Report No. 64803-AL


Albania
Out-of-Pocket Payments in Albania’s Health
System
Trends in Household Perceptions and Experiences 2002-2008


March 28, 2011

Human Development Department
Europe and Central Asia Regional Office




Document of the World Bank
                                                                             Page ii


                                         Albania
                   CURRENCY AND EQUIVALENT UNITS
                               (as of 30 January 2011)
                          Albanian Lek    =   100.00
                                  US$     =   0.96
                                  EUR     =   0.72

                         WEIGHTS AND MEASURES
                           <type weights/measures>

                               ABBREVIATIONS

HII                   Health Insurance Institute
INSTAT                Albanian Statistic Institute
IDRA                  Institute for Development Research Alternatives
GDP                   Gross Domestic Product
GPF                   Governance Partnership Facility
GNP                   Gross National Product
MoES                  Ministry of Education and Science
MoH                   Ministry of Health
PHC                   Primary Health Care




                   ECA Vice President         :    Phillippe H. Le Houerou
             ECCU4 Country Director           :    Jane Armitage
                   HD Sector Director         :    Mamta Murthi (Acting)
                 HDE Sector Manager           :    Jesko Hentschel
         HD Country Sector Coordinator        :    Daniel Dulitzky
                           Task Leader        :    Truman Packard
                                                                                             Page iii


                                       Acknowledgements
The World Bank would like to thank the Albanian Statistics Institute (INSTAT), for their
technical input in development of the governance and accountability modules for the 2008
LSMS. It is important to stress, from the outset, that the period of analysis of this report extends
only to 2008, and therefore, does not capture the impact of the Government of Albania‟s recent
laudable efforts to lower the financial vulnerability of households to the cost of healthcare –
particularly the burden on poor households. The results in this report provide a useful “baseline�
to measure the likely beneficial impact of recent changes.
This report is a product of the World Bank‟s Accountability for Better Governance Program
(P107759) of analytical advisory assistance to the Government of Albania, and was supported
with financial assistance from the Governance Partnership Facility (GPF). The report was
prepared by a team of researchers and World Bank staff: Meltem Aran (Consultant, St.
Antony‟s College, University of Oxford), Sonila Tomini (Consultant, Maastricht Graduate
School of Governance), and Truman Packard (Lead Economist, Human Development
Economics, Europe and Central Asia Regional Department, World Bank). The team received
technical support and input from Lorena Kostallari (Senior Operations Officer, Human
Development, World Bank) and Gentjana Sula (Operations Officer, Human Development,
World Bank) who also managed important discussions and exchanges of information with the
Government from the World Bank‟s Office in Tirana. Administrative and logistical support was
provided from Washington, by Katerina Timina (Program Assistant, ECSHD) and in Tirana by
Elda Hafizi (Program Assistant, ECCAL).
The team worked under the direction of Tamar Manuelyan Atinc and Mamta Murthi (Prior
and Current Directors, Human Development, Europe and Central Asia, World Bank); Jane
Armitage (Country Director, Western Balkans, World Bank ); Gordon Betcherman and Jesko
Hentschel (Prior and Current Sector Managers for Human Development Economics, Europe and
Central Asia, World Bank); Camille Lampart Nuamah and Kseniya Lvovsky (Prior and current
Country Managers, Albania, World Bank), and Daniel Dulitzky (Country Sector Coordinator for
the Western Balkans, Human Development, World Bank).
The Peer Reviewers of this report are, Magnus Lindelow (Senior Economist, Human
Development, East Asia and Pacific Region, World Bank), Mattias Lundberg (Senior
Economist, Human Development Network, World Bank), and Waly Wane (Senior Economist,
Poverty Reduction and Economic Management, Africa Region, World Bank), who provided
helpful guidance at the outset of the AAA program and held the team to a high technical standard
of quality.
Special thanks are extended to Clelia Rontoyanni (Public Sector Specialist, Poverty Reduction
and Economic Management, World Bank) for collaboration with the team during preparation of
the Albania Country Governance Review, and for her suggestions that have strengthened this
report, and to Dena Ringold (Senior Economist, Office of the Chief Economist, Human
Development Network, World Bank) for ensuring consistency between this report and wider
efforts in the World Bank‟s Human Development Network to improve governance and
accountability in the social sectors.
                                                                                                                                                Page iv



                                                           Table of Contents
Part I. Introduction, Motivation and Summary of Findings ........................................................... 1
   1.a. Paying Out of Pocket: Financial Vulnerability and Obstacles to Good Health ................... 1
   1.b. Principal Findings: Out of Pocket Spending, Informal Payments and Impoverishment in
   Albania ........................................................................................................................................ 3
   1.c. The Prevalence Informal Payments in Albania: Summary of Findings ............................... 5
Part II. Albania‟s Health Sector: Structure, Image and Household Experience ............................. 8
   2.a. Albania‟s Health Sector ....................................................................................................... 8
   2.b. Household Perceptions of Governance & Informal Payments .......................................... 12
   2.c. From Perceptions to Experience of Informal Payments .................................................... 16
Part III. The Impoverishing Effects of Health Costs, Formal and Informal ................................. 18
Part IV. The Determinants of Informal Payments in Albania‟s Health System ........................... 24
   4.a. Determinants of Informal Payments for Outpatient Services ............................................ 25
   4.b. Determinants of Informal Payments for Inpatient Services ............................................... 26
   4.c. Insights for Policy Makers ................................................................................................. 27
References ..................................................................................................................................... 29


                                                                 List of Tables

Table 1. Opinions About Informal Payments and How to Solve the Problem ............................. 15
Table 2. Average Per Capita Expenditures on Health and Non-Health Items ............................. 18
Table 3. Financing Budget Shares On Health And Non-Health Item........................................... 19
Table 4. Incidence and Impact of Catastrophic Health Payments with respect to total expenditure.
....................................................................................................................................................... 21
Table 5. Poverty Based on Consumption, Gross and Net of Spending on Health ........................ 22
Table 6. Explanations for Informal Payments and Related Observations .................................... 24


                                                                List of Figures

Figure 1. Health Spending as a Percentage of GDP in 2008, Albania and Comparator Countries 9
Figure 2. Health Spending by Source in 2008, Albania and Selected Countries.......................... 10
Figure 3. Reported Corruption and Bribes in the Health Sector in 2008 in Albania and
Comparator Countries ................................................................................................................... 11
Figure 4. Monthly Salaries of GPs (in 000 Leks) ........................................................................ 12
Figure 5. Are Medical Doctors and Teachers Honest or Corrupt? ............................................... 13
Figure 6. How Transparent are the Ministry of Health Compared to the Ministry of Education
and Science?.................................................................................................................................. 13
Figure 7. Household Perceptions of Changes in Informal Payments from 2005 to 2008 ............ 14
                                                                                                                                                 Page v


Figure 8. Household Perceptions of Formal and Informal Out of Pocket Health Care Costs in
2008............................................................................................................................................... 15
Figure 9. Change in Reported Payment of Informal Gifts to Medical Staff, 2002-2008, by
Quintile of Household Consumption ............................................................................................ 16
Figure 10. Poverty Impact of Out of Pocket Payments for Health Services................................. 23


                                                                Appendix One

Table A.1.1. Population Catchment Area of PHC Facilities, by Region, 2003 ............................ 32
Table A.1.2. Regional Distribution of Hospitals Providing Inpatient Care in Albania, 2003 ...... 32
Table A.1.3. Respondent‟s Priority for Improvement in their Health Facility ............................. 33
Table A.1.4. Responses to LSMS Question “If this health facility received LEK 10 million, on
which activity would you want most of it spent?� ........................................................................ 33
Table A.1.5. Responses to LSMS Question “In your opinion, is the situation better or worse in
terms of corruption cases [in the health sector] comparing today and three years ago?� ............. 34
Table A.1 6. Responses to LSMS Question “Which one of the following do you agree with in
terms of unofficial out-of-pocket fees paid at health facilities in Alba?� ..................................... 34
Table A.1.7. Share of Patients who Made Informal Payments for Outpatient Services 2002-2008
....................................................................................................................................................... 35
Table A.1. 8. Share of Patients who Made Informal Payments for Inpatient Services 2002-2008
....................................................................................................................................................... 35
Table A.1.9. The mean of the amount paid informally per outpatient visit in Albanian Leks
(series with 2002 prices ................................................................................................................ 36
Table A.1.10. The mean of the amount paid informally per day hospitalized in Albanian Leks
(series with 2002 prices) ............................................................................................................... 36
Table A.1.11. Respondents‟ Evaluation of Health Care Facilities in 200 .................................... 36
Table A.1.12. Respondents‟ Evaluation of the Health Care Costs in 2008 .................................. 38


                                                                Appendix Two

1. 1. Determinants of informal gifts .............................................................................................. 39
1. 2. Impact over the years ............................................................................................................ 40


Table A2.1. Two Stage Slection Models for Outpatient and Inpatient Services .......................... 42
Table A2.2. Logit Results (dependent variable =1 if the individual participates in the next year
survey)........................................................................................................................................... 43
Table A2.3. Informal Payments in Outpatient Services: Results from Tobit & Heckman Selection
Models........................................................................................................................................... 45
Table A2.4. Informal Payments in Outpatient Services: Changes Over Time in the ................... 46
Table A2.5. Informal Payments in Outpatient Services: Changes Over Time in Individual
Characteristics Significantly Associated with Making Informal Payments ................................. 47
Table A2.6. Informal Payments in Inpatient Services: Results from Tobit & Heckman Selection
Models........................................................................................................................................... 48
                                                                                                                             Page vi


Table A2.7. Informal Payments in Inpatient Services: Changes Over Time in the Likelihood of
Making Informal Payments........................................................................................................... 49
Table A2.8. Informal Payments for Inpatient Services: Changes Over Time in Individual
Characteristics Significantly Associated with Making Informal Payments ................................. 50


Figure A2.1. Histograms from the predicted scores from the logit models used for matching .... 44
                                                                                                              Page 1




Part I. Introduction, Motivation and Summary of Findings

1.a. Paying Out of Pocket: Financial Vulnerability and Obstacles to Good Health

1.      In many countries absent or poorly functioning prepayment mechanisms for health
care expose families to the financial risks associated with accidents and sickness. In these
countries, a large share of the health services people need have to be paid for out-of-pocket,
sometimes up front at the point of service. Although expecting households to make some
financial contribution for their health care is reasonable --even in wealthy countries with
sophisticated public and private health insurance, and particularly for frequently occurring
conditions that are inexpensive to remedy-- an over reliance on out-of-pocket payments for
health care can lead to impoverishment, particularly for serious, less-frequently occurring
conditions for which the costs of treatment can quickly mount. Health care costs can even rise to
“catastrophic� levels when they force families to significantly compromise their current and
future standard of living, or to divert resources from other basic needs, such education, nutrition
and housing. And if prospective charges for care discourage people from visiting their family
doctor, getting immunized or seeking early treatment when they suspect a health problem, their
financial decisions can have dangerous consequences for public health.

2.      The out-of-pocket costs of health treatment can divert resources from other
spending critical to building and sustaining human capital, increasing the risk of chronic
poverty. In health economics there isn‟t yet a uniformly accepted threshold amount of out-of-
pocket spending that triggers alarm or that unambiguously motivates a policy response.
Naturally, concern rises when families spend a large enough fraction of their budget on health
care as to deprive them of other goods and services. Researchers of this topic have used varying
thresholds from 10 to 25 per cent of total household expenditure, or 40 per cent of “ability to
pay�, that is their total spending minus food expenditures.1 However, concern quickly grows
when households exhaust precautionary savings, and are forced to sell assets, or take loans to
finance the treatment they need. Concern becomes acute when out-of-pocket spending for health
forces households to divest from human capital. And because uninsured health care events can
lead to a loss of income from reduced labour supply or lower productivity, households can suffer
long-term consequences pushing them into a cycle of poverty that can last generations (Baeza
and Packard, 2005).

3.     In the formerly socialist countries of Europe, and especially in the countries of the
Western Balkans, a significant share of out-of-pocket spending for health consists of
“informal�, “under-the-table� or “envelope� payments to physicians and nurses.2
Extensive prior research has documented informal payments, sought to explain the phenomenon,
and shown the deteriorating impact they can have on social welfare. 3 While cross country

1
  See for example Berki 1986, Russell 2004, Wagstaff and Doorslaer 2003, Xu et al. 2003, etc.
2
  According to the definition proposed by Lewis (1999) informal payments in health care are payments to individual
and institutional providers in-kind or cash that are outside the official payment channels, or are purchases that are
meant to be covered by the health-care system.
3
  See for example Gaal, 2004; Delcheva et al., 1997; Vian et al., 2005; Abel-Smith and Falkingham, 1996;
Hotchkiss et al., 2004; Lewis, 2007; Ensor, 2004; Kaser, 1976, Albert et al., 1992; World Bank, 2006.
                                                                                                           Page 2


comparisons have to be interpreted with caution, recent studies by the World Bank show that in
Albania -as a share of total out-of-pocket spending for health- informal payments to medical staff
make up a relatively small portion, when compared informal payments in to neighbouring Serbia
and Kosovo. However, when this finding is examined alongside the much larger share of out-of-
pocket spending in total financing of health in Albania relative in Serbia and Kosovo, and the
much higher incidence of informal payments, concern that these payments are further raising
barriers to care and increasing the financial vulnerability of families grows.4

4.      Does out-of-pocket health spending impoverish in Albania? What weight do
informal payments have in increasing the burden of out-of-pocket health spending? And
why are informal payments so prevalent in Albania‟s health system? These are the
questions that this report attempts to answer. As well as increasing out-of-pocket spending,
informal payments constitute an important part of physicians‟ incomes. As such, many
researchers argue that they contribute to filling a gap in health care financing created by a lack of
adequate resources allocated to public health systems in state budgets. Some specialists have
argued that introducing formal payments, improving the reward mechanisms for medical staff,
increasing providers‟ accountability and empowering patients‟ “voice� can reduce the incidence
of informal payments (Lewis and Pettersson, 2009), although the actual impact of these measures
remains largely untested.

5.       The authorities have even undertaken some measures to make legitimate health
charges more transparent in public health centres and hospitals. However, plans to evaluate
the impact of these measures in particular, and to launch additional reforms to make informal
payments less prevalent, have not yet been made. Where this report can add value, therefore, is:
first, to quantify the importance of out-of-pocket health spending in its various forms; second, to
measure the extent of informal payments and how this varies across types of care, households,
and different regions of the country; and third, where possible to identify what are the most
likely explanations for the problem.

6.      In order to reduce the negative impact on households of out-of-pocket payments for
health and informal payments in particular, the Government is increasing transparency in
the health sector and guaranteeing access to basic care to all. As an initial step toward this
objective, the Government making changes to the health insurance system to grant coverage to
all beneficiaries of the Ndihma Ekonomike, Albania‟s main social assistance program for poor
households. As the analysis in this report shows, informal payments are partly encouraged by an
unclear copayments policy, which often blurs the distinction between formal and informal
payments. Informal payments are also associated with a lax definition of the type of services
that health insurance should cover. Experience elsewhere has shown that it is possible to reduce
informal payments when this issue is addressed as an integral part of health finance reforms of
the type that Albania is undertaking. These reforms set clear limits on what public resources
allocated for health can finance. Defining an explicit package of benefits in this way can help to
improve equity in access and increase accountability for the services specified in the package; as
patients are aware of what services they are entitled to receive and at what prices, the scope for
informal payments is reduced. Definition of an explicit benefit package of health insurance

4
 For prior studies on health access problems created by informal payments and out-of-pocket payments generally,
see Delcheva at al., 1997; Anon, 1999; Falkingham, 1998/99.
                                                                                              Page 3


coverage is a recent key change in Albania‟s health insurance legislation that is likely to have a
strong positive impact. Once the benefit package is created by law, secondary legislation will
define the specific contents of the package and how much it will cost. The current Health
Systems Modernization project would support the design and costing of this package. As is
common in many countries that adopt an explicit package of benefits, a system of formal
copayments would be introduced, including provisions to protect low-income groups by
exempting them from copayments based on some kind of means testing or providing them with
an additional cash benefit to face the increased cost of services. Other measures could be
considered to reduce the extent of informal payments, like allowing health providers collecting
copayments to retain at least part of those revenues, and implementing information campaigns to
inform the population about their rights.

