CHAPTER 3

Decentralization, Fiscal Effort, and Social

Progress in Colombia at the Municipal
Level, 1994–2009: Why Does National
Politics Matter?

Fabio Sánchez Torres and Mónica Pachón

This chapter is a reduced version of the Interamerican Development Bank

(IDB) Working Paper No. IDB-WP-396, available at http://bit.ly/1AK0PfO.
The authors would like to thank the Global Development Network and all the
participants at the workshops held during the course of this project for their support
and comments. They would specially like to thank Ramona Angelescu, Guillermo
Perry, Matthias Krause, Jean Paul Faguet, and Ben Ross Schneider for their very
useful and insightful commentsthat helped advance the ideas that were put forth.
The authors also thank Sandra Margarita Zamora, Susana Rubinstein, and Santiago
de la Cadena for their research assistance, and finally they acknowledge the help
given by the Social Sciences Faculty at Universidad de los Andes.

F.S. Torres (*)

School of Economics, Universidad de los Andes, Bogotá, Colombia
M. Pachón
School of Political Science, Government and International Relations,
Universidad del Rosario, Bogotá, Colombia

© The Author(s) 2017 67

G. Perry, R. Angelescu Naqvi, Improving Access and Quality of
Public Services in Latin America, Latin American Political Economy,
DOI 10.1057/978-1-137-59344-3_3
68 F.S. TORRES AND M. PACHÓN

1   Introduction
Decentralizing policies has been advised and encouraged as a way to bring
governments and public goods delivery closer to the people. It is expected
that voters would elect local politicians who have a better understand-
ing of local needs, and once in office, would allocate the budget to bet-
ter accommodate these needs. Faguet (2004), for example, found that
after decentralization reforms, the Bolivian local governments allocated
more funds to those public goods that the local population needed the
most. Melo (2005) and Faguet and Sánchez (2009) found evidence that
decentralization played a major role in increasing access to education in
Colombia. Yet any evidence of the impact of decentralization on social
progress has shown mixed results (Bardhan and Mookherjee 2006). As
has been argued by Weingast (2009) and others, there are many factors
that may jeopardize the efficient allocation of resources, prompted by a
competitive political process. Among these are the asymmetries and lack
of information in relation to the delivery of public goods at the local level,
the presence of interest groups, weak political parties, as well as clientelism
and corruption—all leading to the slowing down of social progress.
This chapter explores the effect of electoral competition—both local
and national, at the local level—on the incentives to build fiscal capac-
ity and provide public goods such as education and water. The research
hypothesis is that political competition at both these levels gives a boost to
municipal decentralization as measured through local fiscal capacity. Our
findings support this hypothesis. For instance we find that fiscal capacity is
the fundamental variable that explains the differences in sector-wise per-
formance across local governments.
Our findings are consistent with the recent literature on decentraliza-
tion, which stresses the importance of mobilizing local resources to foster
social progress and economic development. As Weingast (2009, 280) has
recently suggested, it is necessary to take into account the type of incen-
tives—in the local political context—that transfers generate without the
existence of fiscal effort, and to not only consider the “importance of
transfers for mitigating horizontal and vertical imbalances.” In the absence
of a local fiscal effort, local authorities may not necessarily act to maximize
social welfare. Also, in such an instance, voters may not care to hold them
accountable. Our paper argues that political competition has a significant
impact on the provision of public services through fiscal effort. When
municipalities are not controlled by a small, regional elite segment but,
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 69

rather, the municipal electorate supports a larger number of candidates,

local politicians cannot rely on national politicians to obtain resources
from the national government. They thus have more incentives for mak-
ing a greater effort in tax collection at the local level. To follow Gadenne
(2011), we expect that, by paying more local taxes, people will have greater
incentives to monitor the local politicians, and electoral accountability will
improve as the population becomes aware of public policy performance.
When local governments involve themselves in policies and actions to
raise their own taxes—in many cases encountering harsh political process—
they become more accountable to their citizens. The voters would want
to have better control over the allocation of the new funds; they would
demand greater efficiency in spending and would pressure the authorities
for a budget allocation closer to the population’s needs. In a pure, local
fiscal framework, citizens internalize their marginal cost of taxes and con-
sequently push for an allocation that better fits their needs, particularly
in the sectors that yield the highest marginal benefit. Such sectors would
be those that provide public goods that are relatively less accessible to the
local population. Thus, in municipalities that make concerted efforts to
raise their fiscal capacity, more rapid progress would be expected, espe-
cially when considering the indicators of social development, as the citi-
zenry would demand better responsiveness and efficiency in spending.
In addition, we suggest that electoral competition has a mediated impact
through fiscal capacity on the distribution of expenditures, leading to posi-
tive outcomes of local resource allocation in education, water, and sewerage
services. As political competition increases, so does awareness about citizens’
needs and the competing parties’ commitment to campaign promises that
appeal to a larger electorate. In the absence of such competition, we could
expect political leaders to reward their loyalists through a more patronage
approach involving exchange of goods and services (selective incentives) for
their political support. This, in turn, could adversely impact fair and efficient
allocation of valued resources. A political process is efficient when it facilitates
the correspondence, in relative terms, of need-based resource allocation. It
is also expected that with a relatively higher correspondence between real
needs and budget allocation, expenditure efficiency must rise.
We analyze this relationship in two sectors: education, and water and
sewerage—two areas of policy in Colombia where the municipality has
substantial jurisdiction. Access to education has increased enormously in
the past two decades (Rodriguez 2010; Faguet and Sánchez 2008), and
the differences in the growth of enrollment rates across municipalities
70 F.S. TORRES AND M. PACHÓN

are quite significant. Moreover, the quality of education seems to remain

stagnant, as shown by the different international tests such as the PISA
(Program for International Student Assessment) and TIMSS (Trends in
International Mathematics and Science Study).
The water and sewerage sector has undergone major institutional changes,
with an important part of their delivery being handled by private or semi-private
firms (Silva 2007; Granados 2008). Currently, more than 35 percent of munici-
palities have effected changes in their water provision services and have put
these in the hands of private companies or specialized providers. Nevertheless,
there is a great deal of discussion regarding the impact of such changes on the
growth of coverage, and on the quality of water and sewerage services.
To carry out our research, we used a rich data set containing detailed
information on sector-wise performance of all the municipalities in
Colombia from 1994 to 2012; information on land tax was gathered
from the National Planning Department and cadastral information from
the Geographical Institute Agustín Codazzi, as well as electoral results for
the councils, mayors, and House of Representatives across the country
from 1994 to 2009 from the Colombian Electoral Office (Registraduría
Nacional del Estado Civil). To complement the quantitative analysis, we
also conducted several interviews of politicians from municipalities where
we could observe the variations in the outcomes as well as in the indepen-
dent variables, mainly with regard to fiscal effort and electoral competition.
This paper is divided into six parts. The second section briefly summa-
rizes the evolution of the decentralization process in Colombia, giving an
account of three aspects of the process: the evolution of local finances, the
quality and coverage of education, and the provision of water and sewer-
age services. The third part is the theoretical framework and a summary
of the more general expectations. The fourth part describes the empirical
methodology linking the political and local state capacity (measured as
fiscal effort), and then local state capacity to service delivery outcomes.
The fifth section shows the results, and the sixth is the concluding section.

2   A Short Account of Colombia’s

Decentralization Reforms
The first steps of Colombian fiscal decentralization reforms occurred at the
end of the 1950s. As reported by Junguito and Rincón (2004), the 1958
Constitutional Amendment assigned at least 10 percent of the national
budget to education expenditure, thus marking the beginning of “a for-
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 71

mal scheme of transfers.” The next steps included the 1968 Constitutional
Amendment that created the “Situado Fiscal”—Law 33 of 1968—that
initiated the sharing system and the sales tax cession, and Laws 46 of 1971
and 14 of 1983 that aimed to reinforce the municipal and departmen-
tal taxes. Colombia, nonetheless, remained very centralized politically, as
designed by the 1886 Constitution. All governors in Colombia were presi-
dential appointees, and, in turn, were in charge of naming all the mayors
in the municipalities. None of them had fixed terms. As a result, local
political careers of most candidates depended on their ties to the regional
and national leaders.
The most significant departure from the 1886 Constitution was the 1986
Constitutional Amendment, which, for the first time in the twentieth cen-
tury in Colombia, stated that mayors should be popularly elected for two-
year terms, with no possibility of immediate re-election.1 Consistent with
the 1986 Constitutional Reform, the 1991 Constitution ushered in a new
stage of decentralization, establishing the rules that would allow the citizens
greater say in public policy as well as to oversee the functioning of their politi-
cally elected leaders. Thus, governors also became popularly elected.
Political decentralization was complemented with fiscal decentralization.
Both types of decentralization augmented—according to Falleti (2010)—
the autonomy of local and regional politicians from the central govern-
ment, which brought about a significant change in the inter-­governmental
balance of power for governors and mayors. For that purpose, the 1991
Constitution introduced a new scheme of transfers which developed into
Law 60 of 1993—with a very precise set of formulae—in which the central
government’s current revenues (mainly national taxes) were to be shared
with the departments and with municipalities. Transfers were distributed
on the basis of unmet basic needs to all municipalities; then, municipalities
had to distribute their own expenditure across sectors, each one with a
fixed percentage. Under these rules, there were few incentives for munici-
palities and departments to increase their revenue-generation capacity.
The established institutional framework also defined the distribution of
natural resource royalties among departments and municipalities (Articles
360 and 361).2
In an effort to better align incentives of politicians to improve their
policy performance, the Law 715 determined that the distribution of the
transfers would be based on the coverage and growth of the health and
education services provided by the territorial entities, and not by popula-
tion and poverty rates—criteria that had been established by Law 60 of
72 F.S. TORRES AND M. PACHÓN

