect of health card ownership and there iscombination effect between health card ownership and program duration onutilization of modern contraceptives controlling for socio-demographic variables.Consistent with model 1, 2, and 3 household who have health cards are more likelyhave use modern contraceptive. Consistent with model 3, the year dummy show thathousehold in 2000 are less likely use modern contraceptive. In this model also can beseen that there is no socio-demographic variables have significant effect to delivery inutilization of modern contraceptives.

Model 5 show that there is positive effect of health card ownership anddummy of program intervention periods, but there is no combination effect between

health card ownership and program duration on utilization of modern contraceptivescontrolling for economic variables. Consistent with model 1, 2, 3 and 4, householdwho have health cards are more likely have use modern contraceptive. Consistent withmodel 3 and 4, the year dummy shows households in 2000 are less likely use moderncontraceptive. The model also shows positive effect of having television and access toimprove water on utilization modern contraceptives.

Model 6 show that there is positive effect of health card ownership anddummy of program intervention periods, but there is no combination effect betweenhealth card ownership and program duration on utilization of modern contraceptivescontrolling for all variables including knowledge of health facilities, socio-demographic and economic variables. Consistent with model 3, the model also show

positive effect of knowledge of where is public hospital, midwife, traditionalpractitioners and posyandu on utilization modern contraceptives. Consistent withmodel 5, the model also shows positive effect of having television and access toimprove water on utilization modern contraceptives.

Model 7 show that there is effect of health card ownership and but there is nocombination effect between health card ownership and program duration onutilization of modern contraceptives controlling for selected knowledge of healthfacilities variables, socio-demographic variables and economic variable.

Table 4.7: Adjusted Probability of having Health card to utilization of moderncontraceptive

Have health card Number ofObservation

AdjustedProbability

Standard Error

No (0) 5,299 0.5334 0.0279Yes (1) 1,051 0.5730 0.0634

LR Chi2 5.37Prob>Chi2 0.0205


23/26

3rd


Table 4.8: Interaction effect between health cards ownership and year dummy

Variables N Mean Std.Dev. Min Max

ie 6,350 -0.0469 0.0036 -0.0517 -0.0410se 6,350 0.0339 0.0016 0.0311 0.0358z 6,350 -1.3788 0.0462 -1.5010 -1.3164

Adjusted probability result of health card ownership to modern contraceptive

use show that 57% household that have health card will use modern contraceptive, itis clearly confirm that having health card is one of factor affecting the decision to usemodern contraceptive during the crisis.

Figure 4.1: Interaction Effect Figure 4.2: Z-statistics

From figure 4.1 can be seen the magnitude of interaction effect in wide rangeand vary depend on level of each covariates. The mean of interaction effect is -0.046(Table 4.8). At figure 4.1 can be seen that the interaction effect can be found widelyalthough none of them statistically significant according figure 4.2. It might behappen because from 1997 to 2000 is crisis periods, therefore the time effect of

program intervention was neutralized by economic shock because of crisis.

From table 4.6 model 7 can be seen that both health card ownership and yeardummy are highly statistically significant. However, the interaction variable is notstatistically significant. From model 7, could be concluded that there is no interaction.It might be happen because from 1997 to 2000 is crisis periods, therefore the timeeffect of program intervention was neutralized by economic shock because of crisis.

The other possible answer why the interaction effect are not is significant tocontraceptive use are stability of contraceptive use in Indonesia. This argument issupported by Strauss et al (2002) and Frankenberg et al (1999) who found that theeconomic crisis did not affect a lot on the change of contraceptive uses. It means thatneither economic crisis nor health card program have significant effect to utilizationmodern contraceptives.

Conclusion and Recommendation

Chapter V is the conclusion of this study. There are two topic of this chapter,finding of the study and implication of this study


24/26

224 | 3rd


Conclusion

This paper not only found that the effectiveness of the program should be increasebut also can be detect some part of the program that need to be improved. Someimportant point for the analysis as follow:

x Health card program have less performance on targeting and distribution.

x Health card program did not affect to secure adequate antennal care ofpregnant women.

x Knowledge where midwives is very important in securing access to adequate

antenatal care.

x Health card program did not affect giving more access on delivery to public

health facilities

x Knowledge of where is public hospital is very important to improve access on

delivery in public facility.

x Health card program have positive effect on modern contraceptive use but the

interaction impact between health card ownership and year dummy is not

significantly improve utilization modern contraceptive.x Knowledge where public hospital, midwife, traditional practitioners and

posyandu is very important to improve the utilization of modern contraceptive.

x Interestingly, economic factors are not major constraint for reproductive health

access. It is need more study to answer why it is happened.

Recommendation

For Further Social Safety Net Program

1. Improve the quality of targeting and distribution for further social safety net

on health program. Program implementer should have individual level data

base of targeted recipient of safety net and did let local level official decided

the recipient freely.

2. Improve the quality monitoring for implementation of further social safety net

on health program to avoid misallocation.

For Further Research

3. Conduct study with more sample size of pregnant women or who have

delivery, especially to investigate the impact social safety net on health to

antenatal care and place delivery.

4. Conduct further study on stable macroeconomic situation, to reduce the bias

because of the effect of business cycle fluctuation. It will be better if the studyconducted with randomized evaluation methods to avoid selection bias.