7.      The analysis in this report covers the period 2002-2008 and for this reason and does
not take into account important measures taken since such as changes to health insurance.
The report exploits a specially designed Health Governance and Accountability module that was
added to the 2008 wave of the Albania LSMS to gain additional insight into the determinants of
informal payments and household perceptions and experiences of governance in the health sector
more generally. From the outset, it is important to point out the limitations of the analysis
presented in this report. The data from the new LSMS module cannot be used to establish causal
links to judge the outcomes of any particular policy. Some measures have been taken to try to
overcome these limitations using statistical techniques, but these cannot substitute for developing
a strategy to evaluate the impact of structural reforms such as the new health insurance law and
putting it in place before their implementation.

8.       This report is structured in four parts. Following the motivation, introduction and
summary of the report‟s findings in Part I, Part II sets the institutional context for readers not
already familiar with Albania‟s health system, and presents data showing public perceptions of
health care and answers to subjective questions included in the Health Governance and
Accountability module of the 2008 Albania LSMS. These data have been combined with prior
waves of the LSMS to show how the incidence of informal payments has changed in recent
years. Part III shows the impact out-of-pocket payments –both formal and informal- have on
household consumption and the incidence of poverty. Part IV presents the results of statistical
analysis, using techniques to match like-households over time, in order to identify significant
factors at the household level that are associated with informal payments and how these have
changed during a period of structural changes in the health system. The most important findings
in this report can be summarized below.

1.b. Principal Findings: Out of Pocket Spending, Informal Payments and Impoverishment in
Albania

9.      Although out of pocket spending for health services in Albania declined overall, the
poorest households remain the most financially vulnerable to the cost of health care. In
2002 about 23 per cent of the population paid out-of-pocket costs for health care that exceeded
10 per cent of their total per capita budget. Encouragingly, the incidence declined to 17.6 per
cent in 2005 and further to 13.3 in 2008. This decline is most likely related to the general fall in
poverty during the period, since the actual amount people paid for health services out of pocket
increased in real terms. However, the income gains for households in the poorest quintile were
                                                                                           Page 4


not as great as for other groups. Throughout the period 2002-2008, the level of out of pocket
payments made by people from households in the lowest quintile were more likely to reach
catastrophic levels (using the 10 per cent threshold). People in this category experienced a less
rapid decline in the incidence of catastrophic out of pocket payments. The share of individuals
from households in the lowest quintile who paid more than 10 per cent of their total spending on
out-of pocket charges for health services, was about 30 per cent in 2002 and 2005 and declined
to 20 per cent in 2008, less than the decline for wealthier groups. Even when alternative
spending thresholds are used the incidence of catastrophic out-of-pocket expenditures continues
to be much higher for households in the lowest quintiles.

10.    Out-of-pocket health expenditures contribute to poverty among Albanian
households. While the poverty headcount decreased from 2002 to 2008, when out of pocket
payments are taken into account, the post-payment poverty headcount is much higher. Taking
out of pocket spending on health into account, poverty increases by 6.5 percentage points in
2002, by 5.2 in 2005 and by 1.7 in 2008.

11.     Over the six year period analyzed in this report, the overall incidence of informal
payments declined. Contrary to Albanians‟ general perceptions of malfeasance in the health
sector, this report finds that the incidence of informal payments captured by the LSMS survey
actually declined along with out of pocket payments generally for the population as a whole.
However, the prevalence of informal payments differed significantly between outpatient and
inpatient services. Informal payments were more limited in outpatient than in inpatient care:
28.1 per cent of outpatients made informal payments in 2002, and 19 percent made informal
payments in 2008. For inpatient services the corresponding numbers are 59.7 per cent in 2002
and 43.9 per cent in 2008.

12.     But just as the poorest patients are still more likely to pay out of pocket, they are
also more likely to pay informally for health services. Informal payments for health services
are widely reported across households from all quintiles of consumption. Although there was a
general decrease in reported informal payments in the period from 2005 to 2008, their incidence
declined less among the poorest households. In fact for outpatient services from 2005 and 2008
respondents in the poorest quintile report the highest incidence of making informal payments.
The same holds, although to a lesser extent, for inpatient services. While the absolute value of
informal payments per capita increased substantially over the years from 220 lek in 2002 to 384
lek in 2008 -a real increase of almost 75 percent - the amount that a person from a household in
the lowest quintile pays informally increased almost five times over the same period (from 121
leks in 2002 to 777 leks in 2008).

13.     The incidence of informal payments for inpatient services declined less and
payments were -on average- much higher than for outpatient services. The average amounts
paid informally for inpatient care per day, were higher than those paid for outpatient care per
visit. Although there was a decline in real terms in the average amount paid informally for
outpatient services, this was accompanied by an increase in the amount of informal payments for
inpatient services between 2002 and 2005, which was followed by only a slight decrease
between 2005 and 2008. As with the incidence of informal payments, people in poorer
consumption quintiles bore the heaviest burden: in 2008 households in the poorest quintile (for
outpatient services) and the second poorest quintile (for inpatient services), made informal
                                                                                                     Page 5


payments that were higher than the average amount and also higher than the informal payments
made by households in all other quintiles of the consumption distribution.



1.c. The Prevalence Informal Payments in Albania: Summary of Findings

14.     The prevalence of informal payments in Albania‟s health system is difficult to
explain with precision. Instead of treating informal payments purely as a coping mechanism
that underpaid medical staff are forced to resort to, they can also be examined as an opportunity
response. In settings where structures for monitoring and accountability are absent or weak, and
where medical staff face little likelihood of detection and sanction, the gains from accepting
under-the-table payments far exceed the income and reputation losses of malfeasance. Medical
staff have powerful information advantages that can be used to elicit “gifts� as long as this gap
between gains and losses remains. Indeed, some experts propose that the incidence and amount
of informal payments can be used as proxy indicators (along with others) to measure the degree
to which governance structures are failing in health systems (Lewis and Petterssen, 2009).

15.    In Europe and Central Asia, informal payments originate mainly from the legacies
of „Semashko‟ health systems that prevailed in most socialist planned economies. The lack of
resources for health care, cultural norms, and weak governance have all been identified as
contributing factors to the prevalence of informal payments (Gaal and McKee, 2005). From the
academic literature, three main explanatory models have been proposed to explain informal
payments, and are hotly debated (Tomini and Maarse 2009): social-cultural norms;
underfunding; and weak governance.

                 The “social-cultural norms� model refers to the practice of gratitude
                  payments observed in many countries. Patients express their gratitude
                  by giving small gifts or payments to physicians and other medical staff.
                  The culture of gift giving is deeply rooted in the behaviour of patients in
                  more cohesive societies that place relatively greater importance on
                  reciprocity. Some researchers have argued that these gifts positively
                  influence patient-physicians relations.5

                 The “underfunding� model links the occurrence and the amounts
                  paid informally to a lack of resources in the health care system.
                  Across Central and Eastern Europe, the transition from a planned to a
                  market economy brought real declines in government health spending.6
                  Oversized and overstaffed hospitals became a burden on government
                  budgets and in many transition countries allocations for health are still
                  low. The lack of resources for equipment, medication and staff, could be
                  motivating hospital administrative and medical staff to elicit informal




5
    See for example Balabanova and McKee, 2002.
6
    Lewis 2000; Ensor 1997; World Bank 1996; Ensor and Savelyeva 1998; Balabanova and McKee, 2005.
                                                                                                            Page 6


                 payments.7 In these settings, patients often pay informally to “jump the
                 queue�, or to receive higher quality services.8

                The “weak governance� model links informal payments to a lack of
                 effective control and accountability in the health sector. Missing or
                 ineffective structures for monitoring service provision and holding
                 providers accountable create an environment that can encourage
                 malfeasance among medical staff. In the absence of effective regulatory
                 and control mechanisms medical professionals can use their superior
                 information and bargaining power to increase their earnings. Indeed, if
                 informal payments total to an amount much greater than a nurse or
                 physician‟s salary, it becomes difficult to consider these solely as “cost
                 contributions� or a coping strategy to make up for underfunding from
                 formal sources (Ensor, 2004).9.

16.     While the academic literature remains undecided whether informal payments are
more a reflection of social norms, resource shortages, or weak governance, Albanian‟s have
clear views on the subject. Among respondents to the 2008 LSMS, 85 to 91 percent agreed or
strongly agreed that informal payments should be eliminated and health staff should be penalized
for requesting payments and gifts. There is a higher incidence (two times higher) and higher
amounts of informal payments for inpatient services than for outpatient services (reported
informal payments for inpatient services can be three to five times higher per day hospitalized).
The higher incidence and the amount of “gifts� for inpatient services could reflect deeper
gratitude given the greater severity of health problems treated in hospital than in out-patient care.
However, the amounts paid are high relative to the average wage in the country, and bear little
relation to the economic means of the “giver�. In fact, for both inpatient and outpatient services,
patients from households in the lower quintiles are more likely to make informal payments and
also to pay higher amounts.

17.     Resource scarcity -underfunded facilities; lack of materials; and underpaid staff-
are likely to be important drivers of informal payments. Where informal payments follow a
“fee-for-service� pattern they could be helping to close the gap created by lack of resources from
formal channels - such as allocations from the government‟s budget, or co-payments that have
been set too low. However, supporting evidence for the underfunding explanation that can be
drawn from the LSMS data is mixed. Patients from mountainous areas and from Tirana
generally pay lower amounts in informal payments -both for outpatient and inpatient services-
than do those living in central or coastal areas. This finding is supportive of the underfunding
hypothesis, as Tirana has the highest concentration of resources (for example of hospital beds, at
401 per 100,000 inhabitants, as compared to the national average of 303 per 100,000
inhabitants). However, these resources are mostly concentrated in tertiary care hospitals treating
people from different regions. In 2003 the number of inhabitants per PHC facility in Tirana was
3005, much higher than the national average of 1440. So even though they are endowed with

7
  Lewis, 2000; World Bank, 1995.
8
  Lewis, 2000; Viana et al., 2006; Liaropoulos et al., 2008; Ensor, 2004; Balabanova and McKee, 2005, Belli et al.,
2001.
9
  See Gaal and McKee, 2005, Ensor and Witter, 2001; Lewis, 2002, Miller et al,, 2000; Thompson and Xavier, 2002.
                                                                                             Page 7


more resources, these are far more intensively used, giving rise at times to scarcity and rationing
comparable to other parts of the country. Furthermore, medical staff are typically paid less in
Tirana than those working in mountainous areas, where higher formal remuneration is given to
compensate for their rural assignments. The differences in the incidence and amounts paid
informally across regions could also simply reflect local conditions: mountainous areas have the
highest poverty rates in the country.

18.    People in the poorest households are more likely pay informally for health care than
those with greater information and means to hold providers accountable. Households in the
lowest consumption quintiles bear the heaviest cost in informal payments for health services.
These groups are more likely to make informal payments and also to pay higher amounts both in
absolute terms and relative to their total consumption expenditure. Since people in these groups
are more likely to lack information, and are less likely to have access to mechanisms with which
to hold service providers accountable, the pattern of informal payments appears to indicate that
weak governance could be significantly contributing to the problem.

19.    Tracking the incidence and magnitude of informal payments over time, although
underfunding is a problem, the persistence of informal payments for the poorest suggests
weak governance may also be important. In the last section of this report, the data from the
2002, 2005, and 2008 waves of the Albania LSMS are pooled and examined using statistical
matching techniques that allow an examination of the individual characteristics of people who
make informal payments for health services and how these have changed in recent years. The
findings suggest that although scarcity of resources in the health system is a contributing factor,
weak governance is likely also to be contributing to the prevalence of informal payments
Albania‟s health sector.
                                                                                                             Page 8




Part II. Albania‟s Health Sector: Structure, Image and Household
Experience

2.a. Albania’s Health Sector

20.     The health care system in Albania has its roots in the Soviet “Semashko� model, and has
suffered from many of the same problems that affect health service delivery other Central and
Eastern European countries, such as over-supply of facilities and medical doctors, low wages
and overly high rates of specialization (Healy and McKee, 1997; Lewis, 2002). Under the
planned economies, the health sector was underfinanced as a “non-productive sector� (Nuri and
Tragakes, 2002). The funds allocated for investments in health technology were historically low.
The Government‟s priority in health was to provide universal access to primary and secondary
care. This policy led to the construction of a large network of health posts or health care centres,
and a relatively large number of local and regional hospitals, which were typically overstaffed
and relied mostly on outdated equipment.

21.     The reforms to the health sector that followed immediately after the fall of the communist
regime focused mostly on reorganizing responsibilities over health care centres. During the
1990s some administrative responsibilities and ownership of many primary health care facilities
were shifted to the local level. However, human resource policies and financing for hospitals
remains centralized. Presently, most care centres and clinics are owned by local governments
while only hospitals remain owned by the Ministry of Health (MoH).10 The Primary Health Care
(PHC) policy developed by the Government in 1997 (World Bank, 2005) aimed to maintain a
health post and a health centre per commune. Outpatient care is delivered mainly through these
PHC facilities but also at polyclinics in urban areas. The main reform measures in PHC have
sought to transfer the financing of the sector to a Health Insurance Institute (HII). Recurrent
expenditures and wages of medical staff are now financed directly through the HII.

22.     Albania spent around 6.8 per cent of its GDP on health in 2008, the last year covered by
the analysis in this report. This level of spending is comparable to other countries in the region
(see Figure 1), however, only half of this amount is publicly financed. The share of out-of-pocket
expenditures remains high in Albania when compared to similar countries (Figure 9). For
households in the lowest quintile the share of out of pocket spending on health is as high as 50
per cent of the total monthly per capita expenditure per one episode. Such high levels of out-of-
pocket spending for health services can create barriers to access and, in some cases, prevent people
moving out of poverty.



10
  Hospitals are still primarily public. The MoH is the owner and administrator of all hospitals In 2008, there were
41 operative hospitals, of which: 4 university hospitals in Tirana, 11 regional hospitals, and 22 district hospitals
(MoH, 2009). Interventions in the hospital network have mainly sought to improve the infrastructure of the sector
and during 2002-2008 period of analysis little happened in terms of reforming the provider financing.
                                                                                                                                                                                                                                                                                                                 Page 9


Figure 1. Health Spending as a Percentage of GDP in 2008, Albania and Comparator
Countries
        12,0



        10,0



         8,0



         6,0



         4,0



         2,0



         0,0
               Turkmenistan




                                                                  Romania




                                                                                      Tajikistan




                                                                                                                                                            Ukraine




                                                                                                                                                                                FYR Macedonia
                                                                                                                                           Czech Republic
                                                                                                                                  Poland
                                                                            Estonia




                                                                                                                                                                                                                                Croatia




                                                                                                                                                                                                                                                     Georgia




                                                                                                                                                                                                                                                                            Greece




                                                                                                                                                                                                                                                                                                                  Moldova
                                                                                                                                                                                                                     Slovakia




                                                                                                                                                                                                                                                               Montenegro
                                           Kazakhstan




                                                                                                                                                                                                                                                                                              Bosnia and Herz.
                                                                                                                                                                                                           Hungary
                                                                                                               Latvia
                                                                                                   Lithuania




                                                                                                                                                                                                                                                                                     Serbia
                                                        Armenia




                                                                                                                                                                                                Bulgaria




                                                                                                                                                                                                                                          Slovenia
                                                                                                                        Belarus




                                                                                                                                                                      Albania
                              Azerbaijan




                                                                                                                        Source: WHO database

23.     Public health spending is financed with social health insurance contributions and general
tax revenues. But given the prevalence of informal employment relationships in the Albanian
economy, only about a third of the active labor force contributes to health insurance.11 The
Health Insurance Institute is responsible for contracting health services with PHCs and also
selected services from hospitals. Although the funding of PHC is done through the HII, most of
the funds (almost two thirds of the total budget) still come from the state budget. Funding for
hospitals in 2008 was still channelled through the MoH.




