W_  F2L2
POLICY RESEARCH WORKING PAPER   2492
Health Insurance Reform                                                Argentina, Brazil, Chile, and
Colombia have reformed the
in Four Latin American                                                  ways health insurance and
Countris .health care are organized and
delivered, have extended
formal coverage to previously
Theory  and  Practice                                                   marginalized groups, and
have tried to finance this
extension fairly. Each has
William Jack                                                            reformed health insurance
differently.
The World Bank
Development Research Group
Public Economics
November 2000



POLICY RESEARCH WORKING PAPER 2492
Summary findings
Jack examines public economics rationales for public         delivered, have tried to extend formal coverage to
intervention in health insurance markets, draws on the       previously marginalized groups, and have tried to finance
literature of organizational design to examine alternative   this extension fairly.
intervention strategies, and considers health insurance        Colombia instituted an implicit two-tiered voucher
reforms in four Latin American countries-Argentina,          scheme financed through a proportional wage tax.
Brazil, Chile, and Colombia-in light of the theoretical         Chile's financing mechanism is similar but the
literature.                                                  distribution of benefits is less progressive, so the net
Equity has been the main reason for large-scale public     effect is less redistributive.
intervention in the health insurance sector, despite the       Argentina's remodeled obras system went halfway: the
well-known failures of insurance and health care markets    financing base is similar and there is some implicit
associated with imperfect information.                       redistribution from richer to poorer obras, but the
Recent reforms have sought less to make private            quality of insurance increases with income.
markets more efficient than to make public provision            On the face of it, Brazil's health insurance system is
more efficient, sometimes by altering the focus and          less redistributive than those of the other three countries,
function of existing institutions (such as the obras         as no tax is earmarked for financing health insurance.
sociales in Argentina) or by encouraging the growth of       But taxes paid by higher-income taxpayers are not
new ones (such as Chile's ISAPREs).                          reduced when they choose private insurance,
Generally, these four Latin American countries have        highlighting the problem of examining the health sector
reformed the ways insurance and care are organized and       independent of the general tax and transfer system.
This paper-a product of Public Economics, Development Research Group-was prepared as part of a regional analysis
of social risk management in Latin America and the Caribbean. Copies of the paper are available free from the World Bank,
1818 H Street NW, Washington, DC 20433. Please contact Hedy Sladovich, room MC2-609, telephone 202-473-7698,
fax 202-522-1154, email address hsladovich@worldbank.org. Policy Research Working Papers are also posted on the Web
at www.worldbank.org/research/workingpapers. The author may be contacted at wgj@georgetown.edu. November 2000.
(49 pages)
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about
development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The
papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this
paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the
countries they represent.
Produced by the Policy Research Dissemination Center



Health Insurance Reform in Four Latin American
Countries: Theory and Practice'
William Jack2
1 This was prepared as part of a regional analysis of social risk management in the Latin
America and Caribbean region of the World Bank. Thanks, without implication, to
Girindre Beeharry, Chico Ferreira, Charlie Griffin, Jeff Hammer, Jeni Klugman, Jerry
LaForgia, Maureen Lewis, Andre Medici, and Miryan Sato for discussions.
2Room ICC 580, Department of Economics, Georgetown University, Washington, D.C.,
20057. Email: wgj@georgetown.edu






1 Introduction
Health care expenses and lost labor earnings due to illness represent a major source of risk for
individuals and families. Exposure to such risks is costly in itself (if individuals are risk averse),
but can also have long-term effects especially on the poor. Selling assets, withdrawing children from
school to care for ill parents, and exiting the labor market can leave low-income families trapped in
poverty.
Protection against the risks of ill health can be achieved by reducing the size and variability of the
underlying stochastic process, for instance by improving public goods that affect health outcomes
(pollution, etc.), and by spreading risks across individuals. This paper will address the second
of these risk reduction strategies, with particular emphasis on the design and organization of the
relevant institutions in Latin America.
Given its income, Latin America is on average a relatively healthy part of the world. For example,
Figure 1 shows data from 157 countries that reported information on GDP per capita and life
expectancy in 1997 (Latin American countries are represented as squares, the others as diamonds').
Visual inspection of the simple logarithmic regression indicates that most Latin American countries
have life expectancies at birth. equal to or significantly above those predicted by their income levels.
Health outcomes measured by infant mortality rates are less positive however, as shown in Figure
2. Similarly, the distribution of health care services within Latin American countries is skewed to
the better off. Recent analysis undertaken by PAHO, UNDP,2 and the World Bank, (Greene et
al. (1999)) indicates wide disparities in both health needs and access to medical care across income
groups. This, coupled with incomplete and fragmented insurance coverage, led several countries in
the region to adopt wide-ranging health sector reforms in the latter part of the 1980s and the 1990s
that continue today.
Large scale changes in health insurance and health care markets inevitably involve significant
'Source: World Development Indicators, 1999, World Bank.
2Pan American Health Organization, and United Nations Development F'rogram, respectively.
1



Long life?
90
80 -
70 -
( 60 -
a)
40-
30
0           10000        20000         30000         40000
GDP per capita
Figure 1:
2



Safe birth?
250 -
*0 200
E
? 150
2 100 
.2  50 
0                       w     1          .0              I
0          10,000        20,000        30,000       40,000
GDP per capita
Figure 2:
3



public intervention. The purpose of this paper is first, in section 2, to examine the arguments in
favor of such intervention from a public economics perspective. Having identified market failure and
redistributional rationales for public intervention, section 3 addresses the important issue of how the
government should intervene. This is effectively a question of organizational design, incorporating
ideas from industrial organization, contract theory, and the theory of the firm. In this light, section
4 undertakes a detailed examination of the reforms pursued in Colombia, Argentina, Brazil, and
Chile. These countries reflect a variety of routes toward the goals of expanding formal insurance
coverage and improving the efficiency of service delivery in their health systems. These directions
of reform are interpreted within the framework developed in sections 2 and 3. Section 5 concludes.
2 Reasons for government intervention in the health sector
The theoretical literature on the performance of insurance markets is well developed. Not all of
the market failures that may arise in such markets, however, necessarily justify public intervention.
This section examines the efficiency and equity reasons for intervention in health insurance markets,
paying specific attention to the informational constraints facing governments.
2.1 Market failure in the health insurance sector
It is useful to briefly review sources of inefficiency in the delivery of health insurance, and to examine
the extent to which public intervention can correct the associated market failures. The inefficiencies
derive mainly from information asymmetries and imperfect competition, and less from standard
public goods and externality characteristics.
Asyrmmetric information - moral hazard
The role of information in the performance of insurance markets has been widely appreciated. In
the health insurance literature, Feldstein (1973), Pauly (1968), and Zeckhauser (1970) showed how
asymmetric information at the ex post stage - that is, after an insured event has occurred - can
4



reduce the efficiency of equilibrium insurance contracts. This moral hazard occurs when insurance
contracts are written on the basis of endogenous incurred expenses and not on the basis of exogenous
health needs. This kind of insurance leads to over-consumption of care, the distortionary costs of
which are offset by reducing the level of insurance.
A similar inefficiency results from "ex interim" moral hazard, when precautionary actions can
be taken after the insurance contract is signed, but before uncertainty is resolved. In this case,
the inability to make insurance contracts contingent on such actions reduces the optimal level of
insurance. In both cases, the individual is second-best optimally exposed to some risk.
Within a partial equilibrium model neither source of moral hazard argues for public intervention,
unless one assumes unrealistically that the government has better information than private insurers.
On the other hand,Greenwald and Stiglitz (1986) showed that, taking a general equilibrium approach,
there may be a role for government intervention, even when the govemment does not have an
information advantage vis-A-vis private insurers. Their argument is simply that, through it's powers
of taxation and subsidization, the government can encourage desirable ex interim actions by altering
the prices of goods and services that have non-zero cross elasticities of demand with such actions.
Thus, taxing cigarettes may reduce the (assumed unobservable) ex interim action of smoking, thus
mirroring an efficient insurance contract. The scope for such Pareto improving intervention becomes
more limited as private insurance contracts become more sophisticated, e.g., by disallowing benefits
to smokers.
Asymmetric information -  adverse selection
While moral hazard derives from asymmetric information that is generated after contracts are
signed, adverse selection occurs in markets when information is held asymmetrically at the date
of contracting. A competitive insurance market in a population with heterogeneous ex ante risk
characteristics may perform inefficiently if insurance contracts cannot be differentiated on the basis
of these risks.
When only a single insurance policy is available, Akerlof's (1970) lemons problem may emerge,
5



with a proportion of individuals choosing not to purchase insurance. On the other hand, if multiple
contracts are available, then even when risk characteristics are unobservable it becomes possible for
insurers to charge low-risk individuals lower prices. All individuals will have some insurance in an
equilibrium (Rothschild and Stiglitz (1976)), but two problems may arise: first, the good risks will
not have full insurance, and second, an equilibrium may not exist.
Evidence of the lemons-type of equilibrium (in which a fraction of the population is uninsured)
has been provided in a number of studies (e.g., Cutler and Zeckhauser (1997), and Cutler and Reber
(1998)). The relevance of the Rothschild-Stiglitz equilibrium is possibly more debatable, since it is
typically high-risk individuals who have trouble obtaining full insurance against health risks, and
not low risks, as their model suggests. However, since risk is correlated with other determinants
of insurance coverage (such as income, education, etc.) it is probably imprudent to dismiss the
underlying model.
Government intervention cannot easily correct these market failures. In both models, universal
and uniform coverage can be mandated, but the resulting resource and risk allocations are not
Pareto-comparable with the initial equilibrium.
Imperfect competition
The models of adverse selection reviewed above identified failures of competitive insurance mar-
kets. But even in the absence of adverse selection problems, insurance markets may yield socially
sub-optimal resource and risk allocations if firms have market power. Such market power may de-
rive from information imperfections on the demand side, say contributing to switching costs (which
make it difficult for new firms to attract customers). Also, increasing returns in administrative costs
suggest that a somewhat concentrated industry is likely to be observed in equilibrium.
In standard industrial organization models, while market power typically leads to allocative in-
efficiency, what competition there is is generally welfare improving. However, in insurance markets
with information asymmetries, competition may sometimes have negative effects on allocative effi-
ciency. For example, when insurers are faced with a heterogeneous risk population they will have
6



incentives to sell policies only to low-risk individuals - i.e., those individuals to whom it is cheap to
provide insurance. If they cannot offer different policies to different risk types, then they may lower
the quality of the policies they do sell to make them sufficiently unattractive to high-risk individuals
(Jack, forthcoming). This kind of selection incentive might suggest public intervention to control the
extent, or at least type, of competition in the insurance market. I will refer to this effect as active
selection - the selection by firms, because of their policies, of good risks, as opposed to adverse
selection - the selection by firms, in spite of their policies, of bad risks.
Consumer protection
Quality of care
The examples of moral hazard above concentrated on the behavior of consumers when they are
at an informational advantage vis-A-vis the provider of insurance. One response by insurers is to
try to improve the information they have about consumers, by undertaking "utilization reviews" -
essentially checking that doctors are not providing "too much" care. In order to motivate doctors,
insurers may indeed give them stakes in the insurance company, converting it to a managed care
organization.
Such an organizational design is efficient, as long as the information asymmetry is removed -
that is, as long as information about health status (and the effects of medical interventions and
their costs) is held symmetrically by all parties. In practice the physician is the primary source of
this information, so that when acting as the patient's agent he confers an information advantage
on the patient vis-A-vis the insurer. However, when acting for the insurer, the physician may put
the patient at a disadvantage, and warranted treatments could be withheld. The usual competitive
forces that induce firms to keep quality high may not work well in this situation, and quality of
care could suffer. One can appeal to the literature on consumer protection and safety standards in
support of the role of government in markets with uninformed consumers.
Financial regulation
Insurance companies perform similar functions to banks. Banks facilitate inter-temporal trades
7