1993.3 Also, if one considers that the length of terms of mayors and gover-
nors was, at the time, three years (from 1994 till 2003) with no possibility
of re-election, political incentives were also not aligned to pay the costs for
raising taxes, without enough time to deliver better public services.
Thus, the 1991 Constitution initiated a process of decentralization
focused on the local provision of goods and services and on transfers,
but neglected the sub-national governments’ (SNGs) generation of their
own resources. As could be expected, during the 1990s, departments and
municipalities financed most of their expenditures through transfers from
the central government instead of making an effort to build a local tax
base. This resulted in significant vertical imbalances and, in most cases,
in fiscal deficits4 that threatened fiscal sustainability and macroeconomic
stability (Sánchez and Zenteno 2011).5
Municipal spending increased from 3.0 percent to 6.8 percent of GDP
between 1994 and 2009, while its own revenues rose from 1.4 percent to 2.5
percent of the GDP during the same period. Thus, municipal vertical imbal-
ances rose—a fact that may have distorted the incentives toward efficiency and
responsiveness of local governments (Sánchez and Zenteno 2011; Sánchez
et al. 2012). At the municipal level, almost 90 percent of the tax revenues
are represented by the Property and Land Tax, Commerce and Industry Tax,
and the gasoline surcharge. Between 1996 and 2000, per capita municipal
taxes did not present significant changes. From 2001, they began to steadily
increase, particularly reflecting the behavior of the property and industry and
commerce taxes. The dynamism in tax collection can be attributed to the tax
reforms, such as Law 488 of 1998 and Law 788 of 2003, which increased the
base for some sub-national taxes such as the gasoline surcharge.

2.1  Decentralization Impact on Education: Better Coverage,

Deficient Quality
Access to education, particularly quality education, is a proven instru-
ment for increasing employment opportunities, which in turn will improve
lifetime income levels, improve health status, and lower pregnancy rates.
Recognizing this fact, the Constitution of 1991 established the right to
education for all citizens, compulsory for all children from 6 to 15 years of
age. Although additional resources have been allocated for improving the
coverage and quality of education (Rodriguez 2010), there are still enor-
mous differences in the enrollment rates and quality of education among the
Colombian municipalities. Some of these differences are explained by struc-
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 73

tural factors such as poverty and wealth distribution, while others are related
to regional and local aspects such as tax capacity and political processes.
Faguet and Sánchez (2008) and Melo (2005) have shown that from
1993, student enrollment in public schools rose significantly. Thus, the
overall enrollment in schools—as a percentage of the population—grew
steadily from 22 percent to 25 percent from 1993 to 2009, while public
school enrollment increased from 14 percent to 21 percent, indicating
that in net terms most of the new students joined the public school system.
While Faguet and Sánchez (2009) state that the allocation of municipal
resources may be the key factor in explaining the differences in enrollment
growth, Melo (2005) argues that the increase in coverage may have been
accomplished at the expense of quality.
Although national transfers go strictly by percentages when it comes to
allocating money for expenditure on a sector, the local administration reserves
the right to spend their own resources as they deem fit, be it on infrastructure,
educational material, or on additional teachers, besides the ones hired by the
department. The decision in terms of what to spend on is expected to have an
impact on the coverage and quality of education. Also, municipal tax capacity
further determines the amount that local governments may freely invest from
their own resources. The evolution in spending came about, first, during the
early decentralization of the 1990s when funds from local resources financed
around 8 percent of the total education outlays. This proportion dropped to
2 percent—in part as a consequence of the increase in central government
transfers—and rose to more than 10 percent around 2008. Thus, after 2002
and coinciding with the Constitutional Reform of the central government
transfer system, as well as the enactment of Law 715, the proportion of edu-
cational spending from local resources began to increase.
In terms of quality, the education scenario seems to remain stagnant
as demonstrated by the different international tests such as PISA and
TIMSS, and by student performance in the Colombian national tests. For
example, most of the public schools rank lower in test distribution (ICFES
2009). Nonetheless, the evidence for Colombia is mixed and the different
methodologies adopted have been subject to criticism.6

2.2  Decentralization Impact: Mixed Results on the Provision

of Water Services
The issue of water and sewerage is particularly important in Colombia, as it
involves different levels of government and the public and private sectors.
74 F.S. TORRES AND M. PACHÓN

In this regard, the central government transfers the resources to be spent on

water and sewerage to the local governments, and at the same time the local
service is provided by the municipality or by a public, semi-­private, or a solely
private entity. Thus, it emerges that there is an interplay of different actors that
brings about different outcomes depending upon the local political setup, the
origin of the provider—public or private—and the local institutions.
Historically, water and sewerage expansion has been affected by politi-
cal and electoral interference, given the political gains as represented by
the provision of such essential services. By the late 1980s the situation of
service provision resembled what Spiller and Savedoff (2000) have called
a “low-level equilibrium,” in which tariffs were low and did not cover
the costs associated with the expansions in coverage and service quality.
Additionally, the system lacked a pricing mechanism to allow the rational-
ization of consumption. As part of this scenario, the system was reformed
within the framework of decentralization.
The reform transferred to the municipalities both the public works-­
related operations and the management of the service. Simultaneously, the
resources transferred from the central government to the municipalities
were increased as means of supporting the latter’s autonomy, and to help
develop their new responsibilities. Under this new framework, the central
government was responsible for the planning, regulation, oversight, and
control of the services, while the provision was carried out by a provider,
which could be one of the following: (1) public service companies incorpo-
rated as public limited liability companies (S.A. ESP, Empresas de Servicios
Públicos), (2) municipalities as direct providers, (3) government-­managed
industrial and commercial companies (EICEs, Empresas Comerciales e
Industriales del Estado), (4) marginal or independent producers, or (5)
organizations authorized to provide services in rural areas or specific
urban areas.7 Accordingly, under the new institutional framework the
municipalities were autonomous although the central government contin-
ued to be the main source of financing for the sector’s investments.8 When
observing the evolution of average own expenditures on the water sector,
the trend was similar to that observed in the education sector: the more
the national transfers, the smaller the investment of own resources in the
sector. After the 2001 reform, incentives were transformed and the trend
changed, increasing the average of own resources utilized.
The pace of transformation of the municipal providers into a business-­
like management model in the case of water supply and sewerage ser-
vices has been rather sluggish. In fact, the majority of direct providers
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 75

are still the municipalities—representing nearly 35 percent of them (Silva

2007). Law 142, by incentivizing the replacement of the municipalities
as direct providers by Companies of Public Utilities, intended to improve
the coverage and quality of the services, which would translate into bet-
ter indicators of the quality of life. Analyzing the influence of governance
on the performance of water and sanitation (WS) services in developing
countries, particularly the Colombian case, Krause (2007) found that
­low-­quality governance of sub-national governments compromises the
internal efficiency of service delivery and widespread access to the ser-
vices. The results obtained have yielded some evidence that private sector
participation (PSP) enhances the internal efficiency of service providers.
When a local manager was asked about the municipal regime before it was
transformed to an ESP (public sector enterprise), he replied,

Back then, all the jobs that had to do with the service delivery were political
quotas. So, the “escobitas” [cleaners], you go, you go… as well as the per-
son charging for the service were political quotas. People that had no clue of
what they were doing were put in these positions in the municipality’s small
company. There were no clear policies or goals, the service was, in practical
terms, free. People paid 20 pesos for the water.9

However, the transformation per se has not been shown to be enough

(Domínguez and Uribe 2005; Krause 2007). The evidence on the advances
in provision and quality in the sector is mixed. As for coverage, census data
reveal that water coverage rose from 60.5 percent to 64 percent in the munic-
ipalities that did not reform and from 78 percent to 78.5 percent in the ones
that did. Granados (2008) shows that water and sewerage coverage was less
in the municipalities with private provision, while the reduction in infant mor-
tality was slower.10 In contrast, Barrera and Oliveira (2007) found positive
effects on coverage and health as a result of the involvement of private parties
in the provision of these services, particularly in urban areas. Prasad (2006)
indicates that the studies that focused on the performance of the provider
companies from a microeconomic point of view—analyzing the efficiency and
productivity indicators—are not conclusive in terms of the effects of private
capital involvement. Gomez-Lobo and Meléndez (2007) also obtained mixed
results on evaluating the PSP in the water and sewerage sectors. Concerning
the affordability of the service, researchers found no statistically significant
effect of PSP. Nonetheless, PSP does seem to increase the quality of service,
measured as continuity of service, and sewerage connection rates.
76 F.S. TORRES AND M. PACHÓN