11
  Public health expenditures are financed by social health insurance contributions and revenues from tax
collections. The health insurance contributions are collected by the tax authority and amount to 3.4 per cent of the
gross salaries of formally employed individuals.
                                                                                                                                                                                                                                                                                         Page 10


Figure 2. Health Spending by Source in 2008, Albania and Selected Countries
(Out of pocket spending is the darkest portion of the bars)
         100%




          80%




          60%




          40%




          20%




           0%




                                                                                                                                                                                                                                    Moldova




                                                                                                                                                                                                                                                                            Azerbaijan
                                                                                                                                    FYR Macedonia
                                                                                                   Hungary
                Czech Republic




                                                                                                                                                                                              Latvia




                                                                                                                                                                                                                                                        Albania
                                 Slovenia




                                                                   Montenegro




                                                                                                             Slovakia




                                                                                                                                                                                                                 Bosnia and Herz.
                                                                                                                                                                                     Greece
                                            Croatia




                                                                                          Poland




                                                                                                                                                                 Serbia




                                                                                                                                                                                                       Ukraine




                                                                                                                                                                                                                                                                  Georgia
                                                                                                                                                    Kazakhstan




                                                                                                                                                                          Bulgaria




                                                                                                                                                                                                                                              Armenia
                                                                                                                        Lithuania
                                                                                Estonia
                                                      Romania




                                                                                                                                                                                                                                                                                         Tajikistan
                                                                Out-of-Pocket expenditure                    General government expenditure on health                                         Other private expenditure on health

                                                                                                               Source: WHO database

24.     Health insurance in Albania should cover most of the formal costs of primary health care
and all the costs of hospital care. Patients are formally required to pay small, fixed co-payments
per visit in PHC or for specialized treatments in hospital care, but until 2008 the amounts
required were still very low. Despite the fact that the law states that all the citizens 12 should be
covered by health insurance, surveys show that only about 40-45 per cent of the population
report having a booklet showing they are covered (World Bank, 2006). Moreover, there are
significant disparities in coverage across regions of Albania (e.g. only 20 per cent of the
population in mountainous areas is covered).

25.     The most prominent issue among households when it comes to health services is that of
high informal payments (see also Figure 3). Informal payments are the manifestation of weak
governance that Albanians experience on a frequent basis. The problem is recognized in the
National Health Sector Strategy (2004) as widespread both in inpatient and outpatient health
care. The World Bank‟s Albania Health Sector Review (2005) shows that informal payments
constituted almost one quarter of total costs paid in inpatient services and are also consistently
high in outpatient services.




12
   Apart from people paying health insurance through formal employment (or self declarations), the law provides
coverage also for other groups like; children below age one, unemployed, social assistance recipients, soldiers,
students, pensioners and war veterans, pregnant women, and other vulnerable categories.
                                                                                                                                                                                                                   Page 11


Figure 3. Reported Corruption and Bribes in the Health Sector in 2008 in Albania and
Comparator Countries
          80%



          70%



          60%



          50%



          40%



          30%



          20%



          10%



          0%




                                                                                                          Kosovo




                                                                                                                                            Moldova
                                                                            Turkey
                                Croatia

                                          Albania




                                                                                                Poland




                                                                                                                                                      Serbia




                                                                                                                                                                               Romania

                                                                                                                                                                                         Greece
                                                                                     Bulgaria




                                                                                                                   Ukraine




                                                                                                                                                                   Lithuania
                FYR Macedonia




                                                                                                                             Russian Fed.
                                                        Bosnia and Herz.




                                                                                                                                                                                                  Czech Republic
                                                    Extremely corrupt                                    Requested a bribe                                     Paid a bribe


                                                                           Source: Transparency International, 2008

26.      Reforms in the health sector from 2002-2008 focussed on primary health care and
outpatient services, consolidating the financing of the PHC through HII (but limited mainly in
the salaries of GPs) and transferring PHC capital investments to local governments. Up to 2008,
little had been done to reform inpatient care, although contracting of inpatient services through
HII was implemented in 2009-2010. Total health financing channeled through HII is steadily
increasing.
                                                                                                      Page 12




Figure 4. Monthly Salaries of GPs (in 000 Leks)
                60

                50

                40

                30

                20

                10

                 0
                      2000     2001    2002     2003   2004     2005    2006     2007     2008
                     Mounthly Salaries of GPs     Mounthly Salaries of GPs (real terms 2000 prices)
                          Source: Team‟s estimates based on World Bank (2006) and HII


27.    The measure many specialist believe is most likely to lower people‟s experience of
informal payments in the health sector, has been the increase in the salaries of medical staff.
Between 2002 and 2005, the increase mainly benefited GPs (paid through the HII), and the
average salary level of GPs increased between 2000 and 2008 both in nominal value and real
terms. Salaries of medical staff in the inpatient health sector have also risen over the same period
and especially since 2005. However, the average wage of medical specialists in hospitals still
remains below the average wage of the medical staff in PHCs.

2.b. Household Perceptions of Governance & Informal Payments


28.      Among Albania‟s public institutions, health institutions rank significantly low in public
perceptions governance. On a scale of 0 to 100, where 0 is very honest and 100 is very corrupt,
the public gave an average score of 62.4 to 20 institutions and groups in a survey collected by the
Institute for Development Research Alternatives (IDRA) in the first quarter of 2010, indicating
high levels of perceived corruption (IDRA, 2010). Religious leaders were perceived as the most
honest with a score of 30, and customs officials were the most corrupt with a score of 84. In
Figure 5 the corresponding scores for health institutions are presented next to those for education
institutions as a benchmark, given the similar importance of education to families. Medical
doctors and teachers lie in between two extremes, but on very different ends of the spectrum.
Teachers scored 46, while medical doctors scored 79 in the IDRA survey of perceptions. These
perceptions vary between respondents who work in the public sector and those in the general
population. However, they have remained fairly stable in the five years these surveys have been
conducted.
                                                                                                                                                                                                                                              Page 13


Figure 5. Are Medical Doctors and Teachers Honest or Corrupt?
                                                                        a. Perceptions Held by the General Public                                                                           b. Perceptions Held by Public Sector Workers




                                                                                                                                       Very honest (0), Very corrupt (100)
                 Very honest (0), Very corrupt (100)

                                                       100                                                                                                                   100
                                                        90                                                                                                                    90
                                                        80                                                                                                                    80
                                                        70                                                                                                                    70
                                                        60                                                                                                                    60
                                                        50                                                                                                                    50
                                                        40                                                                                                                    40
                                                        30                                                                                                                    30
                                                        20                                                                                                                    20
                                                        10                                                                                                                    10
                                                         0                                                                                                                     0
                                                              2005          2006           2008           2009           2010                                                       2005         2006          2008            2009          2010

                                                                        Public school teachers          Doctors                                                                             Public school teachers          Doctors

                                                                                          Source: Institute for Development Research Alternatives

29.     Consistent with improvements in governance in public administration generally,
perceptions of the government agencies charged with managing health and education services
have also improved. A transparency indicator, also calculated by IDRA, measures perceived
transparency, where 0 is not at all transparent and 100 is fully transparent. The average score for
public institutions was 38.7. The Ministry of Education and Science (MoES) scored above the
average with 44, and the Ministry of Health (MoH) scored 40. In both cases, there has been a
significant improvement in perceived transparency since 2005 (Figure 6).


Figure 6. How Transparent are the Ministry of Health Compared to the Ministry of
Education and Science?
                                                                      a. Perceptions Held by the General Public                                                                             b. Perceptions Held by Public Sector Workers
                                                        100                                                                                                                   100
                                                                                                                                  Not at all transparent (0), Fully
     Not at all transparent (0), Fully




                                                         90                                                                                                                    90
                                                         80                                                                                                                    80
                                                                                                                                         transparent (100)
            transparent (100)




                                                         70                                                                                                                    70
                                                         60                                                                                                                    60
                                                         50                                                                                                                    50
                                                         40                                                                                                                    40
                                                         30                                                                                                                    30
                                                         20                                                                                                                    20
                                                         10                                                                                                                    10
                                                          0                                                                                                                     0
                                                               2005           2006           2008            2009          2010                                                      2005          2006          2008            2009         2010

                                                                Ministry of Health               Ministry of Education                                                                 Ministry of Health            Ministry of Education


                                                                                          Source: Institute for Development Research Alternatives


30.     In contrast with the education sector, there have been no specific reform initiatives
directly aimed at improving governance in the health sector, other than the increase in salaries of
medical staff. Motivated by concern for the prevalence of informal payments in the sector, a
special Health Governance and Accountability module was added to the 2008 wave of the
Albania LSMS. Starting with questions about how respondents perceive “corruption� to have
changed in the three year period prior to the survey13 (examples such as informal payments, other
bribery, collusive practices in procurement, etc. were included in the question), there is strongly
significant variation across groups of different consumption levels; according to what type of
health facility they visited; and where in the country they live.

13
     “In your opinion, how has the situation of corruption in the health sector changed in the last three years?�
                                                                                                                                                                       Page 14


31.     Most of the significant variation in responses across consumption groups indicates little
had changed in people‟s perceptions. Indeed, 11 percent of respondents from households in
consumption quintile three indicated that the prevalence of informal payments had worsened.
This dismal pattern in people‟s perceptions is even more notable in the significant variation in
responses across different health facilities, with the largest share or respondents who indicate a
change, reporting a worsening of conditions in private hospitals. However, the most significant
variation is in responses across Albania‟s regions. The largest number of people who report
improvements live in the Coastal region, but even there the majority report the prevalence of
informal payments is the same or has gotten worse. In the Central region, more people report the
situation getting better; in the Mountain region there are slightly fewer who report an
improvement than those who report a worsening; but in Tirana substantially more of those who
report that the situation has changed, say the prevalence of informal payments in the health
sector has gotten worse.

Figure 7. Household Perceptions of Changes in Informal Payments from 2005 to 2008
                       “In your opinion, how has the situation of corruption in the health sector changed in the last three years?�
  General perceptions of corruption in the health                                                                                 Corruption in health facilities
                      sector
                                        60                                              Percentage of respondents who used   60
   Percentage of respondents who used




                                        50                                                                                   50


                                        40                                                                                   40
                                                                                                  health facilities
             health facilities




                                        30                                                                                   30


                                        20                                                                                   20


                                                                                                                             10
                                        10

                                                                                                                              0
                                        0
                                                                                                                                   Better    The same   Worse       Unknown
                                             Better   The same   Worse   Unknown




                                                                  Source: Team estimates using LSMS 2008

32.    Compared to formally-defined and listed user fees/copayments, respondents clearly
consider the costs they have to pay informally for health services to be high: 45 percent of
respondents consider informal payments (including “gifts�) to be “high� and “very high�, while
only 30 percent consider formally defined user fees to be high and very high. The largest group
among those who consider costs to be high and very high, receive care from public polyclinics:
while 37 percent think formal user fees are high and very high, 48 percent think that informal
payments are high and very high.
                                                                                                                                                                                               Page 15


Figure 8. Household Perceptions of Formal and Informal Out of Pocket Health Care Costs
in 2008
                                            Official “formal� health charges                               Money and gifts demanded by health care staff
                                       35                                                                                                  35




                                                                                                  Percentage of respondents using health
  Percentage of respondents who used




                                       30                                                                                                  30

                                       25                                                                                                  25
            health services




                                                                                                                 facilities
                                       20                                                                                                  20

                                       15                                                                                                  15

                                       10                                                                                                  10

                                       5                                                                                                       5

                                       0                                                                                                       0
                                            Very high   High   Affordable      Low     None                                                        Very high   High   Affordable      Low      None




                                                                            Source: Team estimates using LSMS 2008

33.     As discussed previously, there is debate among specialists whether informal payments for
health services in developing and transition countries are more a reflection of social-cultural
norms (gift giving out of gratitude for professional services, in what are relatively more
homogenous and cohesive societies, that are strongly reliant on reciprocity); resource scarcity
(underfunding of facilities and underpayment of medical doctors); or weak governance made
more prevalent by absent of poorly functioning monitoring and accountability structures. This is
an active debate even among researchers who focus narrowly on the Western Balkans and
Albania in particular. However, most people who responded to the Albania LSMS have clear
and unambiguous views about informal payments. No matter what their position in the
distribution of respondents by consumption, 85 to 91 percent of those surveyed agreed or
strongly agreed that informal payments should be eliminated and health staff should be penalized
for requesting payments and gifts.

Table 1. Opinions About Informal Payments and How to Solve the Problem
 “Which one of the following statements do you agree with in terms of unofficial, private out-of-pocket fees paid at
                                          health facilities in Albania?�

                                                                                                                                                               Strongly            Neither agree
                                                          Strongly agree             Agree             Disagree
                                                                                                                                                               disagree            nor disagree
“Private fees must be eliminated and doctors/health care staff should be penalized for requesting such fees�
                                                               54                     38                                                   3                      2                     3
“The Government should increase salaries of health care staff and private fees will automatically be reduced�
                                                               31                     50                                                   12                     3                     4
“Unofficial private fees should be made official and publicly announced�
                                                               21                     29                                                   24                    19                     8


                                                                            Source: Team estimates using LSMS 2008

34.    There is greater difference of opinion as to what can be done to solve the problem of
informal payments. Some respondents support increasing the salaries of health care staff as a
way to reduce the likelihood they will ask for informal payment. Among the responses, there are
                                                                                                                                                                   Page 16


also indications of demand for the greater certainty that would come from making informal fees
official and providing people with prior information about what the health services will actually
cost them. The health sector authorities have made this information available in health centers
and hospitals in the months since the survey was collected. However, the deployment of this
measure was not done in a randomized manner that would easily allow evaluation of its impact.

2.c. From Perceptions to Experience of Informal Payments

35.     Shifting away from perceptions to households‟ reported experience seeking health
services, although the incidence of the informal payments remained high between 2002 and
2008, contrary to Albanians‟ general perceptions of “corruption� in the health sector, the
reported incidence of informal payments actually declined along with out of pocket payments
generally. However, the prevalence of informal payments differed significantly between
outpatient and inpatient services. Informal payments were more limited in outpatient than in
inpatient care: 28.1 per cent of outpatients made informal payments in 2002, and 19 percent
made informal payments in 2008. For inpatient services the corresponding numbers are 59.7 per
cent in 2002 and 43.9 per cent in 2008.

36.     Informal payments were widespread across households in different quintiles of
consumption. Although there was a general decrease in reported informal payments in the
period from 2005 to 2008, their incidence declined less among the poorest households. In fact
for outpatient services from 2005 and 2008 respondents in the poorest quintile report the highest
incidence of making informal payments. The same holds, although to a lesser extent, for
inpatient services.

