osman zaim and müge diler bilkent university department of economics
DESCRIPTION
The Impacts of Health Sector Reform on the Efficiency and Productivity of Public and Private Hospitals in Turkey. Osman Zaim and Müge Diler Bilkent University Department of Economics. Social Security and Health Care (Pre-Reform). - PowerPoint PPT PresentationTRANSCRIPT
The Impacts of Health Sector Reform on the Efficiency and Productivity of Public
and Private Hospitals in Turkey
Osman Zaim and Müge DilerBilkent University
Department of Economics
Social Security and Health Care (Pre-Reform)
∆ A rather dispersed and fragmented social security and health delivery system
Social Security Association - SSA(1946): operates both as an insurer and a health care provider for its members (private sector and blue-collar public sector workers). Provides health services
through its own hospitals.
Pension Fund for Civil Servants (1950) The members and their dependents are provided basic health care services at MoH (i.e. public) hospitals.
Social Insurance Agency of Merchants, Artisans and the
Self- employed (1971). Purchases health delivery services from SSA.
Green-Card (1992) For the extremely needy
Social Security and Health CoverageTABLE: 1 - THE POPULATION COVERED BY SOCIAL INSURANCE PROGRAMS (1996-2005)
INSTITUTIONS 2000 % in 2000 2005 % in 2005I. THE PENSION FUND OF CIVIL SERVANTS IN TOTAL 8.230.201 14,75 9.270.512 13,89
1. Active Insured 2.156.176 2.402.409
2. Pensioners (retired, invalid, widow, widower, orphan) 1.296.935 1.595.973
3. Dependants (1) 4.777.090 5.272.130
II. THE SOCIAL INSURANCE INSTITUTION IN TOTAL 32.192.374 57,71 41.166.730 61,69
1. Active Insured 5.283.234 6.965.937
2. Voluntary Active Insured 843.957 266.558
3. Active Insured in Agriculture 184.675 178.178
4.Pensioners (retired, invalid, widow, widower, orphan) 3.339.327 4.308.186
5. Dependants (1) 22.541.181 29.447.871
III. THE SOCIAL SECURITY INSTITUTION OF CRAFTSMEN,
TRADESMEN AND OTHER SELF-EMPLOYED IN TOTAL (2) 15.036.318 26,96 15.990.253 23,96
1. Active Insured 2.181.586 2.103.651
2. Voluntary Active Insured 254.960 239.388
3. Active Insured in Agriculture 876.148 1.011.333
4.Pensioners (retired, invalid, widow, widower, orphan) 1.277.444 1.600.294
5. Dependants (1) 10.446.180 11.035.587
IV.THE PRIVATE FUNDS IN TOTAL 323.569 0,58 306.169 0,46
1. Active Insured 78.495 75.552
2. Pensioners (retired, invalid, widow, widower, orphan) 71.266 76.027
3. Dependants (1) 173.808 154.590
VI. GENERAL TOTAL 55.782.462 100 66.733.664 100
1. Active Insured 9.699.491 11.547.549
2. Voluntary Active Insured 1.098.917 505.946
3. Active Insured in Agriculture 1.060.823 1.189.511
4. Total Active Insured 11.859.231 13.243.006
5. Pensioners (retired, invalid, widow, widower, orphan) 5.984.972 7.580.480
6. Dependants (1) 37.938.259 45.910.178
V. SOCIAL INSURANCE COVERAGE WITH RESPECT TO HEALTH
SERVICES (3) 54.938.505 66.467.106
VII. RATIO OF INSURED POPULATION (Percent) 82,2 92,0VIII.RATIO OF INSURED POPULATION COVERED BY HEALTH SERVICES (Percent)80,9 91,7IX. TOTAL POPULATION 67.893.000 72.520.000
Unsustainable Deficits∆ The broadened coverage of the social security system has been
achieved at the expense of ever growing deficits since the early 1990s
Unequal Health Service Delivery
Institution Number of Hospitals
Number of Beds
Beds Per Constituency
Number of Specialists
Specialists Per Constituency
Number of Practitioners
Practioners Per Constituency
MoH 751 87709 76 13837 367 28983 218
SSA 118 28517 536 4801 6705 3311 9723
University 43 24754 - 8586 - 8760 -
Private 239 11922 27 8665 37 2870 113
Pressure Groups
∆ IMF∆ World Bank∆ EU : Association Council Decision 3/80 requires
the application of social security schemes of the Member States of the European Communities to Turkish workers and members of their families
RESPONSE: A Social Security Reform consisting of 4 main components complementing each other
∆ The first component is the setting up of a single retirement insurance regime that includes short and long term insurance branches other than health insurance.