3   Conceptual Framework: The Necessary Link

between Political Competition, Local State
Capacity, and Policy Outcomes
In the literature different governance factors have been considered to
influence the provision of water and education, such as transparency,
accountability, technical and fiscal capacity, quality of the local bureau-
cracy, participatory mechanisms, competitive political processes, among
others (Andrews and Shah 2003; Krause 2007). In this paper, we focus
on the effect that national electoral competition has on fiscal effort at the
local level and thereby on public service delivery.
There are various arguments in the literature that tie local political com-
petition to better policy outcomes (Weingast 2009; Besley et al. 2010).
The basic argument is very simple—the more the political actors with real
chances to win office, the better should be their performance and qual-
ity of candidates to avoid being ousted from power. The presence of an
opposition too plays an important role in the provision of information to
the citizenry, lowering the probability of the misuse of public resources.
However, for political competition to make a difference at the local
level, political actors need to have enough autonomy for their decisions
to make a difference in outcome. If there is no autonomy and transfers
are only weakly related to sub-national income growth, the incentives for
better performance can rapidly vanish. In Weingast’s words, “Elections in
the presence of fiscal dependence and opportunism become a means of
political control rather than of citizen expression” (Weingast 2009, 280).
In other words, under a soft budget constraint, politicians would be able
to spend more than they collect, and by doing so, they could effectively
protect or appeal to a certain constituency that they consider pivotal to
their ascent in their political careers.
The budget constraint is also dependent on the availability of resources
for the national government, which does not entail asking for more taxes
from the citizens at the local level. When the regional or national elite can
help ease the budget constraints, their role becomes more significant in
determining policy outcomes. This dependency allows for the national elite
to control the local elite, forcing the latter to make sub-optimal decisions for
their population, or provide them with additional resources in exchange for
political rents and patronage. As a council member interviewed clearly stated,

A good administration in this municipality is done with gestión.11 That is

why I said we would be among the best Santa Bárbara has had, as we have
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 77

a direct connection with León Darío, [the brother of the elected mayor],
in the Chamber of Deputies [former mayor, elected three times non-­
consecutively]. Also, on top of that, he belongs to the Third Committee,
which you already know is the one that decides over the Budget and that
gives him a number of advantages, like it is being the rapporteur, for which
they are given some additional incentives compared to other members.12

There are costs associated with each of the two options to soften the bud-
get constraint. Lobbying for additional national resources implies an elec-
toral compromise and exchange for support with the regional leader who
has the leverage to deliver additional resources. Possibly too, the mayor
would need to prioritize the legislator’s electoral interests instead of his
own. Also, if the municipality is controlled by someone from among the
regional elite, that person will take advantage of this closeness to extract
whatever resources possible to deliver to his or her electorate. In places
where there is greater competition for the votes, and consequently ties to
the regional political network are thinner, the local politician may have
reduced chances of gaining access to such resources. As ­ competition
increases at the regional level for national office, politicians too are
required to maximize their vote share, and may opt for a campaign that
appeals to voters by providing national public goods and policies instead
of localized or appropriable goods (Cox 1987).
Local politicians can also choose to increase their resources by increas-
ing their fiscal revenue. An increase in property and land tax—namely the
updating of the local cadaster—is one of the most important fiscal policy
decisions for a municipality to increase its resources and ease its budget
constraint. Although updates are mandatory at least once every five years,
they can even be done as often as every year. Local politicians consider the
decision to update a difficult one, with associated costs such as the loss
of popularity, which could be significant at an early stage of one’s career.
This update—a clear action of the fiscal effort—determines the property
and land tax base, and therefore its evolution. The lack of update brings
about an undervaluation of the local properties in the local cadaster and
consequently may lead to a tax collection below its potential.
Thus, in municipalities where politics would be captured by one or a
few groups, politicians would most likely lobby for more national resources
instead of raising taxes. Consequently, instead of improving their career
chances through the provision of public goods, they would provide pub-
lic goods and rents through client-based practices that do not necessarily
match the needs.
78 F.S. TORRES AND M. PACHÓN

When asked about the cadastral update, a local politician from El Peñol,
a small town in Antioquia that frequently updates the cadaster, argued,

People do not like that the administration updates the local cadastre. It is
clearly an unpopular measure and if you do it as a mayor you lose popularity.
But you have to do it, you have to respect the law and do it.

When asked about the reasons as to why updating the local cadastre every
five years was not done by a great number of municipalities despite their
obligation to do so, the Mayor of Monterrey, Casanare, said:

If you tax newly established enterprises, you can do it and you will improve
the collection. Despite the update, evasion is fairly generalized and it is dif-
ficult to force people to pay. Nonetheless, if you tell me I am required to
update, I will do it only in the first year. I am finishing my term, and I think
it would be political suicide to do it at another time.

Notwithstanding the political costs to update, it has been shown that

the benefits to the municipality, as mentioned previously, are far greater
than the ones received from other sources (Perry and Olivera 2009). For
example, it has been shown that municipalities with royalties do not have
better social indicators than those without royalties despite the signifi-
cant difference in resources (Perry and Olivera 2009; Economía Urbana
2012). Hence, money that is collected from the citizenry may prove more
beneficial in terms of the outcomes produced than money acquired from
other sources. Thus, when local governments raise more local resources,
have some autonomy to optimize their finances, and are able to impact
policy outcomes, political competition could explain improvements in
policy outcomes. Since local politicians are to blame if things go wrong,
incentives are aligned for both politicians and voters to make an additional
effort in contributing to the public good, as well as exercising their right
to take stock of their performance.
On evaluating a program in Brazil that invests in the modernization of
local tax administrations, Gadenne (2011) found that the increase in local
taxes, prompted by the program, brought up educational enrollment levels
as well as the number of schools built with greater efficiency than did central
governments transfers. Gadanne stated that since citizens have better infor-
mation on taxes than on transfers, rent-seeking opportunities of politicians
are considerably diminished, leading to better spending of the resources.
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 79

In Colombia, as we have observed in the previous sections, municipali-

ties had a soft budget constraint until 1997, the year in which significant
restrictions were imposed on their capacity to be in debt. Following this,
the municipalities, in order to increase their execution capacity, needed to
increase their fiscal capacity. Thus, we argue that national political com-
petition at the local level affects both the effort of the municipalities to
increase their autonomy by raising more revenue and the decisions neces-
sary to improve the service delivery. Local fiscal capacity depends not only
on the wealth and economic activity of the municipality, but also on the
fiscal effort, defined as policy actions that the local government under-
takes to augment local revenues. The fiscal dimension of the State—under-
stood as its ability to tax income and wealth—has recently been regarded
as fundamental for the delivery of public goods, policies promoting devel-
opment, and the implementation of distributive policies (Careaga and
Weingast 2003; Besley and Persson 2010; Cárdenas 2010). In this paper
we allege that local fiscal capacity is also essential for the delivery of local
public goods within a context of political and fiscal decentralization as
experienced by Colombia after 1991. Thus, to understand the interplay
between fiscal capacity and political competition it is crucial to understand
why the improvement of the welfare of the population varies within and
across Colombian municipalities.
We consider two policy sectors that are crucial to the welfare of the
population: education and water. In both these sectors, the municipality
plays a leading role vis-à-vis their performance. In education, for example,
mayors are in charge of setting need-based priority levels in schools with
regard to investment, as well as the type of investment, whether the need
is for infrastructure, educational material, feeding programs, or for more
teachers, to mention just a few. Equally, in the water sector, the mayor is
responsible for taking a decision as to who will provide the service, who
will monitor the provider, and whether the coverage should be increased
to new areas of the municipality. The council and the mayor have the
added responsibility of defining the levels of cross-subsidy from high- to
low-income levels.
The following section gives the methodology and measurement strat-
egy to establish the link between political competition and local state
capacity (measured as fiscal effort), and also to establish the link between
local state capacity and service delivery outcomes.
80 F.S. TORRES AND M. PACHÓN

4   Methodology

4.1  Link between Political Competition and Local State

Capacity: How Often Do Mayors Update Their Municipality’s
Local Cadastre?
To operationalize the decision that best shows the intent of the mayors
to have financial autonomy in spending and prioritizing the available
resources, we focus on the decision to update the cadastre. The updates of
the local cadastre allow us to calculate the underestimation of the values of
local properties—both urban and rural. Thus, once the undervaluation of
properties is estimated it will be used as the instrumental variable of local
tax revenues in the econometric exercises aimed at explaining coverage
and quality outcomes in education and water. In fact, tax revenues may
be endogenous to the provision of education and services; local officials
may decide to increase local taxes when they face higher demand for such
public goods.
Cadastral information indicates that the municipalities take on the task
of cadastral updating approximately nine years after the last update took
place. Nevertheless, 20 percent of the municipalities undertake updates
after five years or less following the last urban update, or seven years or
less of the last rural one. In order to determine the variables explaining the
cadastral updates we estimate a number of hazard models with the follow-
ing specification that includes the contextual variables that may affect the
decision to update, as well as the political competition variables to observe
whether it has the effect we have so far described:

UPDATE i ,t ( = 1) = β1 × Years to last Updatei ,t + β 2 × Political

Variablesi ,t + β 3 × Socio Economic
(3.1)
Variablesi ,t + β 4 × Coverage / Quality Public
Goodsi ,t + Tt + DPk ,

where UPDATEi,t equals one in the years in which the local cadastre
(either urban or rural) has been updated, and zero for the other years. Tt
and DPk stand for time dummies and departmental fixed effects, respec-
tively. It is expected that the longer it takes to update the local cadastre,
the greater the likelihood of updating (β1 > 0). We would also expect that
the situation of coverage or quality of the local public goods should not
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 81

be related to the cadastral updates (β4 = 0) and that, as mentioned earlier,

the political context, particularly political competition, would influence
the cadastral updates.
After estimating the cadastral update hazard model, the undervaluation of
local properties can be assessed. Such undervaluation depends on the num-
ber of years when there was no cadastral update and the number of updates
within a given period. In other words, as the cadastre becomes older, the
values and the number of properties tend to be lower than they should be.
Thus, the lack of updates erodes the tax base, with tax collection remaining
far below its potential. To determine the effect of the lack of updates on the
value of properties, we first estimate a panel fixed effect model for the per
capita value of municipal properties. Such value will be determined by the
structural characteristics of the municipality—such as GDP per capita, pov-
erty rates, concentration of land, participation of urban population, among
others—and by the number of years to the last update, as well as the number
of updates undertaken. Thus, the per capita value of properties is estimated
with the following fixed effect panel equation (see appendix for details):13

Value ofPropertiesit = ζ 1 × Local Structural Variablesi ,t

+ θ1 × Number of year to last
(3.2)
cadastral updatei ,t + θ 2 × Number
cadastral updatesi ,t − to + Tt + σ i + υi ,t .