Figure 9. Change in Reported Payment of Informal Gifts to Medical Staff, 2002-2008, by
Quintile of Household Consumption
                                                            a. Outpatient health care (per visit)
  Mean of the amount paid informally per outpatient visit                                       Mean of the amount paid informally per outpatient visit as
                                                                                                      share of per-capita household consumption
                              0                                                                                         0
                             -10
                                                                                                                      -0.2
                                                                                        Percentage point difference




                             -20
   Difference in 2002 Lek




                             -30
                                                                                                                      -0.4
                             -40
                             -50                                                                                      -0.6
                             -60
                                                                                                                      -0.8
                             -70
                             -80                                                                                       -1
                             -90
                            -100                                                                                      -1.2
                                   Poorest   2   3      4     Wealthiest       All                                           Poorest   2   3     4    Wealthiest       All
                                                                           households                                                                              households


                                                 b. Inpatient health care service (per day hospitalized)
             The mean of the amount paid informally per inpatient                                              The mean of the amount paid informally per inpatient
                                admission                                                                     admission as share of per-capita household consumption
                                                                                                                                                          Page 17


                           1000                                                                                   0
                            500                                                                                   -5




                                                                                   Percentage point difference
                              0                                                                                  -10
  Difference in 2002 Lek



                                                                                                                 -15
                            -500
                                                                                                                 -20
                           -1000
                                                                                                                 -25
                           -1500
                                                                                                                 -30
                           -2000
                                                                                                                 -35
                           -2500                                                                                 -40
                           -3000                                                                                 -45
                           -3500                                                                                 -50
                                   Poorest   2   3   4   Wealthiest       All                                          Poorest   2   3   4   Wealthiest       All
                                                                      households                                                                          households



                                                     Source: Team estimates using LSMS 2008

37.    The average amounts paid informally for inpatient care per day, were higher than those
paid for outpatient care per visit). Although there was a decline in real terms in the average
amount paid informally for outpatient services, this was accompanied by an increase in the
amount of informal payments for inpatient services between 2002 and 2005, which was followed
by only a slight decrease between 2005 and 2008. As with the incidence of informal payments,
people in poorer consumption quintiles bore the heaviest burden: in 2008 households in the
poorest quintile (for outpatient services) and the second poorest quintile (for inpatient services),
made informal payments that were higher than the average amount and also higher than the
informal payments made by households in all other quintiles.
                                                                                               Page 18




Part III. The Impoverishing Effects of Health Costs, Formal and
Informal
38.      In health economics there isn‟t yet a uniformly applied threshold of out-of-pocket
spending above which costs are considered “catastrophic�. Household spending on health care is
treated as catastrophic if it exceeds some fraction of household incomes or total expenditures,
usually measured over a year (Berki, 1986; Russell, 2004; Wagstaff and Doorslaer, 2003; Xu et
al., 2003). Concern rises when households spend a large enough fraction of their budget on
health care as to deprive the consumption of other goods and services. This concern grows when
households exhaust precautionary savings, and are forced to sell assets, or take loans (O‟Donnell
et al., 2005). Concern for the amount of out-of-pocket spending becomes acute when it forces
households to divest from human capital (for example, by delaying or stopping educational
attainment, and in the extreme, constraining food consumption). Because uninsured health care
events can lead to a loss of incomes from reduced labour supply or lower productivity,
households can suffer long-term consequences pushing them into a “trans-generational cycle� of
poverty (Baeza and Packard, 2006).

39.     To be fully appreciated, the impact of informal payments has to be set in a context of the
impact that all out-of-pocket health spending has on households. Previous analytical work on
out-of-pocket spending in Albania and other countries in the Western Balkans shows that
“catastrophic� out-of-pocket payments are common (Bredenkamp et al., 2008). In this section of
the report, the analysis is extended using the last three waves of the Albania LSMS.

Table 2. Average Per Capita Expenditures on Health and Non-Health Items
                                                                 Per capita                 Per capita
            Per capita  Per capita   Per capita
                                                    Per capita      total                       net
               gross      health      informal                                Overall per
                                                      health    expenditure                consumption
           consumption expenditures payments in                              capita health
                                                   expenditures s in private                  (health
              (health   on formal public health                              expenditures
                                                   on transport    health                    payments
           payment inc) payments    care services
                                                                   sector                       exc)
                                             Year 2002
Lowest
             4296.69       475.14       121.17        101.82       144.49      493.94       3802.75
quintile
   2         6230.03       659.04       165.96        188.14       183.04      653.82       5576.20
   3         7915.78       591.48       177.93        128.94       265.70      698.51       7217.28
   4        10317.47       777.19       229.42        248.82       236.42      780.42       9537.05
Highest
             18077.7      1041.31       318.72        286.27       471.78      1071.82     17005.89
quintile
 Total      11092.29       767.10       220.98        193.93       298.44      808.30      10284.00
                                              Year 2005
Lowest
             4394.93       496.82       152.64        104.60       157.97      519.44       3875.50
quintile
   2         6446.95       639.36       278.78        128.55       179.59      617.97       5828.98
   3         8698.88       749.24       267.69        127.79       271.39      763.44       7935.44
   4        11147.78       753.25       252.24        141.84       264.33      708.37      10439.42
Highest
            19054.34      1041.77       348.80        202.30       381.92      902.40      18151.94
quintile
 Total      10801.00       748.55       256.89        136.39       267.80      723.05      10077.95
                                              Year 2008
                                                                                                            Page 19


 Lowest
                5438.88         865.67          777.96         182.29          203.18         849.31         4589.57
 quintile
    2           7655.03        1039.79          208.58         150.78          327.82         866.62         6788.40
    3           9520.95         972.28          230.17         186.80          307.83         769.02         8751.93
    4          12837.11        1494.32          426.61         288.23          727.13        1463.46        11373.65
 Highest
               21422.69        1797.93          336.94         236.36          658.98        1520.30        19902.39
 quintile
  Total        11923.66        1269.22          384.54         208.19          473.96        1125.65        10798.01
Note: Expenditures are given as averages for each of the categories only for those households which have actually
spent for that particular category.
Per capita health expenditures on formal payments include out-of-pocket expenditures on medical fees, laboratory
works and drugs purchased in public outpatient and inpatient services (expenditures incurred in hospitals outside
from Albania are omitted in the table).
Informal payments in health care include gifts paid to medical staff in public outpatient and inpatient services.
Expenditures in private sector include out-of-pocket expenditures on medical fees, laboratory works, drugs
purchased and gifts to medical staff in private services.


40.     Average per capita spending on health and how this has changed from 2002 to 2008
(shown in 2002 lek) is shown in Table 2. The gross per capita consumption of households who
reported out-of-pocket payments, increased in real value from 11,092 lek in 2002 to 11,923 lek
in 2008. Total out of pocket spending for health was on average 808 lek in 2002 and increased
to 1,126 lek in 2008. The same trend is observed for almost all items of spending (formal
payments, informal payment, transport and expenditures on private health care providers and
pharmaceutical purchases ). Formal payments in the health sector increased the most over the
years. Formal health payments per capita increased by about 500 lek from 2002 to 2008 (from
767 lek to 1,269 lek) while spending on health provided by the private sector increased by 175
lek over the same period (from 298 leks to 473 lek). The absolute value of informal payments per
capita increased substantially over the years from 220 lek in 2002 to 384 lek in 2008, a real
increase of almost 75 percent.

41.     While the increase in the amount of total out-of-pocket payments in the period to 2008
seems worrying, a more worrying trend is the distribution of informal payments between
households in different quintiles of consumption expenditure. The amount that a person from a
household in the lowest quintiles pays informally has increased almost five times over the years
(from 121 leks in 2002 to 777 leks in 2008). During the same period the increase in the amount
paid informally by people in households in higher quintiles was far more moderate.

Table 3. Financing Budget Shares On Health And Non-Health Item
            Per capita Per capita                                Per capita
                                     Per capita    Per capita
           consumptio     health                                  private     Overall per     Per capita
                                      informal       health
            n gross of expenditures                             expenditures capita health consumption net
                                    payments in expenditures on
              health    on formal                                in private expenditures of payments
                                    health care    transport
            payments payments                                      sector
                                                 Year 2002
Lowest
              0.04          0.08          0.07            0.09            0.07           0.07            0.04
quintile
   2          0.08          0.16          0.12            0.19            0.10           0.14            0.07
   3          0.11          0.15          0.17            0.17            0.18           0.16            0.11
   4          0.20          0.22          0.24            0.26            0.19           0.22            0.20
Highest
              0.57          0.39          0.40            0.29            0.47           0.41            0.57
quintile
                                                                                                            Page 20


  Total      100.00        100.00        100.00         100.00           100.00         100.00          100.00
                                                    Year 2005
Lowest
              0.06          0.11          0.10            0.15            0.07           0.10            0.06
quintile
   2          0.11          0.17          0.23            0.21            0.12           0.16            0.11
   3          0.15          0.20          0.20            0.21            0.19           0.20            0.15
   4          0.23          0.21          0.16            0.21            0.22           0.21            0.23
Highest
              0.45          0.31          0.32            0.22            0.40           0.33            0.46
quintile
 Total       100.00        100.00        100.00         100.00           100.00         100.00          100.00
                                                    Year 2008
Lowest
              0.08          0.13          0.30            0.21            0.08           0.13            0.08
quintile
   2          0.12          0.13          0.12            0.14            0.12           0.13            0.12
   3          0.16          0.18          0.13            0.16            0.15           0.17            0.16
   4          0.22          0.25          0.24            0.25            0.34           0.28            0.22
Highest
              0.42          0.31          0.20            0.24            0.30           0.30            0.43
quintile
 Total       100.00        100.00        100.00          100.00          100.00         100.00          100.00
Note: Per capita health expenditures on formal payments include out-of-pocket expenditures on medical fees,
laboratory works and drugs purchased in public outpatient and inpatient services (expenditures incurred in hospitals
outside from Albania are omitted in the table).
Informal payments in health care include gifts paid to medical staff in public outpatient and inpatient services.
Expenditures in private sector include out-of-pocket expenditures on medical fees, laboratory works, drugs
purchased and gifts to medical staff in private services.

42.     Table 3 shows the distribution of health or non-health expenditures across households in
each consumption quintile. Generally the poorest -people in households in the the lowest
consumption quintile- spend much more in non-direct health costs, i.e. transportation or informal
payments, when compared with other items. This category of spending has increased
substantially over the years for households in the lowest quintile. In 2008 almost 30 per cent of
the total amount of informal payments and 21 per cent of all transportation costs were paid by
people from households in the lowest quintile. These shares have grown from 7 and 9 per cent,
respectively, in 2002. In 2008 people from the poorest households faced higher out-of-pocket
payments and faced greater barriers to health care than they did in 2002.

43.     When total expenditure is used as the denominator the most common threshold employed
in the economic literature to measure “catastrophic� spending is 10 percent – the threshold at
which prior research has found households can be forced to sacrifice basic necessities (Pradhan
and Prescott, 2002; Rason, 2002; Wagstaff and van Doorslaer, 2003). Table 4 presents
alternative measures out-of-pocket payments for health care for each quintile of household
consumption, over the three waves of the Albania LSMS. The headcount (incidence) in this table
is the share of individuals for whom the proportion of out-of-pocket payments for health (shown
as a percentage of total spending), exceeds a number of set thresholds, ranging from 5 per cent to
25 per cent.

44.    In 2002, 22.6 per cent of the population paid out-of-pocket costs for health services that
exceeded 10 per cent of their total per capita budget. Encouragingly, this incidence declined to
17.6 per cent in 2005, and further to 13.3 in 2008. The largest drop occurred between 2005 and
2008 and may be related to the general fall in poverty during that period (INSTAT, 2009).
However, as observed previously, households in the lowest quintile seemed to have suffered the
most from catastrophic payments (according to this threshold) throughout the period of analysis.
                                                                                             Page 21


The share of people from households in the lowest quintiles who paid more than 10 per cent of
their total expenditures on out-of pocket payments for health services, was about 29-30 per cent
in 2002 and 2005 and decreased only to 20 per cent in 2008.

Table 4. Incidence and Impact of Catastrophic Health Payments with respect to total
expenditure.
   CATASTROPHIC PAYMENTS                            THRESHOLD BUDGET SHARE
           MEASURES               Thresh. 5%     Thresh. 10%    Thresh. 15%       Thresh. 25%
                                           Year 2002
Headcount (H)
    Lowest quintile                  45.2               29.9         20.6             12.2
    2                                41.1               26.7         18.2              8.9
    3                                37.2               24.1         15.1              8.6
    4                                33.8               20.6         13.4              6.5
    Highest quintile                 25.3               11.7          7.3              4.1
    Total                            36.5               22.6         14.9              8.1
Overshoot (O)                         5.4                4.0          3.1              2.0
Mean positive overshoot (MPO)        45.2               29.9         20.6             12.2
                                            Year 2005
Headcount (H)
    Lowest quintile                  42.9               28.7         20.6             13.2
    2                                39.1               22.3         14.4              5.8
    3                                33.6               18.5         12.8              6.4
    4                                25.6               11.8          6.7              2.8
    Highest quintile                 17.2                6.8          3.5              1.7
    Total                            31.6               17.6         11.6              6.0
Overshoot (O)                         3.9                2.7          2.0              1.2
Mean positive overshoot (MPO)        12.4               15.6         17.4             19.9
                                            Year 2008
Headcount (H)
    Lowest quintile                  28.9               20.0         13.8             10.0
    2                                25.0               14.8         10.7              6.3
    3                                22.4               12.0          7.9              4.5
    4                                20.1               11.3          7.9              3.5
    Highest quintile                 15.9                8.4          5.0              2.9
    Total                            22.5               13.3          9.1              5.4
Overshoot (O)                         5.0                4.1          3.5              2.8
Mean positive overshoot (MPO)        22.1               30.8         38.9             52.2


45.     Table 4 also shows measures of “catastrophic overshoot� for each of the years examined.
This measure represents the average extent by which payments (as a proportion of total
expenditure) exceed the respective threshold. In other words it measures the intensity of
catastrophic out-of-pocket payments. This measure is important as it complements the
headcount measure of catastrophic payments (the incidence). The intensity of catastrophic
payments drops as the threshold is raised from 5 to 25 per cent of total consumption expenditures
throughout all years. As these numbers show, the average out-of-pocket amount paid as a share
of total expenditures has increased sharply over the period 2005 and 2008. This reinforces the
finding regarding the dramatic increase in certain categories like the formal payments and the
expenditures on private health care.
                                                                                           Page 22




Table 5. Poverty Based on Consumption, Gross and Net of Spending on Health
                                                                            NORMALISED MEAN
                             POVERTY                       NORMALISED
                                           POVERTY GAP                      POSITIVE POVERTY
                             HEADCOUNT                     POVERTY GAP
                                                                                   GAP
                                           Year 2002
Pre-payment headcount          26.58           300.25           6.14               23.10
Post-payment headcount         33.07           489.49          10.01               30.26
Poverty impact- percentage
                                6.49           189.24           3.87               7.17
point change
Percentage change              24.42           63.03           63.03               31.03
                                           Year 2005
Pre-payment headcount          19.45           215.71           4.41               22.68
Post-payment headcount         23.79           336.57           6.88               28.92
Poverty impact- percentage
                                4.34           120.86           2.47               6.25
point change
Percentage change              22.33           56.03           56.03               27.54
                                           Year 2008
Pre-payment headcount          12.19           110.18           2.25               18.47
Post-payment headcount         15.80           296.11           6.05               38.31
Poverty impact- percentage
                                3.61           185.93           3.80               19.84
point change
Percentage change              29.60           168.75          168.75             107.38

46.     Table 5 shows that out-of-pocket health expenditures have increased the percentage of
poor Albanian household. The poverty headcount has decreased from 2002 to 2008, but the post-
health payment poverty headcounts are much higher if out-of-pocket spending for health care is
taken into account. Poverty increases by 6.49 percentage points in 2002, by 5.23 in 2005 and by
1.69 in 2008. The poverty gap also increases when out-of-pocket spending for health is
accounted for.