∆ The second is the creation of a General Health Insurance towards financing the provision of a high quality health service for all population, which is fair, equal, protective and curative.
∆ The third is the gathering of social benefits and services that are currently being carried out in a scattered manner and establishment of a system based on objective benefit criteria and which is accessible by all groups who are in need.
∆ The fourth component is the creation of a new institutional structure,which will ensure the provision of above mentioned services in a modern and efficient manner
Steps and Measures of Reform∆ April 2003 : as a result of the protocol signed between MoH and the
Ministry of Finance, civil servants are allowed to benefit from private health institutions.
∆ July 2003 : MoH, Ministry of Labor and Social Security and the Ministry of Finance signed a collective utilization protocol that enabled the members of Soc. Ins. Agency of Tradesman, Pension fund of Civil Cervants, active public employees and green card holders to benefit from SSA’s hospitals, and that enabled members of SSA to benefit from MoH (public) hospitals. The execution date of the protocol throughout Turkey was Jan 2004.
∆ Feb 2005: with law no 5502, ownership of all SSA’s hospitals are transferred to the MoH. Hence, with this final step unification process of the reform has been completed.
∆ All security institutions are united under the name of Social Security Institution (SSI), 2005
∆ Currently all patients are covered by SSI and all patients could benefit from either MoH (public) hospitals, YOK (university hospitals) or private hospitals, operating under the administration of MoH
Likely effects of reform on hospital performance
∆ Significant efficiency differentials among hospitals during the pre-reform must have been reduced after reform leading to a more uniform health service delivery. (Reduced efficiency in hospitals owned by SSA and increased efficiency in hospitals owned by MoH, Private Hospitals and University Hospitals as rationing of patients with respect to membership is alleviated)
∆ Significant after reform productivity increase in the health service delivery to meet the health needs of the increased patients with health insurance coverage (new-entrants).
∆ Significant scale adjustments in relatively smaller private and university hospitals as they have started to serve members of SSA after 2003.
∆ Some administrative effect after 2005 as ownership of hospitals changed hands.
Model: Bootstrap DEA-Bootstrap Malmquist
∆ An output orientated variable returns to scale (VRS) DEA model is employed
} producecan ),{(set oduction Pr yxRyx qp
} producecan ),{( yxRyx qp } producecan ),{( yxRyx qp
}),({)(set Output yxRyxY q
0 є Y(x) that's inaction is possible
non zero output levels can not be produced from zero levels of inputs
Y(x) satisfies strong disposibility of outputs and inputs
Y(x) is closed, bounded and convex.
Farrell MeasureThe efficient boundary of the output correspondence set
}1each for )( ),({)( θxYyxYyyxY
}1),(max{
:as defined is efficiencyoutput of measure Farrell )(any For
kkk
kk
xYy
yx
n
iii
n
iiik
n
iiikk
k
nixxyy111
*
*
],,1each for 0,1,,max{
problem LPby obtained becan unit producingeach For
Bootstrapping-General Idea
In bootstrapping, by repeatedly simulating or mimicking the data generating process (DGP) through resampling with replacement and through applying the original estimator to each simulated sample, we could approximate the sampling distributions of the original estimator.