Both the number of years to the last cadastral update and the number of
cadastral updates are indicators of the local fiscal effort and should affect
local tax revenue. The first is positively related and the second, negatively
related to undervaluation.
Once Eq. (3.2) is estimated, we calculate the undervaluation of proper-
ties by using the following equation:

Estimated Under Valuation i ,t = θ1 × Number of year to

last cadastral updatei ,t
(3.3)+ θ × Number cadastral updates.
2

Calculations based on the information from the cadastral office indicate

that the average undervaluation of properties is about 0.2 log points
82 F.S. TORRES AND M. PACHÓN

(about 20 percent) with a standard deviation of 0.25 log points. The dis-
tribution goes from −0.06 to nearly 0.9 log points.

4.2  The Effect of Local State Capacity on the Efficiency

of Service Provision Systems in Education and Water

4.2.1 The Model for Enrollment Rates and Quality of Education

As already mentioned, we suggest that higher fiscal effort of local govern-
ments should reflect itself in higher level of public goods provision. In this
case, we expect that when the citizens pay their dues they are in a better
position to demand that the administration respond to their most urgent
needs, and, in turn, the administration should be able to maximize its
electoral advantage by responding to its electorate. We first address the
education sector.
Then, the influence of the fiscal effort on the educational indicators is
estimated using the following model:

Yi ,t = α1 × Fiscal Capacity ( Taxes ) + α 2 × CG Transfersi ,t

+ α 3 × Royalttiest + α 4 × Political Variablesi ,t (3.4)
+ α 5 × Socio Economic Variablesi ,t + θi + Tt + ei ,t ,

where Yi,t represents an educational outcome—enrollment rates or qual-

ity of education—and the expressions on the right are the explanatory
variables. The symbol θi stands for the municipal fixed effects, while Tt are
year dummies. It is expected that α1 > α2 and α1 > α3 as the efficiency of
the locally raised funds should be greater than the efficiency of the central
government transfers and royalties.
Nevertheless, the coefficient α1 may be biased as both local taxes and
the educational indicators may be related to the omitted variables that
change over time and hence are not fully captured by the municipal fixed
effects. One example of an omitted variable, for instance, is the price of
local production that may affect both local taxes through larger economic
activity and educational outcomes of such enrollment rates. In order to
correct the likely bias in the estimator we instrument the tax capacity vari-
able. The instrumental variable used is the estimated undervaluation of
the properties of the local cadastre, explained above. Such undervaluation
depends exclusively on the frequency and number of cadastral updates.
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 83

Conceptually, undervaluation is not related to the educational outcomes;

it is only through greater tax capacity that a municipality may attain it
when its tax base is enhanced by an administrative action.
Therefore, the first stage of the educational outcome equation is done
using the following equation:

Fiscal Capacity ( Taxesi ,t ) = γ 1 × Estimated Under Valuation i ,t

+ γ 2 × CG Transfersi ,t + γ 3 × Royaltiesi ,t
+ γ 4 × Political Variable si ,t + γ 5 × Socio
Economic variablesi ,,t + ψ i + Tt + ei ,t .
(3.5)

It is then expected that γ1 < 0 since an increase in the undervaluation

for properties would reduce both the property and land tax base and, con-
sequently, the potential tax collection.

4.2.2 The Model for Coverage and Quality of Water

As in the case of education, the estimated model lets us determine the impact
of the fiscal effort and the structure of the water sector on the indicator of
coverage and quality of water at the municipal level. In the case of water
coverage, we use census data for 2005 at the municipal level, and hence the
models to be estimated are cross-section ordinary least squares (OLS) and
instrumental variable ones. Thus, the model has the following structure:

Wi = α1 × Fiscal Capacity ( Taxes )i + α 2 × CG Water Transfersi

+ α 3 × Pollitical Variablesi + α 4 × Political Variablesi
(3.6)
+ α 5 × Socio Economic variablesi + α 6 × EICEs + α 7 × ESP
+ α 8 × Years since reform + ei ,

where Wi stands for water coverage in 2005. For the rest of the ­variables
we computed their average for the period 1994–2005. The variable EICEs
expresses whether water is provided by a government-owned company,
ESP expresses whether it is provided by a mixed or private firm, and year-­
since-­reform stands for the number of years during which the EICEs or
the ESP has been delivering the service.
84 F.S. TORRES AND M. PACHÓN

5   Results

5.1  The Link Between Political Competition and Local Tax

Capacity
The results of the hazard model are presented in Table 3.1. Column 1
shows that the cadastral update is time-dependent. That is, the older the
last update the greater the likelihood of having one in the current year.
In column 2, some political variables to measure political competition are
introduced in the model: the effective number of parties in the munici-
pality, and two dummies identifying mayors from the traditional political
parties—Liberal and Conservative. The effective number of parties mea-
sures the local political competition,14 and the dummies for the traditional
parties are to determine if mayors who have a national network have less
incentive to update. To measure the degree of political competition at the
local level for national office we include two variables: GINI15 for party
share in the House votes (column 2), and the effective number of can-
didates in the municipality competing for the House of Representatives
(column 3). Both these variables indicate the degree of political com-
petition at the local level, for a seat in the Chamber of Representatives
(House). The variables capture the political control that a politician exerts
in a particular municipality. For instance, if the entire share of the votes
for the House in a given municipality is obtained by a particular politician
it means that such a politician controls that municipality. In that case the
mayor would try to maintain his or her political stronghold by bringing
in additional public goods financed through central government resources.
Hence, the local politicians—mayors and council representatives—would
not be so hard pressed to raise local taxes.
Although all the variables have the expected direction, only the GINI is
significant and is negatively related to the decision to update. This means
the fewer the parties in the municipality competing for departmental
votes, the fewer the incentives to update the cadastre (our measure of fiscal
effort). Equally, seen from the perspective of the effective number of can-
didates weighted by their support within the municipality, the expectation
that greater competition is positively related to the decision to update is
confirmed.16 This result confirms that the regional political context within
which the municipality’s politics happens is fundamental to understand-
ing the decision to update the local cadaster. The results, however, should
be viewed with caution as we are assuming that the greater concentration
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 85

Table 3.1 Determinants of cadastral update

The table has a weird format for the presentation of the results.

Variables Cadastral updating LOGIT (marginal coefficients)

Years to last update

Urban 0.1500*** 0.0092*** 0.0093*** 0.166***
(0.0119) (0.0008) (0.0008) (0.0136)
Rural 0.0775*** 0.0057*** 0.0058*** 0.109***
(0.0099) (0.0006) (0.0006) (0.0115)
Political
Representative 0.0006 0.00065 0.0115
number of parties (0.0018) (0.0018) (0.0327)
Mayor from −0.0022 −0.00204 −0.0588
Liberal party (0.0062) (0.0063) (0.116)
Mayor from −0.0017 −0.00137 −0.0321
Conservative party (0.0067) (0.0068) (0.124)
Gini of party share −0.6134*** −0.64577***
(0.1714) (0.1735)
Effective number 0.386***
of House (0.110)
candidates
Decentralization
Per-capita transfers −0.0251*** −0.02374*** −0.426***
(ln) (0.0065) (0.0066) (0.119)
Royalties −0.00022 −0.0002 −0.00191
(0.0006) (0.0006) (0.0106)
Students– −0.0302 −0.354 (0.940)
population ratio (0.0523)
Public test score 0.0922 (0.0803) 1.6 (1.450)
Departmental Yes Yes Yes Yes
fixed effects
Year effects Yes Yes Yes Yes
Observations 7826 7826 6497 6497

Source: See appendix

*,**,***: Coefficients significant at the 10 %, 5 %, and 1 % levels, respectively. Standard errors in
brackets
Per-Capita GDL (ln), Poverty Rate, Gini of Land Value, and Urban Population were included as
controls

of regional votes would soften the local budget constraint. We do not

have an exact measure of the resources that the regional politicians in the
National Congress may have delivered to the local governments.
86 F.S. TORRES AND M. PACHÓN

In column 3, we control for socio-economic variables the results of

which are not shown due to space constraints. Besides, the per capita
transfers from the central government are strongly and negatively related
to cadastral updates, indicating that local fiscal effort may be curtailed if
the municipality is financed through other sources. Finally, the last column
indicates that education enrollment rates and quality of education are not
related to cadastral update, implying that no fiscal effort is forthcoming
from local governments because they are trailing in these indicators.

