Working Paper No. 153

Mexico’s Decentralization at a Cross-Roads

by

Alberto Diaz-Cayeros*

José Antonio González**

Fernando Rojas***

* Assistant Professor, Department of Political Science, University of California, Los Angeles** Senior Research Associate and Latin American Research Coordinator, Center for Research on EconomicDevelopment and Policy Reform, Stanford University*** Lead Public Sector Management Specialist, World Bank

August 2002

Stanford University John A. and Cynthia Fry Gunn Building

366 Galvez Street | Stanford, CA | 94305-6015

Abstract+

This paper focuses on two inter-connected developments that account for the main

challenges of Mexico’s decentralization today. First, we argue that the basic political incentives were misaligned with the decentralizing principles for at least the first two phases of the process of decentralization in the early eighties to the mid nineties. Second, we point out the impact Mexico’s long history of single-party rule has had on the development of federalism. We argue that democratic institutions and processes in Mexico have become important stumbling blocks, rather than promoters of an efficient and equitable decentralization process.

The paper illustrates these arguments with an analysis of four key moments of change in the history of federalism in Mexico: the 1980 Pacto Fiscal, the 1992 National Agreement to Modernize Basic Education, the establishment of a targeted poverty reduction program PROGRESA, and the current Fox administration’s efforts to increase local governments’ accountability for results. These key reforms created a confusing picture of responsibilities with no clear responsibilities at any level of government for key public services. In addition, the system of transfers that has evolved through these reforms is not providing the best incentives for efficient tax collection, expenditures, or service provisions. Our main contribution to the large existing literature on federalism in Mexico is to highlight the importance of considering Mexico’s political system in assessing the prospects for reforming fiscal federalism in that country. While much of the past literature has been focused on the debate over whether more or less control over resources should be decentralized to the states, we argue that the primary focus should be on the design of transfer systems to improve the use of whatever level of resources are decentralized. With these lessons in mind, the final section of the paper discusses in detail ongoing marginal reforms and how they can be strengthened.

+ *This is a paper prepared for the conference on fiscal federalism, Center for Research on Economic Development and Policy Reform, Stanford University.

2

CONTENTS

1. Introduction

2. Political Institutions as Makers or Breakers of Federalism in Mexico

3. The first phase of Mexico’s recent decentralization and federalization: A

Transfer Lead Decentralization

4. The second phase of Mexico’s recent decentralization and federalization

policies: Deconcentration, partial decentralization and confusing

responsibilities

5. The third phase of Mexico’s recent decentralization and federalization

policies: The politics of transformation of social investment funds

6. The fourth phase:

3

1. Introduction

The World Bank recently published a book on decentralization in Mexico that

starts with the following premise: “Democratization, decentralization, development.

These three sequential forces have swept the world over the last decade and have redrawn

the maps of politics, power, and prosperity.” (World Bank 2000). The assumption has

been that democratization leads to decentralization, and the later to development. In all

fairness the publication and the Bank have been careful to point out that not all

decentralization is good for development.1 Since democratization is taken as a normative,

positive development in the world, little examination is made about the specific

institutional characteristics of democracies that have pushed for greater decentralization.

Likewise, the possible reverse causality (how and when decentralization strengthens

democracy) is not generally explored. Furthermore, the connection between

decentralization and expenditure efficiency or investment is taken for granted.

In practice, not all democracies have equally auspicious forces driving a

decentralization process that can improve accountability, equity and government

performance. From the citizen’s point of view, the specific benefits that decentralization

can bring about depend on the organization of the system of representation and the

institutional characteristics of government policy formation and implementation.

Moreover, decentralization appears to be path-dependent, i.e., the costs and benefits of

decentralization depend largely on the selection of entry points and intermediate goals for

the phased-in transition from the centralized state to a decentralized public sector. Not all

processes of decentralization lead to development or strengthened democracy. While

there is no standard path towards decentralization, it appears that countries often select

sub-optimal transition paths –or paths that deviate from democracy, development and

decentralization altogether. Similar to tax reform –but even more so in the case of

decentralization—countries rarely plan their path towards a final version of the

decentralized state. More frequently, decentralization is the result of internally-

1 See ABCDLAC, Remy Prud’homme "On the Dangers of Decentralization" enWorld BankResearch Observer, 10(2), pp. 201-220, 1993; Jonathan Rodden (1999) "Soft Budget Constraintsand Deficits in Decentralized Fiscal Systems", American Political Science Association Meeting,Atlanta., Ter-Minassian;

4

inconsistent accumulated reforms, adopted over time in the name of democracy or

development, in response to balances of political forces that change over time.

The literature on decentralization in Mexico is extensive and it covers a wide

range of topics –usually from a normative perspective. The World Bank publication is an

excellent attempt to take a normative picture of fiscal decentralization at a point in time.2

The publication makes a diagnosis of the situation in 1999 and proposes reforms --where

Mexico should be if the country wants to enhance equity, efficiency or fiscal

sustainability—that the country has found difficult to either pass or implement during the

last three years. However, the book rarely discusses the interaction between politics and

decentralization or the phases of decentralization Mexico has gone through. There is no

attempt to identify the underlying constraints that determine what can be effectively done

under Mexico’s current political context or to propose strategic paths for moving from A

(the current situation) to B (the proposed intermediate goal in the process of

decentralization). We believe the discussion of current obstacles and resistances to

reform, or the proposal of strategic path scenarios that reduce the costs or increase the

benefits of the transition towards decentralization in the short term, is precisely what

advice-seeking countries need the most. Understanding the deeper political and

institutional constraints that effectively shape decentralization --and limit their

development impact— is essential for recommending (path-dependent) decentralization

in a given country.

Rather than an attempt to look at the subject of fiscal federalism broadly or

comprehensively, the hypothesis of this paper is centered on two inter-connected

developments that account for main challenges of Mexico’s decentralization today: i) the

basic political incentives were misaligned with the decentralizing principles for at least

the first two phases of the process of decentralization –the period of time that goes

roughly from the early eighties to the mid nineties. Although the mood of the country

during the last five or so years is gradually leading to a re-alignment of politics with the

most fundamental principles of decentralization, the cumulative effect of more than

twenty years of reform create a formidable barrier to streamlining the process of

decentralization; and ii) that the particular features of democratic institutions and

2 See also Arellano, Rodriguez, Rodriguez & Ward, Diaz-Cayeros (2001),

5

processes in Mexico have become important stumbling blocks, rather than promoters of

an efficient and equitable decentralization process. We argue that the existing literature

which explores the subtleties of fiscal federalism in Mexico has been thorough, but has

generally missed a critical political economy issue, namely, that Mexico’s political

system does not allow for a comprehensive reform of fiscal federalism at this point.

Therefore, we end up accepting that Mexico’s decentralization reform has to follow an

incremental path –and discuss in detail ongoing marginal reforms and how they can be

strengthened.

We fully share the view that sooner or later Mexico shall take a step back from

fine tuning fiscal federalism. However, different from what most fiscal experts

(including the World Bank book) recommend, we hold that the issue in Mexico is not

whether more or less revenues (or sources of revenues) should be immediately

decentralized to subnational governments to alleviate vertical or regional imbalances.

Although fiscal dependency is unusually high for a federal state, Mexican states and

municipalities currently receive a large amount of fiscal transfers, both conditional and

unconditional. The problem is that the country does not appear to have the necessary

institutional developments for a well-designed transfer system (or redistribution of

revenues sources for that matter) to lead to an efficient and equitable decentralized

allocation of resources. In our view, the central problem of fiscal federalism in Mexico

is that subnational governments do not have clear jurisdictions over virtually any policy

area, the federal government always holds a crucial piece of the process and the key

mechanisms to hold either the federal government or the subnational governments

accountable are very imperfect. If something goes wrong, subnational government

usually blame the federal government, and in many cases the federal government steps in.

The federal bargain in Mexico is set up in such way that there are no credible

punishments and therefore states have no incentives to hold up their end of the bargain.

Responsibilities need to be made clearer so that different government levels can be held

accountable.

Rather than the usual debate on more/less fiscal decentralization, we are

proposing that decentralization and fiscal federalism reform, like all other democratic

political institutions in Mexico, need to be adapted to a new environment of multiparty

6

competition and accountable centers of fiscal power, so that actors have the right

incentives to carry out their jobs in a complex decentralized federal system. If political

institutions remain as they are, greater decentralization will be counterproductive: rather

than producing immediacy or proximity between citizen demands and resource

allocation, it will more likely widen the gap between who collects and pays taxes and

how, and where budgets are spent and who administers these budgets, reducing

accountability and increasing regional tensions.

The country is currently undergoing a slow, uncertain transition in the

decentralization process. The overwhelming majority of the literature agrees that

democratization in Mexico has created a demand for decentralization and that the latter is

leading (or will lead) to more equitable and efficient public spending. By comparing

Mexican federalism with that of other Latin American countries, there is a sense that the

country is set on an inevitable path to greater decentralization. Subnational governments

are spending ever- increasing shares of total public expenditures and appear to be

managing more and more sector responsibilities. However, although the process was set

in motion more than 20 years ago, decentralization in Mexico started with the wrong foot

and the country has not been able to straighten the process ever since. Today’s

decentralization lacks both a cohesive design and a credible implementation strategy.

To understand the current decentralization framework it is necessary to look at the

recent evolution of decentralization in Mexico. A thorough historical recount of the

relationship between states and the federal government is a long one and well beyond the

scope of this paper.3 We propose, instead, to highlight four key moments and policy

reforms placed one on top of each other giving rise to the confusing framework that

exists today.

Since the Mexican Revolution, there had been a steady centralization of taxing

powers due to the lack of tax enforcement by subnational governments. The third

National Tax Convention in 1947 established a relatively centralized system that: (i)

Limited local taxes to property and other minor taxes, eliminating the local taxes on trade

and industry and transferring to the federal government the power to collect foreign trade

3 See Careaga and Weingast (2001) and Courchene, Diaz Cayeros, and Webb 2000 for a good historicalsummary of fiscal relations from the early 1920s to 1980.

7

tariffs, natural resources, and other activities including banks, insurance, electricity,

tobacco, gas, telecommunications etc.4 (ii) Gave states revenue shares from federal

excise taxes from natural resources and from certain other goods. (iii) Introduced a

national sales tax administered by the federal government, where the tax rate would be

shared with the states. States could opt to abandon their own sales taxes and join into the

federal system. (iv) Gave exclusivity to the federal government to tax income. (v) The

contribución federal, a transfer from states to the federal government, was phased out.The system functioned well under the rule of the Partido Revolucionario

Institucional (PRI). Subnational governments increasingly delegated financial powers to

the federal government in exchange for patronage, and political careers in the national

governments. The tax structure suffered only various marginal changes until 1980. The

most important development during the years between the establishment of the Sales Tax

in 1947 and 1980 was the gradual incorporation of states into the federal sales tax, until

all of them had abandoned their sales taxes by 1973.