47.    The catastrophic impact that out-of pocket expenditures have on households is illustrated
in Figure 10. The graphs show the impact of out-of-pocket spending pre and post expenditure
per capita, and comparing it with the food poverty line (the horizontal line representing amounts
3,047 lek per capita). Out-of-pocket expenditures for health services can drive individuals into
poverty. This is true not only for the lowest quintiles but also for all the others.
                                                                                                                      Page 23


Figure 10. Poverty Impact of Out of Pocket Payments for Health Services
                                                        Poverty impact on Pen's Parade
                                                              Per capita consumption in 2002




                                50000


                                           0
                    -50000




                                               0      20              40             60           80            100
                                                            Cumulative % of pop. ranked by income

                                                   Pre-payment consumption            Food poverty line = 3047ALL
                                                   Post-payment consumption


                                          Year 2002
                                                           Poverty impact on Pen's Parade
                                                              Per capita consumption in 2005
                                  50000


                                           0
                           -50000




                                               0       20             40             60           80           100
                                                            Cumulative % of pop. ranked by income

                                                   Pre-payment consumption             Food poverty line = 3047ALL
                                                   Post-payment consumption


                                          Year 2005
                                                           Poverty impact on Pen's Parade
                                                              Per capita consumption in 2008
                                  50000


                                           0
                           -50000
               -150000-100000




                                               0       20             40             60           80           100
                                                            Cumulative % of pop. ranked by income

                                                   Pre-payment consumption             Food poverty line = 3047ALL
                                                   Post-payment consumption


                                          Year 2008
                                                                                                              Page 24



Part IV. The Determinants of Informal Payments in Albania‟s
Health System
48.     The theoretical framework and hypotheses presented in this report are based on models
that seek to explain the occurrence and extent of informal payments to medical staff. Three
possibly overlapping explanations have been presented in the literature: social-cultural norms,
underfunding, and weak governance, following Gaal and McKee (2005) and Tomini and Maarse
(2009). It is difficult to conclusively distinguish between these explanations for informal
payments. Table 6 presents what researchers trying to examine the relative importance of each
explanation for informal payments might observe.

Table 6. Explanations for Informal Payments and Related Observations
MODEL                PRESUMED CAUSAL FACTORS                    OBSERVATIONS
1. Social-cultural   1. Culture of gratitude & gifts in         1. In kind and money gifts are paid to medical staff
norms                   cohesive societies that place greater   usually after services are received.
                        relative importance on reciprocity.     2. Patients report that gifts to medical staff are
                                                                voluntary.
                                                                3. A relatively moderate incidence of such gifts,
                                                                usually of modest value.

2. Underfunding      1. Demand-side factors                     1. Unequal amounts paid for inpatient and
                       - free heath care access                 outpatient health services.
                                                                2. Gifts resemble a „fee for service‟ model (patients
                     2. Supply side factors                     use informal payments as top up or to receive better
                       - Inadequate funding                     quality services).
                       - Low salaries of health care staff.     3. Being covered by health insurance will lower the
                       - Investments oriented more towards      probability of paying informally.
                          access rather than quality of
                          services.
3. Weak              1. Lack of monitoring and               1. High incidence of explicitly requested payments.
governance               accountability.                     2. Probability of paying informally is greater the
                     2. Weak rule of law and low             less knowledge and bargaining power held by the
                         probability of sanction if          patients.
                         malfeasance is detected.            3. Low numbers of prosecuted cases of corrupt
                                                             medical staff.
Source: Adapted from “Tomini, S., and Maarse H. (2009). How do patient characteristics influence informal
payments for inpatient and outpatient health care in Albania: Results of OLS and probit models using Albanian
LSMS 2005�

49.     The last part of this report presents the results of more in depth statistical analysis of
individual determinants of the amount paid informally to medical staff, and the average treatment
effects for individual variables between those who pay and those who do not, over the period
2002 to 2008. The estimation procedure followed and full estimation details are provided in
Appendix Two.
                                                                                                               Page 25


4.a. Determinants of Informal Payments for Outpatient Services

50.      Results for outpatient health care services are presented in Table A2.3 of Appendix 2. 14

51.      The incidence of informal payments has decreased over the period of analysis, however a
there is a less clear trend in the amount of informal payments. . They increased significantly
between 2002 and 2005, and decreased slightly between 2005 and 2008 (though the effect is not
statistically significant). Residents of coastal areas are more likely to pay compared to the
residents of central areas, while residents of mountainous areas are less likely to pay.

52.      Older and single people pay higher informal payments, which could reflect their higher
predisposition of getting ill (typically, older people have a higher demand for health care, see
Grossman, 1972), or higher incomes. Being covered by health insurance has a statistically
significant and negative effect on the incidence of informal payments and the effect is highly
significant in both models. Nuclear families without children (including only the head and
spouse) generally pay higher amounts than extended families or other families with children,
suggesting again a possible link between perceived higher income and amounts paid informally.
Medical staff may not have information about the consumption level of the people they treat, but
they may use certain characteristics as “proxies� for the affordability of such payments. Tomini
et al., (2009) show that medical providers have information on the patient‟s willingness to pay
and that they may use it to impose higher amounts of informal payments than what would
generally be expected. What is surprising is that there are not significant differences between
patients from households in the upper and lower quintiles if compared to those in the middle
quintile (almost all coefficients are not statistically significant).

53.     Examining like-groups over time (see Table A2.4 and Table A2.5 in Appendix 2), there
was a decrease in the amount paid informally between 2002 and 2005, followed by an increase in
the amount of informal payments between 2005 and 2008 (though not statistically significant).
The intended or unintended effect of structural reforms to the health sector during this period on
informal payments (financing health services and wages directly from the health insurance
institute, and increasing the wages of GPs) may have been greater in the earlier part of the
period. The most significant changes in the characteristics associated with making informal
payments for out-patient services, took place between 2002 and 2005. Changes are less clear
between 2005 and 2008.

54.      The share of people in the highest quintile who made informal payments fell between
2002-2005 and 2005-2008. However, the opposite happened for the share of people in quintiles
1 or 2 who made informal payments. The share making informal payments from quintile 1
initially fell (though the difference is not statistically significant) and then rose significantly
between 2005 and 2008. The overall trend 2002-2008 is an increase in incidence for people from
all the consumption quintiles, except those in the highest quintile.



14
  The results for the effect of individual characteristics on the amount paid (expressed as the natural logarithm) are
compared between two models (the left-censored tobit and the Heckman selection model). Most results are
consistent regardless of the model used, but the selection model gives a better overview of the separate effects on the
incidence and amount paid informally over the years.
                                                                                               Page 26


55.     The share of patients paying informally over the years in each region does not follow a
clear pattern. Residents of the coastal region and of Tirana were less likely to pay in 2008 when
compared to 2002, but the residents all other regions were more likely to pay. In mountainous
areas, residents were less likely to pay in 2002-2005, but more likely to pay between 2005-2008.

56.     Another interesting observation is the effect of health insurance coverage on the
likelihood of making informal payments. Between 2002 and 2005 people without insurance
were less likely to make informal payments, while people with normal health insurance were
more likely to pay. The overall effect between 2002-2008 shows a decline in the likelihood of
paying among individuals covered by health insurance, but the difference is not statistically
significant. Perversely, in a system with weak monitoring and accountability structures, being
covered by health insurance could increase the likelihood of individuals making informal
payments simply by lowering some of the barriers that would dissuade similar uninsured people
from even seeking services in the first place.

4.b. Determinants of Informal Payments for Inpatient Services

57.      Table A2.6, A2.7 and A2.8 In Appendix 2 shows the results for tobit and selection
models for the amount paid as informal gift to medical staff in inpatient services. Again, the
models can show the differences that exist depending in the factors associated with incidence of
informal payments and the amount paid informally (expressed here as the natural logarithm). The
characteristics associated with the amount paid here are less clear (fewer variables appear
statistically significant) which could be interpreted as confirmation that informal payments for
inpatient services are far more prevalent than in the outpatient services.

58.     The results show that the incidence of informal payments paid for inpatient services has
decreased between during both periods 2002-2005, and 2005-2008 (though only the later change
is statistically significant). Similarly to payments for outpatient services, the amount paid
informally does not necessarily follow the same pattern. The results show that the amounts paid
informally have increased significantly. Residents of coastal areas were more likely to pay and
also to pay higher amounts when compared to the residents of central areas. Residents of
mountainous areas (and Tirana) were less likely to pay and paid smaller amounts. This finding is
similar to the regional patterns of informal payments for outpatient services and may again relate
to the poverty levels in these areas (as discussed previously).

59.     People with chronic illnesses pay lower amounts informally, which may reflect that those
who know how the system works (and how much the medical staff “request�) can better estimate
the amount requested by the physicians, and may even be in a better (more informed) bargaining
position (see also Tomini et al 2009). Health insurance reduces the propensity of paying
informally for inpatient services (if compared to having no insurance). In contrast to informal
payments for outpatient services, here the lowest quintile is less likely to pay and paid lower
amounts informally (although the effects is not statistically significant). The results again
confirm greater prevalence/ubiquity of these payments for those seeking inpatient services,
compared to outpatient services.

60.    The matching findings for like-patients reported in Table A2., show that the amounts paid
informally for inpatient services over the period of analysis is similar to the results for outpatient
                                                                                                              Page 27


services. There was a fall in the amount paid informally for inpatient services between 2002 and
2005, but this was followed by an increase in the amount of informal payments between 2005
and 2008 (though not statistically significant if Kernel matching is used). This change makes the
overall effect between 2002 and 2008 positive but statistically not significant. Again, it appears
as if there were more incentives to reduce informal payments between 2002 and 2005, than there
were between 2005 and 2008.

61.     Few of the individual characteristics that are significantly associated with the likelihood
of making informal payments for inpatient services changed over the period of analysis. The
residents of rural areas were more likely to make informal payments for inpatient services in
2008 than were urban residents. People living in Tirana were also less likely to make informal
payments. Older people were less likely to pay, particularly between 2005-2008. The pattern of
change in the likelihood of making informal payments for inpatient services again disfavours
people from the poorest households. People from households in the lowest quintiles are more
likely to pay in 2008 when compared to 2002 (the effect significant for people in quintile 1) and
the highest quintiles are less likely to pay (though the effect is not significant).

4.c. Insights for Policy Makers

62.     Does the prevalence of informal payments in Albania‟s health sector reflect social-
cultural norms? This explanation is the least supported in the data. A higher incidence (two
times higher) and higher amounts (three to five times higher per day hospitalized) of informal
payments are observed for inpatient than for outpatient services. The amounts of these informal
payments are considerable when compared to average wages in the country and moreover not
directly linked to the economic status of the giver. For both inpatient and for outpatient services,
people from households in the lower quintiles of consumption are at times more likely to give
and also give higher amounts of informal payments.

63.     Do informal payments reflect resource shortages in the health sector? The evidence
supporting this explanation is mixed. Patients from mountainous areas and Tirana pay generally
lower amounts (for both outpatient and inpatient services) than people from the central or coastal
areas. Supporting the “underfunding� explanation, Tirana is relatively “resource rich�, with the
highest concentration of hospital beds (401 per 100000 inhabitants, as compared to the national
average of 303 per 100000 inhabitants). However these beds are mostly located in tertiary care
hospitals treating people from different regions. Thus in some important aspects, Tirana‟s
medical infrastructure is relatively more burdened, and resource shortages and the need for
rationing are greater. In 2003 the number of inhabitants per PHC facility in Tirana was 3005,
much higher than the national average of 1440 (MoH, 2003). Furthermore, medical staff in
Tirana are usually paid less than the those who work in the remote mountainous areas.15
Mountainous areas compare well with the national averages (see Tables A1.1 and A1.2) and they
certainly have better paid medical staff. But the differences can also be simply because of the
specific conditions of such areas (mountainous areas have the highest poverty rates in the
country and poverty in Tirana has increased in the last years – INSTAT 2008). Finally, one of

15
  In both outpatient and inpatient services physicians are paid better if they accept to work on mountainous areas,
but anecdotic evidence shows that hospitals are understaffed and people are often addressed in the regional
hospitals.
                                                                                             Page 28


the most visible reforms in the health sector in recent years relates directly to the “underfunding�
hypothesis: the increase of GPs wages has done little to reducing informal payments.

64.     Does the prevalence of informal payments in the health sector reflect weaj governance in
an institutional environment with absent or ineffective monitoring and accountability structures?
It is reasonable to expect that in an environment with poorly functioning monitoring and
accountability structures, the most vulnerable people will bear the consequences of any
malfeasance of service providers. People in the lowest quintiles bear the heaviest cost of
informal payments (they are often more likely to pay and also pay higher amounts both in
absolute terms and relative to their total spending). This grim and persistent pattern may indicate
that those who lack information, empowerment and access to mechanisms with which to hold
service providers accountable continue to bear the heaviest burden imposed by informal
payments to staff in Albania‟s health system.

65.     The evidence from Albania LSMS 2002-2008 suggests that while scarcity of resources
seems to be an important contributing factor, weak governance in health care are likely to be just
as important. The past reforms and reorganisation of the health sector has only managed to
achieve small reductions in the amounts and incidence of such payments. Empirical evidence
shows that even as the wages of primary care medical staff were increased threefold over the
years, informal payments remain wide spread. The poorest members of the society remain less
protected against such payments.

66.     For this reason, addressing only the scarcity of resources may not be sufficient to the
limitation of the problem. Limited international evidence has shown that the introduction of
formal fees as well as provider accountability mechanisms and other incentives to providers may
limit the phenomenon. While these are certainly important steps to consider, they should be
accompanied by additional measures. The experience of other countries suggest that linking the
formal payments to bonus incentives for medical doctors, advertising the formal fees, etc could
increase the effectiveness of the policy measures against informal gifts.

67.     Encouragingly, the Government of Albania has already embarked on a process of
increasing transparency in the health sector and guaranteeing access to basic care to all, in order
to reduce the negative impact on households of out-of-pocket payments and informal payments
in particular,. As an initial step toward this objective, the Government making changes to the
health insurance system to grant coverage to all beneficiaries of the Ndihma Ekonomike,
Albania‟s main social assistance program for poor households. As the analysis in this report
shows, informal payments are partly encouraged by an unclear copayments policy, which often
blurs the distinction between formal and informal payments. Informal payments are also
associated with a lax definition of the type of services that health insurance should cover.
Experience elsewhere has shown that it is possible to reduce informal payments when this issue
is addressed as an integral part of health finance reforms of the type that Albania is undertaking.
These reforms set clear limits on what public resources allocated for health can finance.
Defining an explicit package of benefits in this way can help to improve equity in access and
increase accountability for the services specified in the package; as patients are aware of what
services they are entitled to receive and at what prices, the scope for informal payments is
reduced.
                                                                                    Page 29




References
-   Abel-Smith, B., and J. Falkingham. 1996. Financing Health Services in Krygyzstan: The
    extent of private payments. Report for the British Government Know-How Fund (KHF).
    London School of Economics, London, U.K.
-   Albert A. Bennett C, Bojar M. 1992. Health care in Czech Republic. JAMA 267 (18):
    2461-66.
-   Andrew Dabalen, Talip Kilic and Waly Wane (2008). Policy Research Working Paper
    4783. The World Bank.
-   Anon, 1999. Paying for health care. Sofia, Bulgaria: Klan.
-   Baeza C, Packard TG. 2005. Beyond Survival: Protecting Households from the
    Impoverishing Effects of Health Shocks in Latin America. Washington, DC: World
    Bank.
-   Balabanova, D., & McKee, M. (2002). Understanding informal payments for health care:
    the example of Bulgaria. Health Policy, 62, 243–273.
-   Belli P. (2002). Formal and informal household spending on health: a multicountry study
    in Central and Eastern Europe.
-   Berki, S. E. 1986. “A Look at Catastrophic Medical Expenses and the Poor.� Health
    Affairs 138–45.g
-   Bredenkamp, C., Mendola, M., and Gragnolati, M. 2008. The impoverishing effect of
    health expenditures: evidence from the Western Balkans. Enhancing efficiency and
    equity: challenges and reform opportunities facing health and pension systems in the
    Western Balkans. HNP Discussion paper 46706.
-   Delcheva, E., Balabanova, D., & McKee, M. (1997). Under-the-counter payments for
    health care: Evidence from Bulgaria. Health Policy 42, 89-100.
-   Ensor, T. (2004). Informal payments for health care in transition economies. Social
    Science & Medicine 58, 237–246.
-   Ensor, T., & Savelyeva, L. (1998). Informal payments for health care in the Form Soviet
    Union: some evidence from Kazakstan. Health Policy and Planning, 13(1), 41-49.
-   Ensor, T., & Witter, S. (2001). Health economics in low income countries: adapting to
    the reality of the unofficial economy. Health Policy 57, 1-13.
-   Falkingham J. 1998/99. Barriers to access? The growth of private payments for health
    care in Kyrgyzstan. EuroHealth 4(6): 68-71.
-   Gaala, P., & McKee, M. (2005). Fee-for-service or donation? Hungarian perspectives on
    informal payment for health care. Social Science & Medicine, 60 1445–1457.
-   Lewis, M. (2000). Who is paying for health care in Eastern Europe and Central Asia?
    World Bank.
                                                                                      Page 30