5.2  The Positive Impact of Local Tax Capacity on Service

Provision in Education
Table 3.2 presents the results of the OLS and the instrumental variables
(IV) panel models for student enrollment in public school as a proportion
of the population. The OLS model indicates that an increase of 1 percent
on local taxes augments the enrollment rate by 0.007 percentage points,
the result being similar to the 0.008 increase in enrollment obtained
through central government transfers. The first stage of the IV model in
column 2 indicates that, as expected, the cadastral undervaluation of local
properties is negatively related to local per capita tax revenue. In fact, if
the estimated undervaluation rises by 1 percent, tax revenue would fall
by 0.11 percent. The first stage also reveals that royalties and intra-party
competition are positively correlated with local per capita taxes. The sec-
ond stage indicates that changes in tax capacity as prompted by changes in
tax effort do indeed positively affect student enrollment. According to the
coefficient, a 1 percent positive exogenous variation of local taxes increases
by 0.021 the proportion of the population enrolled in public schools. Or
in relative terms, if per capita taxes go up by one standard deviation (in log
= 1.26), the educational enrollment rate would sizably increase by 0.58
(0.021 × 1.26/0.063) standard deviations.
As for the other variables, the coefficient of the central governments
transfer is equal to 0.008 and similar to the one obtained using the OLS
panel model. The impact of royalties on coverage is zero. It is apparent
that the taxes coefficient is greater than that of the transfers. This result may
intuitively suggest that higher taxes lead to greater awareness among citizens
with regard to the allocation of the local budget, which, in turn, calls for
greater efficiency in the local public sector.
Finally, the political variables are only weakly correlated with the enroll-
ment rates in public schools. The local effective number of parties (ENP)
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 87

Table 3.2 Determinants of public school enrollment in Colombian municipali-

ties, 1994–2009
Variables The table has a weird format for the presentation of the results

OLS I2SLS IV

Decentralization
Per-capita taxes (ln) 0.00708*** 0.0213* (0.0133)
(0.000587)
Per-capita transfers (ln) 0.00816*** 0.017 (0.108) 0.00815***
(0.000628) (0.000702)
Royalties 0.000113 0.004*** 0.0000509
(0.0000722) (0.00121) (0.0000935)
Political
Effective number of parties 0.0185 (0.0186) 0.00254**
(0.00116)
Squared effective number of −0.0038* −0.000254*
parties (0.0023) (0.000149)
Intra-party competition −0.0000939 0.0059*** −0.000109
(0.0000828) (0.0015) (0.000120)
Proportion of council members 0.00156 −0.043 (0.028) 0.00377**
from mayor's party (0.00151) (0.00180)
Council members re-election 0.0000613 0.0015 (0.0014) 0.0000438
average (0.0000880) (0.0000924)
Mayor from Conservative party 0.00191** −0.00264 0.00181*
(0.000955) (0.0161) (0.000928)
Mayor form Liberal party 0.00105 −0.0308* 0.00123 (0.00107)
(0.000955) (0.0162)
Constant 1.436***
(0.0293)
Instrument
Cadastral undervaluation −0.1114***
(0.0242)
Municipal fixed effects Yes Yes Yes
School fixed effects
Year fixed effects Yes Yes Yes
F-test for instruments 21.05
Prob > F 0.000
Endogeneity test 21.11
Chi-sq(1) p-value 0.000
Observations 11,146 11,135 11,135

Source: See appendix

*,**,***: Coefficients significant at the 10%, 5%, and 1% levels, respectively. Panel regressions with robust
standard errors. Standard errors in brackets
Poverty Rate, Gini of Land Value, Population (ln), Teacher–Student ratio (lagged), No Official Students–
Population ratio (lagged) were included as controls
Enrollment: Groups by Municipalities: 917/906/906; Quality: Groups by Schools:
5482/5373/5373/5458/5346/5436
88 F.S. TORRES AND M. PACHÓN

correlates positively with enrollment, as does intra-party competition,

although the latter is not statistically significant. This is a result that seems
to hold throughout the analysis, suggesting that local politics has less impact
than the logic behind national politics as mediated by fiscal autonomy.
Table 3.3 presents both the OLS and IV panel model for the quality of
education. The variables on the right are calculated both annually and as six-
year averages (as it is assumed that the quality of education at the end of high
school has its foundations in at least the previous six years). Quality is mea-
sured as the average school score of the senior high school students in the
test SABER 11, relative to the test score of the average private high school of
the department where the public school is located. The departmental aver-
age score was used, given that in many Colombian municipalities no private
education is offered. The dependent variable is then

 School Score kJ , I ,t 
Y jk,i ,t =   ,
 Average private Score k
 

where j denotes the public school, i the municipality, t the year, and k the
department where it is located.
Columns 1 and 4 show the OLS panel estimates of the relationship
between per capita taxes and relative score. Such estimates reveal a positive,
statistically significant correlation between the local taxes and test scores in
public schools, especially in the six-year average. Moreover, a strong cor-
relation is found between central government transfers and relative public
scores, indicating that municipalities spend some of these resources on
goods that help to raise the quality of education. It is noteworthy that
the effect of local taxes on quality is substantially greater than the effect
of transfers. The effect of royalties on the quality of education is nega-
tive, coinciding with the previous findings that null the effect of royalties
on the welfare of the population (Gaviria et al. 2002; Perry and Olivera
2009). Column 2 presents the first stage of the quality of the education
model. As expected, the estimated undervaluation of the local properties
negatively impacts the local taxes. The political variables, again, do not
exhibit a consistent relationship with the per capita taxes. For example,
the variable-measuring intra-party competition is positive but with a low
real impact, and low re-election average. This last variable would suggest
that the more senior and successful the council representatives are at get-
ting re-elected, the lower the educational score. Moreover, the mayor’s
Table 3.3 Quality of education: Ratio of SABER 11 of public to private schools
Quality-ratio of SABER 11 of public to private schools

OLS I2SLS IV OLS I2SLS IV

Variables Annual Last six years average

Per-capita Taxes −0.0001 0.0120** 0.000675* 0.0192**

(ln) (0.000329) (0.0008) (0.000393) (0.00771)
Per-capita 0.00242*** −0.0236*** 0.00271*** 0.00257*** −0.0227*** 0.00300***
Transfers (ln) (0.000188) (0.0030) (0.000231) (0.000205) (0.00276) (0.000274)
Royalties −0.000149*** 0.00006 −0.000151*** −0.000270*** 0.00160*** −0.000302***
(0.0000131) (0.0005) (0.0000131) (0.0000137) (−0.00051) (0.000041)
Political
Effective Number −0.000500* 0.03045*** −0.000847*** −0.00121*** −0.03633* −0.000553
of Parties (0.000275) (0.0044) (0.000320) (0.000422) (0.0057) (0.000513)
Squared Effective −0.000701*** −0.030*** −0.000346 −0.000034 0.00294*** −0.000085*
Number of (0.0000231) (0.0037) (0.000284) (0.000042) (0.00056) (0.000048)
Parties
Intraparty 0.000108*** 0.00047 0.0001*** 0.000093*** −0.0018*** 0.000128***
Competition (0.00003) (0.00047) (0.00003) (0.00003) (0.00041) (0.000033)
Proportion of −0.00657*** −0.0247*** −0.00614*** −0.00710*** −0.0572*** −0.00605***
Council Members (0.000707) (0.0032) (0.000745) (0.000790) (0.01067) (0.000924)
from Mayor’s
Party
Council Members −0.000682*** 0.0247*** −0.000948*** −0.000501*** 0.0148** −0.000759***
Reelection (0.000201) (0.0032) (0.0000924) (0.000132) (0.00178) (0.000173)
Average
Mayor from 0.00148*** 0.0154** 0.00128*** 0.00302*** 0.0212*** 0.00260***
DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA...

Conservative (0.000415) (0.0066554) (0.000432) (0.000468) (0.00178) (0.000512)

Party
89

(continued)
Table 3.3 (continued)
90

Quality-ratio of SABER 11 of public to private schools

OLS I2SLS IV OLS I2SLS IV

Variables Annual Last six years average

Mayor form 0.00333*** 0.0277*** 0.003*** 0.00356*** 0.0361*** 0.00289***

Liberal Party (0.000353) (0.00567) (0.000391) (0.000388) (0.0052) (0.000488)
Constant 1.023*** 1.010***
(0.0196) (0.0199)
Instrument
F.S. TORRES AND M. PACHÓN

Cadastral −0.0965*** −0.069***

Undervaluation (0.0082) (0.007)
Municipal Fixed
Effects
School fixed Yes Yes Yes Yes Yes Yes
effects
Year Fixed Effects Yes Yes Yes Yes Yes Yes
F-test for 136.86 97.74
instruments
Prob > F 0.000 0.000
Endogeneity Test 136.96 97.8
Chi-sq(1) p-value 0.000 0.000
Observations 41,316 41,207 41,207 40,827 40,715 40,715

Source: See appendix

*,**,***: Coefficients significant at the 10%, 5%, and 1% levels, respectively. Standard errors in brackets
Per-Capita GDL (ln), Poverty Rate, Gini of Land Value, and Urban Population were included as controls
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 91

membership to the Liberal or Conservative traditional party is positively

correlated with higher per capita taxes as well as higher relative scores.
Most of the political variables, however, are not statistically significant.
Column 3 presents the second stage of the model. It can be seen that
the exogenous variations of the local per capita taxes affect the relative
score of the public schools positively and significantly. In fact, a change
in one standard deviation of per capita taxes prompts a fairly substantial
increase of 0.5 (0.0116 × 1.27/0.04) standard deviations in the relative
test score of the public schools. In this regard, the local fiscal effort would
be manifest in better quality of education. Columns 4 to 6 display similar
results but using the variables on the right averaged over the previous six
years under the assumption that the quality of education takes several years
to improve (or worsen). The results obtained are quite the same.

