Anda di halaman 1dari 7

HEALTH ECONOMICS

Health Econ. 21: 55–61 (2012)


Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/hec.1811

HOW CAN WE INCREASE RESOURCES FOR HEALTH CARE IN THE


DEVELOPING WORLD? IS (SUBSIDIZED) VOLUNTARY HEALTH
INSURANCE THE ANSWER?

JACQUES VAN DER GAAG* and VID STIMAC


Amsterdam Institute for International Development, Brookings Institution, Washington, DC, USA

1. INTRODUCTION

In 2001, the Commission on Macroeconomics and Health estimated that a country needs to spend $34 per per-
son per year to make a package of basic health services available to all its citizens (WHO, 2001). In today’s
dollars, that number is probably closer to $50. Most low-income countries are currently spending only half
of that or even less. About 50% of this is public money. The main question in the health economic field for
low-income developing countries is thus how can we increase total resources for the sector? A simple answer
would be to increase foreign aid for health to those countries that are too poor to finance all health care from
domestic resources. On the surface, it appears that the international community has been highly successful in
achieving that goal. Foreign aid (official development assistance (ODA)) for health increased from $5.6 billion
in 1990 to 21.8 billion in 2007 (Ravishankar et al., 2009).
Although these are big numbers, we have to keep in mind that on a per capita basis, the amount of foreign
aid for health remains very low (about $4 per year for the average country receiving donor money for health).
Still, the creation of the Global Fund, the activities in the health field of the Gates Foundation, and the Global
Alliance for Vaccines Initiative (as well as the US President’s Emergency Plan for AIDS Relief program and
other bilateral aid efforts) have led to a significant increase in global resources for health care in the developing
world. Those new resources are mostly channeled through the government budgets of the recipient countries.
Perhaps surprisingly, the question remains: did this increase lead to a commensurate increase in healthcare
resources at the country level? This question comes to mind once one remembers one of the strongest and best
documented ‘laws’ in the health economic literature, namely that per capita health expenditure at the country
level is almost completely determined by per capita GDP, no matter what type of health system we are talking
about, or what health policies are put in place.
In this short note, we will take a closer look at the impact of ODA for health on per capita health spending at
the country level. Our results suggest the existence of significant crowding out effects. We then ask how these
crowding out effects can (at least in part) be avoided and put (subsidized) voluntary health insurance forward as
a possible solution. We will discuss the significant implementation problems inherent in rolling out voluntary
health insurance in low-income countries and conclude with some suggestions on how to move forward.

*Correspondence to: Amsterdam Institute for International Development, Brookings Institution, Washington, DC, USA. E-mail:
jvandergaag@brookings.edu

Copyright © 2011 John Wiley & Sons, Ltd.


56 J. VAN DER GAAG AND V. STIMAC

2. DETERMINANTS OF HEALTH EXPENDITURES

The economic literature about the determinants of health expenditures (at the country level) dates at least back
to 1977 when Joseph Newhouse published a short article in the Journal of Human Resources (Newhouse,
1977). Using cross-sectional data for OECD countries, Newhouse showed that variation in the countries’ in-
come level (GDP per capita) can explain more than 90% of the variation in the countries per capita expenditures
for health care. The implied income elasticity of the relationship between health expenditures and income is
larger than one. Thus, one should expect that with income growing, health care will take an ever larger part
of the financial resources available in a country, a phenomena we are now, three decades later, all familiar with.
In the decades after the publication of Newhouse’s article, many researchers have reexamined the tight relation-
ship between a country’s income level and healthcare expenditures. They investigated the effect of fee-for-service
versus capitation systems, various institutional and regulatory arrangements, and the relative roles of the public and
private sectors. They use new and expanded data sets and more sophisticated estimation techniques.
An excellent review of this literature (Gerdtham and Johnsson, 2000) is published in the Handbook of
Health Economics. We will not repeat all their results here, except for the general conclusion:

A common and extremely robust result. . .is that the effect of per capita GDP (income) on expenditure is
clearly positive and significant, that the income elasticities are. . .close to unity or even higher than unity.
This result appears to be robust to the choice of variables included in the estimation models, data, the choice
of conversion factors and the methods of estimation. (Gerdtham and Johnsson, 2000, p. 45)

To illustrate the strength of this result, we estimated a log-linear equation explaining the variation in per
capita healthcare expenditures as a function of per capita income for a sample of 146 countries using 2006 data.
The results are given in Table I.
Clearly, ‘the first law of health economics’ is still valid: the impact of income on expenditure is very large
(96% of the cross-country variation is explained by income alone), and the income elasticity is very precisely
estimated and above one. Later on, we will investigate what this means for the ability of development aid to
increase overall spending on health, but first, we will re-estimate this equation for sub-Saharan countries only,
where development aid constitutes a relatively large part of the government budget.

3. HEALTH EXPENDITURES IN SUB-SAHARAN AFRICA


Table II shows cross-sectional results for 43 sub-Saharan African (SSA) countries for the year 2006. The in-
come elasticity is slightly lower than for the overall sample but not statistically different from 1.0. Eighty-six
percent of the variation in health expenditures per capita is explained by variation in GDP per capita.
It stands to reason that governments that put a high value on health outcomes will spend relatively more on
health care. The question is whether the increase in public expenditures (either autonomous or induced by foreign
aid) leads to an increase in total health expenditures. To test for this, we add the public share in overall healthcare
financing to the equation. As we see in Table III, the effect of more government financing on total expenditures is
zero. It appears that there is complete crowding out of private expenditures when governments expand their financ-
ing. Although some crowding out was to be expected (Cutler and Gruber, 1996; van der Gaag and Stimac, 2008),

Table I. Log gross health expenditures/capita 2006, full sample


Log GDP/capita 1.10 ***
(0.02)
Constant 3.66 ***
(0.15)
N 148
R-squared 0.96
***Indicates significant at 1% level.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. 21: 55–61 (2012)
DOI: 10.1002/hec
HOW CAN WE INCREASE HEALTH CARE RESOURCES IN THE DEVELOPING WORLD? 57

Table II. Log health expenditures/capita 2006, sub-Saharan Africa


Log GDP/capita 0.97 ***
(0.06)
Constant 2.83 ***
(0.38)
N 43
R-squared 0.86
***Indicates significant at 1% level.

Table III. Log health expenditures/capita 2006, sub-Saharan Africa


Log GDP/capita 0.97 ***
(0.07)
Log public share of health expenditures 0.001
(0.17)
Constant 2.82 ***
(0.62)
N 43
R-squared 0.91
***Indicates significant at 1% level.

the fact that complete crowding out appears to be taking place is worrisome if more financial resources for health
care are necessary to cover growing health needs. Simply expending public financing is not going to do it.
This brings us to the other option for increasing financial healthcare resources: foreign aid (ODA). In
Figure 1, we show how large the share of foreign aid is in overall healthcare financing in SSA.
In an earlier paper (van der Gaag and Stimac, 2008) using data of all countries that receive ODA for health, we
found no effect of ODA on overall healthcare resources. We argued that ODA of about $4 per capita per year (av-
eraged over all recipient countries) may have been too small to detect any affect. As we see in Figure 1, ODA for
health is relatively high in SSA countries. Because almost all ODA goes to the government, one would expect the
public share of overall financing to go up with an increase with ODA. In Table IV, we show whether this is true.
Indeed, ODA does increase the public share (as one would expect because ODA is such a big part of public
expenditures in SSA) but not dollar for dollar. The results imply (for the median SSA country) a ‘leakage’ at the
government level of about 25%. And as we have seen in Table III, even an extra government effort to increase
spending on health does not have a positive impact on the overall level of resources. ‘Leakage’ at the
government level and crowding out of private resources by an increase in public money appear to greatly
hamper our ability to increase overall resources for health in low-income countries.1