25/26

3rd


Bibliography

Achmad, I. and Westley, S.B. (1999) Indonesian Survey Looks at AdolescentReproductive Health East West Center, Asia Pacific Population and PolicyNumber 51 October 1999

AUSAID (2002) The Impact of Asian Financial Crisis on the Health Sector in

Indonesia http://www.ausaid.gov.au/publicationsAngrist, J and Pischke, J (2008) Mostly Harmless Econometrics: An Empiricists

Companion Princeton University PressBhatia, M., Yesudian, C., Gorter, A., Thankappan, K (2006) Demand Side for

Reproductive and Child Health Service in India Economic and Political WeeklyJanuary 21, 2006

BPS (2003) Indonesia Demographic Health Survey 2002-2003 Indonesian CentralBureau of Statistics, National Family Planning Coordinating Board, Ministry ofHealth, and Macro International, December 2003

Ellis, R and McGuire, T (1993) Supply Side and Demand Side Cost Sharing inHealth CareJournal of Economic Perspectives Volume 7, Number 4-Fall 1993-Pages 135-151

Frankenberg, E., Smith, J.P., and Thomas, D (2002) Economic shocks, wealth andwelfare February 2002

Frankenberg, E., Beegle, K., Sikoki, B., and Thomas, D. (1998) Health, FamilyPlanning and Wellbeing in Indonesia during an Economic Crisis: Early Results

from the Indonesian Family Life Survey RAND Labor and Population ProgramWorking Paper Series 99-06

Frankenberg, E., Thomas, D., and Beegle, K. (1999) The Real Costs of IndonesiasEconomic Crisis: Preliminary Findings from the Indonesia Family Life SurveysRAND Labor and Population Program Working Paper Series 99-06

Frankenberg, E., Sikoki, B., and Suristiarini, W.. (1998) Contraceptive Use in aChanging Service Environment: Evidence from Indonesia during the Economic

Crisis Studies in Family Planning 2003 ; 34[2]: 103-116Hotchkiss and Jacobalis, S. 1999. Indonesian health care and the economic crisis: ismanaged care the needed reform?Health Policy 46: 195216

Johar, Melliyani (2007) The Impact of the Indonesian Health Card Program: aMatching Estimator Approach School of Economics Discussion Paper 2007/ 30,University of New South Wales

Lanjauw, P., Pradhan, M., Saadah, F., Sayeed, H., and Sparrow, R. (2001) Poverty,Education and Health in Indonesia: Who Benefits from Public Spending?December 2001

Pitoyo, A.J. 2007. Indonesian Health Standards: The Evidence of Dynamic of HealthConditions from Indonesian Family Life Survey. Paper presented at IPSRInternational Conference on Understanding Health and Population Over Time:

Strengthening Capacity in Longitudinal Data Collection and Analysis in Asia andthe Pacific Region at Royal Benja Hotel, Bangkok. Thailand, May 24-25, 2007.

Pritchett, L., and Suryahadi, A (2002) Targeted Programs in an Economic Crisis:Empirical Findings from the Experience of Indonesia SMERU Working Paper,SMERU Research Institute, October 2002

Ravallion, Martin (2001) The Mystery of Vanishing Benefit: An Introduction ofImpact Evaluation The World Bank Economic Review Vol 15 No. 1 115-140

Ravallion, Martin (2008) Evaluating Anti Poverty Program The Handbook ofDevelopment Economics Vol 4, Edited by Paul Scultz and John Strauz


26/26

226 | 3rd


Saadah, F., Pradhan, M., and Sparrow, R. (2001) The Effectiveness of the HealthCard as an Instrument to Ensure Access to Medical Care for the Poor during the

Crisis Paper prepared for the Third Annual Conference of the GlobalDevelopment Network, Rio de Janeiro, Brazil, December 912, 2001, atwww//gdnet.org

Saadah, F., Pradhan, M., and Sparrow, R. (2007) Did Health Card Program ensure

access to medical care for the poor during the Indonesia crisis? The WorldBank Economic Review vol. 21, no. 1, pp. 125150

Somanathan, Aparnaa (2008) The Impact of Price Subsidies on Child Health CareUse: Evaluation of the Indonesian Health Card The World Bank, PolicyResearch Working Paper 4622, May 2008

Sparrow, Robert (2006) Health, Education and Economic Crisis: Protecting thePoor in Indonesia PhD Dissertation, Vrije University Amsterdam, theNetherlands

Sparrow, Robert (2008) Targeting the Poor in Times of Crisis: The IndonesianHealth Card.Health Policy and Planning no. 23, 2008 pp 188199

Strauss, J., Beegle, K., Dwiyanto, A., Herawati, Y., Pattinasarany, D., Satriawan, E.,Sikoki, B., Sukamdi., Witoelar. (2002) Indonesian Living Standards Three Year

after Crisis: Evidence from the Indonesian Family Life Survey RANDCorporation

United Nations (2003) Indicators for Monitoring the Millennium DevelopmentGoals: Definitions, Rationale, Concept and Sources The United NationsDevelopment Group, United Nations, New York, 2003

Waters, H., Saadah, F., and Pradhan, M. (2003) The impact of the 199798 EastAsian economic crisis on health and health care in IndonesiaHealth Policy andPlanning 18(2), pp 172-181

WHO (1998) Health Implications of the Economic Crisis in South-East AsianRegion World Health Organization, Report of Regional Consultation Bangkok,Thailand, 23-25 March 1998

Wooldridge (2005) Introductory Econometrics: A Modern Approach 3rd editionThomson Learning

World Bank (2008) Investing in Indonesias Health: Challenges and Opportunitiesfor Future Public Spending Health Public Expenditure Review 2008

Yoddumnern-Attig, B., Guest, P., Thongthai, V., Punpuing, S., Sethaput, C.,Jampaklay, A., et al (2009) Longitudinal Research: A tool for Studying SocialChange Institute for Population and Social Research, Mahidol UniversityThailand.

evaluating the impact of health card program on access to reproductive health services: an...

Documents