5.3  The Positive Effect of Local Tax Capacity on Provision

of Water Service
The results of the OLS estimation presented in column 1 of Table 3.4
show that neither the local taxes effort nor royalties have any significant
correlation with water coverage, whereas central government water trans-
fers do. In fact, according to the OLS model an increase of 1 percent in
the central government water transfers during the period 1994–2005 saw
an increase in the proportion of municipal water coverage by 0.054 points.
The measure of the number of years following the reform or the existence
of an EICEs in the municipality is also not statistically significant in the
coverage equation. The existence of an ESP, on the contrary, is associated
with greater coverage.
The IV estimations are in columns 2 and 3 of Table 3.4. The first stage
of the model indicates that the average undervaluation of properties, as
expected, negatively affects the average per capita taxes. The poverty rate,
also as expected, is negatively correlated with tax capacity, while the GINI
coefficients for land ownership and population exhibit a positive correlation.
The second stage in column 3 shows that an exogenous variation of the
fiscal effort positively and significantly impacts water coverage. Thus, if per
capita taxes increase by 1 percent (over 1994–2005), then the proportion
of water coverage increases by a sizable 0.15 points. In the IV model, the
water transfers did not turn statistically significant, as the existence of the
public or private character of the providers did not explain the increase in
water coverage. In conclusion, the differences in the fiscal effort are quite
92 F.S. TORRES AND M. PACHÓN

Table 3.4 Water coverage for 2005 in the Colombian municipalities

Variables The table has a weird format for the presentation of the
results

OLS I2SLS IV

1994–2005 average

Decentralization
Per-capita taxes (ln) −0.0125 0.160*** (0.0520)
(0.00800)
Per-capita transfers (ln) 0.0548*** 0.733*** −0.0720 (0.0462)
(0.0240) (0.0915)
Royalties 0.000422 0.0385*** −0.00683**
(0.00213) (0.00826) (0.00327)
Water institutions
Number of years after the 0.00532 −0.0057 0.00675 (0.00417)
reform (0.00350) (0.0136)
EICEs municipality 0.0381 0.145 (0.0970) 0.00399 (0.0306)
(0.0250)
ESP municipality 0.0576*** 0.121 (0.0839) 0.0361 (0.0264)
(0.0217)
Political
Effective number of parties −0.0208 0.089 (0.197) −0.0235 (0.0604)
(0.0505)
Squared effective number of 0.00773 −0.0225 0.00983 (0.0107)
parties (0.00898) (0.0349)
Council members re-election −0.00144 −0.00157 −0.00131 (0.00235)
average (0.00199) (0.00767)
Intra-party competition 0.00247 −0.00663 0.00293 (0.00221)
(0.00183) (0.00724)
Proportion of council −0.103** 0.076 (0.206) −0.0759 (0.0627)
members from mayor’s party (0.0523)
Mayor from Conservative 0.023 (0.0328) −0.268** 0.0468 (0.0409)
party (0.129)
Mayor form Liberal party 0.0773** 0.085 (0.132) 0.0374 (0.0410)
(0.0334)
Constant 0.540*** −9.163*** 2.068*** (0.536)
(0.238) (0.891)
Instrument
Cadastral undervaluation −0.775***
(0.127)
F-test for instruments 37.21
Prob > F 0.000

(continued)
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 93

Table 3.4 (continued)

Variables The table has a weird format for the presentation of the
results

OLS I2SLS IV

1994–2005 average

Endogeneity test 17.929

Chi-sq(1) p-value 0.000
Observations 949 904 912

Source: See appendix

*,**,***: Coefficients significant at the 10%, 5%, and 1% levels, respectively. Panel regressions with robust
standard errors. Standard errors in brackets
Socio-economic variables such as Poverty Rate, Gini of Land Value, and Population(ln) were included as controls

relevant when it comes to explaining the disparities in water coverage across

Colombian municipalities. Again, as seen in the previous models for educa-
tion, the impact of local politics cannot be observed with the variables that we
measured. Although we know from qualitative accounts and more detailed
fieldwork that local politics does matter, the variables—as presented here—
do not capture the impact directly. Indirectly, however, one could interpret
the efficiency differences in the impact between national and local resources
available to municipal administrations as evidence of there being a virtuous
cycle between fiscal autonomy and the outcome of public service delivery.
The variables for quality water are also explained by municipal fiscal
effort as shown in Table 3.5. The dependent variable used is the com-
pliance of local drinking water with the potable parameters during the
period 2006–2009. In this case, the dependent variable is Wi,t where i is
the municipality and t the year.
Column 1 of Table 3.5 presents the results of the OLS fixed effect panel
model, showing that per capita taxes are not statistically related to the
quality of water, while per capita transfers in fact are. Column 2 displays
the first stage of the regression, revealing—as expected—that the under-
valuation of the properties negatively impacts taxes. The second stage, in
column 3, shows that an exogenous variation of per capita taxes leads to
improved water quality. Both per capita transfer and per capita royalties are
not significant in explaining water quality. The model also gave estimations
using random effects (last three columns), and although the coefficient for
Table 3.5 Determinants of quality of water, 2006–2009
94

Variables Local drinking water compliance

OLS I2SLS IV OLS I2SLS IV

Fixed effects Random effects

Decentralization
Per-capita taxes (ln) 2.193 (1.913) 98.97** 3.577*** 18.09*** (4.698)
(47.08) (0.744)
Per-capita transfers (ln) 5.287*** 0.0311** 2.616 1.951* 0.235*** −2.092 (1.663)
(1.528) (0.0156) (2.582) (1.139) (0.0268)
F.S. TORRES AND M. PACHÓN

Royalties 0.169 (0.184) −0.00307 0.31 0.179 (0.109) 0.0226*** −0.152 (0.154)
(0.00199) (0.283) (0.00251)
Water institutions
Number of years after 0.539 (1.119) 0.0131 −0.397 0.0349 −0.00474 0.000841 (0.226)
the reform (0.0118) (1.655) (0.240) (0.00523)
EICEs municipality 3.002 (2.355) 0.175*** 0.839 (2.378)
(0.0511)
ESP municipality 0.976 (2.111) 0.0104 0.938 (1.982)
(0.0446)
Political
Effective number of −14.16 (11.24) 0.171 (0.123) −31.72 −4.875 −0.701* 7.31 (9.476)
parties (18.26) (8.738) (0.213)
Squared effective 15.83 (11.24) −0.155 (0.123) 31.53* 5.578 (8.723) 0.748*** −7.321 (9.524)
number of parties (17.98) (0.212)
Council members −0.955 (1.394) 0.0222 −3.242 −0.594 0.113*** −2.145** (0.892)
re-election average (0.0151) (2.33) (0.772) (0.0178)
Intraparty competition 2.130*** 0.0112 0.235 0.129 (0.444) 0.0104 −0.357 (0.464)
(0.676) (0.00722) (1.280) (0.0106)
Variables Local drinking water compliance

OLS I2SLS IV OLS I2SLS IV

Fixed effects Random effects

Proportion of council 13.20*** 0.00118 10.91* 10.31*** −0.0627 10.31*** (3.039)

members from mayor’s (4.018) (0.0432) (6.079) (2.907) (0.0693)
party
Mayor from −6.407*** −0.0245 −3.652 −3.083** −0.132*** −0.926 (1.531)
Conservative party (1.967) (0.0208) (3.242) (1.361) (0.0324)
Mayor from Liberal −2.235 (1.944) 0.00847 −3.269 −2.524* 0.121** −4.467***
party (0.0209) (2.914) (1.381) (0.0333) (1.527)
Constant 121.4 (109.1) 94.45*** −4.633*** 171.1*** (26.33)
(12.91) (0.285)
Instrument
Cadastral −0.086*** −0.385***
undervaluation (0.030) (0.0422)
Municipal fixed effects Yes Yes Yes No No No
F-test for instruments 8.37 83.22
Prob > F 0.0039 0.000
Endogeneity test 9.046 11.245
Number of 934 861 861 933 920 920
municipalities
Chi-sq(1) p-value 0.0037 0.000
Observations 3417 3209 3209 3414 3222 3222

Source: See appendix

*,**,***: Coefficients significant at the 10%, 5%, and 1% levels, respectively. Panel regressions with robust standard errors. Standard errors in brackets
DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA...