Simar and Wilson (1998, 2000)In this section, we briefly discuss the relationship between the efficiency score obtained
from DEA model, k̂ and bootstrapped efficiency score *ˆk .Note that the true population
values for Ω, Y(x) and ∂ Y(x) could not be observed, but for any given sample of
observations S = {(xi, yi) │i = 1,....,n }, the sample estimators of )(ˆ,ˆ xY and k̂ could
be derived. Within the true world, k̂ is an estimator of θ based on the sample S,
generated from some DGP, P. On the other hand, in the bootstrap world, *ˆk is an
estimator of k̂ based on the sample S* generated from )(ˆ SP . Then, if bootstrap is
consistent the following relationship holds: ** )ˆˆ(~)ˆ( SS kkkk . In other words,
the known bootstrap distributions will duplicate the original and unknown sampling distributions of the estimators. This relationship would, then lead to a very crucial result:
the bias of the original DEA estimator k̂ , )ˆ(bias ks, kksE , that could be calculated
by using its bootstrap counterpart, )ˆˆ(iasb̂ *ks, * kks
E which is already known. Thus,
bias corrected estimates which is denoted by *ˆk could be obtained as
kskk iasb,
**
ˆˆˆ
Therefore, bootstrapping estimates the bias corrected efficiency score for each hospital within some range known as confidence interval
Malmquist Productivity indexThe Malmquist index is based on the concept of output distance
function Dt0
1
}))/,(:inf{),(( 0 ttYtXtYtXtD . Using the output
distance functions, Fare, Grosskopf, Norris and Zhang (1994) define the Malmquist output-based productivity as:
M X Y X YD X Y D X Y
D X Y D X Yt t t t t
t t t t t t
t t t t t t01 1 1 0
1 10
1 1 1
0 01
1 2
( , , , )
( , ) ( , )
( , ) ( , )
/
or equivalently as
MD X Y
D X Y
D X Y D X Y
D X Y D X Yt
t t t
t t t
t t t t t t
t t t t t t01 0
1 1 1
0
01 1
0
01 1 1
01
1 2
( , )
( , )
( , ) ( , )
( , ) ( , )
/
Data ∆ Original data set contains 1150 hospitals (public, private, SSA, and
university). Source is MoH.∆ We have excluded those that are specialized in just one field of
medicine such as mother and child health care, physiotherapy and rehabilitation, mental disorders, eye care, oncology, cardiology, urgent care and traumatology.
∆ Some had missing data∆ A careful outlier analysis, Wilson (1993)∆ Finally, with the elimination of outliers in the data, a balanced
panel of 441 hospitals (281 MoH, 85 SSA, 45 university and 30 private hospitals) in the pre-reform period and 415 hospitals (338 MoH, 47 university and 30 private hospitals) in the post-reform period is obtained
Variables
Output variables∆ outpatient visits, ∆ number of small, medium and big surgeries separately, ∆ number of births ∆ total inpatient daysInput variables∆ number of beds ∆ number of specialists ∆ number of practitioners
Results-Efficiency Comparisons
2001 2002 2003 2004 2005 2006 Institution DEA DEA* DEA DEA* DEA DEA* DEA DEA* DEA DEA* DEA DEA*
MoH (merged) 1.619 1.771 1.631 1.782 1.716 1.863 1.442 1.574 1.451 1.610 1.270 1.370 MoH (indep.) 1.410 1.564 1.428 1.595 1.482 1.649 1.343 1.486 1.401 1.576 1.282 1.401 SSA (revolved) 1.102 1.251 1.099 1.260 1.121 1.297 1.215 1.364 1.357 1.533 1.163 1.267 SSA (merged) 1.250 1.397 1.280 1.437 1.316 1.475 1.430 1.581 - - - - Univ. 1.213 1.335 1.319 1.483 1.192 1.326 1.170 1.301 1.260 1.430 1.184 1.299 Private 1.338 1.564 1.371 1.624 1.393 1.648 1.369 1.583 1.260 1.486 1.266 1.448 OVERALL 1.338 1.492 1.363 1.534 1.390 1.560 1.317 1.464 1.370 1.548 1.251 1.370
Elimination of inefficiency differentialsCI Width for Bias Corrected Efficiency Scores of 2001
1,000
1,200
1,400
1,600
1,800
2,000
MoH (merged) MoH (indep.) SSA (revolved) SSA (merged) Univ. Private
institution
ub
lb
dhatbc
CI Width for Bias Corrected Efficiency Scores of 2003
1,000
1,200
1,400
1,600
1,800
2,000
2,200
MoH (merged) MoH (indep.) SSA (revolved) SSA (merged) Univ. Private
institution
ub
lb
dhatbc
CI Width for Bias Corrected Efficiency Scores of 2004