(saving and dissaving) implemented through contemporary inter-personal trades (lending and bor-
rowing), while insurance companies facilitate trades between uncertain states of nature implemented
through inter-personal pooling of current risks.
The transformation function of banks (converting short-term savings into long-term project
funds) results in a mis-match between the time profiles of assets and liabilities in banks' balance
sheets, introducing a moral hazard problem with regard to the behavior of asset managers. Similarly,
insurance companies tend to collect premiums in advance of the resolution of uncertainty, and may
optimally build up resources in order to self-insure against systemic risks. In this case, insurance
company managers must choose how and where to invest these funds. A feature of health insurance
that is not common to banks in this regard is the fact that the real value of the firm's assets is
a function of two factors: the quality of financial investments, and the costs of providing medical
services.
Dewatripont and Tirole (1994) use an incomplete contracts model to show that bank managers
can be given appropriate incentives to perform by transferring control from equity holders (who have
relatively weak incentives to interfere with management) to debt holders (whose incentives to inter-
fere are stronger) when bank performance as measured by the value of assets is poor. This allocation
of authority is in place of a more sophisticated, but infeasible, explicit performance contract. The
role of government is then to act as a representative of small, uncoordinated debtors, and the theory
rationalizes public take-overs of distressed banks as a means of providing incentives to managers.
A similar role can be ascribed to the government with respect to health insurance regulation. If
the value of a firm's assets falls enough, the government may wish to intervene and take over the
administrative functions (maybe contracting out such administration to another healthy insurer).
Life-time insurance and non-diversifiable risks
A final important failure of insurance markets is that they often do not provide life-time insurance.
Since individuals' health needs exhibit a degree of autocorrelation, insurance that is actuarially fair
only on an annual basis exposes the individual to a high variance of medical costs over the lifetime.
8



Part of the reason it is difficult to provide life-time insurance is that the future development of
medical care prices is itself uncertain, due mainly to the vagaries of technological and epidemiological
dynamics. These components of risk are systemic, so cannot easily be insured against (except inter-
temporally, across generations). The government faces the same constraints as private insurers
in terms of the nature of the systemic risks, but may be in a better position to facilitate inter-
generational trades due to its power of taxation.
2.2 Equity
Efficiency is a relatively uncontroversial goal of economic policy and organization. However, in the
health care and health insurance sectors, equity is equally, if not more, important in shaping policy.
This may be because individuals are more willing to accept differences in general income levels than
in the status of health across individuals.
At a conceptual level, having high medical needs, or being at high risk of needing medical at-
tention, reduces an individual's available budget set. In consequence, the government may wish to
redistribute resources between individuals with identical money incomes. In practice, the redistri-
bution from healthy to sick is often attempted by imposing uniform prices for health services and
for health insurance across individuals. Of course, charging uniform prices for doctor visits is not
redistributive if the sick must visit the doctor more often than the healthy.
On the other hand, in insurance markets, uniform pricing of insurance policies across individuals
is a feasible tool for redistribution from low risks to high risks. However, requiring private firms to
community rate may only exacerbate adverse selection and active selection problems that already
exist.
It is important to note that, even if selection issues were unimportant, it would still be only
second-best to require uniform insurance pricing. The first-best policy, of course, is to redistribute
income (lump-sum) from low risks to high risks, and to require each individual to buy insurance at the
actuarially fair price (i.e., to allow price discrimination by firms). Such redistribution is notoriously
9



difficult, and even more so when income inequality itself is high, as it is in many Latin American
countries. Indeed, a social welfare maximizing planner will likely wish to redistribute from rich to
poor, and from less risky to more risky. This multi-dimensional problem 6f redistributive taxation
is difficult, even when the government restricts itself to simple linear tax and transfer systems (Jack
(1999)).
Henriet and Rochet (1999) have recently analyzed the optimality of a uniform public insurance
system within the context of a similar multi-dimensional redistribution problem. In the absence
of moral hazard, they find that a comprehensive insurance policy providing full insurance to all
individuals is part of an optimal tax and insurance system. This result relies to some extent on the
assumption that individuals face the same distribution of losses, albeit with different probabilities.
As the demand for health care is income elastic, however, the financial cost of care consumed when ill
is an endogenous function of income (and is in fact limited by the individual's available income). This
suggests that the poor may prefer additional income transfers and less extensive public insurance to
being offered the same level of coverage as the non-poor. Alternatively, the rich may be willing to
pay somewhat higher taxes in order to have more comprehensive insurance than the poor.
This discussion leads us to expect that in the presence of health risk and income differentials
across the population, and in the absence of first-best redistributive taxation, governments will
likely wish to couple a progressive general tax source (e.g., an income tax) with a system of health
insurance (privately or publicly supphed - more on this later) that delivers subsidized insurance to
the poor, but allows coverage to increase with income.
One means of effecting such a graduated insurance profile is to have a mixed public/private system
of insurance, in which the government provides (or mandates) a given base level of insurance, and
individuals are permitted to top up their coverage through private purchases, or to opt out of the
public system and purchase private insurance. For example, Besley and Coate (1991) have shown
that public provision of insurance (of low enough quality) can be used as a redistributive tool as
long as individuals have access to supplementary private coverage. Using a political economic model
10



with three classes of voters, Gouveia (1997) also establishes the use of opting out as a means of
implementing non-uniform insurance coverage in the presence of a progressive income tax. Such
a system is supported by the rich, since the cost of purchasing their preferred level of (privately
provided) insurance is less that the taxes they would pay for a similar level financed through taxation.
The poor also support the mixed system because they have a lower demand for coverage - they
would prefer to save some of their taxes that would finance better coverage in order to purchase
other goods.3
3 The nature of government intervention
The arguments above have suggested that either due to market inefficiencies (mainly adverse se-
lection) and redistributive concerns, governments may wish to control individuals' choices about
insurance in certain ways. What the discussion does not tell us is how such control over choices
should be effected. In this section we examine the design of public interventions recognizing that
the productive efficiency of insurance is a function of the administrative costs incurred and the costs
of providing covered services. Costs of provision of insurance and medical care can be controlled
through explicit contractual arrangements between insurer and provider, organizational choices (e.g.,
vertical integration), and competition. These three avenues are considered in turn.
3.1 Motivating physicians
In many markets, providers of goods or services are paid on the basis of outputs, not inputs, providing
strong incentives for efficient production. It is well understood however that insured medical services
3There is a growing literature on normative and political economic models of the public provision of private goods
- see Blomqvist and Christiansen (1999) for a useful summary. Uniform public provision of a private good, such as
health insurance, can be welfare improving to the extent that imposing quantity constraints on some individuals'
choices weakens self-selection constraints in the optimal income tax problem. These ideas are further elaborated on
in Blomqvist and Christiansen (1998a) and (1998b).
11



are nearly never compensated in this fashion. For example, an insurance policy that reimburses
incurred costs gives the provider no incentive to choose an efficient input mix.
The underlying problem here is one of motivating physicians and hospital administrators. When
neither health output nor physician effort is directly purchasable (i.e., contractible) then effort
incentives may be difficult to generate without exposing the physician to undesirable risk. At two
extremes are the salaried physician and the decentralized fund holder. Under the first arrangement,
an insurer (possibly the government) pays the physician a fixed amount, independent of his supply
of effort, and reimburses non-physician expenses (e.g., laboratory tests). In the absence of ethical
concerns or job satisfaction (which are likely to be significant in many instances), doctors will tend to
substitute out of personal effort and into complementary inputs. On the other hand, a decentralized
fund holder, who is given a fixed budget to finance all incurred costs (including the cost of his own
effort) will have strong incentives to choose the right input mix.
The trade-off, as in any moral hazard problem, is between incentives and risk. A salaried physi-
cian faces little personal financial risk associated with expensive cases, since he is reimbursed for
other costs. On the other hand, a fund holder with good effort incentives must bear the full costs of
treating expensive cases, and so might be exposed to considerable risk. This generates a potential
alternative source of inefficiency: if providers attempt to reduce their risk exposure by inducing
hard to treat patients to switch physicians, then these active selection attempts can increase equi-
librium costs, or reduce quality. If instead the physician is provided with some insurance against
hard-to-treat cases directly by the payer, then his incentives to perform are weakened.
It should be clear from this discussion that the trade-offs between risk and incentives at the
provider level are not specific to the public sector. Private insurance companies also must induce
physicians to implement insurance contracts at minimum cost, and so face a similar optimization
problem. Some endogenous differences across the public and private sectors may emerge however that
imply different contractual relationships between payers (the government, or insurance companies)
and providers of medical care. For example, if public insurance is aimed at the poor, then in the
12



absence of accurate eligibility tests self-selection constraints may require that the public system
provide a relatively low quality of service. Inducing low effort from physicians who provide services
to the public system (they might still be in the private sector, not public sector employees) might be
easier than inducing the higher effort, and hence service quality, characteristic of private insurance
companies who, in equilibrium, serve the non-poor. One might expect then to see relatively flat
incentive schemes for physicians serving the public insurance system, and steeper compensation
schedules employed by private insurers.
Another reason that public and private insurance systems may provide different incentive schemes
to physicians is that physicians may represent a heterogeneous group. If they differ in their aversion
to risk, in their ethical priorities, or in their job satisfaction, it may be optimal to offer one kind of
compensation contract to one group of physicians, and another kind to a second group.
Once again, however, these arguments suggest reasons that we might see different ways of paying
physicians emerge in the delivery of health insurance, but they are not necessarily the outcome of
a public-private mix of insurance provision. For example, it is easy to imagine a purely private
market becoming vertically segmented, with some plans targeting low-demand consumers (i.e., the
relatively poor), and others targeting the non-poor, with a corresponding range of provider payment
mechanisrns.
3.2 Purchaser-provider split versus vertical integration
The previous section examined issues of how to pay physicians, under the assumption that such
payment mechanisms were implemented through explicit contractual agreements. Instead of writing
a detailed contractual agreement between insurer and physician, the two parties might decide instead
to integrate into a single organization, and to rely on bargaining protocols to determine the allocation
of rents. T)raditionally, in many countries in Latin America and elsewhere, public insurance systems
have been vertically integrated to a significant degree. Recent reforms however have focused on
separating the functions of insurance and provision, through the so-called purchaser-provider split.
13