Socio-economic variables such as Poverty Rate, Gini of Land Value, and Population(ln) were included as controls
95
96 F.S. TORRES AND M. PACHÓN

per capita taxes is lower than with fixed effects it is greatly significant. As
in the IV fixed effects model, the coefficients for per capita transfers and
royalties are not statistically different from zero. While the political vari-
ables have no clear impact in the fixed effects model, some of them turn
significant in the random effects model. The Squared Effective number
of parties has a negative impact suggesting that a limited or fragmented
local party system could negatively affect the provision of quality water.
Thus, we find convincing evidence that greater local fiscal effort is linked
to greater coverage and better quality of water. As with the education sec-
tor, both national transfers and royalties seem to have a limited impact on
the provision of the goods, for different reasons. National transfers, on the
one hand, may be too strict and rigid, while royalties may be too flexible.

6   Discussion and Conclusions

This chapter is an attempt to study the interaction between a range of
governance structures and variables with policy outcomes in two sectors:
education and water. We were particularly interested in the effect of politi-
cal competition on the provision of public goods in a decentralized con-
text. Indeed, during the last two decades, Colombia moved from being an
extremely centralized political and fiscal system to a highly decentralized
environment where local actors have the ultimate responsibility of exe-
cuting policy. From the theory behind the decentralization reforms, one
should have expected that the closeness between the elected officials and
citizens would have brought virtuous cycles of representation in which
public goods provision could have flourished in a very natural way around
the territory. However, results have not been uniform around the country,
and even with abundant resources, some municipalities have been unable
to achieve decent socio-economic indicators.
What is the underlying reason for these differences across the Colombian
municipalities? Our argument has two parts. The first part suggests that the
local authorities are faced with the dilemma of how to get things done.
They can either rely on their regional and national networks to search
for resources in exchange for political support, or they can build their
own political support by raising taxes, thereby increasing their political
autonomy and capacity to deliver. National politicians too are faced with
a dilemma. When they get their votes across different municipalities and
on more diverse issues, they can compete over votes in two forms: one via
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 97

providing rents, and the other via national policy programs. Whenever they
control a territory big enough to get elected, they will choose to provide
rents. This in turn is a tempting option for the local politician, who knows
he or she will assume a cost if he or she decides to raise local taxes. Thus,
in the absence of competition, the traditional political networks will keep
transferring the resources to the municipalities, and the mayor will con-
tinue delivering to his political network his political support and that of his
voters. If there is competition at the national level, and there is no obvious
relation to the politicians at the municipal level (House members), the
mayor would feel more pressured to increase his own resources in order
to get things done and fulfill his prospective ambition. Thus, the greater
the diversity and political competition at the national level, the greater are
the incentives for the local political actors to increase their fiscal capacity.
Equally, this virtuous use of the resources that are directly assumed by the
local population may result in greater accountability, as the population is
empowered to demand better public provision of these goods.
In line with this first part of the argument, we find that national political
competition at the local level matters insofar as it provides mayors with the
right incentives to build their own fiscal capacity. Consequently, it is not
the local political context that matters; what matters is the type of political
networks within which local authorities perform to increase their capacity.
We operationalize this capacity and the fiscal effort, and show that greater
competition for national office at the local level, both among parties and
candidates at the local level, both for parties and for candidates, matters
when it comes to mayors’ decision to update their local cadaster—the
single most important decision in terms of increasing their own resources.
We also find that the resources resulting from taxes are more gainfully
employed toward better provision of services as well as improved quality of
water and education in Colombia, compared to resources provided by the
national government in the form of transfers and royalties with specified uses.
Although we could not measure the impact of local politics in the provi-
sion of these services through the variables included in the model mea-
suring the politics at the local level, the differences across these diverse
funding sources in terms of efficiency suggest that there is more virtuous
use of the resources that are directly assumed by the local population. This
result holds true for both the sectors, which have very different structures.
For royalties, results even show a negative correlation with the quality of
education.
98 F.S. TORRES AND M. PACHÓN

Thus, further research is required to understand how the government

can prevent the municipalities from falling on a more client-based path,
and instead opt for one on which they can build their own capacity to
respond to the needs of the population. It is also important to continue
searching for the links between national and local political competitions,
as from our results it would seem clear that holding elections at the local
level does not automatically lead to better policy outcomes.

7   Appendix 1: Data sources

Variable Source

2005 Water coverage 2005 Colombian Population Census, National Statistics

Administrative Department (DANE)
Quality-local drinking water Territorial Development Office, National Planning
compliance Department (DNP)
Public school enrollment Planning Office of the Education Ministry and DANE
Quality—ratio of SABER 11 Scores from the SABER 11 test, Colombian Institute for
of public to private schools Education Evaluation (ICFES)
Per-capita taxes Municipal Budgetary Executions, DNP
Per-capita transfers Municipal Budgetary Executions, DNP
Royalties Municipal Budgetary Executions, DNP
Number of years after the Public Services Superintendence
reform
Poverty rate For 1993 and 2005 the data were taken from the 1993
and 2005 Population Census done by the DANE. For the
other years the poverty rate was calculated by CEDE
Gini of land value The data were taken from the Geographical Institute
Agustín Codazzi and calculated by CEDE
Population 2005 Colombian Population Census, DANE
Effective Number of Parties Electoral data for each of the years since 1994 from
Registraduría Nacional del Estado Civil
Council members re-election Electoral data for each of the years since 1994 from
average Registraduría Nacional del Estado Civil
Intra-party competition Electoral data for each of the years since 1994 from
Registraduría Nacional del Estado Civil
Proportion of council members Electoral data for each of the years since 1994 from
from mayor’s party Registraduría Nacional del Estado Civil
Cadastral undervaluation The data were taken from the Geographical Institute
Agustín Codazzi and calculated by CEDE
Teacher–student ratio (lagged) Planning Office of the Education Ministry
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 99

Variable Source

6-year average teacher–student Planning Office of the Education Ministry

ratio (lagged)
No official students– Planning Office of the Education Ministry
population ratio (lagged)
Per-capita transfers (ln) Territorial Development Office, National Planning
Department (DNP)
Royalties (ln) Territorial Development Office, National Planning
Department (DNP)
Number of years after the Public Services Superintendence
reform
Students–population ratio Planning Office of the Education Ministry
Public test score Colombian Institute for Education Evaluation (ICFES)

Notes
1. The periods of mayors and governors changed to three-year term after
1994, and to four-year terms in 2003.
2. Before the reform, royalties would automatically be transferred to munici-
palities in which the extraction of resources was done. In June 2011 the
Royalties Reform was approved in Congress, transforming the way in
which they were regionally allocated. Instead of just transferring those to
municipalities, sub-national entities need to compete for the resources by
presenting projects which, in turn, need to be approved by entities with
representation from the national, departmental, and local tiers of
government.
3. In order to increase the efficiency of resource transfers, a reform of the
system was undertaken in 2000. In the first place, it created a revenue shar-
ing system (Sistema General de Participaciones, or SGP) that fixed the
amount of resources to be transferred and established a 2 percent annual
growth in real terms. In addition, the formula to allocate the resources
across sub-national entities based on Law 60 of 1993 was also changed by
the Laws 715 of 2001 and 1176 of 2007.
4. According to Rodden (2002), large and persistent deficits occur when the
sub-national governments depend strongly on inter-governmental trans-
fers, and have, at the same time, free access to credit, generating fiscal
indiscipline.
5. As a matter of fact, the territorial debt rose from 1.1 to 3.5 percent of the
GDP between 1990 and 1999 (Ministerio de Hacienda 2014). In an effort
to better align incentives of politicians to improve their policy perfor-
mance, the Law 715 determined that the transfer’s distribution would be
100 F.S. TORRES AND M. PACHÓN

based on the coverage and growth of the health and education services
provided by the territorial entities, and not by population and poverty
rates—criteria that had been established by Law 60 of 1993. Concerning
the generation of own resources, Congress approved Law 488 of 1998
whereby the base for some sub-national taxes such as the register tax was
increased, and Law 1111 of 2006, which raised cigarette taxes. In order to
augment health revenues, Decree 127 of 2010 increased the rates of the
departmental cigarette and liquors tax, as well as VAT for beer and gam-
bling. Furthermore, the gasoline surcharge was unified, and a surcharge on
petro-diesel was levied.
6. According to Rodriguez (2010), the lack of adequate data may explain
why the results have suggested both positive and negative impacts of
decentralization. Using panel data techniques, she evaluates the effects of
decentralization on the quality of education in public schools, concluding
that reforms increase the gap in the results of standardized test preparation
applied in public and private schools. She argues that the results are driven
neither by the lack of transfers from central government nor by the lack of
investment of resources in the sector. They are mainly driven by the
increased enrollment of poor students in public schools. When this factor
is controlled for, it is found that the public education system serves a larger
number of students, besides offering better quality education.
7. Law 142 of 1994, Article 15. Other changes introduced by Law 142 were
(a) the definition of a pricing regime based on the cost of providing the
service, (b) the creation of a control mechanisms of the provider perfor-
mance enforced by citizens, (c) establishment of a regime of free enterprise
which constitutes the base for the entry of the private sector as a provider,
and (d) for the implementation of management control and internal con-
trol systems within the provider companies. See Krause (2007) for a com-
plete explanation of the differences between EICEs and ESP.
8. Also, the Colombian system of user’s fees follows a cross-subsidization
approach in which residential users from low socio-economic strata (1, 2,
and 3) receive discounts in their fees, which are covered by fees charged
from the high socio-economic strata (5 and 6) as well as by the commercial
and industrial users. Due to the deficit nature of the scheme in most
municipalities, central government transfers partially finance these subsi-
dies. The remaining resources cover a fraction of the investments needed
to provide the services, either through direct subsidies to the provider—
that could be municipality itself—or through the delivery of physical infra-
structure (Silva 2007).
9. “En ese entonces, casi que todos los cargos adscritos a la prestación de
estos servicios eran de cuotas políticas. Entonces, que las escobitas: vaya
usted, vaya usted; que la persona que se encargaba de la facturación.
DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 101