4. HOW CAN CROWDING OUT BE AVOIDED?

Crowding out takes place at all levels of decision making. At the donor level, increased ‘specialization’ leads
some donors to pull out of specific sectors that have become the ‘responsibility’ of other donors. It is, for
instance, striking that the African Development Bank in 2008 committed just $36 million to new healthcare
projects for sub-Saharan countries, out of an overall portfolio of more than $2 billion. This is a continent that
suffers from half of the global disease burden and spends just 1% of global health resources.
At the government level, there is leakage of various kinds, the most benign being a substitution of domestic
resources from sectors that receive development aid to other sectors. And in the market place, we find evidence
of significant crowding out of private resources by public resources.

1
Note that these cross-sectional regressions do not prove anything and suffer from some data problems and methodological issues, such as
the possibility of endogenous ODA. However, the results are so strong that the direction of the effects is likely to be as shown. More de-
tailed research on the underlying causes of these results would be very beneficial.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. 21: 55–61 (2012)
DOI: 10.1002/hec
58 J. VAN DER GAAG AND V. STIMAC

Source: OECD data base


Figure 1. Share of foreign aid in overall healthcare financing in sub-Saharan Africa. Source: OECD database

Table IV. Log public share of health expenditures, sub-Saharan Africa


Log GDP/capita 0.10 *
(0.05)
Log external resources health/capita 0.23 ***
(0.07)
Constant 2.94 ***
(0.07)
N 43
R-squared 0.36
*Indicates significant at 10% level; ***indicates significant at 1% level.

Under these circumstances, increasing overall resources for the healthcare sector is a daunting task but not
necessarily an impossible one. One option would be to put conditionalities in place, for instance, in the form of
matching grants, to try to avoid crowding out (or ‘leakage’) at the government level; although, in general, these
types of conditionalities are hard to enforce. Moreover, because almost all aid money goes to the government, it
would not solve the largest crowding out problem.
The introduction of voluntary (possibly subsidized) health insurance for low-income households may be a
way to significantly reduce crowding out. Starting from the observation that in low-income countries, private
contributions constitute about 50% of all financial resources for health care, it stands to reason that govern-
ments’ and/or donors’ efforts to increase overall resources should do everything possible to keep these signif-
icant private resources within the system. As we have seen, that is currently not the case. Unfortunately, these
private resources consist almost entirely of out-of-pocket expenditures, thus putting low-income households at
risk of falling below the poverty line because of unexpected high expenditures for health care. Health insurance
is the obvious answer to deal with this problem: it keeps the private (household) money in the system while
greatly reducing the risk that households face as a result of health shocks. Unfortunately, insurance is rare in
low-income countries and, it proves to be difficult to implement it successfully and in a sustainable way.
A recent paper in Health Economics provides a good example: it describes a carefully designed randomized
experiment involving health insurance for informal sector workers in Nicaragua (Thornton et al., 2010). The
project is implemented with the explicit goal to shed light on three of the most pervasive problems that practi-
tioners in this field face: low uptake, difficulties with payment collection and re-enrollment, and adverse selec-
tion. The results of the study are not encouraging: uptake was indeed very low (just 20%), and re-enrollment
was just 10% of initial enrollment. Adverse selection did not appear to be an issue.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. 21: 55–61 (2012)
DOI: 10.1002/hec
HOW CAN WE INCREASE HEALTH CARE RESOURCES IN THE DEVELOPING WORLD? 59

This lackluster effective demand for health insurance contrasts starkly with the growing literature on willing-
ness-to-pay (WTP) for insurance (van der Gaag, 2009; Gustafsson-Wright et al., 2007; van der Gaag and
Rinke de Wit, 2011). A typical WTP study will show potential uptake of 80% or higher. The top quintile of
the income distribution does, understandably, not show much interest. The bottom quintile, although eager
to participate, vastly overestimates its ability to pay, but the middle three quintiles show realistic results, with
a WTP premiums that amount to between 4% and 6% of overall consumption levels (in line with their average
out-of-pocket outlays for health care).
Furthermore, where low cost health insurance has been rolled out on a large scale, mostly in middle-income
countries, the results are generally very positive (Escobar et al., 2010). But the roll-outs of relatively small
health insurance projects for low-income informal sector workers do not always live up to the expectations
raised by the WTP literature. Why is that? There are at least five reasons for this, none of which is sufficiently
understood, and all warrant more (and in some cases different) research.