These reforms effectively require that explicit contractual arrangements govern relationships between
insurers and providers.
On the other hand traditionally, private insurance was of the fee-for-service type (i.e., indemnity
plans), whereby a physician would send a bill to the insurer for covered services. This is one kind
- indeed a common one - of explicit contract. Over time, however, private insurance companies
have moved towards a more integrated organizational structure, bringing physicians in-house, or at
least adopting long-term contractual relationships with them.4 This apparent anomaly between the
evolution of the organization of public and private systems can be understood in a number of ways.
Soft budget constraints
First, both institutional developments might represent attempts to provide physicians with
stronger incentives, within constraints that differ between the public and private sectors. For exam-
ple, it might be difficult for a government bureaucracy to commit to funding a public sector employee
prospectively, thereby making him the residual claimant. Thus, if future renegotiation in the event
of high costs makes it difficult to credibly threaten punishment, endogenous soft budget constraints
limit incentives. Conversely, if political pressure is likely to force governments to expropriate profits
(or to renegotiate future contracts) in the event of low costs, incentives will again be dulled. By con-
tracting explicitly with the private sector - i.e., by adopting a purchaser-provider split - a public
payer may be able to harden what would otherwise be an endogenously soft budget constraint.
On the other hand, the easiest way for a private payer to provide a physician with incentives
might be to make him a shareholder in the insurance firm. (Of course, there are obvious free-rider
problems when physicians are paid on the basis of group profits and not individual contributions
thereto.)
Common agency in public institutions
4It is important not to trivialize the nature of private insurance organizational structures. In reality a wide range of
organizational forms exists, ranging from for-profit firms with share-holder doctors to non-profit HMOs with salaried
providers.
14



A second way to understand the opposing directions of reforms in the public and private sectors is
to look more closely at the nature of the purchaser-provider split arrangements. In particular, instead
of contracting directly with individual physicians (as we assumed in the preceding paragraphs), a
public insurance system might sign contracts with groups of physicians - indeed often with managed
care organizations. This suggests that the function that is being contracted out from the public
system is the management of physician services. Having a formal arm's length contract between
the public sector and the manager of physician services might be an effective way of improving
the incentives of such a manager. For example, building on the work of Williamson (1985) and
Holmstrom and Milgrom (1990), Dixit (1997) has shown that when an agent (here the manager)
reports to multiple non-cooperative principals with heterogeneous objectives, a negative externality
exists amongst the principals that results in low-powered incentives for the manager.5 The implicit
side contracts that may easily exist in a political environment support such a view of the constraints
on incentives of public sector managers (but less so for individual physicians). Requiring an explicit
contract could facilitate the cooperation of the principals, removing the externality, and leading to
higher-powered incentives for the manager.6
Explicit arm's length contractual arrangements might also be a way of limiting the scope of a
manager's activities. The usefulness of limiting objectives of public sector decision-makers has been
suggested by Tirole (1994) and formally elaborated in a model of career concerns by Dewatripont,
Jewitt, and Tirole (1999). Providing a manager with a well-defined "mission" again makes it easier
to induce effort. In Dixit's analysis the narrowing of a manager's objectives is effected by reducing
the number of competing principals to whom the manager answers.
sThe model relies on the manager taking a number of actions or efforts. Principals preferences over the outcomes
of different actions differ. Each principal will then propose a contract that offers the agent more insurance by making
payment negatively correlated with performance in dimensions that the principal does not care for. All principals act
the same, resulting in a dilution of incentives.
6Contracting out to a private manager may not be necessary. Corporatization - an intermediate step - may be
sufficient.
15



Contractual incompleteness
An alternative literature examines issues of contracting out versus internal provision - i.e.,
vertical integration - starting from the presumption that contracts are necessarily incomplete.
Even if choices are observable by both parties to a contract, if they are not verifiable and if the
contract is consequently unenforceable, then institutional arrangements can have substantive effects
on incentives. In particular, ownership of productive assets can matter when explicit contracts are
unavailable - private contractors are those who own the assets they use to produce services (e.g.,
hospitals), while public servants do not have the implied control rights over asset use.
Hart, Shleifer, and Vishny (1997) present a model of service provision when quality and cost
are non-contractible.7 By definition, public sector employees cannot retain ownership of any quality
innovations they generate, while private sector providers can. Incentives for quality innovations are
thus greater in the private sector. On the other hand, incentives for cost reduction are also greater
in the private sector, but cost control is associated with lower quality. Thus the private sector will
always (in the model) produce at lower cost, but could produce higher or lower quality. When one
of the ways of reducing costs is to actively select easy-to-treat patients ahead of expensive cases, the
social aspects of quality can be severely effected by incentives to control costs.
One might be willing to argue that innovations in medical care are very important, while those
in insurance administration are less so. This argument favors private provision of physician services,
under the condition that active selection could be controlled adequately. However, in countries with
large sections of the population uninsured against health needs responsive to standard treatments,
innovation in insurance delivery may have high social payoffs, in which case public provision may
then be favored.
71f cost is non-contractible, then procurement contracts like those studied by Laffont and Tirole (1993) are not
feasible, and a fixed price contract must be used.
16



3.3 Competition
Competition within the public sector, in the private sector, and between the two, can provide
incentives for quality provision and cost-reducing effort on the part of medical care practitioners. The
important feature of competition is the discipline it allows consumers to place on service providers.
Competition within government: quasi-markets
Competition among suppliers should not necessarily be identified with private supply. Indeed,
the UK government has attempted to induce competition among public providers by developing the
so-called quasi-market (Le Grand (1991)). Even when consumers do not face financial incentives to
choose wisely between suppliers, they might still induce effort and quality provision if their choices
affect the payoffs to providers. Thus Halonen and Propper (1999) model the impact of competition
on quality when providers are paid by a public sector payer on behalf of consumers who are free to
choose their supplier. The essential feature of their model is that when providers' objectives are not
coincident with consumers' (on average), allowing consumer choice can help to realign providers'
incentives. The benefits of competition are of course limited by the elasticity of demand.
Private sector competition
Competition can improve incentives, but can also have negative effects. The discussion above
centered on public sector providers' incentives to attract patients through quality improvements. An
important aspect of quality is that, while it is reasonable to assume that all individuals value more
quality than less, the willingness to pay for quality varies with health needs. In general, a hospital is
likely to face a higher elasticity of demand with respect to quality from patients with high needs than
from patients with low needs, assuming an alternative source of supply exists. Competition can then
result in a kind of "race to the bottom," in which all hospitals try to deter high cost patients from
seeking treatment by under-providing quality. Such negative effects of competition would not arise
if hospitals were able to charge sufficiently higher amounts for treating expensive patients. Similar
forces are likely to be at work in private insurance markets.
17



In some private health insurance markets consumers are required to commit ex ante to limit
their ex post choices, thus weakening the competitive pressures they can exert on physicians. This
occurs under various managed care arrangements (HMOs, PPOs, etc.) These restrictions reduce ex
post demand elasticities, allowing either prices to be higher, or more likely, quality to be lower.8
Public-private competition
A common argument in favor of large purchasing groups is the monopsony power they can wield
in negotiating supply contracts. However, as Propper and Green (1999) have recently pointed out,
there is no particular merit in such actions - market power is inefficient be it on the supply side
or the demand side. They suggest that under such arrangements staff will either be of poor quality
(good staff will be driven from the market by low wages), or employment contracts will permit
outside earnings with little or no monitoring, weakening incentives for performance of primary job
tasks.9 Introducing competition from the private sector may thus lead to higher public sector wages
and costs, but with a net welfare gain. The lesson from this analysis is simply that focusing on
budgetary impacts alone is not sufficient to determine welfare effects, especially if rationing costs
(i.e., waiting times, search costs) are included (Danzon (1992)).
Alternatively, introducing public provision into a private market may be beneficial if it serves
the purpose of making a minimum quality standard credible. For example, Ronnen (1991) has
shown in a model of vertical product differentiation that a minimum quality standard can raise the
quality provided by all market participants (even those who would have met the standard in the
absence of regulation), and lower equilibrium hedonic prices (i.e., prices adjusted for quality). An
appropriately chosen standard, by restricting product differentiation and intensifying ex post price
competition, makes all consumers better off. Such a standard could in principle be imposed by
government, without recourse to public provision. If quality is difficult to monitor however, and if
lapses are costly to penalize, public provision of the standard quality at a minimal price could act
8For a review of models of strategic competition with horizontal product differentiation, see Tirole (1989).
9Such arrangements are often found in academic institutions, particularly ones funded from public sources.
18



as a substitute for direct monitoring. In equilibrium, private sector quality would be higher than
that in the public sector, and any increase in prices would not be enough to outweigh the positive
effects of higher quality.
The idea that public and private sector quality can act as strategic complements should be
viewed with some caution. Standard equilibrium analysis (Hammer (1997)) suggests the need to
anticipate potential crowding out of private sector supply by public sector provision. In the case of
non-differentiated goods this is indeed appropriate, and one expects that if the publicly provided
quality level was too high then private supply would dry up.
4 Institutional evolution in Latin America - some examples
Over the past twenty years several Latin American countries have embarked upon wide-ranging
reforms of their health insurance and delivery systems. Generally, among the higher-income countries
there has been a move toward extending explicit insurance coverage to those outside the formal labor
market. At the same time, these countries have examined the ways in which insurance and health
care have been delivered, and instituted reforms that are meant to improve the allocative and
production efficiency in the sector. They have tended to move away from integrated public provision
of insurance and care, in the direction of more decentralized provision that relies, to varying extents,
on private sector involvement. Major health insurance reform, like health care itself, appears to be a
luxury good, and the poorer countries in the region have focused on more basic challenges in terms
of primary care delivery. Below I will review the experiences of four countries that have adopted
significant reforms: Colombia, Argentina, Brazil, and Chile.10
10For a more complete survey of health systems in Latin America see Beugoa et al. (1998).
19



4.1 Colombia
The health sector reforms initiated in the early 1990s under the direction of Minister of Health
Londofio represent possibly the most ambitious policy interventions undertaken since the implemen-
tation of large-scale social security mechanisms of the 1950s and 1960s, both in Latin America, and
outside the region. Before the reforms, Colombia's public health delivery system was centralized,
budget-financed, and poorly organized and consisted of two uncoordinated bodies: (1) the social
security institutions (SSIs), that provided subsidized health services, and hence insurance, to formal
sector workers; and (2) the Ministry of Health, that provided public health inputs, and subsidized
hospital care for those who could not, or would not, use the SSI system. This sub-section wil not
present a full history of the "pre-existing condition" of the Colombian system or the reforms, but
will examine how the reforms conform with some of the general organizational structures discussed
in sections 2 and 3..
Consistent with the discussion in the first section of this paper, the general goal of the Colombian
reforms appears to have been to ensure a basic level of coverage for all individuals, that could be
improved upon for those willing and able to pay more. The redistributive nature of the reforms is
manifest in two aspects: first, the tax base used to finance the new system is at least proportional,
if not somewhat progressive (it consists of a payroll tax plus general revenues); and second, since in
the absence of the reforms many informal sector workers and their families were formally uninsured
the provision of a basic level of insurance at minimal personal cost acts like a negative income tax."1
There has indeed been a marked increase in formal coverage of the population, particularly among
lower-income groups. Figure 3 (composed from Cuadro No. 7 of Uribe et al. (1999)) shows coverage
rates by income quintile in 1993 and 1997. Overall, the proportion of individuals with insurance
doubled in this period from 23.7 percent to 57.2 percent, with the largest proportionate gains among
1 l The uninsured did receive health care services under the old system, mostly from public hospitals and clinics. An
assumption often made is that formal insurance will lead individuals to consume necessary care earlier, to have access
to better quality care, and to face lower personal costs.
20