1,000
1,200
1,400
1,600
1,800
2,000
MoH (merged) MoH (indep.) SSA (revolved) SSA (merged) Univ. Private
institution
ub
lb
dhatbc
CI Width for Bias Corrected Efficiency Scores of 2006
1,000
1,200
1,400
1,600
1,800
2,000
MoH (indep.) MoH (merged) SSA (revolved) Univ. Private
institution
ub
lb
dhat.bc
Productivity trendsOVERALL MALMQUIST EFFCH TECHCH PUREEFFCH SCALEEFFCH
2001 1 1 1 1 12002 1,02794139 0,98675 1,041744 0,981485085 1,0053647142003 1,01686862 0,967 1,051571 0,962545119 1,0046281272004 1,097248455 1,021803 1,073835 1,015991249 1,0057205622005 1,118525199 0,951401 1,175661 0,973139082 0,9776617492006 1,116363526 1,078487 1,03512 1,063204363 1,014373903
MoH (Indep.)2001 1 1 1 1 12002 1,018828019 0,997169 1,02172 0,987775966 1,0095095892003 0,978548186 0,953157 1,026639 0,951350935 1,0018986662004 1,128859867 1,045809 1,079413 1,04985019 0,996150522005 1,169498822 0,976785 1,198149 1,008906033 0,9682583052006 1,130905361 1,100837 1,028012 1,102734294 0,998274313
MoH (merged with SSA)2001 1 1 1 1 12002 1,018090588 0,985278 1,033303 0,992669462 0,9925535432003 1,001892169 0,964624 1,038635 0,943872384 1,021985542004 1,173263579 1,12003 1,047529 1,123209678 0,9971692142005 1,154184494 1,021977 1,129365 1,116383747 0,9154349872006 1,198959407 1,229967 0,97479 1,275401012 0,964376417
SSA (revolved)2001 1 1 1 1 12002 1,061695901 0,987933 1,074664 1,002409863 0,985557592003 1,092368665 0,948945 1,15114 0,983186268 0,9651731862004 0,980231185 0,895088 1,095123 0,907253784 0,9865901272005 0,934785739 0,78877 1,185118 0,814812416 0,9680390662006 1,000816563 0,942499 1,061876 0,936822639 1,006058949
SSA (merged with MoH)2001 1 1 1 1 12002 1,034023993 0,967781 1,068449 0,976323923 0,9912495142003 1,064255533 0,928627 1,146053 0,949954549 0,9775485232004 0,950655105 0,874642 1,086907 0,873931094 1,00081378
UNIV. HOSPITALS MALMQUIST EFFCH TECHCH PUREEFFCH SCALEEFFCH2001 1 1 1 1 12002 1,031216539 0,951672 1,083583 0,919823478 1,0346250892003 1,074098802 1,119746 0,959235 1,017816786 1,1001447172004 1,134230448 1,13179 1,002156 1,036283526 1,0921629012005 1,1659889 1,020875 1,141456 0,957525978 1,0659509912006 1,203300545 1,149505 1,045573 1,017850115 1,1288421
PRIVATE HOSPITALS
2001 1 1 1 1 12002 1,036093776 0,964814 1,073879 0,975749639 0,9887928632003 1,097889877 0,943967 1,16306 0,96073198 0,9825495622004 1,094696588 0,991042 1,104591 0,977036817 1,0143347052005 1,122064003 1,071317 1,047152 1,06203902 1,0082486972006 1,172556883 1,112027 1,054482 1,056728825 1,050595142
Cumulative Malmquist index over 2001-2006
0,8000,8500,900
0,9501,0001,0501,100
1,1501,2001,250
2001 2002 2003 2004 2005 2006
malm overall
MoH(indep.)
SSA(revolved)
MoH(merged)
SSA(merged)
Univ.
Private
Confidence intervals for Malmquist index
CI Width for MALM 2001-2002
0,8400,8800,9200,9601,0001,0401,0801,1201,160
MoH(merged)
MoH(indep.)
SSA(revolved)
SSA(merged)
Univ. Private
institution
malm ub
malm lb
malm
CI Width for MALM 2005-2006
0,000
0,200
0,400
0,600
0,800
1,000
1,200
MoH (indep.) MoH (merged) SSA (revolved) UNIV. Private
institution
malm ub
malm lb
malm
Concluding Remarks∆ All the expected positive effects of reform seem to have been
realized. Slightly reduced efficiency in formerly SSA owned hospitals have been more than offset by increased efficiency in MoH hospitals as well as in private hospitals and university hospitals, leading to a more accessible and higher quality service provision which also reflects itself in patients’ satisfaction surveys (less waiting time by switching from queuing regime to an appointment regime)
∆ Significant scale adjustments in small scale private and university hospitals seem to have taken place
∆ Significant after reform productivity increase in the health service delivery have been instrumental in meeting the health needs of the increased patients with health insurance coverage (new-entrants).