4.1. Supply interventions


Trying to roll out voluntary (private) health insurance in an environment where the government offers ‘free health
care for all’ seems like a waste of time. However, many poor and low-income people vote with their feet and bypass
public health facilities to visit private facilities where they pay a fee for service. Clearly, they are willing to pay for an
upgrade in quality, given the often awful situations in public clinics and hospitals. This quality is not necessarily
medical/technical quality. More convenience (shorter waiting times and more cordial treatment by the administrative
personnel) may already be enough to make the shift from public to private. Any roll-out of voluntary health insurance
under such circumstances should therefore be accompanied with a highly visible increase in the quality of participat-
ing clinics. We do not know how strong this supply intervention needs to be, and it is likely to be context specific.
But it would also be a relatively easy thing to find out by using focus groups discussions and other marketing tools.

4.2. Time preferences and discount rates


Poor households in low-income settings experience a myriad of risks; droughts, floods, unstable income and
employment opportunities, and disease and early death (Fafchamps, 1999). As a result, they may be more tol-
erant to risk (be less risk averse) than individuals who grow up in less risky environments. In addition, very low
and volatile daily (cash) income levels may lead to very high individual discount rates. Low risk aversion and
high discount rates may explain the apparently low effective demand for prepaid health insurance among low-
income people. Furthermore, being struck by a deadly disease (notably HIV/AIDS) may further reduce risk
aversion because those individuals may ‘value the future less’ (Lammers, 2008). The impact of the level of risk
aversion and the variation in individual discount rates in the developing world on intertemporal decision mak-
ing has been extensively researched for agricultural purposes (Binswanger, 1980; Duflo et al., 2008). Much less
is known about the impact on decision making in other areas, such as buying prepaid health insurance. Further
research in this area, through experimental research and real-life field pilots, is likely to increase our ability to
design and roll out health insurance products that are more attractive to the poor.

4.3. Endogenous rationality


Thornton et al. (2010) observed that households do not enroll in health insurance even when their actual health
expenditures exceed the insurance premium. Such apparently ‘irrational’ behavior is not uncommon in low-
income settings. Although rational choice theory has been a powerful corner stone of (development) economics,
an increasing number of experimental studies show that the generalized axiom of revealed preference (GARP)
is often violated (Sippel, 1997). Interestingly, evidence also suggests that such GARP violations are less com-
mon in environments that resemble conventional (competitive) market situations (List and Millimet, 2008). A
recent study by Cecchi shows that sesame farmers and brokers, who are used to rely on personal contacts for
their trade, significantly reduce ‘irrational behavior’ (i.e., GARP violations) after just 1 day of exposure to real

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. 21: 55–61 (2012)
DOI: 10.1002/hec
60 J. VAN DER GAAG AND V. STIMAC

market competition in an experimental auction setting (Cecchi, 2010). Thus, rational decision making can be
‘learned’, either in experimental or in real-life settings. As for the latter, when it comes to health insurance,
researchers and practitioners would be well advised to be patient. (Large) health expenditures are rare events,
so the benefits of prepaid insurance accrue only slowly over time. Extensive information campaigns up-front,
and targeted follow-up campaigns during the roll-out of the insurance, may be necessary to speed up the learn-
ing process. It may easily take a few years before individuals find out for themselves how beneficial formal in-
surance can be before they give up their so-called ‘irrational’ behavior.