Colombia Health Insurance:
Proportion of population with coverage
70%
60%-
50% 
40%-
30%-
20%-
10%  -
0% -
Quintile 1    Quintile 2    Quintile 3    Quintile 4    Quintile 5
Income class
* 1993 * 1997
Figure 3:
the poor.
At the same time, the reforms attempt to improve incentives for productive and allocative ef-
ficiency, by encouraging alternative provider payment systems and allowing consumer choice, re-
spectively. Integration of a market mechanism for delivery with redistributive objectives presents
policy-makers with complex issues of design. Formally, there are three main groups of actors: con-
sumers, purchasers, and providers. A fourth agent is the government, that, while retaining active
roles in policy-making, regulation, and public health initiatives, acts as a budget balancer (in that
it allows payments made by any one of the other three not to equal those received by the recipient).
An implicit two-tiered voucher scheme
Insurance coverage is allocated to consumers through two regimes with separate but related fi-
21



nancing arrangements. I will discuss these below, but it turns out these schemes are approximately
equivalent to a simple two-level voucher system, although no formal voucher scheme exists. Effec-
tively, members of one group of families (those with workers in the formal sector) receive a voucher
for insurance that covers a wide range of services at notionally high quality, while individuals in the
complementary set (essentially the poor) receive a voucher for a less generous package of insurance.
Members in the first group are said to be in the "contributory regimen," and those in the second
are referred to as participating in the "subsidized regimen."
On the financing side, participants in the contributory regimen pay a 12 percent payroll tax
to help finance health care. Although this tax is earmarked for health services provision, because
its payment is non-discretionary and the value of the implicit voucher received upon payment is
unrelated to the amount of tax paid, we can, without loss of generality, consider the tax to be part
of general revenue.12 That is, the distortionary effects of financing should not be considered to be
offset by the fact that individuals receive something - insurance - in return for paying the tax.
Participants in the subsidized regime also contribute financial resources to support health in-
surance costs. These contributions are means tested (they may be zero in many cases) and not
directly related to the cost of the coverage they receive. Since members of the subsidized regimen
are required to purchase insurance, the contribution acts as part of the general tax base in terms
of its distortionary effects. Thus despite the nominal link between financing and coverage at the
individual level, it is appropriate to treat financing and demand as independent.
The next issue to discuss is where and how the implicit vouchers can be "spent". Participants
in the contributory regimen can use their vouchers to buy insurance from so-called EPSs (Empresas
Promotores de Salud), which are essentially private sector insurance companies. The EPS can cash
in the voucher with the government (via the FOSYGA - Fondo de Solidaridad y Garantfa) for a
fixed amount, known as the UPC, which is adjusted for some of the risk attributes of the consumer.
12There are of course questions of optimal taxation to consider - is a wage tax more or less efficient than another
revenue source, for example? I will not get in to this here, however.
22



Participants in the subsidized regimen can use their implicit vouchers to purchase (less generous)
insurance either from EPSs, or from ESSs (Empresas Solidarias de Salud). ESSs are medical care
purchasing organizations that subnational governments must set up to facilitate the coverage of the
self-employed and non-salaried workers who may not be able to access EPSs (say due to remoteness).
Because the implicit vouchers have fixed monetary values there is little effective price competition
amongst EPSs and ESSs. They are allowed to charge a price lower than the UPC, but none of the
price reduction accrues to consumers, so such a policy has little if any effect on demand. Instead,
the EPSs compete on the basis of the level of insurance they provide, as determined by copayment
rates, quality of service, and extra services offered. Thus, the standard packages of services defined
for participants in each regimen act as basic plans, which can be supplemented by insurers to attract
clients.'3 This kind of arrangement encourages efficient provision of insurance, since the insurer gets
to retain any efficiency gains.
However, there are two potential problems with using consumer demand to discipline insurance
providers. First, if demand is not elastic in response to quality changes (i.e., if individuals can-
not easily judge quality differences, and if they do not have easy access to alternative providers),
incentives to control costs might outweigh incentives to improve or maintain quality. Given these
determinants of demand, it is arguable that demand by participants in the subsidized regimen will
be less quality elastic than that of participants in the contributory regimen, so that the quality of
insurance provided to the poor will be reduced further.
Second, quality competition can naturally induce active selection efforts on the part of EPS/ESS
managers, who might design policies to attract inexpensive clients, perhaps by bundling extra ser-
vices, for which an additional premium can be charged, with elements of the standard package. If
the standard package is provided at low quality, but additional services are provided at high quality,
then it may be possible for firms to attract relatively profitable clients. For example, La Forgia
13The standard package in the contributory regimen is referred to as the POS (Plan Obligatorio de Salud), and
that of the subsidized regimen is the POS-S.
23



(1998, p. 257) reports concern over "the practice of EPSs to integrate the POS [standard package]
with complementary plans, thereby undermining competition for a homogeneous service plan." Both
of these concerns can in principle be addressed by regulating more closely the quality of insurance
provided, but there are obvious limits to the efficacy of such policies.
An intriguing aspect of the Colombian system is that two basic levels of insurance are mandated
for separate sections of the population. Mitigation of adverse selection problems and public provision
(or at least mandatory consumption) of a private good to facilitate redistribution as discussed in
the first section of the paper each imposed a single insurance package on all individuals. The
resolution of this apparent anomaly is that although a base level of insurance might be useful
in redistributing income from rich to poor, within the group of individuals who would purchase
supplementary insurance under such a system significant heterogeneity of risks will likely persist.
In order to facilitate such voluntary supplementation while avoiding the potential adverse selection
problem within this (wealthier) group, another specific insurance contract is imposed.'4
The supply side
The efficiency of provision is determnined by how providers are paid, and how they are organized.
This section briefly reviews the financial arrangements that govern incentives and the degree of
integration of insurance and provision.
Demand versus supply subsidies
On the supply side, EPSs and ESSs contract with hospital and physician groups, including
private sector IPSs (Instituticiones Prestadoras de Servicios), and formerly public sector, but now
autonomous, ESEs (Empresas Sociales del Estado). The financial aspects of such contracts are
less precisely regulated than are the demand side transactions between consumers and EPSs/ESSs,
14 A simpler explanation for the use of two levels of mandatory insurance is that it is too expensive for the ggvernment
to finance the POS for the poor. Intentions are to phase the POS-S up to the POS over time. However, if it were
only a matter of money, the question would then arise as to why a uniform coverage plan, below the POS but above
the POS-S, was not initially introduced for all individuals, and then phased up over time, with the same time profile
of budgetary expenditures. Presumably the answer to this question lies in the political realities of reform.
24



although the law attempts to encourage innovative payment methods to encourage provider effort
and efficiency (e.g., capitation, DRG-based payments).
One aspect of the reforms that has received some considerable attention is the fact that medical
care providers are paid on the basis of demand. That is, insurers contract with providers to supply
services for the covered clients, based on expected (in the case of capitated payments) or realized
(in the case of fee-for-service and DRG-based contracts) use. In the past, when the government paid
for services, payments were nearly always purely prospective, deriving from budgetary allocations
to hospitals, provincial health ministries, etc. The assumption of many commentators appears to be
that such a transition is undeniably beneficial.
However, there are two dimensions along which we might expect the alternative regimes to
differ. First, in terms of the use of a fixed revenue base, demand-based financing would appear to
offer the prospect of a more efficient allocation. On the other hand, due to the well-documented
price elasticity of demand for health care, prospective budgetary allocations have the advantage of
imposing a cap on aggregate spending. Indeed, before the reforms Colombia exhibited a relatively
low level of health expenditure amounting to around 4 percent of GDP in 1990, but recently this
has exploded to nearly 11 percent of GDP in 1998. There is some suggestion that this expenditure
increase has been caused by the introduction of demand subsidies without the removal of supply
subsidies (so hospitals receive budgetary funds and can additionally charge EPSs and ESSs for
services provided), so that the welfare implications are unclear.'5 Nonetheless, to achieve allocative
efficiency both within the health sector and across sectors, provider payment mechanisms must
be carefully designed, incorporating both demand side responsiveness to needs, and constraints on
15The welfare effects derive from two sources: the extent to which the additional expenditure finances inefficient
production versus private consumption (of medical personnel or hospital owners); and the shadow cost of public funds
That is, if the additional expenditures are not matched by increased service levels (which could be of little value),
they represent pure transfers with little efficiency effect. Except, that is, if the shadow cost of public funds is not
considered. When the distortionary costs of raising the additional revenues are incorporated, the welfare impact of
the increased spending, no matter what it finances, can be large and negative.
25



over-use.
Purchaser-provider split
The formal separation of purchaser and provider is mixed in the Colombian system. On the one
hand, under the proposed reforms (when fully implemented) the government will have little direct
role in the provision of either health insurance or health care. The FOSYGA acts as a clearing house
for taxes paid by individuals and transfers made to EPSs and ESSs. Apart from monitoring the
quality of the goods (i.e., insurance packages and medical services) provided, the public sector does
not actively perform any purchasing role, this function being delegated to consumers.
On the other hand, the degree of integration of the delivery of insurance and medical care varies
widely. Some purchasers (EPSs and ESSs) contract at arm's length with provider networks (IPSs
and ESEs), while others effectively own such networks, along the lines of HMOs. Even if an EPS
owns or has close contacts with a particular provider network, it is required by law to offer the
services of at least one other IPS to consumers. On the face of it, since consumers have free choice
of EPS, requiring that each EPS offers services through at least two IPSs seems redundant. One
possible explanation for this regulation is that it aims to increase ex post quality competition (among
providers), under the assumption that it is costly for individuals to changes EPSs, but less so for
them to switch between IPSs offered by any given insurer.
4.2 Argentina"6
Compulsory tax-financed health insurance has been delivered in Argentina by a plethora of for-
profit and non-profit institutions in conjunction with a predominantly private supply of physician
and hospital services for many years. On the face of it, this might appear to be a model with
high potential, dealing with insurance market failures (e.g., adverse selection) through compulsory
participation, equity concerns through progressive taxation, and efficiency of delivery through the
power of decentralized provision of insurance and services. However, a number of factors have meant
16Much of this sub-section relies on World Bank (1997, 1999).
26



such potential has not been realized, and have led to major reforms of the sector in the mid- and
late- 1990s.
Descriptive analysis
Argentina's formal health insurance system in the late 1980s was well-developed. All employees
were obliged to be insured by the so-called obra social that covered the sector in which they worked.
In essence, the obras sociales were, and remain, non-profit insurance companies owned by the relevant
labor union. There are about 360 such obras, covering about 10 million individuals and their families.
In addition, each of the 24 provinces of the country operates an obra protincial, covering about 5
million public sector employees and their dependents. Obras are funded on the basis of (non-
discretionary) payroll taxes.
Retired workers and pensioners - about 4 million individuals - received health insurance cov-
erage through the Integrated Program of Medical Care (PAMI, Programa de Asistencia Medica
Integral), operated by the National Social Service Institute for Retirees and Pensioners (INSSJP,
Instituto Nacional de Servicios Sociales para Jubilados y Pensionados). These services were funded
by payroll taxes and taxes on pension benefits. The obras and PAMI combined covered about 61
percent of the population. Another 2 million people were covered by private for-profit, pre-paid
insurance plans (pre-pagas), and a further 1 million received insurance through small mutuales, of
which there were around one thousand. The distribution of coverage across types of insurance is
shown in Figure 4. The 25 percent of the population who lacked formal insurance would typically
seek care through public hospitals operated by provincial and municipal governments.
Most individuals with formal insurance sought the care of private sector providers, although
obras provided on average 10 percent of services in-house. Some obras in fact contracted out the
insurance function as well, paying Empresas Administradora de Prestaciones (EAPs) a capitation
payment per member: the EAPs then contracted with provider networks for service provision. Tra-
ditionally providers were compensated on a fee-for-service basis, although diagnosis-based payment
and capitation have become more common.
27