The first key reform that shaped the modern decentralization framework was the

overhaul of the sales tax and the introduction of the National Value Added Tax (or the

IVA as it is known in Mexico) through the Sistema Nacional de Coordinación Fiscal

(often referred to as Pacto Fiscal) in 1980. The states were asked to joint voluntarily andall of them did, because it was in their best interest. The states gave up their share of

revenues on the sales tax, and they eliminated some remaining state excise taxes in

exchange for a share (initially 13%, gradually increased to 20%) of unconditional

revenue transfers (participaciones) from the VAT, the federal income tax, and some oil

fees in the middle of massive oil boom . The derivation component of revenue from the

sales taxes was changed to a more equalizing formula, which was only marginally tied to

fiscal effort).

The remarkable aspect of this reform is that there was not an accompanying

transfer of responsibilities, partly because in the beginning the idea was that states were

going to receive the same amount of resources previously received. However, the

amount of participaciones has increased drastically, without corresponding revenue

responsibilities or precise expenditure requirements in mind. States began programs that

4 Art. 73 as cited in World Bank 2000.

8

in many cases duplicated federal programs without any clear division of responsibilities.

Section 3 argues the role of unconditional revenue sharing transfers, coupled with othertransfers that came later, made Mexico’s decentralization primarily expenditure based.

The second key reform moment came in 1992 with the National Agreement to

Modernize Basic Education. In the agreement, the federal government attempted to

transfer both the education responsibility and the corresponding share of the federal

expenditure budget to the states. In practice, the 1992 reform of education was more

deconcentration than decentralization –since the states received most of the education

transfers fully earmarked including the detailed payroll. Deconcentraton of education

was precedent-setting for other sectors: a similar model was followed in health; in water,

the federal government delegated the distribution of water to municipalities but it retained

jurisdiction of water as a resource. Section 4 argues that the so-called (sector)

decentralization reforms of the early nineties were, in fact, deconcentrations with unclear

division of responsibilities across levels of government. Weak accountability replaced

largely centralized –yet more clearly accountable—sector expenditures. This reform, on

top of the reform which centralized the collection of the VAT in exchange for

unconditional transfers placed key public sector activities – health education and water –

in jeopardy because of the lack of accountability for the ultimate delivery of the service.

The third key reform moment is the undoing of another key reform that occurred

in the late 1980s. Namely, the transformation of the social investment programs and

funds that were first established as part of the macroeconomic stabilization and structural

reform program of the late 1980s and early 1990s. By the late 1990s, there was an

increasing consensus that there was too much discretionality in --too large a share of-- the

federal spending targeted to subnational governments. In particular, the consensus was

that there was too much discretionality in the expenditures decisions of the poverty relief

program Solidaridad. The Zedillo administration began to transform these funds into

formula based transfers to states and municipalities, while initiating an ambitious

household targeted poverty program, PROGRESA. Section 5 argues that it has been and

will continue to be difficult to completely transform these funds into incentive compatible

formula-based conditional transfers. The reason is that, as it usually happens with fiscal

transfers, the aportaciones quickly developed their own stakeholders that will resist their

9

realignment with a cohesive package of inter-government incentives for expenditure

efficiency and accountability. Section 5 discusses the political mechanisms that exist to

transform the aportaciones towards inter-government, result-oriented agreements, as atransitional phase towards enhanced efficiency and clearer division of responsibilities in

Mexico.

The fourth reform moment is today and the way the Fox administration is trying

to address the issue of decentralization. The diagnostic is clear and has been assumed by

the government. The current decentralization arrangement is a mosaic of different

measures taken on top of each other, creating confusion in almost every aspect of

decentralization. To deal with this problem, the Federal government is mainly trying to

use two instruments: conditional grants and credit lines as incentives for inter-

government agreements (convenios) by which subnational governments effectively

assume result-oriented responsibilities tied to additional resources.

The central hypothesis of this work is that the biggest difficulty to change the

current fiscal arrangement in Mexico is political. Section 2 argues that like other

federations, the political institutions of the Mexican federal arrangement produce a statusquo bias. Multiple veto players must be made better off in order to allow for any changein the current system.5 This status quo bias is a safeguard that seeks to protect the rightsof heterogeneous states. The prospects for reform of the current system depends on the

possibility of constructing a large enough political coalition that may agree on a

redistribution of the power centers of sector decisions and the regional allocation of

federal resources that generate a more equitable and efficient system, while also giving

credit and other political gains to the members of the reform coalition. In the current

political conditions such coalition is highly unlikely to form. In fact, our analysis

suggests that not only is there a status quo bias, but that the few incremental changes thathave been possible point in the direction of decreasing accountability and exacerbating

the problems of the Mexican federal pact.

The paper is organized as follows. Section 2 provides some background on the

political institutional features of Mexico’s federalism that have generated a political

impasse in the budgetary processes, which hinders the reform of the system of

5 On veto players see Tsebelis, 1990.

10

intergovernmental transfers and prevents a significant overhaul of Mexico’s fiscal

federalism. It provides a cooperative game theory analysis of the coalitional dilemmas the

Mexican legislature faces for the future, if it is to reform the federalist arrangement. The

basic conclusion from that analysis is that the only feasible coalitions in the Mexican

legislature, given the institutional structures discussed before, are extremely oversized

ones, likely to make only minor incremental changes. Section 3 summarizes the current

inter-government fiscal system and the subnational borrowing regulatory framework.

Sections 4 through 6 then the 4 key reform moments of the process of decentralization,

respectively emphasizing that a) Mexico’s decentralization began by depriving of vigor

states own revenue sources, thereby widening the gap in the benefit principle between

who taxes and who receives the benefits of public action, and weakening government

responsiveness and accountability; b) the process of deconcentration in education, health

and water provision, highlighting that the reform and benefits of decentralization are yet

to be reaped in those areas; c) the growth of broadly defined earmarked fiscal transfers

during the last few years is a result of the process of discussion and approval of the

federal budget. Since the Partido Revolucionario Institucional (PRI) lost its majority in

the Chamber of Deputies in 1997, approval of the federal budget has been made

contingent on coalition formation around the provision of ever larger transfers to states

and municipalities; and d) current federal attempts to re-establish hard budget constraint,

clear division of responsibilities and accountability via subnational borrowing regulations

and inter-government performance agreements and provides a feasible agenda for short-

term reform that may build the basis for sustainable reform of Mexico’s federalism in the

medium and long term. Section 7 provides some brief conclusions.

11

2. Political Institutions as Makers or Breakers of Federalism in Mexico.Mexico’s federalism has fundamentally been shaped by the political and

institutional environment where path dependent policies have been embedded. This

section attempts to succinctly explore the link between democratization and

decentralization as it is specifically expressed in Mexico. We highlight the features in the

institutional configuration and political landscape that, we believe, limit the possibilities

for further reform of Mexican federalism.

Mexico is organized as a federal republic with 31 states and a Federal District

located in Mexico City. The 1917 Constitution clearly provides for separation of

legislative, judiciary and executive powers under a presidential system of government,

along lines similar to the United States constitutional system on which it was originally

inspired upon. However, the specific political institutions organizing the Mexican

legislatures and the regulation of its federal pact are dramatically different from those

found in the US. The differences are critical for the design of decentralization, because of

the effects they have on political accountability and representation. The first part of this

section describes the political institutions currently operating in Mexico and the way

democratization has affected their role in the decentralization process. Using a game

theoretic framework, the second part of the section argues that given the political

institutions, it is highly unlikely that a major decentralization reform will be approved by

the legislature. Instead, only marginal reforms can be expected in the future, based on

broad state coalitions.

Political Institutions in Mexico and the Annual Budget Approval RitualWithin a bicameral system, the Chamber of Deputies is elected every three years,

with no immediate reelection, through a mixed system of single member districts (SMD)

and proportional representation (PR) –two fifths of the 500 member body are made up by

these compensatory seats in five multi-state districts.6 The same ballot is cast to vote for

both the single member district candidate and the proportional representation party list.

Such an organization of the legislature tends to increase the number of political parties

6 For discussions of the Mexican political arrangement see Diaz-Cayeros and Magaloni 2001; Corneliusand Craig, 1999.

12

represented, given the relatively high mean district magnitude, and mitigate the

particularistic (or personal) incentives of legislatures, in favor of a greater partisan

discipline.

Since there is no immediate reelection, legislators do not have incentives to

provide targeted benefits to their narrow district jurisdictions, but rather to focus on

passing legislation that will benefit the jurisdiction where their future career objectives

are found: usually the state congress, a municipal presidency, the Senate, or the

governorship of their state. Even the proportional representation legislators elected

through party lists, who usually occupy some of the important leadership positions in the

legislature, are often concerned about advancing local careers in their states. These

localist incentives are, however, tempered by the partisan incentives of the proportional

representation lists.

This legislative career behavior is quite different from the one that used to prevail

when the PRI was a hegemonic party, virtually uncontested in its nominations for any

public office. During the heyday of PRI hegemony, the legislature would rubberstamp

any bill submitted by the president, who was also the leader of their party, because the

president controlled all the nominations to future elective posts, or appointments to the

federal bureaucracy. With the onset of democratization, starting in the 1980s, but most

clearly after 1997, legislators from all parties in the lower house are torn between

incentives to remain loyal to partisan mandates, in order to secure a nomination in the

proportional representation lists controlled by the party leaderships, or to follow localist

careers aligned with the governors and the local parties in their states. Haggard, Garman

and Willis (2001) have established that there is a direct link between a push towards

greater decentralization and the local nomination of elective posts in party systems in

Latin America. In their argument decentralization is more prevalent when nominations

are controlled by local party organizations because the fate of politicians depends on

citizen demands at the local level. The Mexican case is ambiguous in this context,

because deputies are not just responsive to partisan or local incentives, but to a mix,

given by the mixed electoral system, the non-reelection clause and the fused ballot for

both proportional representation and single member district races. Moreover, the internal

organization of the lower chamber favors strong party leadership through committee

13

assignments. But perhaps the most important effect of the electoral system in the lower

house of the legislature is that it promotes the fragmentation of the body, with the

representation of at least three effective political parties. The fragmentation has been

observed at least since 1988, although it is not until 1997 that its political implications

have been most obvious: the party of the president does not control the majority in the

chamber of deputies. The political science literature has stressed that this is the Achilles

heel of the presidential system of government as compared to parliamentarism. Divided

governments can make the cooperation between the president and the legislature very

problematic, but more importantly, a combination of presidentialism and multipartyism is

likely to be destabilizing, given the difficulty for reaching compromises and cooperation

between a fragmented legislature and a minoritarian president.7

In contrast to the U.S. Senate, which is quite symmetric in terms of its power vis-

à-vis the lower chamber, and where the representation of states is equal, the Mexican

Senate is extremely weak and the representation of states is diluted by partisan

considerations. The Mexican Senate has significantly less formal powers than the lower

chamber, because it has no authority over voting the federal expenditure budget. Its

members are elected through a mixed system with a relatively high district magnitude

(for an upper chamber in a federal system). Voters cast only one ballot, which elects four

senators for six years, concurrent with the presidential election. A two senator formula is

elected in each state through a plurality rule; a third Senate seat is filled by the first

minority, that is, the first candidate in the formula that was in second place; and the last

senator comes from an at-large national proportional representation party list.