4.4. Trust and social capital


Individuals who grow up in low-income settings do have an experience with informal insurance mechanisms,
which are quite different than the formal insurance discussed in this note (Fafchamps, 1999). Unfortunately,
these informal arrangements are less efficient than formal insurance. Moreover, they break down in the case
of covariate risks (such as a drought or an epidemic). Still, mutual informal savings arrangements, voluntary
exchange of labor services (especially in agriculture), loans from (and to) neighbors and family members, cash
holdings, and stocks of food supply are all examples of informal insurance mechanisms to protect against the
effects of external shocks (and to smooth consumption over time). The closer knit a society is, that is, the more
social capital it possesses, the stronger the protection (Gustafsson-Wright, 2007).
It would be worthwhile to investigate to what extent this ‘fabric of society’ can be harnessed to increase the
understanding of the benefits of formal insurance, to help increase enrollment and re-enrollment, or even to en-
force (group based) mandated insurance. A similar mechanism has been successfully employed in the field of
microfinance, where the strength of the groups results in very low default rates.

4.5. Logistics
Finally, we need to draw attention to the logistics of rolling out health insurance in resource poor settings. From the
definition of a household (to identify household members who are coinsured) to the administration of insurance
claims and subsequent payments, to the prevention of outright fraud, to the ease with which people can enroll
and re-enroll, the practical implementation problems loom large. Chances are that this, in addition to their limited
ability to raise the necessary resources, is one of the reasons why governments so often fail to deliver on their prom-
ise to ‘provide health insurance for the entire population’. The private sector may be better positioned to, for in-
stance, experiment with new technologies to overcome some of these implementation problems. In Africa, millions
of poor households now have access to banking services by cell phone. A similar development in the insurance
sector may greatly facilitate enrollment and re-enrollment, as well as solve other administrative problems. For that
to happen, more field experiments are necessary, with robust evaluations of their success or failure.

5. CONCLUSION
Low-income countries do not have enough domestic resources to provide a basic package of health services to
the entire population. Many have to rely on significant foreign aid for health and, given current growth fore-
casts, will have to do so for years if not decades to come. The vast majority of the aid resources go to the public
sector. Through a combination of ‘leakage’ at the government level and crowding out of private resources, the
effect of aid on overall resources is being greatly reduced. This, in part, may explain the dismal record of ‘aid
effectiveness’. We argue that much more attention needs to be given to the interactions among the various
sources of financing health care (public, private, and donor) and put the introduction of voluntary health insur-
ance forward as a partial solution. Insurance reduces the risks that households face as the result of health shocks
and keeps the private resources for health in the system. Nevertheless, the introduction of private voluntary
health insurance faces many problems (Schellekens et al., 2009; Escobar et al., 2010). We listed a few and sug-
gested research strategies to get a better understanding on how to overcome them. Recent research using

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. 21: 55–61 (2012)
DOI: 10.1002/hec
HOW CAN WE INCREASE HEALTH CARE RESOURCES IN THE DEVELOPING WORLD? 61

randomized controlled experiments have made some progress (Thornton et al., 2010; Banerjee and Duflo,
2011; Karlan and Appel, 2011.), but the projects often have a limited focus or a short time horizon, while a
comprehensive approach and patience are necessary for the sustainable implementation of health insurance.
The complete failure of a top–down government-driven approach to health sector development in low-
income countries is sufficient reason to look for alternative approaches. Private sector initiatives are showing
up at an increasing rate (International Finance Corporation, 2011). The donor community has become increas-
ingly generous towards the healthcare sector. It would be good if some of that donor money would be directed
to stimulate private sector initiatives so that the government and the private sector can start working together to
improve the often dismal health situation in low-income countries.2 Rigorous evaluation of the success or fail-
ure of such initiatives should be an integrated part of the exercise.