Argentina Health Insurance:
Types of coverage
25%
32%
10%
13%           16%
*Obras sociales         *Obras provinciales
OPAMI                   a Pre-pagas and mutuales
* Uninsured
Figure 4:
28



Individuals covered by PAMI were assigned a family doctor, who was supposecd to act as a
gatekeeper and was an INSSJP employee. Other providers (specialists, hospitals, diagnositicians,
etc.) are paid with a mix of capitation payments and fee-for-service contracts.
Recognizing the wide variation in both wages and health risks by occupation and hence the wide
variation in per capita contributions across obras, the Administracion Nacional del Seguro de la
Salud (ANSSAL) in theory provided redistribution both from rich to poor obras, and from low-risk
to high-risk obras. ANSSAL also provided a regulatory structure within which the obras operated.
Outcomes
The main problems identified with the provision of health insurance were consumer dissatisfaction
with poor quality (i.e., allocative inefficiency), financial instability (deriving from unsustainable
production inefficiency), and inequity. We address each of these in turn, and describe the associated
reforms.
Allocative inefficiency
Discipline on providers comes either from the demand side through consumer choice, or from
effective regulation. Under the old system, formal sector workers were not permitted to exercise
choice on any dimension. They could not choose how much to pay for insurance, because the
payroll tax was non-discretionary (and opting out was not permitted), or from whom to purchase
it. INSSJP beneficiaries were not permitted to choose their family doctor. While there are well-
understood limits to the efficacy of consumer demand responsiveness in health care and insurance
markets, the abolition thereof is unlikely by itself to improve matters.
Alternatively, the regulatory apparatus of ANSSAL and INSSJP was not sufficient to ensure
quality coverage in the absence of consumer choice. Some regulatory failures were obvious, for
example the ability of obras to engage in activities not related to health insurance (like recreation
facilities). Others derived from the political use to which executive positions in the regulatory
institutions were put and the scope for corruption that emerged.
Financial instability
29



Financial instability derived both from the revenue and cost sides. Total wage taxes (of which
health insurance contributions constituted about 20 percent) amounting to 56 percent of the gross
wage were deemed to be excessive and an impediment to macroeconomic performance and interna-
tional competitiveness and were reduced during the early 1990s, cutting into the health insurance
tax base. The 1995 financial crisis and subsequent economic slump did not help matters, to say the
least. Of course, as a policy instrument, such tax cuts should not be seen as a cause of production
inefficiency, but they did contribute to mounting debts amongst the obras. On the cost side, weak
management, excess staffing levels, and rising medical care costs contributed more fundamentally to
the inefficiency of resource use.
The power of the trade unions, as owners of the obras, was important in this respect. World
Bank figures show average beneficiary-staff ratios of about 130, with some obras reporting fewer
than 10 beneficiaries per employee. Despite the high level of unemployment, and the associated
low shadow wage, the enormous distortionary costs associated with such labor intensive production
would appear to outweigh the positive welfare effects through their impact on the unemployment
rate.
Weak management of the obras may indeed be a misrepresentation. In fact, managers exercised
significant control of resource uses and policy choices, usually in favor of the controlling union and
not necessarily in the long-term interests of the workers. Without any incentive to control costs
-both because there were no shareholders, and because the government imposed a soft budget
constraint - otherwise productive effort was diverted to political maneuvering and rent-seeking.
Even with little consumer-imposed discipline, one might have expected the obras to aspire to
productive, if not allocative, efficiency. As non-profit organizations any surpluses generated would
have to be distributed in kind (this may explain the involvement of the obras in non-medical ac-
tivities), but this should not have dampened cost-consciousness unduly. Instead, costs escalated
while service quality remained stagnant at best. In fact, this situation arose because, far from being
residual claimants vis-a-vis the net revenue streams generated from insurance provision, obras faced
30



very soft budget constraints due to easy access to government credit. Since individuals had no pos-
sibility of switching obras, even in the event of financial collapse, it was difficult for the government
to commit to allowing them to go under. Indeed it was far more politically palatable (and in the
short run, possibly more efficient) to extend credits and permit continuation of coverage with the
existing obra than to close down an insurer and assume responsibility for coverage of its members.
Thus the lack of formal and credible bankruptcy procedures meant the threat of shut-down was an
ineffective incentive mechanism.
Redistribution
About 90 percent of payroll tax contributions were retained by the obra in which an individual
worker was enrolled and the remaining 10 percent of revenues passed on to ANSSAL for redistri-
bution. This redistribution was meant to be both from rich to poor obras, and from low-risk to
high-risk obras. In principle, such redistribution would allow the different obras to provide broadly
similar levels of coverage.
However, in 1994 per-beneficiary own revenues varied from $5 per month for the poorest obra
to $80 per month for the richest. Only a token impression could hope to be made in such gap-
ing inequality with the resources available to ANSSAL from the payroll tax. In practice, things
were worse, with per-beneficiary transfers being only slightly negatively correlated with own income
(World Bank (1997)), suggesting only a marginal, if any, impact on inequality of health insurance
coverage among workers.
From this discussion it can be inferred that obra revenues took the form of a more or less fixed
payment per person plus a proportion of the average wage in the sector. Thus there was some
redistribution within each sector, but not across sectors. This can be compared with the payment
received by Colombia's insurance providers - the EPSs - whose per capita revenue is a fixed amount
(based on the UPC), with no proportional component. At the other extreme, Chile's ISAPREs (see
below) are funded more or less in proportion to each individual's wage, with no fixed component.
The other redistributive function of the ANSSAL was meant to allow obras with different under-
31



lying costs, associated with different risk profiles, to provide similar service levels. Since membership
in a specific obra was mandatory, and obras could not reject participants, risk profile differentials
contributed only to inequality, and did not have any direct efficiency impacts associated with active
risk selection etc.17 To the extent that individual risk is correlated with personal income, the offset-
ting risk adjustments could thus in principle have been incorporated directly into the income-based
redistributive mechanism. Instead, transfers were made not on the basis of ex ante risk, but on the
basis of ex post realized costs, in particular for certain procedures that were considered high-cost
(although some, like prostheses, were cheap but still included).
Current and future reforms
Over the long term, insurance reform is aimed at attaining universal coverage through competitive
provision of at least a minimum level of insurance. The existing fragmented structure - consisting of
a public system for pensioners and retired individuals, a non-competitive but decentralized system
of obligatory insurance for formal sector workers, an unregulated competitive system for others
able to pay, and a public hospital system providing insurance through the provision of low quality
services - is to be unified across consumers. That is, the characteristics of insurance providers and
the environment in which they operate, are to be independent of the identities of the individuals
they cover. This represents a rational separation of the organization of production (determined
by internal efficiency considerations), and the allocation of consumption (determined by allocative
efficiency and equity concerns). The major incentive instruments are to be consumer choice and,
where this is ineffective or imperfect, regulation, both of quality and financial soundness.
17Risk adjusters, even with mandatory insurance and no choice of insurer, could have efficiency effects in the labor
market. For example, consider a situation in which individuals are taxed a fixed amount, and must obtain coverage
from their employers. High-risk individuals will prefer more insurance and lower wages to low risks, and the two may
sort themselves into different occupations, especially if cash wages within an occupation are inflexible. Risk adjusters
can remove the link between wages and the value of insurance, and allow a more efficient allocation of labor across
sectors. When individuals pay a fraction of their wage for insurance (and not a fixed amount) this story has to change
a bit.
32



Competition among obras is being introduced, and by mid-1999 most people were aware that
they had a choice of insurer. Consolidation in the industry is clearly required, and has been partially
achieved through mergers of obras.'8 Insurers are required to offer a standard health benefits package
(the PMO - Programa Medico ObligatorTio). As well as representing a minimum level of insurance
that may facilitate redistribution in general, the PMO has enabled consumers to make relatively
easy comparisons between obras.
Competition between obras and pre-pagas is so far limited. Individuals who are eligible for cover-
age by an obra are not permitted to obtain formal insurance from a pre-paga. However, some schemes
have developed whereby an obra will subcontract with a pre-paga, thus allowing an individual to
formally remain in the obra sector while effectively receiving insurance from outside.
Through the solidarity fund, obras with low incomes are assured of a minimum level of funding.
Of the individual contributions of 8 percent of wages, the obra receives 90 percent.'9 The additional
10 percent is used to finance transfers to poor obras in order to bring them up to a minimum of $40
per family per month.20 Thus the net price paid by individuals is still strongly positively correlated
with their wage, so it is likely that some equilibrium vertical segmentation by quality will result.
Part of the reform process is to change the $40 per family minimum to a risk-adjusted (by age and
sex) per person amount. This would increase the redistributive role the solidarity fund (assuming
noone got less than $40 under the new scheme), reduce the proportion of the wage tax directly
retained by the obras, and reduce the potential for vertical segmentation of the market.
4.3 Brazil2l
Health insurance in Brazil is something of a mix between a nominally comprehensive public system
and a large and active private system. Unlike Argentina, Chile, and Colombia that have tried in var-
l5The number of registered obras fell from 360 in 1997 to 294 in 1999.
"5For white collar workers the retained portion is 85 percent.
20The 10 percent also funds transfers for high complexity and long-term treatments, as well as some other transfers
to obras in "special" situations.
21 Much of this section is based on information reported in World Bank (1994) and Lewis and Medici (1998).
33