Personas que no tenían ni idea que tenían que hacer en un cargo de esos y
era así como se colocaban las posiciones de los cargos en una empresa de
pequeña estructura que tenía el municipio. No había políticas claras, no
había directrices, el servicio era prácticamente regalado: la gente pagaba 20
pesos por el agua.” Manager, Water Company of el Peñol, Antioquia.
10. Referring to the municipalities in which the service is provided by private
parties, the manager of the Water Company of El Peñol argued; “Los alcal-
des generan compromisos con ese tipo de gente y vienen, ese modelo es
perverso, muchas veces esa gente viene y esa gente de lo que tratan es de
escurrir al usuario. Finalmente vienen, lo clavan con unas tarifas exorbitan-
tes, le sacan el jugo al negocio, no hacen ninguna inversión y salen y se van
con la plata. Entonces yo pienso que una entidad pública, siempre y
cuando, se logre manejar con cierta autonomía administrativa, financiera,
con cierta independencia del tema político, si se maneja así se puede mane-
jar con unos criterios de rentabilidad más que, o sea una rentabilidad
económica que le permita ser auto sostenible en el tiempo.”
11. This is the term used by politicians to refer to the effort to get money from
the national government.
12. “Básicamente una buena administración en este municipio se hace con
gestión. Por eso les decía yo ahora que aspiramos a ser una de las mejores
administraciones que haya tenido el municipio de Santa Barbara por la
coyuntura que tenemos a León Darío en la Cámara. Porque aparte de
todo, él pertenece a la comisión tercera, que como ustedes bien saben es la
de presupuesto y eso le da ciertas ventajas frente a algunos representantes
porque ser ponentes y participar en las ponencias del presupuesto, les dan
algunos incentivos.”
13. Several specifications were estimated for the years to the last cadastral
update: lineal, quadratic, cubic, and diverse forms of dummy variables for
the number of cadastral updates.
14. The measure of the effective number of parties is the inverse of the
Herfindal Index to measure the competition among party shares of votes.
The unit of analysis is the political party.
15. The GINI is a measure of statistical dispersion and measures the concentra-
tion across political parties. Thus, if one party controls all the votes for the
House, the “political concentration” is highest (GINI of 1). If more par-
ties do, then the “political concentration” goes down.
16. This is the same measure as the effective number of parties, but instead of
taking the party as a unit of analysis each candidate is counted
independently.
102 F.S. TORRES AND M. PACHÓN

References
Andrews, M., and A. Shah. 2003. Towards Citizen Centered Local-level Budgets
in Developing Countries. In Handbook on Public Sector Performance Reviews,
vol 3, ed. Shah Anwar. Washington, DC: The World Bank.
Bardhan, P., and D. Mookherjee (ed). 2006. Decentralization and Local
Governance in Developing Countries: A Comparative Perspective. Cambridge,
MA: MIT Press.
Barrera, F., and M. Oliveira. 2007. Does Society Win or Lose as a Result of
Privatization? The Case of Water Sector Privatization in Colombia. RES
Working Papers 3230, Inter-American Development Bank, Research
Department.
Besley, T., and T. Persson. 2010. State Capacity, Conflict, and Development.
Econometrica, Econometric Society 78(1): 1–34.
Besley, T., T. Persson, and D.M. Sturm. 2010. Political Competition, Policy and
Growth: Theory and Evidence from the US. The Review of Economic Studies
77(4): 1329–1352.
Cárdenas, M. 2010. State Capacity in Latin America. Economía—Journal of the
Latin American and Caribbean Economic Association 10(2): 1–46.
Careaga, M., and B. Weingast. 2003. Fiscal Federalism, Good Governance, and
Economic Growth in Mexico. In Search of Prosperity: Analytic Narratives on
Economic Growth, ed. Dani Rodrik. Princeton, NJ: Princeton University Press.
Cox, G. 1987. The Efficient Secret: The Cabinet and the Development of
Political Parties in Victorian England. Cambridge, MA: Cambridge
University Press.
Domínguez, C. and E. Uribe. 2005. Evolución del servicio de acueducto y alcan-
tarillado durante la última década. Documentos CEDE 2005-19, Universidad de
los Andes-CEDE, Bogotá.
Economía Urbana—Centro Nacional de Consultoría. 2012. Una Evaluación del
Impacto de las Regalías en 80 Municipalidades. Informe de Consultoría,
Departamento Nacional de Planeación, Bogotá, Colombia.
Faguet, J. 2004. Does Decentralization Increase Government Responsiveness to
Local Needs? Evidence from Bolivia. Journal of Public Economics 88(3–4):
867–893.
Faguet, J., and F. Sánchez. 2008. Decentralization’s Effects on Educational
Outcomes in Bolivia and Colombia. World Development 36(7): 1294–1316.
———. 2009. Decentralization and Access to Social Services in Colombia.
Documentos CEDE 005401, Universidad de los Andes-CEDE, Bogotá.
Falleti, T. 2010. Decentralization and Sub National Politics in Latin America.
New York: Cambridge University Press.
Gadenne, L. 2011. Tax Me, But Spend Wisely: The Political Economy of Taxes,
Theory and Evidence from Brazilian Local Governments. Mimeo. Paris School
of Economics.
 DECENTRALIZATION, FISCAL EFFORT, AND SOCIAL PROGRESS IN COLOMBIA... 103

Gaviria, A., J. G. Zapata, and A. González. 2002. Petróleo y Región: El caso del
Casanare. Cuadernos de Fedesarrollo, No. 8, Bogotá, Colombia.
Gomez-Lobo, A. and M. Meléndez. 2007. Social Policy, Regulation and Private
Sector Water Supply: the Case of Colombia. Working Papers wp252, University
of Chile, Department of Economics, Santiago de Chile.
Granados, C. 2008. Contribuyen las reformas al progreso social?, el caso de los
servicios de acueducto y alcantarillado y la mortalidad infantil en Colombia
1990–2004. Documentos CEDE 2008-07, Universidad de los Andes-CEDE,
Bogotá.
ICFES. 2009. Informe de resultados nacionales, Saber 3, 5, y 9. http://www.
icfes.gov.co/investigacion/informes-de-resultados-de-evaluaciones-­­
nacionales/saber-3o-5o-y-9o
Junguito, R., and H. Rincón. 2004. La política fiscal en el siglo XX en Colombia.
Borradores de Economía, 318, Banco de la República, Bogotá.
Krause, M. 2007. The Political Economy of Water and Sanitation in Developing
Countries: Cross-country Evidence and a Case Study on Colombia. Dissertation
for the degree of Doctor of Economics, Department of Economics, Justus-­
Liebig-­Universität Gießen.
Melo, B. 2005. Impacto de la Descentralización Fiscal sobre la Educación Pública
Colombiana. Borradores de Economía 002802.
Ministerio de Hacienda. 2014. Responsabilidad fiscal subnacional descentralización
en Colombia: Quince años de consolidación. http://www.minhacienda.gov.
co/portal/page/portal/HomeMinhacienda/asistenciaentidadesterritoriales/
Publicaciones/Libros/15%20anios.pdf
Perry, G., and M. Olivera. 2009. El impacto del petróleo y la minería en el desar-
rollo regional y local en Colombia. Documentos de Trabajo No. 2009/06, CAF-­
Caracas, Venezuela.
Prasad, N. 2006. Privatisation Results: Private Sector Participation in Water
Services After 15 Years. Development Policy Review 24(6): 669–692.
Rodden, J. 2002. The Dilemma of Fiscal Federalism: Grants and Fiscal Performance
Around the World. American Journal of Political Science 46(3): 670–687.
Rodriguez, C. 2010. Correspondencia de economía política y decisiones de edu-
cación en las familias después de la descentralización. El Trimestre Económico
LXXVII (4), no. 308, pp. 977–1006, October–December.
Sánchez, F., and J. Zenteno. 2011. Descentralización y sostenibilidad fiscal subna-
cional: el caso de Colombia. Documentos CEDE 2011-37, Universidad de los
Andes-CEDE, Bogotá.
Sánchez, F., I. España, and J. Zenteno. 2012. Sub-National Revenue in Latin
America and the Caribbean Countries: The Case of Colombia. Mimeo, BID-
Red de Centros de Investigación, Washington.
Silva, J. 2007. Reformas Estructurales en el Sector de Agua Potable y Saneamiento
Básico en Colombia, 1990–2006, Mimeo. Washington, DC: World Bank.
104 F.S. TORRES AND M. PACHÓN

Spiller, P.T., and Savedoff, W.D., 2000. Lost Water: Institutional Compromises for
the Supply of Public Health Services. Agua perdida: compromisos institucionales
para el suministro de servicios publicos sanitarios. Inter-American Development
Bank. Mimeo.
Weingast, B. 2009. Second Generation Fiscal Federalism: The Implications of
Fiscal Incentives. Journal of Urban Economics 65(3): 279–293.