This means that half of the senators are responsive to partisan influences, in terms

of their placement in the proportional representation lists or at the top of the state

formula. In the time of PRI hegemony the Senate was a particularly attractive post, from

which politicians sought nominations to their state governorships (around a third of the

governors had just previously held a Senate seat). However, the greater partisan makeup

of the body, due to the mixed system gradually introduced since 1988, together with the

fact that the Senate has no influence on how the federal budget is allocated to the states,

7 The literature on this debate is enormous, but a good summary can be found in Linz 199x. The linkbetween multipartyism and parliamentarism as a difficult combination can be found in Mainwairing, 199x).

14

means that Senators do not constitute the “natural” defenders of state interests in the

federal pact.

The peculiar combination of local vs. partisan incentives for politicians in the

lower and upper chambers in Mexico suggests that paradoxically, deputies might be

slightly more likely to serve state interests than senators. The entire legislature is plagued

with collective action problems both on its partisan and territorial dimensions: partisan

fractions and state delegations have no incentive within the current system to support

presidential bills, but rather to push for amendments that might increase the share of local

government transfers in the federal budget. This collective action problem emerges from

a critical formal procedural issue reflecting on the power of the president that emerges

from the fact that the Senate cannot vote on the expenditure budget. The Constitution

establishes that presidential veto powers operate only when a bill is voted by both

Chambers. Since the Senate is not involved in the budget bill, a strict interpretation of the

Constitution implies that the president cannot veto the budget approved by the lower

chamber. In practice, the question of whether the president can veto a budget is hotly

debated, since historically the fact is that during the 1920s the president did veto

budgetary bills (see Weldon, 1996). Regardless of how such a constitutional controversy

would be settled by the Supreme Court, if a veto ever comes to be exercised, the

uncertainty surrounding this institutional feature seriously diminishes the bargaining

power of the president vis-à-vis the legislature.

Moreover, the fact that there is no established reversion point that allows for the

president to use some budget (the previous year’s, or the proposed bill) in case the

Chamber of Deputies cannot agree on one by December 31st gives an additional leverage

to deputies who can hold back their support for the budget. This means that in Mexico the

executive is very weak in the approval stage of the budgetary process, even though it is

procedurally strong in the formation stage, since the finance minister (Secretaria de

Hacienda y Credito Publico –SHCP) is dominant in that process.

The expenditure budget is approved only by the lower chamber. The revenue law

is approved by both. Specific legislation dealing with the fiscal federal pact, namely, the

Ley de Coordinacion Fiscal (LCF), is voted by both the Senate and the Chamber ofDeputies. The implication of this is that when the budgetary process increases funds

15

transferred to the states, when those changes are incorporated into the ley de coordinación

fiscal, it is very difficult for decreases to be made in subsequent dates. While the

budgetary process determines the size of funds transferred to states and municpalities, the

fact that those funds are tied according to the LCF to the revenue collected by the federal

government (Recaudacion Federal Participable –RFP), changes the formulas and

allocation criteria need to be approved by both the Chamber of Deputies and the Senate.

Once a new fund, allocation criteria or provision has been included in the Ley deCoordinacion Fiscal, it takes both chambers, not just a budgetary appropriation, to makefurther changes. The players of federalism at the state level are embodied in the directly

elected governors, who play the executive role within a system of separation of powers.

Those governors are elected according to a staggered electoral calendar, so that only a

handful of them are elected concurrently with the president and the federal legislature. It

is not until the fifth year in the (six year) presidential term that the majority of the

governors have been elected during the current presidential term. The political

significance of this staggered setup is that governors do not ride federal or presidential

coattails, so their electoral claim to representation is independent from that of the federal

government. This is a key difference with respect to Argentina, Brazil or other federal

systems, , sinced the president lost his clout with governors. In the past, the control that

PRI presidents had over their governors was political, not institutional. To the extent that

most governors are not affiliated to the party of the sitting president (and this is likely to

continue to be a feature in Mexico’s political landscape in the future) cooperation

between executives in Mexico will require federal concessions. Presidents in Mexico

cannot really campaign on behalf of governors, and, to the extent that decentralization

continues, local politics will increasingly be more detached from national political

processes.

The institutional power of governors differs across states, with some being

extremely free to appoint and dismiss their cabinets, and propose legislation, while others

are quite constrained by their legislatures. The power of the governors powers is kept in

check in all cases by unicameral legislatures, and local judiciaries (there are no state

Supreme Courts, though). Except for the informal influence they can exert on their state

deputies or senators, governors have no arena in which they can debate the issues of

16

federalism. Finance ministers from the states, together with representatives from the

federal finance ministry, the SHCP, can discuss fiscal federalism every year in the

Reunion Anual de Funcionarios Fiscales, but their decisions have no political weight.

The municipal level of government has played a key role in Mexican federalism,

notwithstanding that municipalities are not, strictly speaking, partners in the federal pact.

There is no specific mechanism for the representation of municipal interests within the

state level institutional structure, except to the extent that they belong to local

congressional districts. However, municipal interests are primarily protected by the

federal constitution. In contrast to other federal regimes, state legislatures do not

determine the main characteristics and attributes of local governments. Provisions in the

federal constitution determine the composition of municipal councils, the functional areas

where municipalities have exclusive jurisdiction, and the transfer of federal and state

financial resources. State legislatures can decide the characteristics of the property tax,

which is the main source of revenue for municipalities, and can determine the allocation

of some, but not all, the federally mandated transfers across their municipalities.

To sum up, the political institutions in Mexico generate a complex set of veto

players with varying political incentives: 1) the mixed system in the legislature generates

a fragmented lower chamber, where no party is likely to control the majority –multiparty

coalitions must be formed to pass legislation; 2) career advancement motivations lead

deputies to cater to either partisan or state interests –districts are likely to be neglected; 3)

the Senate represents party, rather than state interests –there is no natural forum for the

debate over federalism; 4) the budgetary process involves a weak president, subject to be

held up by individual or groups of deputies –oversized coalitions held together by side

payments are likely to be formed in order to gain budgetary approval; 5) states as

represented in the Senate have no decision power in the expenditure budget approval –but

they may block changes to the status quo in the fiscal federalism arrangement; 6)

municipalities have incentives to press, through their co-partisans in the federal

legislatures, for greater devolution –they have no incentive to align their positions with

their state governments and legislatures.

17

Coalition building strategies and federalism

This section explores the prospects for change from the perspective of the

possible coalitions that may support either a comprehensive reform to Mexico’s federal

pact or less ambitious measures to streamline decentralization –particularly fiscal

decentralization. It discusses how the institutional features reflected in the composition of

the legislature, make coalitions for the reform of federalism very unlikely to form. The

analysis is both institutional and political: it stresses the institutional incentives that

generate or discourage party discipline and catering to regional interests in the Mexican

legislature; while at the same time paying attention to the political configurations that

facilitate or hinder the formation of specific regional coalitions.8

Given the institutional setting in Mexican federalism discussed above, the

question is what are the coalitional strategies that are most likely to emerge from the

configuration of political forces in the federal legislature elected in 2000? The analysis in

this section concentrates in the Chamber of Deputies because of it key role in the budget

process but the problems highlighted only become compounded by the presence of

Senatorial veto powers over changes to legislation. The basic coalition problem in the

years to come in Mexico is the following: Given the fragmentation of the legislature

(where no single political party controls a majority) and the partisan effects of discipline,

reformist minimal winning coalitions are not likely to be formed. Instead, oversized

coalitions that incorporate both partisan and state interests will modify the design of

fiscal federalism only to increase the size of the “pie” to be distributed, without

internalizing the costs of that increase or stimulating efficiency and accountability in

expenditure responsibilities already transferred to states and municipalities.

The idea is that local political actors conceive federal policy (and specifically the

federal budget) as a fixed sum to be divided amongst themselves. The divide the dollar

game that is played out every year in the debates over the role of the federal government

in decentralization and the size and scope of federal transfers makes states better off than

a disagreement outcome. However, states do not internalize the costs of the specific

agreement they reach. This notion is quite distinct from the common view of the Mexican

8 The literature on coalition making in the Mexican legislature is only starting to be generated. The firstcritical contributions are Weldon (2000) and Casar (2000 and 2001).

18

federal agreement as a zero-sum game, usually expressed by governors and other local

political actors. In the zero-sum image, the problem with the distribution of federal funds

is that any formula could be changed, because there will always be a coalition willing to

do so. Instead, the coalitional problem we have in mind is one where resources to the

states increase, and the problem from the point of view of the states is to find ways to

distribute those additional benefits.

Coalition theory in cooperative games provides a useful framework to understand

the conditions under which legislators are likely to vote together approving changes to

the current system of fiscal federalism. The most basic solution concept developed by

that literature, the Minimal Winning Coalition (MWC), suggests that when players must

distribute a fixed pie amongst themselves, they will try to exclude as many potential

partners as possible, while retaining a slim majority. A minimal winning coalition forms,

according to the theory, when the members of the coalition can share the benefits of

cooperation, while excluding those outside of the coalition from the pie to be shared.

Cooperative game theory does not explain the internal mechanisms through which the

coalition distributes and enforces the sharing of the resources they control –it is a

cooperative solution concept.

In practice, minimal winning coalitions are seldom observed in legislative

contexts. The reason is mostly related to institutional features in specific legislatures, and

ideological or issue specific considerations that make some coalition partners more likely

to cooperate with each other. In one extreme, universal coalitions, where all the members

in a legislature share in the benefits of cooperation, can emerge through a process of

logrolling, when the costs of financing the resources available to be shared are not

internalized by the coalition. Logrolling is often described in the Brazilian and the US

legislators as the informal rule of reciprocity that makes legislators vote for each other’s

projects or amendments. Oversized coalitions, short of universalism, are common

empirically. Ideological (or issue position) considerations seem to play a prominent role

in these coalitions, requiring partners to be ideologically contiguous (connected) over an

issue space in order to generate cooperation. Coalition theory is useful, beyond its

specific predictive power, in that it identifies key players and the incentives they have to

cooperate or block legislative changes.

19

With perfect party discipline, in which each party would vote as a block and there

were no state divisions, the MWC that should approve legislative changes in Mexico is

one involving the Partido Accion Nacional (PAN), which controls the presidency, and the

left wing party, the Partido de la Revolucion Democratica (PRD). The empirical problem

with such a coalition is that the PRD has been unwilling to cooperate with the party in

office, for ideological reasons, during both the 1997-2000 legislature and in the current

one. A minimal winning coalition in matters of federalism would not be partisan, but

regionally based. PRI and PAN are closer ideologically, so a coalition between them is

perhaps not so surprising, given that there is already a precedent of legislative

cooperation between both parties at least since 1988. However, the problem of such

coalition, from the perspective of cooperative game theory, is that it is too large. Such

coalition could surely form in matters of policy or nonexcludable public goods provision,

but when the issue is a divisible fund to be distributed, coalitions partners have most to

gain when they keep the coalition as close to 50 percent as possible.