ious ways to implement (at least partially) privately provided mandatory insurance coverage, Brazil
has opted to provide universal public insurance (financed from general revenues) while encouraging
use of the private sector as an alternative. Consumers certainly have the freedom to choose among
private plans, and are able to choose whether or not to use the government system. The big differ-
ence is that when opting out of the public system, individuals do not take their financing with them.
As a result consumer choice provides little incentive for the public system to maintain quality. In
reality the public system acts as a floor, available to all but used primarily by the lower classes.
The health sector reforms have concentrated on the organization of public service delivery, partic-
ularly through decentralization, and the payment of providers by the public system. Until recently,
relatively less attention has been paid to the demand side, either with respect to using demand as
an efficiency-enhancing mechanism by exposing the public sector to competition, or with respect to
consumer protection through regulation of private insurance.
The public insurance system
Until 1993 public insurance was implemented through INAMPS (Instituto Nactonal de Assisten-
cia Medica e Premndencia Social), the medical arm of social security. INAMPS originally covered
contributors to social security, i.e., formal sector workers, but under the new constitution of 1988
and subsequent legislation coverage was nominally extended to the whole population. Having a
universal right to health care, as enshrined in the constitution however, has not helped those in
the northern states who continue to lack adequate medical care facilities. Public health insurance
expenditures are financed through earmarked taxes (e.g., a tax on the gross income of enterprises -
Contribuicao Financeira para o Setor Social, or COFINS, and a 0.3 percent financial transactions tax
- Contribuicao Social sobre a Movimentacao Financeira) and general tax revenues of the federal,
state, and municipal governments.
A comprehensive reform of the health system (Reforma Sanitdria) was instigated in the early
1980s. The first phase of this, in 1984, was the integrated health actions (AIS) reform, which
shifted some supply to under-utilized public hospitals, and coordinated the functions of INAMPS
34



with the Ministry of Health. Later, in 1987-8, a further reform (SUDS - Sistemas Unificados
e Descentralizados de Satide, or Unified and Decentralized Health Systems) led to a transfer of
INAMPS staff and facilities to state health secretariats, with the central agency acting solely as a
funding conduit (and being renamed the INSS - Instituto Nacional de Seguranga Social). This
decentralization reform was partially reversed in the third phase in 1990 under the SUS reform
(Sistema Unico de Satide - Single Health System), with INAMPS reverting to a more centralized
structure, although state governments were given authority to write contracts with private providers.
In 1993 INAMPS was abolished and integrated into the Ministry of Health under the umbrella of
the SAS, Secretaria de A,coes de Satide. The main function of the SAS is to transfer funds to state
health secretariats.22
INAMPS implemented most of its insurance coverage through contracting with private sector
providers, first on a fee-for-service basis, but later using a U.S. Medicare-type prospective payment
system (PPS). There are currently two PPSs in operation, for outpatient and inpatient services
respectively.23 There have been significant problems with data management under these schemes -
indeed, the cost control attributes of prospective payment have probably been ineffective because of
a lack of monitoring and evaluation by the public payer.24
The prospective payment of medical providers and hospitals on the basis of utilization suggests
that the Brazilian government has chosen a limited version of contracting out. This reimbursement
22Some of this transfer is made prospectively on a per capita basis for the purposes of primary care.
231I think the PPS for out-patients is actually a per capita transfer from the central government to the states, which
is then used to finance care. But I'm not sure what the relationship is between usage and payment - i.e., if that is
truely prospective or not.
24Let me explain: if diagnoses are inelastic and verifiable, then PPSs provide good incentives for cost containment
(although they may still overpay for cheap cases, see Chalkey and Malcomson (1999)) However, if the payer does
not observe the validity of the diagnsis or subsequent resource use, providers can increase their revenues by over-
diagnosing. In theory, the incentives for cost control remain, as the provider continues to be the residual claimant,
and the social cost increases only due to the distortionary costs of raising the additional revenues to pay the higher
charges. However if the higher revenues can only be consumed in certain ways (e.g., through the purchase of high-tech
equipment) then production inefficiency adds to the social cost.
35



mechanism does not represent the contracting out of the management of individuals' health care
needs in any meaningful sense, as might be descriptive of capitation funding mechanisms. Nor does
it represent the contracting out of insurance, which again could be implemented through a (risk-
adjusted) capitation payment to an intermediary that would provide coverage for the individual of
agreed services with associated copayments, etc. Thus while the private sector is heavily involved in
the delivery of services financed by public insurance, the incentives of providers to focus on outcomes
(i.e., health) as opposed to inputs, are relatively weak, with predictable impacts on quality and costs.
Private insurance
Private insurance covered about 25 percent of the population in the mid-1990s. The expansion
in private coverage over the past 30 years has been brought about by a number of factors, including
growing incomes (in the 1970s), a tax deduction of out-of-pocket expenditures and premiums, and
the deteriorating quality of the public system.
Private insurance is provided through four alternative types of organizations. The largest, and
historically most important, is the prepaid group practice (PGP), which is similar to the U.S. staff
model HMO (i.e., physicians are salaried employees of the insurer), and had (in 1991) about 47
percent of the private insurance market. Medical cooperatives (like IPA - Individual Practice
Association - models in the United States) restrict use to a set of preferred providers with whom
contracts are written, and captured 28 percent of the market in 1991. Large employers (20 percent
of the market) at times self-insure and offer company health plans, sometimes contracting out the
administrative functions to financial intermediaries. Finally, only a very small proportion of those
covered by private insurance (4 percent) enroll in indemnity plans (i.e., reimbursement insurance).
Regulation of the private insurance market was virtually non-existent until 1998. Indeed, in the
early 1990s a request by the prepaid group practice trade association (ABRAMGE, whose members
represent 45 percent of the PGP market) for public regulation was turned down. Thus exclusions
and restrictions are common, financial soundness unchecked, and fraudulent practices are at least
36



perceived to be prevalent.25 In fact, the poor reputation of PGPs has been the main factor con-
tributing to the relative growth of medical cooperatives and company health plans over the past
10 years, although PGPs still capture the largest portion of the market. Lately however, a number
of regulatory initiatives have arisen. These include a regulatory framework created by the federal
government, and the establishment of a Supplemental Health Department within the Ministry of
Health and a Supplemental Health Board (consisting of representatives from private insurance com-
panies, government, and consumer groups) to improve monitoring and control. An independent
regulatory agency is currently being considered in conjunction with technical assistance from the
Inter-American Development Bank, to cover issues of public information, solvency, accreditation of
plans, quality, consumer rights, and risk management.
Meeting efficiency and equity objectives
It could be hoped that Brazil's universal public insurance system complemented by an active
private insurance option, and its heavy reliance on private sector provision of medical care, could
strike a balance between the efficiency and equity concerns outlined in the first section of this paper.
However, despite the ambitious nature of reforms over the past 15 to 20 years, the system falls
short of achieving this goal. On the one hand, to be meaningfully implemented, a universal base
level of public insurance must be supported by medical care services that are in fact available to all.
However the quality of and access to publicly financed services varies widely, suggesting that the
public system is insufficient to meet the needs of the poor.26
The efficiency attributes of private delivery on the other hand have been compromised by exactly
25The public bad nature of a reputation for poor performance explains ABRAMGE's request to the authorities for
external regulation. Effective self-regulation would have required the establishment of a reputation, that, given the
behavior of some PGPs, was costly to achieve. Using an outsider to gain credibility would have been cheaper, but
was not attractive to the government.
26An alternative interpretation is that the public system is too generous for the non-poor, but given that the poor
often pay relatively high prices to use the private system suggests that any base level of public insurance is essentially
worthless to them.
37



those characteristics of health care and health insurance that suggest caution in relying too heavily
on the market to generate quality and cost savings. Unregulated, private insurance companies
have provided limited insurance (no lifetime coverage, excluded conditions, and other limitations on
coverage), and despite avoiding the fee-for-service paradigm have been unable to control costs.
4.4   Chile27
Chile's health system is one of the more closely and thoroughly studied in Latin America. The
country undertook wide-ranging and innovative reforms of its health and social security systems
starting in the early 1980s partly in response to failures of the bureaucratic centralized regimes
that had prevailed previously. Similar to the arrangements in Argentina, the reformed Chilean
system of compulsory health insurance couples private provision of insurance for some with public
insurance for others. Funding is primarily through a proportional payroll and pensions tax, although
there is no equivalent of Argentina's redistribution fund.28 Colombia's reforms have been more
ambitious, implementing more fully a redistributive allocation mechanism, while privatizing, or at
least corporatizing, insurance provision to a greater degree. The pattern of insurance coverage by
income and risk category that has resulted in Chile has been criticized by some commentators.
Nevertheless this pattern could be appropriate under certain assumptions about the redistributive
capacity of the tax and transfer system.
The mix of public and private insurance
Under the reformed system, formal sector workers and pensioners are required to contribute 7
percent of their incomes to finance health insurance. Each individual has the choice to allocate
his/her contributions to one of about 35 private insurance companies, known as ISAPREs, or to the
Fundo Nacional de Salud, FONASA (National Health Fund). The designated recipient of the funds
then provides insurance coverage for the individual and his/her dependents. Individuals who are
27This sub-section relies heavily on Larrafiaga (1997) and Bitran (1998).
2"There might be a reinsurance fund, but this is different.
38



unemployed or who work in the informal sector are automatically covered by the public insurance
system.
The private insurance sector consists of about 35 ISAPREs,29 covered about 31 percent of con-
tributors in 1995, and accounted for about half of insured medical care spending. ISAPREs can
offer multiple policies, and are free to charge corresponding premiums. These premiums can vary
on the basis of age, gender, and the number of insured, and on the quality and extent of insurance,
for example as defined by the extent of coinsurance and copayments.30 Individuals are permitted to
increase their contributions above the 7 percent of income in order to purchase a higher cost policy.
In 1995 there were close to 9,000 policies offered by ISAPREs, reflecting a near continuum of vertical
differentiation matching the distribution of wages and hence contributions.
One important characteristic of contracts offered by ISAPREs is that they often include stop-loss
components, limiting the financial risk of the insurer to a certain amount (above which the individual
receives no reimbursement). This feature is perverse. Normally we think of optimal insurance as
covering catastrophic events nearly in full, but with demand-controlling cost-sharing in the form of
deductibles and/or coinsurance at lower levels of expenditure. Indeed, the stop-loss feature, more
than explicit restrictions on covered services, limits the extent of insurance actually offered, and
means that the high-risk - especially the elderly - choose not to participate. Table 1 reports the
shares of each age group enrolled in FONASA and the ISAPREs.
Of course, those with low incomes also tend to choose not to participate in the ISAPRE system
because the premiums are unaffordable, and because the public system provides free insurance.3'
Figure 5 confirms this pattern of demand, although it has changed over time. FONASA classifies
29In 1997 there were 34 ISAPREs: 21 were open to any purchaser, and 13 were closed funds, open only to employees
of specific firms.
30ISAPREs cannot discontinue insurance, and can impose at most an 18-month waiting period on clients for pre-
existing conditions.
31 In terms of the choice between private and public insurance, the latter is free even for contributors, since their
contributions are non-discretionary. The premiums charged by private insurers are higher than 7 percent of income,
but are of course discretionary.
39