-   Lewis, M. (2002). Informal health payments in central and eastern Europe and the former
    Soviet Union: issues, trends and policy implications. European observatory on Health
    Care system series. Funding health care: options for Europe.
-   Liaropoulos, L., Siskou, O., Kaitelidou, D., Theodorou, M., & Katostaras, T. (2008).
    Informal payments in public hospitals in Greece. Health Policy 87, 72-81.
-   Miller, W. L., Grødeland, Å. B., & Koshechkina, T. Y. (2000). If you pay, we‟ll operate
    immediately. Journal of Medical Ethics 26, 305-311.
-   Ministry of Health of Albania. The long-term strategy for the development of Albanian
    Health System. April 2004
-   Nuri, B., & Tragakes, E. (2002). Health Care Systems in Transition. Copenhagen,
    European Observatory on Health Care Systems, 4(6).
-   O‟Donnell, O., E. van Doorslaer, R. P. Rannan-Eliya, A. Somanathan, C. G. Garg, P.
    Hanvoravongchai, M. N. Huq, A. Karan, G. M. Leung, K. Tin, and C. Vasavid. 2005.
    “Explaining the Incidence of Catastrophic Payments for Health Care: Comparative
    Evidence from Asia.� EQUITAP Working Paper #5. Erasmus University, Rotterdam,
    Netherlands, and Institute of Policy Studies, Colombo, Sri Lanka.
-   Pradhan, M., and N. Prescott. 2002. “Social Risk Management Options for Medical Care
    in Indonesia.� Health Economics 11: 431–46.
-   Rajeev H. Dehejia & Sadek Wahba, 2002. "Propensity Score-Matching Methods For
    Nonexperimental Causal Studies," The Review of Economics and Statistics, MIT Press,
    vol. 84(1), pages 151-161, February.
-   Russell, S. 2004. “The Economic Burden of Illness for Households in Developing
    Countries: A Review of Studies Focusing on Malaria, Tuberculosis, and Human
    Immunodefi ciency Virus/Acquired Immunodefi ciency Syndrome.� American Journal of
    Tropical Medicine and Hygiene 71(Supp. 2): 147–55.
-   Stiglitz JE. 1988. Economics of the Public Sector (2nd edn). W.W. Norton & Company:
    New York.
-   Thompson, R., & Xavier, A. (2002). Unofficial payments for acute state hospital care in
    Kazakhstan. A model of physician behaviour with price discrimination and vertical
    service differentiation. LICOS Discussion Papers, Discussion Paper 124.
-   Tomini, S., and Maarse H. (2009). How do patient characteristics influence informal
    payments for inpatient and outpatient health care in Albania: Results of OLS and probit
    models using Albanian LSMS 2005
-   Tomini, S., Groot, W., and Pavlova, M. (2009). Paying Informally in the Albanian Health
    Care Sector: A Two-Tiered Stochastic Frontier Model.
-   Vian, T., Grybosk, K., Sinoimeri, Z., & Halld, R. (2006). Informal payments in
    government health facilities in Albania: Results of a qualitative study. Social Science &
    Medicine 62, 877-887.
-   Wagstaff, A., and E. van Doorslaer. 2003. “Catastrophe and Impoverishment in Paying
    for Health Care: with Applications to Vietnam 1993–98.� Health Economics 12: 921–34.
                                                                                       Page 31


-   World Bank. (2006). Albania Health Sector Note. Report No. 32612-AL.
-   World Bank. Averting the old age crisis. Oxford: Oxford University Press, 1995a.
-   World Bank. World development report: Investing         in   health.   Oxford:     Oxford
    University Press, 1993.
-   World Health Organization 2000. World Health Report 2000. Geneva, Switzerland:
    World Health Organization.
-   Wynia, M. K., Latham, S. R., Kao, A. C., Berg, J. W., & Emanuel, L. (1999). Medical
    professionalism in society. The New England journal of medicine 341(21), 1612-1616
-   Xu, K., D. E. Evans, K. Kawabate, R. Zeramdini, J. Klavus, and C. J. L. Murray. 2003.
    “Household Catastrophic Health Expenditure: A Multicountry Analysis.� Lancet 362:
    111–17.
                                                                                                 Page 32



                                              Appendix One

                                           Descriptive Statistics


Table A.1.1. Population Catchment Area of PHC Facilities, by Region, 2003
                                                       NUMBER O F INHABITANTS PER PHC FACILITY
           AREA                  REGIONS                                                     All PHC
                                               Health centers  Health Posts    Polyclinics
                                                                                             facilities
Central                      Berat                 5079            1021           64340         839
                             Elbasan               5668            1570           90684        1213
                             Gjirokastër           2686             752           37610         579
                             Korcë                 5765            1128           53036         927
                             Shkodër               3288            3612          128237        1698
Coastal                      Durrës               16345            2452          122590        2096
                             Fier                  8139            2161          127315        1685
                             Lezhë                 3061            2067           53061        1206
                             Vlorë                 3860            1892           96491        1253
Mountain                     Dibër                 4868            5753           63285        2531
                             Kukës                 4284            2931           37131        1663
Tirana                       Tirana                7034            6101           37369        3005
Albania‟s average                                  5274            2045           62638        1440
Source: Ministry of Health


Table A.1.2. Regional Distribution of Hospitals Providing Inpatient Care in Albania, 2003
          AREA           REGIONS           MOH         HOSPITALS /   HOSPITAL   BEDS /     TOTAL STAFF
                                        HOSPITALS        100000       / BEDS    100000        PER BED
Central                Berat                  4           2.07          431     223.29          1.16
                       Elbasan                6           1.65         1071     295.26          1.00
                       Gjirokastër            4           3.55          372      329.7          1.23
                       Korcë                  4           1.51          835     314.88          1.05
                       Shkodër                2           0.78          653     254.61          1.32
Coastal                Durrës                 2           0.82          544     221.88          1.74
                       Fier                   3           0.78          678     177.23          0.82
                       Lezhë                  3           1.88          334     209.82          1.62
                       Vlorë                  6           3.11          831     430.61          1.02
Mountain               Dibër                  4           2.11          518     272.84          1.18
                       Kukës                  3           2.69          383     343.83          1.17
Tirana                 Tirana                 6           1.00         2399     401.24          1.88
Albania‟s average                            50           1.53         9049      303.6          1.27
Source: Albania Health Indicators, Ministry of Health.
                                                                                                               Page 33


Table A.1.3. Respondent‟s Priority for Improvement in their Health Facility
                                                  COASTAL       CENTRAL       MOUNTAIN         TIRANA        TOTAL
Condition of the health services                   0.16***       0.08***        0.08***        0.08***         0.11
Attention from the health officers                  0.32**       0.27***        0.20***        0.38***          0.3
Official costs of health services                  0.19***         0.22           0.22         0.29***         0.22
Money or gifts demanded                               0.1        0.12***        0.09***          0.09           0.1
Availability of medicine supply and vaccine         0.15*        0.16***        0.18***        0.08***         0.14
Quality of medicine and vaccine supply             0.08***       0.13***        0.22***        0.05***         0.11
Waiting time                                       0.00***         0.01         0.00***        0.02***         0.01
Other                                               0.00***          0.01         0.01**           0.01         0.01
Stars indicate if the mean for the subgroup is significantly different from the mean for all other subgroups. ***
p<.01, ** p<.05, * p<.1



Table A.1.4. Responses to LSMS Question “If this health facility received LEK 10 million,
on which activity would you want most of it spent?�
                                                  COASTAL        CENTRAL      MOUNTAIN        TIRANA         TOTAL
Medicine/vaccine supply                            0.39***        0.33***       0.39***         0.33*         0.35
Doctor/nurses training                             0.13***         0.16*          0.16          0.15          0.15
Condition of facilities                            0.12***          0.1          0.11*        0.06***          0.1
Provide free care for poor                         0.27***        0.35***          0.3         0.28**         0.31
Rehabilitation of the building                     0.08***        0.02***       0.02***         0.06          0.05
Rehabilitation of the building                     0.08***        0.02***       0.02***         0.06          0.05
Salaries of doctors                                0.01***        0.03***       0.02***       0.11***         0.04
Other                                               0.00**          0.01         0.01**         0.01          0.01
Stars indicate if the mean for the subgroup is significantly different from the mean for all other subgroups. ***
p<.01, ** p<.05, * p<.1
                                                                                                                    Page 34


Table A.1.5. Responses to LSMS Question “In your opinion, is the situation better or worse
in terms of corruption cases [in the health sector] comparing today and three years ago?�
                                          BETTER               THE SAME               WORSE              UNKNOWN
Quintiles
    Quintile 1                              0.22                   0.49               0.15                 0.14
    Quintile 2                              0.22                 0.45***              0.15                 0.18
    Quintile 3                            0.17***                0.53***            0.11***                0.19
    Quintile 4                               0.2                  0.46*               0.15                 0.29
    Quintile 5                              0.22                   0.48               0.15                 0.15
    Total                                    0.2                   0.48               0.14                 0.18
Health facilities
    Public hospital                          0.2                   0.52             0.17***                0.11
    Public primary centre                   0.21                 0.53***            0.10***                0.16
    Public polyclinic                       0.22                 0.43***            0.21***                0.14
    Private hospital/clinic                 0.16                   0.48             0.32***                0.04
    Total                                   0.21                   0.51               0.14                 0.14
Stratum
    Coastal                               0.28***                0.46***            0.09***                0.17
    Central                               0.18***                  0.48               0.14                  0.2
    Mountain                              0.19***                  0.48             0.21***                0.12
    Tirana                                0.13***                0.52***            0.21***                0.14
    Total                                    0.2                   0.48               0.14                 0.18
Stars indicate if the mean for the subgroup is significantly different from the mean for all other subgroups. ***
p<.01, ** p<.05, * p<.1

Table A.1 6. Responses to LSMS Question “Which one of the following do you agree with
in terms of unofficial out-of-pocket fees paid at health facilities in Alba?�
                     STRONGLY                                                   STRONGLY            NEITHER AGREE
 QUINTILES                                AGREE            DISAGREE
                       AGREE                                                    DISAGREE             NOR DISAGREE
       Private fees must be eliminated and doctors/health care staff should be penalized for requesting such fees
      1                0.56*              0.35***              0.04                0.02                    0.03
      2                0.52**             0.41***              0.03                0.02                  0.02***
      3               0.50***              0.40*               0.04              0.03***                   0.03
      4                 0.54                0.39              0.04*              0.00***                   0.03
      5               0.59***             0.35***            0.02***             0.01***                   0.03
     Total              0.54                0.38               0.03                0.02                    0.03
      The Government should increase salaries of health care staff and private fees will automatically be reduced
      1               0.23***                0.5             0.19***               0.03                  0.05***
      2               0.27***             0.54***              0.13               0.04**                 0.02***
      3                 0.32                 0.5               0.11                0.03                    0.04
      4               0.36***                0.5             0.09***             0.02***                   0.03
      5               0.41***             0.45***            0.08***               0.03                    0.04
     Total              0.31                 0.5               0.12                0.03                    0.04
                        Unofficial private fees should be made official and publicly announced
        1              0.17***            0.24***            0.27***             0.23***                  0.09
        2              0.19***              0.29               0.24              0.21***                 0.07*
        3                0.22               0.28               0.24                0.17                   0.08
        4                0.21                0.3               0.25               0.17**                  0.07
        5              0.25***            0.33***            0.19***             0.13***                  0.09
     Total               0.21               0.29               0.24                0.19                   0.08
Stars indicate if the mean for the subgroup is significantly different from the mean for all other subgroups. ***
p<.01, ** p<.05, * p<.1
                                                                                                                Page 35


Table A.1.7. Share of Patients who Made Informal Payments for Outpatient Services 2002-
2008
     QUINTILES                  YEAR 2002                      YEAR 2005                      YEAR 2008
          1                  0.228          (0.041)        0.268            (0.042)       0.221           (0.047)
          2                  0.293          (0.042)        0.244            (0.046)       0.158           (0.035)
          3                  0.268          (0.032)        0.203            (0.039)       0.200           (0.040)
          4                  0.327          (0.042)        0.222            (0.035)       0.215           (0.035)
          5                  0.271          (0.028)       0.162**           (0.029)       0.150           (0.032)
  Total population           0.281          (0.021)        0.218            (0.021)       0.190           (0.021)


Nr of observations                   2120                           1463                          1151
Population size                   446938                           312500                       265258
Quintiles refer to total per capita consumption quintiles. Standard errors are in brackets.
Stars indicate if the mean for the subgroup is significantly different from the mean of all other subgroups.
*** p<.01, ** p<.05, * p<.1


Table A.1. 8. Share of Patients who Made Informal Payments for Inpatient Services 2002-
2008
     QUINTILES                  YEAR 2002                      YEAR 2005                      YEAR 2008
          1                 0.479*          (0.063)        0.548            (0.060)       0.465           (0.071)
          2                  0.569          (0.060)        0.602            (0.056)       0.495           (0.085)
          3                  0.654          (0.048)        0.586            (0.061)       0.351           (0.080)
          4                  0.644          (0.049)        0.529            (0.057)       0.445           (0.078)
          5                  0.598          (0.048)        0.517            (0.055)       0.434           (0.086)
  Total population           0.597          (0.030)        0.559            (0.032)       0.439           (0.041)


Nr of observations                   707                            677                           377
Population size                   134282                           125563                       84317.1
Quintiles refer to total per capita consumption quintiles. Standard errors are in brackets.
Stars indicate if the mean for the subgroup is significantly different from the mean of all other subgroups.
*** p<.01, ** p<.05, * p<.1
                                                                                                                    Page 36


Table A.1.9. The mean of the amount paid informally per outpatient visit in Albanian Leks
(series with 2002 prices
Quintiles          Year 2002               Year 2005               Year 2008
     Quintile 1             327.55             (55.92)            305.49       (39.50)          304.38           (42.82)
     Quintile 2            259.31**            (22.23)            292.73       (37.98)          242.6            (34.95)
     Quintile 3             323.27             (53.20)            394.12       (62.54)          265.37           (27.89)
     Quintile 4             384.95             (109.04)           323.7        (31.14)          273.82           (34.65)
     Quintile 5             450.65             (74.50)            321.38       (31.85)          235.61           (28.52)
  Total Population          366.48             (37.98)            326.02       (18.95)          267.2            (15.89)
Quintiles refer to total per capita consumption quintiles. Standard errors in brackets.
Stars indicate if the mean for the subgroup is significantly different from the mean of all other subgroups.
*** p<.01, ** p<.05, * p<.1


Table A.1.10. The mean of the amount paid informally per day hospitalized in Albanian
Leks (series with 2002 prices)
       Quintiles                  Year 2002                           Year 2005                     Year 2008
     Quintile 1             752.31*            (131.34)      939.74**          (205.38)       636.24***       (156.70)
     Quintile 2             789.29*            (115.23)       1091.27          (159.15)        1044.33        (263.67)
     Quintile 3             878.58             (122.16)       1222.04          (247.80)        1241.18        (469.45)
     Quintile 4             1079.34            (184.35)       1592.53          (370.96)        1937.34        (613.74)
     Quintile 5             1911.9             (750.62)      2638.10*          (829.67)        1733.77        (437.79)
  Total Population            1127.76         (180.89)          1418          (172.53)        1271.81      (177.29)
Quintiles refer to total per capita consumption quintiles. Standard errors in brackets.
Stars indicate if the mean for the subgroup is significantly different from the mean of all other subgroups.
*** p<.01, ** p<.05, * p<.1