Table 2.1 applies the concept of MWC to the regional composition of the

Mexican legislature in 2000. Since the legislature contains both single member district

legislators and deputies elected through proportional representation, the analysis takes

each state delegation, and each proportional representation partisan fraction as an

individual player, whose voting weight depends on the number of deputies it contains.

The coalition reported in the table involves the largest states, which can contribute most

legislators to a coalition, together with the PAN and the PRD’s non-territorially based

deputies, elected through proportional representation. Such an alliance would be the

smallest possible group of legislators that can get a bill approved. A coalition like this

would hardly be observed in practice. It includes the president’s home state, but it

excludes most of the states governed by the PAN. That feature would make it unlikely

that the PAN proportional representation deputies would join it. Such coalition would

also require the leadership of the PRD and the Mexico City governor to join, neglecting

prominent leaders in that party, who are currently governors in small and/or poor states

(Zacatecas, Baja California Sur, Tlaxcala, Chiapas). It would also involve incorporating

three of the states with the most powerful PRI governors. Such a coalition would not fly

to the extent that the PAN and the PRD are in opposite ends of the ideological spectrum.

20

Moreover, the exclusion of most states would make such coalition incapable of making

legislative changes that would be approved in the Senate.

Table 2.1. Minimal Winning Coalition: Largest states with PAN and PRDproportional representation deputies

Cumulativepopulation Districts PAN PRD PRI

CumulativeVotes Governor

México 12.84 36 24 1 11 36 PRIDistrito Federal 22.15 30 25 5 0 66 PRDVeracruz 29.54 23 14 2 7 89 PRIJalisco 36.11 19 16 0 3 108 PANPuebla 41.18 15 6 0 9 123 PRIGuanajuato 46.01 15 14 0 1 138 PANSingle MemberDistricts 138 99 8 31 138ProportionalRepresentation 122 82 40 0 122Legislators 260Percent Vote 52.0%

An alternative scenario of an (almost) MWC would be one where the three largest

states find a way to obtain the support of all the PRI and PAN fraction PR legislators, in

order to have the smallest number of players involved (Table 2.2). This scenario is two

votes short of a full majority, but it highlights that the largest states are in a way vital for

the support of any coalition for reform. Again, such coalition would be highly unlikely,

since it involves governors from parties different from that of the president, excluding all

PAN governors. Since the PAN has strong delegations in the SMD of those states, the

coalition does incorporate a majority of its legislators, which suggests those legislators

might be the most torn between supporting changes that benefit their states and voting on

non-territorial lines with their co-partisans.

21

Table 2.2 Almost Winning Coalition: Largest States with PAN and PRIproportional representation deputies

TotalCumulativePopulation Districts PAN PRI PRD

CumulativeVotes Governor

México 12.84 36 24 11 1 36 PRIDistrito Federal 22.15 30 25 5 30 PRDVeracruz 29.54 23 14 7 2 23 PRISingle MemberDistricts 89 63 18 8 89ProportionalRepresentation 82 78 0 160Legislators 249Percent Vote 49.8%

More realistic coalitions would involve oversized groups, which although costly

to their members because they must share limited resources with more members than

strictly necessary to garner a majority, would be more likely to form due to the partisan

and regional incentives. Oversized coalitions would also be necessary in order to obtain

Senatorial support for any change to the Ley de Coordinación Fiscal. Table 2.3 shows acoalition of PAN governors and PAN legislators, which elicits the support of the PRI

deputies elected through proportional representation. This coalition does not include PRD

deputies, regardless of the principle by which they are elected. In this case states are

ranked not by size, but according to the importance of the PAN legislative fraction in

each state. At the top of the list are states where all the legislators from the single member

districts in the state come from the PAN. The last state includes the last governor in a

state with a substantial number of PAN legislators. Such a coalition involves a sharp

break with party discipline for the PRI –it would only make sense to the extent that, for

example, the governor of the State of Mexico would be able to capture the party

leadership in the legislature, and be willing to break the party. Such coalition makes sense

for the PAN, since it includes the overwhelming majority of its legislators. The coalition

thus described would fall short in terms of senatorial support, since it would contain only

10 states.

22

Table 2.3. Coalition of PAN Governors, plus PAN and PRI proportionalrepresentation deputies

Cumulativepopulation Districts PAN PRI PRD

Cumulativevotes Governor

Baja California 2.32 6 6 0 0 6 PANAguascalientes 0.95 3 3 0 0 9 PANGuanajuato 5.78 15 14 1 0 24 PANJalisco 12.35 19 16 3 0 43 PANDistrito Federal 21.66 30 25 0 0 68 PRDMorelos 23.24 4 3 1 0 72 PRIQuerétaro 24.61 4 3 1 0 76 PANMéxico 37.45 36 24 11 0 111 PRIChihuahua 40.52 9 6 3 0 120 PRINuevo León 44.42 11 7 4 0 131 PANSingle MemberDistricts 107 24 0 131ProportionalRepresentation 82 78 0 160Legislators 291Percent Vote 58.2%

An opposite scenario would assume PRI deputies being able to break party

discipline in the PAN (and given its slim majority, being able to command the support of

7 PRD legislators in the included states). The coalition partners described in table 2.4

would have a very slim advantage in the Senate, falling short of a majority by several

votes. Such a coalition is highly unlikely on several grounds. First, the PAN controls the

presidency, which has the power of proposal in the budget bill, which suggests it would

be very difficult for such coalition to articulate a full proposal of reform to fiscal

federalism (although such coalition can surely block changes it does not want). Second,

the coalition’s majority is so slim, that it would easily fall hostage of a few deputies, such

as the PRD defectors or legislators from the two states governed by the PAN. The

important issue that this coalition highlights, however, is that this configuration is the

only one that would include a substantial number of poor states.

23

Table 2.4. Coalition of PRI proportional representation deputies and states withlargest PRI delegations

Cumulativepopulation Districts PAN PRI PRD

Cumulativevotes Governor

México 12.84 36 24 11 1 36 PRIChiapas 16.77 12 1 11 48 CoalOaxaca 20.31 11 1 10 59 PRIPuebla 25.38 15 6 9 74 PRIGuerrero 28.58 10 9 1 84 PRIVeracruz 35.97 23 14 7 2 107 PRISinaloa 38.63 8 7 1 115 PRIHidalgo 40.95 7 7 122 PRITamaulipas 43.72 8 3 5 130 PRINuevo León 47.62 11 7 4 141 PANSan Luis Potosí 50.04 7 3 4 148 PRITabasco 51.96 6 4 2 154 PRIJalisco 58.53 19 16 3 173 PANSingle MemberDistricts 75 91 7 173ProportionalRepresentation 0 78 0 78Legislators 251Percent Vote 50.2%

A more likely scenario of coalition formation is provided by Table 2.5. This

configuration of states is labeled a federalist coalition, because it includes all the PAN

governors, the PRI and PAN deputies in those states, and the PRI and PAN governors in

the richest states. Most crucially, the coalition only excludes a handful of PAN deputies

in states where the party is very weak. This is one of the few coalitions that are small,

while including virtually all PAN legislators. If the PAN attempted to include all of its

legislators it would need to form a universalistic, not a MWC.

24

Table 2.5. Federalist coalition

TotalShare

population Cumulative PAN PRI PRD TotalPRI +PAN Cum.

Governor

DistritoFederal 9.31 9.31 25 0 5 30 25 25 PRDMéxico 12.84 22.15 24 11 1 36 35 60 PRIJalisco 6.57 28.72 16 3 0 19 19 79 PANVeracruz 7.39 36.11 14 7 2 23 21 100 PRIGuanajuato 4.83 40.94 14 1 0 15 15 115 PANNuevo León 3.9 44.84 7 4 0 11 11 126 PANPuebla 5.07 49.91 6 9 0 15 15 141 PRI

Chihuahua 3.07 52.98 6 3 0 9 9 150 PRIBaja

California 2.32 55.3 6 0 0 6 6 156 PANCoahuila 2.38 57.68 4 3 0 7 7 163 PRISonora 2.29 59.97 4 3 0 7 7 170 PRI

Tamaulipas 2.77 62.74 3 5 0 8 8 178 PRISan LuisPotosí 2.42 65.16 3 4 0 7 7 185 PRI

Michoacán 4.25 69.41 3 1 9 13 4 189 PRIMorelos 1.58 70.99 3 1 0 4 4 193 PRIQuerétaro 1.37 72.36 3 1 0 4 4 197 PANAguascalien

tes 0.95 73.31 3 0 0 3 3 200 PANSMD 144 56 17 217 200 200PR 82 0 0 82

TOTAL 28256.4%

The interesting feature of this federalist coalition is that it does include key

players that are contained in specific proposals that have been made to reform some

aspects of Mexico’s fiscal federalism. For example, the World Bank (1998) calculates

that if educational funds in Mexico were distributed according to educational efficiency,

namely students finishing their basic studies, Estado de Mexico, Michoacan, Nuevo

Leon, Puebla and Guanajuato would receive more resources, while Baja California Sur,

Campeche, Chiapas, Guerrero, Oaxaca and Tamaulipas would receive significantly less.

Thus a formula that allocates educational resources to states according to this criteria

could pass the legislature, according to our analysis. Courchene and Diaz-Cayeros

calculate in the World Bank (2000) publication that a major overhaul of the transfer

system would benefit Puebla, Veracruz, Michoacan, Guanajuato, Jalisco, Estado de

Mexico, Chihuahua, Nuevo Leon, the Federal District and maybe Baja California and

25

Sonora. A coalition like the one in table 2.5 would need to receive compensation in order

to pass such a reform, with a much lower transition cost than the one calculated by those

authors for compensating all losing states. The problem with any of those reforms, as can

be clearly seen from table 2.5, is that while changes would generally be supported by rich

states, they would put in question the compensatory features of the federal pact, because

they exclude virtually all the poor states!

Regardless of the specific coalitions that may come to be formed in the Mexican

legislature, three lessons emerge from the analysis presented above in this section. First,

given the partisan incentives in the mixed system of representation, any coalition that

seeks to reform the fiscal federal pact must be oversized. Second, the coalition must cater

to the interests of states such as Mexico, the Federal District and Veracruz, which enjoy

virtually guaranteed inclusion due to their large size. Third, to the extent that party

discipline remains strong, coalitions will be closer to universalist, which reduces the

scope for changes, since everyone would need to be made better off.