Age group
Insurance   0-1   2-14  15-24  25-54  55-64  65+   Total
FONASA    65.8  64.7   58.3   57.0   68.3  79.9   61.8
ISAPREs    24.9  23.9   22.4   26.7   16.5   6.9   23.1
Other         9.3  11.4   19.3   16.3   15.2  13.2   15.1
Total         100   100    100    100    100   100    100
Source: Uribe et al. (1999) Table 2.1
Table 1: Chilean health insurance: Distribution of coverage by age
beneficiaries into four income groups, labelled A (the poorest) through D (the richest). (It also has
a fifth classification, E, for non-beneficiariers, i.e., those enrolled with an ISAPRE, or not formally
registered with FONASA.) In 1994, half of FONASA's beneficiaries came from the lowest income
group, 25 percent from the next group, and around 12-13 from each of the richer two groups C and
D. In 1996 this picture started to change, with individuals from the poorest group representing just
one-third of FONASA enrollees, while the share of group D enrollees doubled to about 25 percent.
The shares of groups B and C however did not conform with this convergent tendency.32
In most dual public-private systems, the coverage offered by the public system is independent
of income, although in practice some correlation is to be expected if public facilities in poorer areas
are of lower quality. However, in Chile this is not the case. In fact, there is in principle a negative
correlation between income and coverage for those choosing the public system because copayments of
between 0 and 50 percent are charged on a means-tested basis for hospital services.33 Thus, holding
risk constant, the value of public coverage falls with income. Since the value of endogenously priced
private insurance increases with income, the income profile of coverage is U-shaped. This design
32This discussion is incomplete as it has not been established that the income categories represent quartiles (i.e.,
with equal numbers of individuals across the population).
33Larranaga (1997). This negative correlation may be offset by the fact that contributors to FONASA are eligible for
services provided by private physicians, with a copayment, whereas the indigent are required to use public providers.
40



Chilean Health Insurance: Distribution of
FONASA by income category
60%
50%  -
40%
30%-
20%  -
10% 
0%                 I              I             l
1994          1995           1996          1997           1998
+ Income group A (poorest)  m  Income group B
-+-- Income group C             -a- Income group D (richest)
Figure 5:
induces more individuals to opt for the private system, but reduces the value of coverage for those
at the margin who stay in the public system.
The resulting pattern of coverage in which those with high risks and low incomes use the public
system and others use the ISAPREs has been criticized as fragmentary. This is certainly the case,
but it does implicitly assume that uniform coverage is optimal. While this is probably true in the
first best (abstracting from differences in risk preferences and demand elasticities), in a second best
world in which the government wishes to redistribute from the rich to the poor and from low risks
to high risks, such a pattern may not be unreasonable.
What is perhaps questionable is the extent to which higher-income individuals are effectively re-
quired to purchase superfluous insurance, or at least insurance of low marginal value. A comparison
with Colombia's more focused system is useful in this respect. Under current practice, Colombian
users of the contributory regimen (corresponding to Chilean ISAPREs) receive a more or less stan-
41



dard insurance package independent of income, implying an in-kind redistribution from high- income
contributors to lower-income contributors, which results in two tiers of more or less uniform health
insurance. In the ISAPRE system however, above the public level of insurance there is a continuum
of insurance qualities, increasing with income. The Colombian arrangement therefore allows for
some redistribution within the group of relatively higher-income individuals. Perhaps more impor-
tantly, since the average value of the implicit voucher used in the ISAPRE system is equal to the
average contribution across all ISAPRE users, there is limited redistribution (through the health
system at least) between groups in Chile. In contrast, because the cost of the standard insurance
package under the Colombian contributory regimen need not (in fact, does not) equal the average
contribution by users, there is scope for further inter-group redistribution.34
A final characteristic of insurance contracts generated under the ISAPRE system is that, since
they are tied closely to wages, quality tends to follow the life cycle pattern of wages, which may be
somewhat different from the time profile of an individual's demand for insurance. Thus it is not just
the case that some individuals with high lifetime earnings will be induced to purchase more insurance
over their life spans than they would desire, but that the pattern of coverage may not match their
pattern of needs. Even if capital markets work well, this problem will persist, unless insurance
companies can write long-term contracts in which high contributions at ages of high earnings but
relatively low risk are made in exchange for lower contributions later in life when earnings are lower
and risks higher.
340f course, to make definitive statements about the extent of redistribution between groups we need to know the
full structure not only of taxes, but other government expenditures. Focus on the earmarked taxes and the provision of
insurance is useful for predicting the effects on distribution of marginal parameter changes. For example, an increase
in the payroll tax rate in Colombia would increase the resources available for both the low quality and high quality
insurance packages. Doing the same in Chile would necessarily increase the quality of the high quality packages, with
probably little effect on that of the publicly provided services. Oin the other hand, Bitran (1998) has found that within
the FONASA-financed public insurance, the incidence of net benefits is reasonably progressive, suggesting a degree
of within-group redistribution through the public system.
42



The organization of publicly provided health care
About 70 percent of Chile's population is covered by the public insurance system, which financed
care delivered through mainly public hospitals and a mix of public and private ambulatory care.
Medical services financed through the public budget are funded primarily from FONASA and the
general budget (for the indigent), with additional resources deriving from copayments by public
patients, payments from ISAPREs that used public facilities, and other sources. FONASA acts
primarily as a decentralized financing agency, collecting contributions and distributing funds to
providers through a network of 26 Health Services. Following a contraction in public health spending
in the 1980s (reflecting in part, but not fully, the growth of ISAPREs during that period), real public
health expenditures more than doubled between 1990 and 1996 (Larraniaga (1997)). Despite this
increase, the perceived quality and adequacy of public services remained stagnant, leading to a recent
debate about reform of the internal organization of the public system. Some have argued in favor
of moving towards internal markets (see section 2) and introducing competitive pressures into the
delivery system, while others - particularly some labor unions and parts of the medical profession
- have resisted what they see as a move towards privatization.
Although a decentralized organizational infrastructure exists, in the form of the 26 Health Ser-
vices, central bureaucratic control of some crucial decisions remains in the hands of the central
authorities, including the Ministry of Finance. In particular, labor inputs and compensation are
dictated from the center and funded directly from the budget. Some central control of manpower
allocation is likely to be necessary in the health system, especially one in which providers are not
necessarily profit maximizers and in which service provision has a redistributive role. However, allo-
cations made centrally tend to be unresponsive to changing cost structures and needs, and provide
little incentive for innovation.
One part of the financial apparatus that does provide incentives for cost-consciousness is the use
of a prospective payment system for financing the use of drugs and material supplies.35 As is well
35This is referred to as the PAD. Less common services are funded on a reimbursement basis from a global budget
43



known, prospective payment gives good incentives for cost control, although as discussed in section
2 it may induce undesirable actions by providers on other margins, the most common identified issue
being quality. However, even in the absence of quality concerns, the cost containment incentives are
only operative if providers can exercise choice over input use. When the major component of costs -
labor - is not under the control of the hospital, the impact of incentives for cost control is reduced
significantly. Indeed, it could lead to a bias in favor of admitting patients with conditions whose
treatment is relatively labor intensive, since the shadow cost of labor to the hospital is effectively
zero, at least in the short to medium run.36
As well as increasing the effectiveness of incentives for cost control by allowing providers (hospi-
tals) to become residual claimants, a movement towards full prospective payment (that would fund
wages as well as other inputs) could also generate a degree of competition similar to that envisioned
in the quasi-market reforms of, for example, the United Kingdom. Whether this competition is
socially beneficial depends, as usual, on the responsiveness of consumer demand to quality, both as
determined by geographical constraints and limitations on information. When it is not, which is
especially likely the case in poor and/or rural areas, direct monitoring and regulation of quality are
necessary.
5 Concluding comments
Despite the well-known failures of insurance and health care markets associated with imperfect
information, the primary motivation for large-scale public intervention in the sector has been equity.
Most often this derives implicitly from a concern for good-specific equity, but it can also be justified
under a mechanism known as PPP. When the global budget constraint binds, such services must be rationed in some
fashion.
3"More specifically, it could lead to a bias in favor of conditions for which the elasticity of substitution between
labor and non-labor inputs is high, in which case the return to a switch towards a more labor intensive production
process is relatively high.
44



as part of a more general second-best redistributive mechanism. Within the context of public
intervention in the pursuit of equity goals, it is reasonable to assert that the reforms of the last two
decades have sought not so much to improve the efficiency of private markets, but to improve the
efficiency of public provision, either through direct use, or mimicking, of such markets. This has
sometimes been achieved by altering the focus and function of pre-existing institutions - e.g., the
obras sociales in Argentina, or by encouraging the growth of new institutions, such as the ISAPREs
in Chile.
Coupled with the reforms of the ways insurance and care are organized and delivered, countries
have attempted to various degrees to extend formal coverage to previously marginalized groups,
and to finance this extension fairly. Colombia can be seen as instituting an implicit two-tiered
voucher scheme financed through a proportional wage tax. Chile's system has a similar financing
mechanism, but the distribution of benefits is less progressive, so that the net effect is in principle less
redistributive. Argentina's remodeled obras system is something of a half-way house: the financing
base is similar, but the distribution of benefits in terms of the quality of insurance increases with
income, but there is some implicit redistribution from richer to poorer obras. On the face of it,
Brazil's health insurance system serves less of a redistributive function that those of the other
countries, to the extent that there is no earmarked tax dedicated to financing health insurance. This
however highlights the limitations of examining the health sector independently of the general tax
and transfer system. The taxes paid by higher-income individuals in Brazil are not reduced when
they opt for private insurance: the real question is then how redistributive is the general tax base,
an issue I have not addressed in this paper.
Finally, let me briefly mention the issue of non-insurable or systemic risks. In health, such risks
are usually associated with disease epidemics and/or natural disasters. However, recognizing the
links between economic opportunities and health status, it becomes apparent that adverse economic
impacts deriving from macroeconomic fluctuations may have important effects on health needs and
the performance of health insurance institutions. An important parameter in an analysis of these
45



links would be the rate at which health status declines in the face of negative income/consumption
shocks, and the extent to which otherwise temporary macroeconomic shocks translate into permanent
decreases in welfare for sub-populations. The ability of a health system to respond quickly to income-
induced health shocks could be an important factor in reducing the effects of hysteresis on poverty.
Bibliography
Akerlof, George (1970): "The market for 'lemons': quality uncertainty and the market mecha-
nism," Quarterly Journal of Economics, 84, pp. 488-500.
Bengoa, R., J.M. FernAndez, R. Nunio, P. Key, and D. Nichol (1998): "A descriptive review of
the health systems of Latin American countries," International Health Partnership.
Besley, Timothy and Steven Coate (1991): "Public provision of private goods and the redistri-
bution of income," American Economic Review, 81(4), pp. 979-984.
Bitran, Ricardo (1998): "Equity in the financing of social security for health in Chile," Bitran y
Asociados and Partnerships for Health Reform Project.
Blomqvist, So3ren and Vidar Christiansen (1998a): "Price subsidies versus public provision,"
International Tax and Public Finance, 5, pp. 283-306.
- - - - --and -         (1998b): "Topping up or opting out? The optimal design of public
provision schemes," International Economic Review, 39(2), pp. 399-411.
- - - - --and -         (1999): "The political economy of publicly provided private goods,"
Journal of Public Economics, 73, pp. 31-54.
Chalkey, Martin and James M. Malcomson (1999): "Cost sharing in health service provision:
An empirical assessment of cost savings," paper presented to the International Seminar on Public
Economics, Belgium.
Cruz-Saco, Marfa Amparo and Carmelo Mes-Lago (1998): Do Options Exist? The reform oj
pension and health care systems in Latin America, University of Pittsburg Press, Pittsburg PA.
Cutler, David and Richard Zeckhauser (1997): "Adverse selection in health insurance," NBER
Working Paper No. 6107, Cambridge, Mass.
46