REFERENCES

Banerjee AV, Duflo E. 2011. Poor Economics. Public Affairs: New York.
Binswanger H. 1980. Attitudes towards risk: experimental measurement in rural India. American Journal of Agricultural
Economics 62(3): 395–407.
Cecchi F. 2010. Does market experience promote rational choice? Experimental evidence from rural Ethiopia, University of
Wageningen, mimeo.
Cutler DM, Gruber J. 1996. The effect of Medicaid expansion on public insurance, private insurance and redistribution.
American Economic Review 86(2): 378–83.
Duflo EM, Kremer J, Robinson J. 2008. Why don’t farmers use fertilizers? Evidence from field experiments in Kenya.
American Economic Review 98: 482–488.
Escobar MC, Griffin CC, Shaw RP (eds). 2010. The Impact of Health Insurance in Low-and Middle Income Countries.
Brookings Institution Press: Washington.
Fafchamps M. 1999. Rural Poverty, Risk and Development. FAO, Center for the Study of African Economies. Oxford
University Press: Oxford.
Gerdtham U, Johnsson B. 2000. International comparisons of health expenditures: theory, data, and econometric analysis.
In Handbook of Health Economics, vol. 1, Culyer AJ, Newhouse JP (eds). Elsevier Science BV: Amsterdam.
Gustafsson-Wright E. 2007. Baring the threads; social capital, vulnerability and the well-being of children in Guatemala.
Ph.D. Thesis, Tinbergen Institute, University of Amsterdam.
Gustafsson-Wright E, Asfaw A, van der Gaag, J. 2009. Willingness-to-Pay for health insurance: an analysis of low cast
health insurance in Namibia, Social Science & Medicine 69(9): 1351–1359.
International Finance Corporation. 2011. Healthy Partnerships: How Governments can Engage the Private Sector to Im-
prove Health in Africa. World Bank: Washington.
Karlan D, Appel J. 2011. More than Good Intentions. Penguin Group: New York.
Lammers L. 2008. HIV/AIDS, risk and intertemporal choice. Ph.D. Thesis, Tilburg University.
List JA, Millimet DL. 2008. The market, catalyst for rationality and filter of irrationality. Journal of Economic Analysis &
Policy 8(1).
Newhouse J. 1977. Medical-care expenditures: a cross-national survey. Journal of Human Resources 12(1): 115–125.
Ravishankar N, Gubbins P, Cooley RJ, Michaud KM, Jamison DT, Murray CL. 2009. Financing of Global Health: tracking
development assistance for health from 1990 to 2007. The Lancet 373: 2113–2124.
Schellekens O, de Beer I, Lindner M, van de Vugt M, Schellekens P, Rinke de Wit T. 2009. Innovation in Namibia, pre-
serving private health insurance and HIV/AIDS treatment. Health Affairs 28(6): 1779–1806.
Sippel R. 1997. An experiment on the pure theory of consumer’s behavior. The Economic Journal 107: 1431–1444.
Thornton RL, Hatt LE, Field EM, Islam M, Solis Diaz F, Azucena Gonzalez M. 2010. Social security health insurance for
the informal sector in Nicaragua: a randomized evaluation. Health Economics 19(S1): 181–206.
van der Gaag J. 2009. Health care for the World’s poorest: is voluntary (private) health insurance an option? In The Poorest
and the Hungry, van Braun J, Vargas Hill R, Pandya-Lorch R (eds). IFPRI: Washington.
van der Gaag J, Rinke de Wit T (eds). 2011. The Okambilimbili Health Insurance Project. ParmAccess: Amsterdam.
van der Gaag J, Stimac V. 2008. Towards a new paradigm for health sector development. Technical Partner Paper #3, The
Rockefeller Foundation, September.
World Health Organization. 2001. Commission for Macro-Economics and Health. WHO: Geneva.

2
Full disclosure: The first author is engaged in the long-term evaluation of voluntary private insurance projects in Kenya and Nigeria, which
are sponsored by the Dutch government through the Health Insurance Fund, see www.pharmaccess.org.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. 21: 55–61 (2012)
DOI: 10.1002/hec

Anda mungkin juga menyukai