26

3. The first phase of Mexico’s recent decentralization and federalizationpolicies: A Transfer Led Decentralization

The Centralization of Tax Collection in 1980.The first key ingredient in the current decentralization panorama was the

introduction of the Value Added Tax, VAT, in 1980 --which centralized most indirect

taxes while at the same time it guaranteed unconditional, almost automatic resources to

the states. The impact of the reform of 1980 on the potential for decentralization reform

in Mexico is still felt today and a secondary purpose of this section is to provide a brief

snapshot of the existing intergovernmental fiscal system. Indeed, when the real balance

of taxing powers and clearly divided responsibilities is taken into account, Mexico’s

fiscal federalism will probably be ranked among the most centralized in the world. It has

not been possible, for instance, to accompany the accelerated process of expenditure

decentralization of the last few years with devolution of fiscal authority to states or

municipalities. This section discusses how, when and why Mexico’s sovereign states

gave up most of their revenue capacity in exchange for a share of federal tax revenues.

The following sections explain how and why Mexico has been trying ever since to

partially redress the picture of weak subnational taxing powers and weak subnational

accountability that emerged from the federal fiscal pact of 1980.

Fiscal centralization in Mexico was the result of a long process through which

states and municipalities gave up their capacity to tax, in exchange for unconditional

transfers from the federal government. The main argument behind such arrangement was

the notion that double taxation was detrimental for economic activity. Multiple tax

jurisdiction as established in the Constitution was gradually eliminated through efforts

that started in the 1920s and culminated with the establishment of the revenue sharing

system, the Sistema Nacional de Coordinación Fiscal (SNCF) and the introduction of the

Value Added Tax in 1980. The arrangement was not imposed by the federal government.

State governments accepted the terms of a specific deal: fiscal authority was given up in

exchange for revenue shares that were not contingent on their own fiscal effort.9

The tax rate in Mexico is low by international standards and equal to 11% of

GDP. Taxes collected at the subnational level as a share of an already low national tax

9 See Diaz Cayeros, 1997.

27

effort is also low by international standards. This result is due to a long history of tax

centralization after the revolution. It was very difficult to get the states to collect taxes so

the federation took it upon itself to centralize the collection of taxes.

The main sources of state funds are payroll taxes and automobile registration.

Payroll taxes produce almost half of the state own tax revenues (45.6% but the share

increases to 63% if one excludes the Federal District). However, state own revenues

account for only 9.2% of total revenues of state governments due to the dominance of

revenue sharing (see discussion on transfers and fiscal dependency rates below).

Furthermore, when tax administration costs are taken into account, state own revenues

amount to only 2.7% of total state net revenues. Twenty-one of the 32 states use payroll

taxes with rates varying from 0.5% to 4% --though most of the rates fall between the 1%-

2% range.10

Relying on the payroll tax has pros and cons. The main problem is that the

payroll is an origin-based tax which may distort the location of economic activity if

levied at different rates. Second since payroll taxes are collected at the employment

location, cities near state boundaries present a problem because people may live in one

state and work on another (as it happens, for instance with the metropolitan areas of the

Federal District and the State of Mexico in the center of the country or Torreon-Gomez

Palacio in the North). The problem is that the taxes go to the state where employment

takes place and the demand for services is in the state where people live. Nevertheless,

the payroll taxes could play a greater role in state revenues if some minor changes were

implemented. The most obvious administrative reform –and one that ia being partially

implemented in Mexico—is that states either double check with the federal social

security registration of entirely abandon their individual tax requirements and regulations

and try to conform more closely with the federal payroll-based contributions through a

surcharge on the existing federal levies.

Automobile registration and user fees are formally federal taxes, but they are

directly and completely assigned to the states. Compliance with these taxes appear to be

improving over the last few years.. As a matter of fact, states are both learning from and

10 Source: World Bank 2000.

28

competing with one another, facilitating tax compliance to taxpayers (through better

billing, bank payments, etc.) and gradually linking services to tax revenues.

Other taxes have been proposed for the state level: Individual income taxes,

excise taxes on alcoholic beverages and tobacco products, corporate income tax, retail

sales tax, a compensating Value Added Tax (VAT). All of these have their advantages

and disadvantages that are well explored in the literature. At this point it appears that

implementation of any of these will not be without serious problems. The main concern

is that different state taxation regimes could affect the allocation of production and could

hinder interstate trade.

Figure 3.1: Breakdown of state own revenue

Sources: INEGI as cited in World Bank 2000

Like states, municipal governments are dependent on federal funds. The main

source of own municipal revenues is the real state property tax comprising 13% of total

net revenue and 74.2% of municipal own revenues. Except for Mexico City, the

collection of property tax is low. Municipal governments have not updated cadastres

29

effectively or frequently. Municipal governments are considering levying excises on

public utilities. Although easily enforced, applying an indirect tax such as this to

corporate consumption of electricity introduces a cascading element that would put

inflationary pressures and distort certain chains of production.

It is useful to remember that the creation of a highly centralized system,

characterized by weak subnational taxing powers and high fiscal and political

dependency was only possible due to the political dependence of governors, municipal

presidents, local and federal deputies and senators on a hierarchical system of career

advancement created by the hegemonic PRI. The party centrally controlled the

nominations that allowed for career advancement, while protecting its members from the

challenges of democratic competition (sometimes through illegal means). The president,

as the leader of the party, used his power and resources in the federal sphere to subdue

local autonomy. Centripetal forces in the regions were not eliminated, though. Hence, the

continuous resurfacing of decentralization demands that has accompanied the process of

democratization since the early nineties –always debilitating the potential of in-depth

decentralization reform for higher subnational autonomy.

This is an enviable position for subnational governments that share a culture of

political patronage based upon pyramidal relations with centralized governments or

political parties. Those governments would prefer to spend without needing to collect

taxes since such governments are not interested in receiving the credit of good

expenditure management or adequate government financing. In single-party systems

revenue is often collected without asking citizens what they want to pay, or what their

money should be spent for. In fact, given the difficulty at eliciting cooperation and

compliance, authoritarian systems often rely on natural resource rents in order to support

their activities.

In multi-party, competitive systems the fiscal challenge for a government is to

find a balance between the taxes that are raised and the public goods that are provided

according to citizen demands. Such a balance can only be reached through transparent

fiscal systems that are perceived by most voters as fair. The budgetary battles of each

year are the expression, in democratic systems, of the political compromises seeking to

keep such balance.

30

Since the distribution of funds in the system of revenue sharing is very inelastic to

local tax collection effort and most local expenditure is financed with federal transfers,

the blame for high taxes, or the lack of financial resources to fulfill citizen demands, can

always be put on the federal level of government. The leverage of states and

municipalities in the budget process is very high, given the peculiar conditions of

budgetary authority previously discussed. Such leverage did not become evident until

after 1997. The conventional discussion of the creation of budgetary item 33 and the

increase of transfers to subnational levels of government has stressed the advantages in

greater transparency and devolution achieved each year.11

A Transfer Lead DecentralizationIn the Mexican federal arrangement, however, state and municipal governments

often spend not according to a political balance of revenue collected and goods and

services provided, but sending funds to those citizens who have the most capacity to

press for them, without taking into account where the funds come from. The federal

budgetary battles have become, in the last years, an expression of the lack of

accountability of local governments. In each budgetary round, since the PRI lost its

absolute majority in 1997, transfer funds have been increased and new ones created in

order to obtain legislative upport for the presidential bill.

Figure 3.2 shows that public expenditures have been decentralized quickly in

Mexico while subnational tax revenues have hardly increased. A few points are worth-

noticing about this figure. First, Mexico’s level of subnational expenditure in 1996 was

still significantly below Brazil’s and Argentina’s or –unitary—Colombia for that matter.

Thus, although the federal fiscal pact of 1980 represented an important move in the

direction of fiscal decentralization, Mexico was still a relatively centralized country –

when measured by the share of subnational treasury disbursements in the country’s total

public expenditure. It is only after the second and third phase of decentralization reform

(see below), with the consolidation of earmarked, formula-based transfers (transfers), that

Mexico’s states and municipalities reach levels of 40 percent and higher of total current

11 See the chapters 3 and 5 in Giugale and Webb, 2000.

31

government revenues. Only then can Mexico be ranked at the level of decentralization

reached earlier by Argentina, Brazil or Colombia.

Second, subnational taxes are well below Argentina’s and Brazil’s –even

Colombia’s. This imbalance presents a potential problem in and of it self because states

are so dependent on the central government for transfers and because of the potential

collection efficiency increases at the state and local level for some taxes has not been

realized. Brazil, for instance, authorizes states to charge the VAT while unitary

Colombia authorizes departments to charge more excises than Mexican states.

Finally, comparative analysis of expenditure decentralization should be taken with

a grain of salt because, as was shown above, at least for the case of Mexico much of this

“decentralization” is really just deconcentration of expenditures; i.e. earmarked transfers

to subnational governments where states and municipalities have no discretion to pursue

higher efficiency by modifying allocation or production functions. Therefore, a

comparison of social expenditures in Brazil and Mexico (or Colombia) mixes apples with

oranges: decentralized expenditures are a different species from deconcentrated

expenditures. We now elaborate further on the first point listed above.

Figure 3.2: Subnational taxes and expenditures shares in Argentina, Brazil, andMexico.

Source: IMF, Government Financial Statistics, as cited in World Bank 2000.

32

Table 3.1 shows expenditure by government level in Mexico between 1989 and

1996. At that time, earmarked transfers, (aportaciones ) were still counted under Federaljurisdiction. According to Table 3.1, the federal government still spends almost three

quarters of public expenditures. However, the share dropped from almost 84% in 1989 to

about 75% in 1996. Most of the gain has been by state governments while municipalities

continue to spend around 4% of total government spending. During this time period, the

states’ administrative expenditures increased from around 30% of their total expenditures

to about 81%.12 No doubt, transfers for health and education are largely responsible for

this increase in current expenditures. Still, the key observation is that the reforms of the

federal fiscal pact and sector deconcentration/decentralization came to guarantee a close-

to-certain level of revenues with no additional fiscal effort and little or no effective

responsibilities in health and education for the state level. In turn, guaranteed income

appears to have stimulated expansion of relatively rigid administrative expenditures. In

this regard, Mexico appears to lag behind Brazil or Colombia that have recently

introduced fiscal responsibility laws that set strict ceilings on subnational administrative

expenditures.

Table 3.1: Expenditure by level of government in México from 1989-1996

Source: INEGI as cited in World Bank 2000.

12 Source: World Bank Estimates with SHCP data.

33

Horizontal disparities in public spending in Mexico have decreased over time but

remain high by international standards. Table 3.2 shows total, current and capital

spending per capita across states. Not surprisingly the Federal District (Mexico City),

has the highest expenditure but excluding this observation, public spending does not

appear to be regressively allocated. The state of Mexico, one of the richest states has a

relatively low expenditure per capita. The capital expenditures per capita present even

greater heterogeneity. The largest capital expenditure (in Mexico City) is over twenty

times the lowest, (state of Mexico).

Table 3.2: State Expenditures per capita in 1997.