Table A.1.11. Respondents‟ Evaluation of Health Care Facilities in 200
                                       OUTPATIENT SERVICE                                 INPATIENT SERVICE
                              Have not paid        Have paid                 Have not paid        Have paid
                                                                     Total                                          Total
                              informal gifts     informal gifts              informal gifts     informal gifts
                                      Condition of the Health services facility
Not satisfactory                   0.11               0.11           0.11          0.16              0.1            0.13
Somewhat satisfactory             0.31*              0.25*           0.29          0.22              0.3            0.26
Satisfactory                      0.24*              0.30*           0.26         0.22*             0.31*           0.26
Good                               0.31               0.31           0.31         0.38*             0.27*           0.34
Excellent                          0.03               0.04           0.03          0.02             0.01            0.02
                                          Attention from the Health Officers
Not satisfactory                   0.09               0.12           0.09         0.06*             0.13*           0.09
Somewhat satisfactory              0.28               0.28           0.28          0.31              0.3             0.3
Satisfactory                       0.29               0.31           0.29          0.25             0.33            0.29
Good                               0.33               0.28           0.32         0.36**           0.22**            0.3
Excellent                          0.02               0.01           0.02          0.02             0.01            0.02
                                                                                                               Page 37


                                  Availability of Medicine supply and Vaccine
Not satisfactory                 0.21***          0.35***        0.24         0.21**            0.35**         0.27
Somewhat satisfactory             0.35              0.38         0.36         0.25**            0.38**         0.31
Satisfactory                      0.23              0.18         0.22           0.23              0.16         0.2
Good                             0.20***          0.09***        0.17         0.29***           0.09***        0.2
Excellent                        0.02***          0.00***        0.02           0.01              0.02         0.01
                                    Quality of Medicine and Vaccine supply
Not satisfactory                 0.26***          0.35***        0.28           0.26              0.29         0.27
Somewhat satisfactory             0.35              0.33         0.34           0.26              0.36         0.3
Satisfactory                       0.2              0.23         0.21           0.24              0.25         0.24
Good                             0.17***          0.08***        0.15         0.22***           0.09***        0.17
Excellent                         0.02              0.01         0.02           0.01              0.01         0.01
Stars indicate if the mean for the subgroup is significantly different from the mean for all other subgroups. ***
p<.01, ** p<.05, * p<.1
                                                                                                               Page 38


Table A.1.12. Respondents‟ Evaluation of the Health Care Costs in 2008
                              VERY HIGH             HIGH          AFFORDABLE             LOW               NONE

                                          Official costs of Health services
Public hospital                 0.07***              0.2               0.22            0.36***             0.15
Public primary centre             0.09             0.20**             0.22*             0.34*             0.15**
Public polyclinic               0.11***           0.26***            0.27***           0.27***            0.10***
Private hospital/clinic           0.12              0.23               0.22            0.22***            0.22**
Total                             0.09              0.21               0.23              0.33              0.14
              Money or other Gifts demanded by doctors and other Health care staff for services
Public hospital                 0.17***           0.33***            0.18***           0.14***            0.18**
Public primary centre           0.12***             0.29             0.22***           0.22***            0.15**
Public polyclinic               0.22***           0.26***              0.2             0.14***             0.17
Private hospital/clinic         0.07***             0.27             0.32***             0.16              0.19
Total                             0.15               0.3               0.21              0.18              0.16
Stars indicate if the mean for the subgroup is significantly different from the mean for all other subgroups. ***
p<.01, ** p<.05, * p<.1
                                                                                            Page 39



                                        Appendix Two

                   Propensity Score Matching and Estimation Results

1. 1. Determinants of informal gifts

       We pool data from LSMS 2002, 2005 and 2008 to check for the main determinants of
informal gifts over time. Albania LSMS it is not a panel survey so we have to rely on cross
section analysis to determine the effect of individuals‟ characteristics on the amount paid
informally to medical staff. Our first goal is to get more insights into the individual
characteristics determining the amount paid as gifts. In this case the amount of informal gifts is
determined by individual characteristics such that:

        yi*  xi' �  � i ,            i=1, 2, 3 …,N                (1)

where, y i refers to the amount paid informally to medical staff in outpatient or inpatient
services, x i denotes a vector of exogenous and observed individual characteristics (see Tables ),
 ï?¢ represents a vector of estimated coefficients corresponding to the such characteristics, and ï?¥ i
is a vector of residuals errors that have a normal distribution.

Not all the patients that visited health care centres reported making informal gifts to medical
staff. Therefore we have a share of people reporting zero payments. We suspect this may be
mainly for two reasons: (1) the left censoring at zero, or (2) a possible selection bias. The left
censoring means that the dependent variable is represented by a latent variable y * that is not
observed when yi  0 . The distribution has full support within the interval (,) but we can
only observe values that are equal or above 0. The observed variable is thus related to the latent
variable through the observation rule:

                   y * if y *  0
                   0 if y * ï‚£ 0                                      (2)

        As we are dealing with informal gifts it is reasonable to imagine a negative demand for
such payments: some people may be aware that they should receive treatment at no costs, or that
they have already contributed to financing public health care either directly by paying health
insurance contributions or through other taxes. The possible selection bias may be a classical
case when the selection and the outcomes are determined by different probability functions. This
is a two part model where the first part is a binary outcome modelled as the probability that the
binary variable is bigger than 0, and the second part uses a linear regression to model the
                                                                                       *
outcome if the outcome is bigger than 0. The model introduces a latent variable y1 and the
outcome y 2 is observed if y1  0 . In our case y1 defines if the patient makes an informal
            *                 *                      *


payment, y 2 defines the amount of the informal payment, and y1  y 2 .
            *                                                   *    *
                                                                                             Page 40



The model includes two equations, the first one being a selection equation for y1 , where

             1     if y1  0
                       *


             0     if y1 ï‚£ 0
                       *
                                                                          (3)

       and the resulting outcome for y 2 , where

               *
              y2    if y1  0
                        *


             -      if y1 ï‚£ 0
                         *
                                                                          (4)

        The selection model is used to check for any biases in the selection of the respondents
(i.e. people that for certain reasons will never make informal payments but that otherwise are
included in the sample). Intuitively, this is less likely since informal gifts are widespread and
there is no obvious evidence that would exclude a certain category of person from paying them.
Nevertheless this test is designed to capture the overall effect that certain observable
characteristics may have on the selection. This is important in order to avoid any biases
associated with the estimates from the tobit model. Results from the two models are compared to
check for inconsistencies.

The satisfaction of the assumptions of normality and homoskedasticity of the tobit model shows
that this model fits the data relatively well (see Tables A5 and A6). The statistically significant
likelihood-ratio test of the selection model shows significant correlation between the errors of the
two parts of the model, confirming that this selection model performs better than the OLS).
However, testing for the robustness of the selection models (using the two step estimation -
Table A6) shows that such models are not as robust as the tobit (especially for the inpatient
services where selection appears to be much smaller). This indicates that the amount paid
informally is determined jointly for different categories.

1. 2. Impact over the years

      In order to isolate the individual determinants and how they have changed over the years
we employ propensity score matching (PSM). PSM helps to identify individuals with similar
characteristics over the different waves of the survey data, in the absence of an experimental set
up, where we would be able to construct counterfactuals. After these counterfactuals are
identified we can then use them to evaluate changes over the years in the characteristics of those
who make informal payments and those who do not. This can help to capture the effect that
public policies and reforms to the health sector over this period might have had on the incidence
or amount of informal payments.

      We assume y1i to be the outcome of the treatment for individual i when this individual is
subject to treatment and y 0i to be the outcome in absence of the treatment for the same individual
(see also Deheija and Wahba 2002; Dabalen, Kilic et al. 2008). The treatment effect for an
individual is:
                                                                                               Page 41



      � i  y1t  y0i                                                          (5)

and population treatment effect can therefore be defined as:

      � i  E( y1i | Ti  1)  E( y0i | Ti  0)                                (6)

where T indicates the treatment, and Ti is equal to 1 if individual i participates in the treatment
and 0 if not. As the treatment here refers to the consecutive year of the survey we do not observe
outcomes that would have materialised if corresponding individuals had not participated in the
treatment E ( y0i | Ti  1) . In our case the treatment involves entire population and therefore the
participation is independent of potential outcomes ( y1i , y0i ï?ž Ti ) . The average treatment effect
for the population treated is:

      �  E ( yi | Ti  1)  E ( yi | Ti  0)                                  (7)

The main motivation for employing PSM techniques is that -in the absence of a panel survey- it
is one of the few effective ways to match individuals with the same characteristics over the
years. The variables used for the matching correspond to observable characteristics of
individuals that are stable over time. Thus PSM approximates an experiment where everyone has
the same probability of participating in the consecutive year, and this probability is balanced
among the treated and not treated, and conditional on observed variables. The randomisation
ensures that the qualities of the treated and not treated are identical in terms of the distribution of
observed characteristics.

We have reason to believe that since all of the waves of the LSMS survey were country
representative, all individuals had the same probability of participating in the consecutive wave
of the survey, and that both treated and not treated came from the same economic environment.
Both the randomisation and the similarities in the background (HIT, 1997) are fulfilled in our
study.
                                                                                                             Page 42


Table A2.1. Two Stage Slection Models for Outpatient and Inpatient Services

                                            OUTPATIENT SERVICE                             INPATIENT SERVICE
                                  Zero or larger                                  Zero or larger
        VARIABLES                                        Log of the amount                                   Log of the
                                payment (selection                              payment (selection
                                                                paid                                        amount paid
                                      model)                                         model)
Constant                         -0.237*    (0.139)      6.500***    (0.445)    0.384*       (0.228)    6.648***       (0.739)
(Year 2002)
Year 2005                        -0.257***     (0.050)   0.274***    (0.094)     -0.049       (0.072)   0.218**        (0.110)
Year 2008                        -0.230***     (0.056)     0.120     (0.108)   -0.289***      (0.088)    0.338         (0.251)
Rural area                         0.013       (0.049)     0.103     (0.065)     0.058        (0.075)    -0.058        (0.125)
(Stratum central)
Stratum coastal                   0.189***     (0.052)   -0.259***   (0.079)     0.202**      (0.086)     0.102        (0.158)
Stratum mountain                 -0.294***     (0.061)    0.233**    (0.110)   -0.326***      (0.078)   -0.482**       (0.210)
Stratum Tirana                   -0.337***     (0.074)    0.311**    (0.137)    -0.305**      (0.125)     0.035        (0.264)
Gender (“0� male, “1�
                                   0.029       (0.042)    -0.004     (0.056)     0.066        (0.065)      -0.096      (0.109)
female)
Age (in years)                     0.002       (0.005)     0.003     (0.009)    0.002         (0.007)      -0.008      (0.016)
Age squared                       -0.000*      (0.000)    -0.000     (0.000)    -0.000        (0.000)      0.000       (0.000)
(Civil status – married)
Civil status - divorced and
                                     -            -       -0.040     (0.069)       -             -         -0.012      (0.156)
widow
Civil status – single                                      0.122     (0.091)                               -0.066      (0.173)
Chronic illness                    -0.000      (0.059)     0.093     (0.076)   -0.183**       (0.091)      -0.029      (0.182)
Health rate (“1� very bad -
                                   -0.041      (0.032)    -0.024     (0.044)    -0.007        (0.052)   0.312***       (0.081)
“5� excellent)
Have health insurance license    -0.233***     (0.048)       -          -      -0.198***      (0.072)        -            -
(Primary/No education)
Secondary education                -0.019      (0.058)    0.145*     (0.076)     0.097        (0.086)      0.021       (0.134)
University education               -0.170      (0.108)   0.345**     (0.159)     0.082        (0.170)      0.332       (0.249)
(Nuclear family with
children)
Nuclear families head and
                                  0.370*       (0.210)    -0.491*    (0.285)    -0.300        (0.372)      -0.220      (0.654)
spouse
Extended families                  0.157       (0.107)     0.012     (0.147)     -0.072       (0.154)      0.352       (0.255)
Quintile 1                         -0.013      (0.072)     0.049     (0.094)    -0.186*       (0.097)      -0.032      (0.186)
Quintile 2                         0.011       (0.069)    -0.108     (0.090)     -0.022       (0.097)      -0.077      (0.148)
(Quintile 3)
Quintile 4                        0.127**      (0.063)    -0.070     (0.088)     0.044        (0.100)      -0.057      (0.152)
Quintile 5                         0.049       (0.065)   0.167**     (0.085)     0.079        (0.103)      0.300*      (0.156)
Employment: public official        0.020       (0.114)       -          -        0.101        (0.175)         -           -
Employment: health
                                   -0.149      (0.166)       -          -       -0.098        (0.206)        -            -
practitioners

/mills                          -1.062*** (0.346)           -           -       -1.118        (0.885)        -            -
Number of observations                             4707                                           1707
chi2                                             253.358                                         195.193
note: *** p<.01, ** p<.05, * p<.1 Standard errors are in brackets.
                                                                                                            Page 43


Table A2.2. Logit Results (dependent variable =1 if the individual participates in the next
year survey)
                                  YEARS 2002 – 2005               YEARS 2005 - 2008               YEARS 2002 - 2008
Constant                        -1.764***      (0.182)          4.799***        (0.280)         3.995***        (0.270)
Rural area                        -0.035       (0.035)           -0.064         (0.046)        -0.131***        (0.047)
(Stratum central)
Stratum coastal                  0.090**          (0.035)        -0.013          (0.046)          0.074         (0.046)
Stratum mountain                  -0.027          (0.036)      -0.250***         (0.044)       -0.282***        (0.046)
Stratum Tirana                   0.113**          (0.046)       0.299***         (0.070)        0.441***        (0.071)
Gender                           0.068**          (0.030)      -0.361***         (0.037)       -0.338***        (0.038)
Age                             0.017***          (0.006)      -0.280***         (0.010)       -0.296***        (0.010)
Age square                         0.000          (0.000)       0.003***         (0.000)        0.003***        (0.000)
(Civil status – married)
Civil status - divorced and
                                  -0.050          (0.080)       4.608***         (0.074)       4.613***         (0.078)
widow
Civil status - single           0.544***          (0.062)      -2.899***         (0.113)       -2.570***        (0.110)
Chronic illness                 0.169***          (0.058)      -0.305***         (0.118)         -0.141         (0.112)
Health rate (“1� very bad -
                                0.341***          (0.037)         0.096          (0.073)       0.385***         (0.069)
“5� excellent)
(No insurance)
Health insurance status:
                                  -0.019          (0.034)       -0.094**         (0.045)       -0.094**         (0.045)
normal
Health insurance status:
                                0.535***          (0.133)       0.510**          (0.214)       1.018***         (0.247)
war/other invalid
(Primary/No education)
Secondary education             0.239***          (0.037)       0.290***         (0.048)       0.507***         (0.049)
University education            0.361***          (0.063)       0.925***         (0.085)       1.235***         (0.091)
(Nuclear family with
children)
Nuclear families head and
                                 0.654**          (0.266)      -0.535***         (0.167)         0.173          (0.226)
spouse
Extended families               0.504***          (0.100)       0.785***         (0.098)       1.389***         (0.135)
Quintile 1                      0.253***          (0.050)        0.147**         (0.061)       0.291***         (0.064)
Quintile 2                        0.072           (0.048)         0.082          (0.060)         0.075          (0.063)
(Quintile 3)
Quintile 4                      -0.239***         (0.046)        0.083           (0.062)       -0.203***        (0.062)
Quintile 5                      -0.479***         (0.046)        -0.020          (0.066)       -0.536***        (0.065)

Log pseudolikelihood                    -17938.641                      -10144.923                     -9934.2702
Nr of observation                          26406                          26127                           25519
Pseudo R2                                 0.0197                          0.4398                          0.4382
note: The results are used to estimate the propensity score for the matching of individuals over the years.
*** p<.01, ** p<.05, * p<.1; Reference categories in italics. Standard errors are in brackets.
                                                                                                                         Page 44