Cutler, David and Sarah Reber (1998): "Paying for health insurance: The Tradeoff between
competition and adverse selection," Quarterly Journal of Economics, 113, pp. 433-66.
Danzon, Patricia (1992): "Hidden Overhead Costs: Is Canada's System Really Less Expensive?"
Health Affairs (Spring) pp. 21-43.
Dewatripont, Mathias, Ian Jewitt and Jean Tirole (1999): 'The economics of career concerns,
Part II," Review of Economic Studies, 66.
Dewatripont, Mathias and Jean Tirole (1994): The Prudential Regulation of Banks, Cambridge
MA, MIT Press.
Dixit, Avinash (1997): "Power of incentives in private versus public organizations," American
Economic Review, 87, pp. 378-382.
Escobar Maria-Luisa,. et al.(1999): unpublished technical annexes to "Chile: Health Insurance
Issues: Old Age and Catastrophic Health Costs," World Bank Report, CL-19940, December.
Feldstein, Martin (1973): "The welfare loss of excess health insurance," Journal of Political
Economy, 81(2), pp. 251-80.
Gouveia, M. (1997): "Majority rule and the public provision of a private good," Public Choice,
93, pp. 221-244.
Greene, Edward, Jose Vicente Zevallos, and Ruben Suarez (1999): Health Systems Inequalities
and Poverty in Latin America and the Caribbean, EquiLAC/IHEP Project.
Greenwald, Bruce and Joseph Stiglitz (1986): "Externalities in Economies with Imperfect Infor-
mation and Incomplete Markets," Quarterly Journal of Economics, 101(2), pp. 229-264.
Hart, Oliver, Andrei Shleifer, and Vishny (1997)
Halonen, Maija and Carol Propper (1999): "The organisation of government bureaucracies: The
choice between competition and single agency," Centre for Market and Public Organisation Working
Paper No. 99/010.
Hammer, Jeffrey (1997): "Economic Analysis for Health Projects," World Bank Research Ob-
server, 12(1), pp. 47-72.
47



Henriet, Dominique and Jean-Charles Rochet (1999): "The redistributive consequences of public
health insurance," paper presented to the International Seminar on Public Economics, Belgium.
Holmstrom, Bengt (1979): "Moral hazard and observability," Bell Journal of Economics, 10, pp.
74-91.
… …(1982): "Moral hazard in teams," Bell Journal of Economics, 13, pp. 324-40.
Holmstrom, Bengt and Paul Milgrom (1990): "Multi-task principal-agent analyses," Journal oj
Law, Economics, and Organization, 7, pp. 24-51.
Jack, William (forthcoming): "Controlling Risk Selection Incentives When Health Insurance
Contracts are Endogenous," Journal of Public Economics.
Jack, William (1999): "Optimal redistributive taxation when individuals differ on two dimen-
sions: The integration of health expenditures into a linear income tax," mimeo.
Laffont, Jean-Jaques and Jean Tirole (1994): A theory of incentives in procurement and regula-
tion, MIT Press, Cambridge MA.
La Forgia, Gerard (1998): "Health Sector Reform: A financial-service flow model and the Colom-
bian case," in Cruz-Saco and Mesa-Lago, eds., pp. 225-266.
Larraniaga, Oswaldo (1997): "Reforms of the health sector in Chile," mimeo.
Le Grand, Julian (1991): "Quasi-Markets and Social Policy," Economic Journal, 101 (405),
1256-67.
Lewis, Maureen and Andre Medici (1998): "Health Care Reform in Brazil: Phasing Change," in
Cruz-Saco and Mesa-Lago, eds., pp. 267-289.
Medici, Andre, Juan Luis Londonio, Oswaldo Coelho, and Helen Saxenian (1997): "Managed Care
and Managed Competition in Latin America and the Caribbean," in Schieber, ed., pp. 215-232.
Pauly, Mark (1968): "The economics of moral hazard: Comment," American Economic Review,
58, pp. 531-37.
Propper, Carol and Katherine Green (1999): "A larger role for the private sector in health care?:
A review of the arguments," Centre for Market and Public Organisation Working Paper No. 99/009.
48



Ronnen, Uri (1991): "Minimum quality standards, fixed costs, and competition," Rand Journal
of Economics, 22(4), Winter, pp.490-504.
Rothschild, Michael and Joseph Stiglitz (1976): "Equilibrium in competitive insurance markets:
An essay on the economics of imperfect information," Quarterly Journal of Economics, 40, pp.
629-649.
Rubinstein, A. and Yaari, M. E. (1983), "Repeated Insurance Contracts and Moral Hazard,"
Journal of Economic Theory, 30, 74-97.
Schieber, George (ed.) (1997): Innovations in Health Care Financing: Proceedings of a World
Bank Conference March 10-11, 199?, World Bank, Washington, DC.
Shleifer, Andrei (1985): "A Theory of Yardstick Competition," The Rand Journal of Economics,
16(3), pp. 319-27.
Tirole, Jean (1989): The theory of industrial organization, MIT Press, Cambridge MA.
Tirole, Jean (1994): "The internal organization of government," Oxford Economic Papers, 46(1),
pp. 1-29.
Uribe J.P., Londoiio B., Jaramillo I. (1999): "Descentralizaci6n y Reforma en los Servicios de
Salud: El Caso Colombiano," World Bank Consultant's Report. Bogota, Colombia. December.
Williamson, Oliver (1985): The economic institutions of capitalism, New York: Free Press.
Wilson, Charles (1977): "A model of insurance markets with incomplete information," Journal
of Economic Theory, 16, pp. 167-207.
World Bank (1994):
… -           (1997): Argentina: Facing the Challenge of Health Insurance Reform, Report
No. 16402-AR.
… (1999): Implementation Completion Report: Republic of Argentina, Health In-
surance reform project, Report No. 19426.
Zeckhauser, Richard (1970): "Medical insurance: A case study of the tradeoff between risk
spreading and appropriate incentives," Journal of Economic Theory, March, 2(1), pp. 10-26.
49






Policy Research Working Paper Series
Contact
Title                            Author                  Date              for paper
WPS2468 Pricing, Subsidies, and the Poor:    Ian Walker           November 2000      S. Delgado
Demand for Improved Water Services Fidel Ordofiez                          37840
in Central America               Pedro Serrano
Jonathan Halpern
WPS2469 Risk Shifting and Long-Term       Mansoor Dailami         November 2000      W. Nedrow
Contracts: Evidence from the     Robert Hauswald                            31585
Ras Gas Project
WPS2470 Are Larger Countries Really More  Stephen Knack           November 2000      P. Sintim-Aboagye
Corrupt?                         Omar Azfar                                38526
WPS2471 Validating Operational Food       Luc J. Christiaensen    November 2000      L. Christiaensen
Insecurity Indicators against a  Richard N. Boisvert                       81463
Dynamic Benchmark: Evidence      John Hoddinott
from Mali
WPS2472 Uzbekistan and Kazakkhstan:       Asad Alam               November 2000      L.Henson
A Tale of Two Transition Paths   Arup Banerji                               84026
WPS2473 Banking Risks around the World:   Luc Laeven              November 2000      R. Vo
The Implicit Safety Net Subsidy                                             33722
Approach
WPS2474 Exports and Information Spillovers   Alessandro Nicita    November 2000      L.Tabada
Marcelo Olarreaga                         36896
WPS2475 Industrial Growth and the Quality of  David A. Grigorian  November 2000      A. Nelson
Institutions: What Do (Transition)    Albert Martinez                      37117
Economies Have to Gain from the
Rule of Law?
WPS2476 Measuring Banking Efficiency in the  Cevdet Denizer       November 2000      M. Dinc,
Pre- and Post-Liberalization     Mustafa Ding                              36233
Environment: Evidence from the   Murat Tarimcilar
Turkish Banking System
WPS2477 Picking the Poor: Indicators for  Norbert R. Schady       November 2000      T. Gomez
Geographic Targeting in Peru                                                32127
WPS2478 Institutions, Politics, and Contracts:  Lorena Alcazar    November 2000      P. Sintim-Aboagye
The Attempt to Privatize the Water   Lixin Colin Xu                         37644
and Sanitation Utility of Lima, Peru   Ana Maria Zuluaga
WPS2479 Estimating the Effects of Corruption: Aminur Rahman       November 2000      S. Powell
Implications for Bangladesh      Gregory Kisunko                            33526
Kapil Kapoor



Policy Research Working Paper Series
Contact
Title                            Author                   Date              for paper
WPS2480 Productivity Growth and Resource    Mubarik Ali            November 2000      D. Byerlee
Degradation in Pakistan's Punjab:    Derek Byerlee                          87287
A Decomposition Analysis
WPS2481 Foreign Direct Investment in Africa:  Jacques Morisset     November 2000      N. Busjeet
Policies Also Matter                                                        33997
WPS2482 Can Institutions Resolve Ethnic    William Easterly        November 2000      K. Labrie
Conflict?                                                                   31001
WPS2483 The Credit Crunch in East Asia:    Pierre-Richard Ag6nor    November 2000     M. Gosiengfiao
What Can Bank Excess Liquid      Joshua Aizenman                            33363
Assets Tell Us?                  Alexander Hoffmaister
WPS2484 Banking Crises in Transition       Helena Tang             November 2000      A. Carcani
Economies: Fiscal Costs and      Edda Zoli                                  30241
Related Issues                   Irina Klytchnikova
WPS2485 Are Corruption and Taxation Really   Raymond Fisman        November 2000      R. Bonfield
Harmful to Growth? Firm-Level    Jakob Svensson                             31248
Evidence
WPS2486 Who Must Pay Bribes and How Much? Jakob Svensson           November 2000      R. Bonfield
Evidence from a Cross-Section of                                            31248
Firms
WPS2487 Finance and Macroeconomic Volatility Cevdet Denizer        November 2000      I. Partola
Murat F. lyigun                            35759
Ann L. Owen
WPS2488 Revisiting the Link between Poverty  Niels-Hugo Blunch     November 2000      H. Vargas
and Child Labor: The Ghanaian    Dorte Verner                               37871
Experience
WPS2489 Banking Crises and Exchange Rate  Ilker Domac              November 2000      A. Carcani
Regimes: Is There a Link?        Maria Soledad Martinez                     30241
Peria
WPS2490 Contractual Savings, Stock, and    Gregorio Impavido       November 2000      P. Braxton
Asset Markets                    Alberto R. Musalem                         32720
WPS2491 Labor Demand and Trade Reform      Pablo Fajnzylber        November 2000      T. Gomez
in Latin America                 William F. Maloney                         32127
,/1