34

Source: IINEGI as cited in World Bank 2000

35

The Nature fiscal transfers: The creation and transformation of the Social InvestmentFund

As a result of a dismally low subnational tax effort, it is not surprising that states

are dependent on federal transfers. Figures 3.3 and 3.4 show the high dependency of

state and municipal governments on federal and state transfers respectively. The states

are ranked according to their degree of poverty.13 Only the Federal District has a sizable

share of self generating revenue –though it still accounts for less than half of total

revenues.

Every federal system is characterized by regional transfers, which bridge the

inevitable gap between substantial subnational responsibilities and subnational taxation.

Indeed, no sub-national jurisdiction with substantial responsibilities is able to fully

finance its own activities. However, the crucial question on the grounds of accountability

is whether, at the margin, new expenditure projects are financed by new taxes or

expansions of the tax base that are locally borne. That is, the benefits of fiscal federalism

are possible when the benefit principle holds at the margin. In Mexico, the federal fiscal

pact (the Sistema Nacional de Coordinación Fiscal, SNCF) that started in 1980

established a hard-to-reverse trend of ever-further weakening the Wicksellian connection

between taxes and expenditures. Local governments have grown used to a system of

revenue sharing that does not reward collection effort, and where the blame for high

taxes, or the lack of financial resources to fulfill citizen demands, can always be put on

the federal level of government. This is a very comfortable position.

13 The index is the Foster-Greer-Thorbecke index with a poverty line seta t two minimum wages.

36

Figure 3.3: Sources of funds for States

Source: Courchene & Diaz-Cayeros (2000), with data from INEGI, SHCP and DOF

Figure 3.4: Sources of funds for Municipal Governments

Source: Courchene & Diaz-Cayeros (2000), with data from INEGI, SHCP and DOF

State Funds 1999

$0

$1,000

$2,000

$3,000

$4,000

$5,000

$6,000

Chiapas

Oaxaca

Zacatecas

Guerrero

Puebla

SanLuisPotosí

Hidalgo

Yucatán

Veracruz

Tabasco

Campeche

Durango

Michoacán

Tlaxcala

Guanajuato

Querétaro

Nayarit

QuintanaRo

o

Tamaulipas

Jalisco

México

Morelos

Aguascalientes

Chihuahua

Coahuila

Sinaloa

NuevoLeón

Colim

a

BajaCa

liforniaSur

Sonora

BajaCa

lifornia

DF

States

Pesospercapita

Conditional

Unconditional

Own Revenue

Municipal Funds 1999

0.00

200.00

400.00

600.00

800.00

1000.00

1200.00

1400.00Chiapas

Oaxaca

Zacatecas

Guerrero

Puebla

SanLuisPotosí

Hidalgo

Yucatán

Veracruz

Tabasco

Campeche

Durango

Michoacán

Tlaxcala

Guanajuato

Querétaro

Nayarit

Quintana

Roo

Tamaulipas

Jalisco

México

Morelos

Aguascalientes

Chihuahua

Coahuila

Sinaloa

Nuevo

León

Colima

BajaCalifornia

Sur

Sonora

BajaCalifornia

DF

States

Pesos percapita

Own revenue Unconditional Conditional

37

Transfers in Mexico have increased in importance recently. In 1999 they

accounted for of almost 40% of the federal budget. There are currently two main types

of federal transfers: unconditional revenue sharing (participaciones) and conditionaltransfers (aportaciones). We describe each in turn. Besides, the federal government offersmatching grants to states and municipalities, generally for subsidized, anti-poverty

programs.

Figure 3.4:

The system of unconditional transfers (participaciones) began in the 1970s buttook its recent form in the early 1980s when the sales tax was transformed to the current

Value Added Tax, VAT. Participaciones are represented by Ramo 28 and represent 47%

of all transfers. The transfer amount is equal to 20% of the federal income tax, the VAT,

and oil fees. In other words, participaciones come from the most important sources of

revenues. 45.17% is allocated on a percapita basis, 45.17% is allocated on a historical

inertial basis, and 9.66% is allocated in a way that compensates the other two. Diaz

Cayeros (1995) demonstrated a gradual convergence across states in per capita terms.

Most of these funds are allocated to the Fondo General de Participaciones and the fundshave unrestricted use.

Composition of Transfer Funds in the Federal Budget of2002

FAEB34%

FAS6%

FIS5%

FAFMDF5%

FAM1%

FAET1%

FASP1%

PAFEF0%

R2847%

38

Conditional transfers or aportaciones (including those for the Federal District inbudgetary item 25) amount to about 53% of total transfers to subnational governments in

2002. The so-called Ramo 33 aportaciones have become a complex amalgam of 8 fundstransfer programs. Nevertheless, one could group them into two broad types of

expenditures. Most of Ramo 33 is made up of funds for tasks that were originally carriedout by the federal government. The most important are education and health

expenditures represented by FAEB (for education) and FAS for health in figure 3.4. The

second is made up of funds whose original intent was to provide matching grants to

finance public works. During the Salinas administration these funds increased in

importance drastically. However, they were criticized for being too discretional. They

were transformed during the Zedillo administration into the Municipal Social

Infrastructure Fund and their allocation was determined by a complex poverty formula.

The per capita variability across states is greater than for participaciones.

Table 3.4: Transfers and Own resources as share of total subnational resources

Source: Ley de egresos as cited in World Bank 2000.

39

The obvious answer to the issue of abating the fiscal dependence of subantional

governments on federal transfers is that subnational taxation has to increase. Indeed,

international experience suggests a variety of creative ways to increase subnational

taxation, including inter-government tax sharing agreements. Although there are several

ways to increase subnational taxation (and the World Bank book of 2000 lists and

proposes the most obvious ones), the truth is that increasing subnational own tax

revenues in Mexico has become politically difficult in Mexico. As a matter of fact, what

is going on is exactly the opposite: there is a natural “race” to the bottom in subnational

taxes. This is especially true if, as it is the case today, citizens cannot observe a

difference in quality with increases in taxes.

Subnational BorrowingStates are demanding access to financial markets to finance their activities.

Analyzing subnational debt numbers in Mexico is a bit deceiving. Table 3.5 show the

level of debt by state since 1994. Total debt increased from 26,958 million pesos in 1994

to 71,627 in 1998. In 1997 this represented 10% of total public debt (including debt by

Banco de Mexico) and about 2% of GDP. (World Bank 2000). Despite the recent

growth, currently the size of subnational debt does not pose a macroeconomic problem

for Mexico. However, the small size of the debt mask deeper fiscal imbalances that have

been erased through repeated federal bailouts. States fiscal deficits have been

systematically larger than the increases in indebtedness. Figure 3.5 shows primary

balances with and without extraordinary transfers from the federal government. In the

first case, states are running a primary surplus while in the second they are running a

deficit. Moreover, around 40% of subnational debt is implicitly guaranteed by the federal

government because it was contracted with Banobras, a national development bank

created for this purpose, before enactment of new regulatory framework for subnational

borrowing (see below) The rest is contracted with commercial banks. Note that

subnational debt is concentrated in a few states notably the State of Mexico, and the

Federal District which alone account for over 53% of total subnational debt in 1998.

40

Since these states are also some of the richest, any bailout of subnational debt will be

regressive.

Table 3.5: Subnational debt by State from 1993-2001DEBT STOCK OF STATES AND MUNICIPALITIES

(year end, million pesos, ranked by burden)

1993 1994 1995 1996 1997 1998 1999 2000 Sep-01

as % ofGDP in2000

M…XICO 2,729 4,843 8,644 13,397 16,610 19,629 22,410 23,342 24,935 4.70% SONORA 2,547 3,150 4,869 6,086 3,672 4,178 4,585 4,936 4,909 3.70% SINALOA 482 874 1,338 1,677 1,931 1,990 1,501 2,449 2,437 2.60% NUEVO LE”N 1,273 2,348 6,427 5,464 6,707 7,990 8,624 8,796 8,719 2.60% B.C.S 261 304 297 351 450 472 517 592 651 2.20% GUERRERO 498 516 858 984 1,169 1,310 1,512 1,599 1,576 1.90% JALISCO 2,261 2,812 3,372 3,876 4,007 4,514 4,749 5,162 5,085 1.70% QUER…TARO 344 1,283 1,090 1,017 1,061 1,216 1,327 1,400 1,402 1.60% DURANGO 297 552 462 607 714 838 860 998 1,148 1.60% CHIAPAS 212 1,025 992 1,088 962 1,067 1,022 999 978 1.20% QUINTANA ROO 437 450 643 740 843 856 762 749 889 1.10% TABASCO 188 518 343 411 432 605 601 602 587 1.00% SAN LUIS POTOSÕ 306 346 426 544 599 661 642 842 815 1.00% B.C.N. 478 1,000 960 1,214 1,380 1,612 1,642 1,775 1,694 1.00% MORELOS 162 144 233 244 365 395 408 508 474 0.80% COLIMA 104 192 263 291 237 186 208 182 167 0.70% CHIHUAHUA 412 922 1,215 1,539 1,689 1,646 1,586 1,537 1,519 0.70% TAMAULIPAS 416 369 532 364 315 271 702 817 627 0.50% COAHUILA 249 516 926 1,116 594 649 787 803 757 0.50% PUEBLA 106 156 321 309 352 657 716 855 994 0.40% GUANAJUATO 280 406 412 465 517 590 559 627 607 0.40% YUCAT¡N 225 305 288 321 372 317 310 189 136 0.30% OAXACA 182 260 147 193 203 362 291 247 242 0.30% NAYARIT 118 223 188 178 115 101 88 92 97 0.30% MICHOAC¡N 101 250 256 252 216 283 293 214 187 0.20% HIDALGO 30 23 14 16 13 10 0 163 137 0.20% CAMPECHE 127 499 461 518 419 222 149 111 90 0.20% AGUASCALIENTES 284 364 308 339 287 197 164 141 123 0.20% ZACATECAS 132 124 381 469 236 98 75 39 13 0.10% VERACRUZ 348 348 379 262 79 23 23 18 606 0.00% TLAXCALA 69 136 53 0 0 0 0 0 0 0.00% Source: SHCP

41

Figure 3.5 Subnational Aggregate Fiscal Deficits 1989-97 with and withoutextraordinary federal transfers.

Source: SHCP as cited in World Bank 2000

Late in 1999, Mexico introduced an innovative framework to regulate subnational

debt. It uses a combination of market discipline and administrative controls to insure

subnational debt remains within indebtedness capacity and stimulates additional fiscal

effort, expenditure efficiency and hard budget constraint. There are six key elements in

the new regulatory framework for subnational borrowing: (i) The federal government

ties its hands and forbids discretionary transfers. (ii) The federal government stops

participating in the elaboration of the legal debt contracts thereby eliminating (or

lowering the channels) for recourse to the federal government. (iii) Subnational debt

becomes subject to borrower concentration limits in the financial sector lowering the

damage that a single state can cause. (iv) Linking capital risk weighting and the credit

ratings assigned to subnational governments. This way the financial sector limits the

access of states with weak financial performance through an innovative market approach.