Figure A2.1. Histograms from the predicted scores from the logit models used for matching

                                            (a) Predicted scores for 2002 and 2005
                                                                    0                             1




                                                6
                                                4
                                 Density




                                                2
                                                0




                                                        .2   .4         .6     .8       .2   .4            .6   .8
                                                                             Pr(year02_05)
                                             Graphs by year02_05




                                           (b) Predicted scores for 2005 and 2008
                                                                   0                              1
                                            6
                                            4
                         Density




                                            2
                                            0




                                                    0              .5               1   0         .5                 1
                                                                             Pr(year05_08)
                                           Graphs by year05_08




                                            (c) Predicted scores for 2002 and 2008
                                                                   0                                  1
                                            6
                                            4
                       Density




                                            2
                                            0




                                                    0              .5               1   0             .5             1
                                                                             Pr(year02_08)
                                           Graphs by year02_08



  Note: Predicted scores are based on the models given in Table A8 above and show the regions of the common
                              support for the matching of individuals over the years.
                                                                                                              Page 45


Table A2.3. Informal Payments in Outpatient Services: Results from Tobit & Heckman
Selection Models
                                                                     HECKMAN SELECTION MODEL WITHOUT
                                          TOBIT MODEL
                                                                                      EXCLUSION
           VARIABLES
                                        Log of the amount         Zero or larger payment         Log of the amount
                                               paid                  (selection model)                  paid
Constant                                1.465**      (0.642)        -0.218        (0.139)       5.692***      (0.258)
(Year 2002)
Year 2005                              -0.736***       (0.150)    -0.258***        (0.050)       0.125**      (0.058)
Year 2008                              -0.670***       (0.197)    -0.228***        (0.056)        -0.037      (0.075)
Rural area                               0.092         (0.146)      0.008          (0.049)      0.147***      (0.050)
(Stratum central)
Stratum coastal                        0.490***        (0.153)     0.190***        (0.052)      -0.165***     (0.055)
Stratum mountain                       -0.876***       (0.183)    -0.296***        (0.061)        0.067       (0.072)
Stratum Tirana                         -1.032***       (0.223)    -0.336***        (0.074)        0.104       (0.090)
Gender (“0� male, “1� female)            0.133         (0.128)       0.031         (0.042)        0.012       (0.045)
Age (in years)                          0.047**        (0.023)       0.002         (0.005)        0.006       (0.008)
Age squared                            -0.001***       (0.000)      -0.000*        (0.000)        -0.000      (0.000)
(Civil status – married)
Civil status - divorced and widow        -0.127        (0.196)         -              -          -0.031       (0.071)
Civil status – single                     0.385        (0.261)         -              -          0.127        (0.092)
Chronic illness                           0.014        (0.176)       0.003         (0.059)       0.078        (0.062)
Health rate (“1� very bad - “5�
                                         -0.148        (0.097)      -0.043         (0.032)       -0.043       (0.036)
excellent)
Have health insurance license          -0.779***       (0.143)    -0.259***        (0.047)             -         -
(Primary/No education)
Secondary education                       0.026        (0.172)      -0.015         (0.058)       0.116*       (0.062)
University education                     -0.430        (0.323)      -0.162         (0.108)       0.213*       (0.128)
(Nuclear family with children)
Nuclear families head and spouse         0.985         (0.622)      0.365*         (0.210)       -0.270       (0.216)
Extended families                       0.553*         (0.319)      0.151          (0.107)       0.095        (0.117)
Quintile 1                              -0.028         (0.213)      -0.011         (0.072)       0.041        (0.077)
Quintile 2                              -0.020         (0.205)      0.012          (0.069)       -0.109       (0.073)
(Quintile 3)
Quintile 4                              0.370**        (0.188)      0.127**        (0.063)       -0.002       (0.067)
Quintile 5                                0.188        (0.192)       0.050         (0.065)      0.194***      (0.069)
Employment: public official               0.133        (0.335)       0.050         (0.111)          -            -
Employment: health practitioners         -0.569        (0.494)       -0.193        (0.165)          -            -

/sigma                                 3.168***        (0.082)         -              -                -         -
/athrho                                    -              -       -0.408***        (0.143)             -         -
/lnsigma                                   -              -       -0.293***        (0.046)             -         -

Log likelihood                             -4272.674                                   -3585.060
Number of observations                        4707                                        4707
 R2 (pseudo)                                  0.032                                         -
rho chi2 (1)                                                                            -0.387**
note: *** p<.01, ** p<.05, * p<.1; Reference categories in italics. Standard errors are in brackets.
                                                                                           Page 46


Table A2.4. Informal Payments in Outpatient Services: Changes Over Time in the
Likelihood of Making Informal Payments
                                                                  OUTPATIENT SERVICE
Year 2002 - 2005
      Nearest - neighbor model                                   -0.217***       (0.048)
      Kernel model                                               -0.208***       (0.043)
Year 2005 - 2008
      Nearest - neighbor model                                     0.015         (0.081)
      Kernel model                                                 0.054         (0.071)
Year 2002 - 2008
      Nearest - neighbor model                                    -0.038         (0.083)
      Kernel model                                                -0.040         (0.079)
note: *** p<.01, ** p<.05, * p<.1 Standard errors in brackets.
                                                                                                              Page 47


Table A2.5. Informal Payments in Outpatient Services: Changes Over Time in Individual
Characteristics Significantly Associated with Making Informal Payments
                              DIFFERENCE                       DIFFERENCE                      DIFFERENCE
                            BETWEEN YEARS           S.E      BETWEEN YEARS          S.E      BETWEEN YEARS           S.E
                               (2002-2005)                     (2005-2008)                      (2002-2008)
Rural area                        -0.008          (0.025)        0.134***          (0.043)         0.077*          (0.041)
Stratum coastal                 0.072***          (0.023)          0.019           (0.042)         -0.005          (0.038)
Stratum central                   -0.030          (0.023)          0.015           (0.039)        0.075**          (0.037)
Stratum mountain                -0.075***         (0.020)          0.048           (0.033)         0.021           (0.036)
Stratum Tirana                    0.034*          (0.021)        -0.082**          (0.042)       -0.091***         (0.035)
Gender                           0.056**          (0.025)        -0.290***         (0.045)       -0.251***         (0.042)
Age                             2.596***          (1.005)         -3.251*          (1.979)       -4.585***         (1.891)
Age square                      167.874*          (89.401)       -359.564*        (181.26)      -536.648***       (167.81)
Civil status - married          0.053***          (0.023)         -0.004           (0.043)       -0.099***         (0.040)
Civil status - divorced
                                  0.010           (0.017)         -0.013           (0.043)         0.020           (0.036)
and widow
Civil status single             -0.038***         (0.014)          0.021           (0.027)       0.088***          (0.025)
Chronic illness                   0.036           (0.023)        -0.088**          (0.042)       -0.095***         (0.039)
Health rate                     -0.157***         (0.040)         0.167**          (0.076)       0.201***          (0.068)
Health insurance status:
                                0.088***          (0.025)         -0.036           (0.045)         -0.039          (0.042)
normal
Health insurance status:
                                -0.029***         (0.012)          0.007           (0.027)         0.001           (0.023)
war/other invalid
No health insurance             -0.058***         (0.024)          0.028           (0.041)         0.038           (0.040)
Without education               -0.052***         (0.010)         -0.024           (0.017)       -0.052***         (0.021)
Primary education               0.059***          (0.025)         -0.023           (0.023)         -0.025          (0.041)
Second education                  -0.005          (0.023)         0.061*           (0.037)         0.049           (0.035)
University education              -0.001          (0.013)          0.044           (0.029)         0.028           (0.023)
Nuclear families head
                                0.015***          (0.005)         0.014*           (0.008)       0.013***          (0.004)
and spouse
Nuclear families couple
                                  0.026           (0.018)          0.036           (0.032)         0.021           (0.030)
and children
Extended families                 0.007           (0.009)         -0.019           (0.024)         -0.006          (0.017)
Quintile 1                        -0.009          (0.017)         0.051*           (0.031)       0.110***          (0.027)
Quintile 2                      0.059***          (0.018)          0.010           (0.033)       0.110***          (0.029)
Quintile 4                        0.013           (0.021)          0.013           (0.038)         0.042           (0.035)
Quintile 5                       -0.050**         (0.023)           -0.037        (0.041)      -0.292***           (0.040)
Note: Results are displayed for the matching using 5-nearest neighbour estimator. The caliper width is 0.01. Stars
indicate : *** p<.01, ** p<.05, * p<.1 Standard errors in brackets.
                                                                                                            Page 48


Table A2.6. Informal Payments in Inpatient Services: Results from Tobit & Heckman
Selection Models
                                                                      HECKMAN SELECTION MODEL WITHOUT
                                         TOBIT MODEL
                                                                                  EXCLUSION
           VARIABLES
                                       Log of the amount         Zero or larger payment        Log of the amount
                                              paid                  (selection model)                 paid
Constant                               3.415**      (1.388)        0.403*        (0.224)      6.789***      (0.490)
(Year 2002)
Year 2005                               -0.090       (0.305)       -0.045       (0.071)        0.220**      (0.110)
Year 2008                               -0.718       (0.492)     -0.271***      (0.087)        0.357**      (0.178)
Rural area                               0.182       (0.321)       0.051        (0.073)         -0.069      (0.112)
(Stratum central)
Stratum coastal                        0.878**       (0.360)       0.170**      (0.085)         0.067       (0.131)
Stratum mountain                      -1.816***      (0.340)     -0.333***      (0.077)       -0.461***     (0.129)
Stratum Tirana                        -1.514***      (0.549)      -0.286**      (0.124)         0.079       (0.206)
Gender (“0� male, “1� female)            0.261       (0.283)        0.060       (0.064)         -0.106      (0.104)
Age (in years)                          -0.015       (0.048)        0.004       (0.007)         -0.008      (0.016)
Age squared                              0.000       (0.000)        -0.000      (0.000)         -0.000      (0.000)
(Civil status – married)
Civil status - divorced and widow       -0.658       (0.470)             -         -           0.010        (0.154)
Civil status – single                   -0.426       (0.539)             -         -           -0.103       (0.169)
Chronic illness                        -0.828**      (0.396)         -0.187**   (0.090)        -0.019       (0.146)
Health rate (“1� very bad - “5�
                                        0.141        (0.224)          -0.026    (0.052)       0.296***      (0.083)
excellent)
Have health insurance license         -0.960***      (0.311)     -0.208***      (0.062)              -         -
(Primary/No education)
Secondary education                     0.443        (0.365)          0.090     (0.085)            0.009    (0.130)
University education                    0.550        (0.716)          0.045     (0.168)            0.308    (0.255)
(Nuclear family with children)
Nuclear families head and spouse        -1.653       (1.663)          -0.347    (0.369)        -0.246       (0.644)
Extended families                       -0.169       (0.683)          -0.057    (0.153)        0.364        (0.258)
Quintile 1                             -0.922**      (0.425)         -0.179*    (0.097)        -0.011       (0.158)
Quintile 2                              -0.142       (0.416)          -0.038    (0.096)        -0.091       (0.151)
(Quintile 3)
Quintile 4                               0.158       (0.427)          0.044     (0.099)        -0.067       (0.153)
Quintile 5                               0.533       (0.436)          0.089     (0.102)        0.296*       (0.156)
Employment: public official              0.649       (0.739)          0.156     (0.148)           -            -
Employment: health practitioners        -0.404       (0.890)          -0.085    (0.172)           -            -

/sigma                                5.061***       (0.135)          -            -                 -         -
/athrho                                   -             -        -1.058***      (0.149)              -         -
/lnsigma                                  -             -         0.476***      (0.052)              -         -

Log likelihood                             -3406.004                                   -2655.230
Number of observations                        1707                                       1707
R2 (pseudo)                                   0.021
rho chi2 (1)                                                                           -0.785***
note: *** p<.01, ** p<.05, * p<.1 Standard errors are in brackets.
                                                                                           Page 49


Table A2.7. Informal Payments in Inpatient Services: Changes Over Time in the
Likelihood of Making Informal Payments
                                                                   INPATIENT SERVICE
Year 2002 - 2005
      Nearest - neighbour model                                  -1.053***       (0.418)
      Kernel model                                               -0.904***       (0.381)
Year 2005 - 2008
      Nearest - neighbour model                                   2.774*         (1.415)
      Kernel model                                                1.920          (1.363)
Year 2002 - 2008
      Nearest - neighbour model                                    1.129         (0.917)
      Kernel model                                                 1.045         (0.869)
note: *** p<.01, ** p<.05, * p<.1 Standard errors in brackets.
                                                                                                             Page 50


Table A2.8. Informal Payments for Inpatient Services: Changes Over Time in Individual
Characteristics Significantly Associated with Making Informal Payments
                              DIFFERENCE                   DIFFERENCE                  DIFFERENCE
                               BETWEEN                      BETWEEN                     BETWEEN
                                                  S.E                          S.E                         S.E
                              YEARS (2002-                 YEARS (2005-                YEARS (2002-
                                   2005)                         2008)                     2008)
Rural area                         0.070         (0.052)         0.197*      (0.110)       0.258*         (0.149)
Stratum coastal                    0.017         (0.036)        0.226**      (0.086)        0.113         (0.120)
Stratum central                    0.042         (0.044)          0.003      (0.097)        0.090         (0.128)
Stratum mountain                   -0.034        (0.052)         -0.149      (0.108)        0.112         (0.146)
Stratum Tirana                     -0.025        (0.037)         -0.080      (0.078)     -0.317***        (0.110)
Gender                             0.031         (0.051)      -0.440***      (0.099)     -0.359***        (0.146)
Age                                1.635         (1.975)         -3.293      (4.224)    -16.765***        (5.730)
Age square                         78.798       (164.58)       -299.339     (356.06)   -168.024***       (517.98)
Civil status - married            0.097**        (0.047)         -0.003      (0.105)       -0.041         (0.141)
Civil status - divorced and
                                   -0.019        (0.028)         -0.023      (0.098)       -0.022         (0.128)
widow
Civil status - single             -0.069*        (0.037)          0.002      (0.076)        0.035         (0.089)
Chronic illness                   -0.098*        (0.051)          0.060      (0.110)     -0.332***        (0.139)
Health rate                        -0.075        (0.090)         -0.173      (0.187)        0.252         (0.260)
Health insurance status:
                                   0.073         (0.052)          0.102      (0.110)       -0.139         (0.149)
normal
Health insurance status:
                                   0.023         (0.024)         -0.083      (0.070)       -0.040         (0.072)
war/other invalid
No health insurance               -0.096*        (0.052)         -0.019      (0.110)        0.180         (0.149)
Without education                  -0.021        (0.015)         -0.004      (0.024)      -0.111*         (0.067)
Primary education                  0.001         (0.048)         -0.053      (0.046)        0.027         (0.139)
Secondary education                -0.001        (0.045)          0.128      (0.094)        0.040         (0.121)
University education               0.021         (0.021)          0.001      (0.051)        0.043         (0.071)
Nuclear families head and
                                   0.007         (0.005)         0.020*      (0.012)        0.015         (0.011)
spouse
Nuclear families couple
                                   0.032         (0.034)         -0.066      (0.078)        0.023         (0.077)
and children
Extended families                  0.001         (0.020)         -0.042      (0.055)        0.044         (0.052)
Quintile 1                         0.030         (0.041)         -0.038      (0.099)     0.256***         (0.100)
Quintile 2                         0.013         (0.041)          0.002      (0.082)        0.063         (0.118)
Quintile 4                         -0.026        (0.042)          0.042      (0.089)       -0.121         (0.126)
Quintile 5                         -0.036        (0.042)          0.017      (0.077)       -0.140         (0.132)
Note: Results are displayed for the matching using 5-nearest neighbour estimator. The caliper width is 0.01. Stars
indicate : *** p<.01, ** p<.05, * p<.1 Standard errors in brackets.