way to compare differences in choice from one country to another and over time within a country. For instance, the Chilean reforms first allowed municipalities
to set wage levels for human resources, but over time
that “decision space” was narrowed by national legislation that restored civil service salary ranges. Using this
method, our team at Harvard along with our counterparts in different countries have also been able to assess the impact on some indicators of performance of
these different types of decentralization. In Colombia
and Chile we were able to show that decentralization
has improved the equity of budgetary allocations for
municipalities—contrary to expectations that decentralization would increase the gap in health spending
between rich and poor municipalities.
However, lack of reliable data on health statistics, service utilization, funding levels, and other indicators of
system performance hampers the ability to demonstrate
the effect of decentralization on performance.
In India, we completed a study of three different Indian
states with different degrees of decision space for local
authorities. But the data on health system performance—
in terms of levels of immunization, prenatal care activities, chronic disease management—were not consistently
What We Need to Know
The increased emphasis on implementing decen- tralization of health services in Latin America nd elsewhere makes it important to develop the evidence needed to evaluate those poli- cies. We need to know what degree of “decision
space” exists, and what kinds of institutional capacities
(skills and experience, funding levels and management
organizations) are effective to achieve policy objectives
that can range from improved responsiveness to local
health problems and priorities, to building greater equality among and within municipalities and developing a
more efficient use of scarce resources.
This research agenda should ideally be undertaken at
the earliest stages of policy changes to establish a baseline from which to judge the outcome of changes. It is
also important, as in the Seguro Popular case, to allow for
the random selection of target areas in the initial phases
of implementation to better demonstrate that the policy
changes, rather than other factors, caused the changes.
The key question remains: what type of decentralization in health services will improve institutions’ ability
to deliver better health outcomes?
To answer this question we need to
learn more about which constituents get
more choice in the decentralization process and, perhaps more important, how
much choice they get over what kinds of
issues. Then we need to develop a metric for what constitutes institutional
capacity at the central and the decentralized levels.
What skills and knowledge are required of administrators? How much expertise in accounting, financial
management, human resources management, and political advocacy are essential? What funding levels are
sufficient from both the central and local sources? What
governance issues about transparency, corruption and
elite capture need to be addressed to make local authority more effective?
Finally, we need better outcome data at the decentralized level to demonstrate how health status, financial risk protection and patient satisfaction change with
the changes in decision space and institutional capacity.
Armed with better studies we should be able to move
beyond the anecdotal stories that have dominated the
policy debates of recent years to shape better policies
to improve health outcomes.
In colombia and chile, decentralization
has improved the equity of budgetary
allocations—contrary to expectations.
accurate enough for their system of routine reporting to
show the impact of different degrees of decision space.
On the other hand, a careful study of the impact of
Mexico’s recent health care reform— Seguro Popular, initiated in 2003—delivered some interesting results. Selecting a random sample of municipalities, the study
demonstrated significant improvements as a result of
decentralization—in one case, a 29 percent reduction
in catastrophic expenditures by Seguro beneficiaries in
only 10 months. The reform was an unusual combination of federal initiatives and funding, along with mandated activities, that required matching funding from
the states with a rather “wide” decision space for local
delivery. The study suggests that a reform structured to
improve institutional capacity at both the center and
the state level can achieve significant benefits.