(v) States can register their loans with the federal government only if they are current in

42

the publication of debt and fiscal deficits. (vi) Federal development banks, namely

Banobras, can only make loans to states when condition five is satisfied and the capital

risk weighting is less than 100%. Although it is still too early to assess the performance

of the new regulatory structure, so far the system is working according to new

regulations. States and larger municipalities are producing better balance sheets in the

expectation of getting a better credit rating and accessing capital markets in more

favorable terms.

Figure 3.6: Subnational Bailouts and new debt.

Incentive Effect of Bailout-0.105633803

AGS

BC BCS

CAM

COA

COL

CHISCHI

DGO

GTO

GROJAL

MEX

MICH

MOR

NAY

NLOAX

PUE

QRO

QR

SLPSIN

SON TAB

TAM

TLAX

VER

YUC

ZAC

-1

-0.5

0

0.5

1

1.5

2

-1.5 -1 -0.5 0 0.5 1 1.5

Increase (decrease) in debt stock 1994-1997

Increase(decrease)indebtstock1997-2000

43

4. The second phase of Mexico’s recent decentralization and federalization policies:Deconcentration, partial decentralization and confusing responsibilities.

In 1989 the Northern state of Baja California elected the first opposition governor

since the PRI was founded in 1929. Soon an increasing number of states and

municipalities opted for governments of party affiliations other than that of the federal

government. States and municipalities would increasingly develop fiscal policies and

development agendas of their own. By 1992 the federal government announced that it

was ready to transfer responsibilities, beginning with social sector responsibilities, to the

country’s 32 federal entities. The country was about to take the first step towards

decentralization in decades.

Education in Mexico is a classic case of deconcentration rather than decentralization.The process of decentralization began in 1992 during the Salinas administration, whenthe Education Ministry (Secretaría de Educación Pública -SEP) signed the NationalAgreement to Modernize Basic Education. The agreement was signed by the nationalteachers union, and the federal and state governments. The agreement appeared at first asa solid start in the direction of decentralizing the educational system, but it falls short ofactually decentralizing education. The federal government remains in charge of settingstandards, developing curricula, teaching programs, teachers training, producingtextbooks, and monitoring and evaluating subnational performance. The federalgovernment finances most infrastructure in the sector, but the states have some say on theway these funds are spent. The key element where the agreement fell short was in thecontrol of operation of the educational system. The federal government delegated to thestates the tasks of constructing and operating the schools and paying the teachers, alongwith the corresponding funds that were previously under federal jurisdiction. Theproblem is that the funds are fully earmarked and most decisions are made by SEP inMexico City. In effect, the states only distribute the checks and execute the plans withearmarked funds already provided by SEP. Even the teacher’s detailed payroll isspecified in the transfer of funds, although states do have some leeway to hire and fireteachers. Negotiations for wage increases are made at the state level, but the state sectionsof the federal teacher’s union do not feel committed by those agreements –the actualincremental wage bill is bargained by the national union in Mexico City. Therefore, mostmeaningful decisions are still made at the Federal level. Ironically, the Municipalitieshave little or no role in the process, although these entities are closest to the actualdelivery of service. The decision to exclude municipalities, where most educationresponsibilities laid until the 1930s, was probably because of the great heterogeneity inthe administrative capacity of local governments. The changes have introducedconfusion between pre-existing state school systems and the federal system, which needto be reconciled. The extent of state systems varies dramatically across states, whichraises an additional issue of horizontal imbalance in the transfer of education funds.Deconcentration of payroll and other operations has created administrative inefficiencieswithout offsetting gains in allocation efficiencies that should come with decentralization.

44

The movement towards decentralization faced four major constraints that ended

up by reducing decentralization to a confusing mix of deconcentration and partial

decentralization. It can be said that the decentralization movement of the early nineties

imprinted the wrong signals in Mexico’s decentralization –and those signals are still

present 10 years later. What the country has done in the second half of the nineties and

the very first years of this millenium can be read as a series of patchy reforms geared to

reversing the problems created by the extreme centralism of the federal fiscal pact

initiated in 1980 and the processes of decentralization of health, education and social

infrastructure, now contained in the federal fiscal pact (or Ley de Coordinación Fiscal,LCF).

The four major constraints faced by the decentralization movement of the early

nineties were:

! The federal fiscal pact of the early 1980s debilitated the states’ taxing powers

while at the same time transferring resources the states had already committed to

a wide range of activities –including health and education expenditures that either

duplicated or were parallel to those of the federal level.

! Political resistance to effectively empowering subnational governments –as

transfer of resources and responsibilities were generally perceived as further

weakening the monopoly power of the PRI party-government

! Trade unions that were used to a large quota-power under the long-standing

corporatist state were not prepared to accept fragmentation of their power to

negotiate collective bargaining agreements on a subnational scale or otherwise

debilitate their power-sharing position. Neither were unions prepared to accept

fiscally weaker managers

! Civil society did not mobilize in favor of decentralization. Service users had no

basis to expect states would perform better --in terms of coverage, quality or

expenditure efficiency—than the federal level.

At the end of the day, Mexico adopted a social sector decentralization scheme that

conveys perverse incentives to all stakeholders: From the revenue point of view, the main

45

concern of the federal level was to equate fiscal transfers with health or education

payroll. Although states are technically contractors of teachers and doctors, federal rules

and collective bargaining agreements between unions and the federation still prevail.

Production functions for health and education were kept highly rigid, thereby preventing

efficiency in production. Parents, teachers or social security users can hardly establish

any connection between paying taxes and state health or education services. As time has

passed since the decentralization agreement of 1992, states have kept demanding

additional resources to ensure payment of wage increases. Other factor inputs –such as

medical or education supplies, equipment or physical facilities have received little

attention. In fact, what many states still do is leave education sector management to trade

union leaders. Supply side focus in management deters attention to educational demand,

performance and monitoring of teacher quality. From the point of view of

expenditure responsibilities, the federal level has not been able to do away with human

resource management or micro-monitoring earmarked transfers for payroll or supplies. In

the case of education, most states keep parallel, separate (federal, state) teacher

administrations. In the case of health, different degrees of dual (federal, state) systems –

created or consolidated through ad hoc measures— offer different service quality and

coverage to one and the same population. Confusing division of responsibilities weakens

accountability and government responsiveness and deters potential local or regional

ownership.

Decentralization in the public health system has been more successful than ineducation. Like in other Latin American countries, the health system in Mexico hasseveral subsystems. The most important player is the Social Security Institute (IMSS)which until the mid 1990s had the pension system and the formal sector employees healthcare under its tutelage. Mexico underwent a pension reform and most of the pensionsystem has been privatized. Today the IMSS covers virtually all formally employedpeople in the country. The ISSSTE is a similar social security institute for publicemployees. The defense, the navy and the national oil company, Pemex, also have similarinstitutions. However, it is worth noticing that neither of the latter institutions have beenreformed. They are highly centralized, and still combine health and a pay as you gopension system both of which run large deficits.

The second major health delivery system is administered by the Ministry ofHealth (Secretaría de Salud –SS) and it covers the population not covered by any of thesocial security institutes above. The idea is that SS caters to poorer segments of thepopulation. The payroll of nurses and doctors within the national health system are

46

decentralized in a fashion similar to that discussed for the case of education. States hadbeen reluctant to take this responsibilities in the early 1990s, but the federal governmentsuccessfully negotiated the deconcentration of health when budgetary item 33 wascreated at the end of 1997. The private sector provides service to those who can afford it.The three systems operate with little coordination.

The IMSS has implemented an impressive decentralization program whichdelegates funds, tasks, and decision power to an increasingly greater number of regionaloffices and administrators. States have spending autonomy of health funds as long asthey meet the minimum federal standards. The problem is that the standards are not clearor easily verifiable. Besides, health transfers do not have incentives to increaseefficiency.

With a few noteworthy exceptions, the country still expects to see the benefits of

decentralized administration of social services. Although politicians may still claim credit

for the country’s growing share of social expenditures at the regional or local levels, the

truth is that there is little or no connection between decentralization and higher coverage

or quality of service. Neither can it be said that current subnational expenditures in health

or education are more efficient than previously centralized expenditures. Moreover, it

would be very hard of a state or local government to claim political credit for better

management or enhanced performance. Citizens are generally apathetic to who is

effectively administering the service. The federal level –although still responsible for

social policy—finds it difficult to monitor or stimulate subnational governments. States

find that accepting responsibility for social services not only leads to cumbersome

administration but elevates their fiscal and political dependency. No wonder some states

like Oaxaca in the south or Aguascalientes in the center of the country, have already

offered to give back social services to the federal government. States now have to

administer responsibilities whose costs and production functions are largely determined

by the federal level. Although states can in theory select and appoint teachers, the whole

civil service system for education is in practice so rigid that states have little or no saying

in the incentives or performance evaluation of teachers.

As a matter of fact, if social sector decentralization is to advance in Mexico, the

country will have to first learn to deal with the transfer of personnel from the national

level to subnational levels of government. Similarly, if the country is to ever assess the

benefits and costs of social sector decentralization, it will be necessary to first ensure that

47

there is clear division of labor and a given level of government can be held individually

accountable for specific tasks of sector management. In sectors that are as labor intensive

as education or health there can be neither subnational government responsiveness to

citizens’ demands nor increased service benefits from decentralization unless personnel

management is fully transferred to subnational governments. In practice, the confusing

mix of deconcentration and decentralization in Mexico’s social services means that

neither the federal nor the state governments can gain –political credit or otherwise—

from hiring and firing decisions or granting performance-based results to teachers.

At first glance, progress in the decentralization of water and sanitation services

has been significant. The process of water decentralization began in 1991 when the

National Water Program transferred responsibilities form the federal government to the

municipalities. 118 out of 135 cities with populations of over 50,000 have autonomous

water utilities, for a third of them municipal governments set water tariffs and for the rest

the state congress does.14 However, this does not mean that water utilities are well

managed. In fact, water utilities are often an important source of income of municipal

governments and therefore the administration is subject to municipal politics; coverage,

leakage, quality, and efficiency varies tremendously.

More importantly for this study, decentralized provision of water and sanitation is

also plagued with confusion and uncertainty. However, the way confusion prevails over

clear division of responsibilities in water and sanitation is different from the ambiguity in

social sector decentralization. In the case of water, the federal level is clearly responsible

for collecting, transporting and selling “block water” (big volumes of water). As a

general rule, the federal government sells water to the states and the states sell water to

municipalities and other local providers. The source of the problem is not the formal

division of responsibility but the lack of enforcement of price for water. The federal

National Commission of Water (Comisión Nacional de Agua, CNA) does not enforcewater payments, thereby leading states to do likewise. In turn, local providers feel little

incentive to effectively apply water charges to final users. Poor cost recovery practices

lead to substantial subsidies at all levels of government. Since perverse incentives are

14 Source, World Bank 2000. Mexico City is a specially complex case where even the national legislaturehas some say in the administration of the water utility.

48

transmitted from the top down, no level of government is effectively accountable for

under-financing –and eventually under-provision or exhaustion of sources—of water.

Overall, there are few if any responsibilities that are fully and exclusively

assigned to states or municipalities. Distribution of responsibilities according to the law is

confusing by itself (see Table 4.1); in practice, the division of labor among levels of

government is even more complicated. In practice, responsibilities cannot be easily

identified or verified. Therefore it is easy for different levels of government to pass the

blame to each other.

Table 4.1: Mexico´s responsibility assignment.ExpenditureFunction

Federal Government State Governments Municipal Governments

Defense 100%Police Federal and border corps.

Federal transfers to stateand municipal policy

Special policy (concurrentwith Federal)State public order andsafety.

Local Policy Corps

Foreign Affairs 100%Labor Policies 100%Monetary andFinancial Policies

100%

Foreign Trade anddiplomatic relations

100%

Mail andTelecommunications

Along with privateproviders

Transportation Most, railroads, ports andairports, have beenprivatized.

Some airports Local publictransportation.

Roads Federal Highwayconstruction andmaintenance (C&M).Rural road financing

State feeder Roads C&M.Maintenance of secondaryfeeder roads.Implementation of ruralroad development

Local streets.

EnvironmentalStandards

National StandardsBiosphere reserves,national monuments andparks

State can adopt their ownstandards, complying withthe federal onesNational Parks

Land use permitsLocal parksSolid waste

Education(massivedeconcentration)

Policies, programs, andnormsEvaluation and audits ofsubnational performanceFinancing throughtransfers (Ramo 33)Labor relations, wage-setting and most training.Federal technicalinstitutes and universities

Operation, administrationand implementation ofmandates and programsMaintenance (concurrent)ConstructionHalf technical schools andsome universities.

Minimal role, schoolmaintenance and someconstruction.

Health Policies and normsthrough the social

Primary care for ruralpopulation and urban

49

security institute.Financing throughtransfers in Ramo 33.Most capital structuredecisions.Evaluation and audit ofsubnational performance.Labor relations and wagedeterminationSecondary and tertiaryhospitals.

poor.Some financing.Administration ofprograms and self-evaluation.Epidemiology andpreventive careReproductive health.

Water andSanitation

Water supply and sewage(concurrent)

Garbage collectionWater supply and swage;some water systems havebeen privatized butmunicipalities retain debtliability.

Source: World Bank 2000 and authors information.

50

5. The third phase: The Politics of Transformation of Social Investment FundsAs part of the macroeconomic stabilization and economic reform process, Mexico

and most of Latin America, experimented with one form or another of social investment

funds. The original idea was that the least well off would bear the heaviest burden of the

economic reforms. In order to alleviate poverty and maintain social cohesion, central

governments created social investment funds that bypassed the regular government

structure and made exceptions to ordinary disbursement and control procedures,

channeling large amounts of funds directly into “social infrastructure investments” that

would benefit the poor. In the case of Mexico, bypassing ordinary procedures only

became a problem with new democracy. Although special funds were targeted to poor

communities, the national government –frequently the presidency itself—kept influence

and control on the funds menu of eligible expenditures as well as individual allocation

decisions. Since no fund of this sort has effectively prevented or eliminated discretionary

government decisions, funds were at odds with decentralization on at least two counts: i)

social investment funds became a way to distort congress approved compensation

formulas and horizontal equalization goals, and ii) funds frequently financed

responsibilities that were being transferred to subnational governments, thereby

debilitating subnational fiscal responsibility and inter-government accountability.

In Mexico, the initial program began under the Salinas administration and was

called Programa Nacional de Solidaridad, Pronasol.15 Although the program was

relatively successful –inasmuch as it reached forgotten communities with social services

and infrastructure—multi-party competition and the pressure for clearer decentralization

rules moved congress to transform the Solidaridad funds into earmarked, formula-baseddiscretionary transfers. Some of the resources from the previous Solidaridad programwere allocated to Progresa; but most of those resources went into the Fondo deInfraestructura Social (FIS).

Since the mid 1990s, before the PRI had lost its majority in the Chamber of

Deputies, the federal government began moving in the direction of achieving greater

transparency in the allocation of resources to subnational governments. In 1996 the bulk

15 As discussed before, the Pronasol funds were initially included in the federal budget under Ramo 26.Beginning 1998, when those funds were transformed into formula-based earmarked transfers, they wereincorporated into Ramo 33.

51

of resources from chapter (or Ramo) 26 of the budget were transformed to a formula-based, poverty targeted transfer with the creation of the Fondo de Desarrollo SocialMunicipal, FDSM. Still, 32 percent of the funds remained allocated on a basis other thanpoverty—due to a peculiar compromise whereby each state was assured 1 percent of the

funds, regardless of its poverty levels. Moreover, distribution of the FDSM resources

from states to municipalities followed a different formula, which heavily weights

population –notwithstanding the fact that the distribution of the federal transfer to the

states was calculated on a poverty-based formula that could be readily extended to and

applied at the municipal level.16 Those “adjustments” to the sophisticated poverty

formula utilized for most of the transfer from the federation to the states can only be

understood as concessions granted to state governments.

In any case, the process of transformation of social investment funds became

more aggressive since the introduction of the FDSM, in the 1997 budget, as primarily a

formula-based, poverty targeted transfer. This process continued in 1998 with the

incorporation of Education, Health, Infrastructure and some other formula-based transfers

into the Ley de Coordinación Fiscal, LCF. However, Fortamun was also created in 1998as a broadly defined earmarked transfer that could elicit support –and overcome political

resistance-- for budget approval. The tendency to add broadly defined transfers continued

in 1999, with the creation of the Public Safety fund, and in 2001 with the introduction of

a special fund for capacity strengthening at the state level.

The 1998 budget gathered previously-dispersed earmarked transfers and special

funds under budget chapter (or Ramo) 33. Education, Health, Social Infrastructure andother pre-existing federal transfers were then incorporated into the LCF. Transfers under

Ramo 33 were denominated aportaciones..17 Although the aportaciones of Ramo 33 arecommonly thought of as formula-driven, the fact is that distribution of the largest

transfers –those for health and education-- obey more the supply-driven historical

distribution of social expenditures in the federal budget than an explicit territorial

compensation or demand-based subsidy formula. More important for our account, an

overlooked aspect of that budgetary process was the creation of a new fund, the Fondo

16 For the best discussion and analysis of this process see Mogollon, 2002.17 See Giugale and Webb (2000) and the references therein.

52

para el Fortalecimiento de los Municipios y el Distrito Federal (Fortamun) into the Leyde Coordinación Fiscal. Fortamun was created to elicit support from the opposition

parties for the approval of the budget. A smaller fund was originally considered in the

presidential bill as part of the strategy by the Finance Ministry to help states in their debt

overhang. According to the original Executive design, the new fund was not going to be

allocated by the SHCP according to debt conditions in each state, but on the basis of

equal per capita terms. After legislative discussions, the size of Fortamun was tied to the

evolution of federal revenues, its permanence was guaranteed by its inclusion in the LCF

(which requires Senatorial approval to be modified), and --although the law states that it

should preferably be allocated for purposes of debt reduction—it can be allocated in

practice to a wide range of purpose, as selected by each individual state.

As can be seen in table A, the legislature made modifications which were

consonant with the logic of the argument we have made. The per capita allocation of

funds for the strengthening of municipalities and the Federal District was increased by

almost 25%. The other feature to note in the table is that the adjustment to various funds

was not related to the dynamics of federal revenues. The adjustment in health and

infrastructure, for instance, went beyond the change in the Recaudacion FederalParticipable. As a matter of fact, from 1998 the monetary amount of the (newly grouped)aportaciones follows a different curve from that of the participaciones. While calculationof the latter depends strictly on the dynamics of federal revenues (since their formula

refers to the Recaudacion Federal Participable), calculation of the amount of the

aportaciones is primarily a function of year-to-year political negotiations between the

Executive, the Legislature and the federal entities (states and the Federal District) at the

time of budget discussions. Ever since, the aportaciones became the wild card to

introduce additional transfers that are not formula based –often not even related to a

transparent criteria. The 1998 budget contained an additional concession to the states,

which was a provision that the Impuesto Sobre Automoviles Nuevos (ISAN), which ispart of revenue sharing, would be fully administered by the states.

53

Table A. Modifications of selected decentralized budgetary items (millions) 1998ApprovedBudget

ProposedBill

Percentagechange

Participaciones 112,403.0 113,438.6 -0.9%Fondo de Aportaciones para laEducación Básica y Norma 67,512.6 67512.6 0

Fondo de Aportaciones para losServicios de Salud 10,546.2 10,808.9 -2.4%Fondo para la Infraestructura SocialMunicipal 9142.3 9,262.6 -1.3%Fondo de Aportaciones para elFortalecimiento de los Municipios ydel Distrito Federal 6732.1 5,400 24.7%

The 1999 budget process saw the creation of an additional transfer to the states,

the Public Safety Fund. While that fund was meant to be formula-based, the criteria used

for the allocation among states did not become clear until some years later. On the other

hand, in order to generate an increase of the Fortamun funds allocated to all states,

congress decided to eliminate the participation of the Federal District in Fortamun. This

was not dissimilar from strategies pursued in the past by the federal government, which

has sometimes used the DF as the federal entity that can bear adjustments that benefit all

states –without generating political turmoil. As discussed under Section 2 above,

differential treatment of the Federal District illustrates the difficulties in constructing a

political coalition that includes the District, the State of Mexico and the State of

Veracruz.

The 2000 budget presents additional evidence as to why we interpret the

adjustment of 1998 as politically motivated, with the intention of garnering the support of

the states for budget approval by the legislature. The adjustments in the revenue side, as

can be seen from the participaciones row in table B, increased funds available by half apercentage point. This same adjustment is observed in the funds for social infrastructure

and Fortamun, which, by legal mandate, have to keep strict correspondence with federal

revenues (the Receta Federal Participable). However, education funds were not adjustedin that proportion, while health took a disproportionate adjustment.18 The funds for public

safety, now firmly embedded in the structure of aportaciones were kept at the presidential

18 The figures for education include the Federal district, which in the presidential bill was still under adifferent budgetary item.

54

requested level. Hence, the 2000 budget confirmed the 1999 precedent as to how fiscal

transfers could be adjusted to please subnational governments (and corresponding

political interests) in the years the states are prepared to negotiate –rather than take the

federal government hostage during the budgetary battle.

Table B: Modifications of selected decentralized budgetary items (millions) 2000

Participaciones 161712.8 160,883.3 0.5%Fondo de Aportaciones para laEducación Básica y Normal 118,404 118,198.7 0.2%Fondo de Aportaciones para losServicios de Salud 20,022.7 20,262.1 -1.2%Fondo para la Infraestructura SocialEstatal 1,937.9 1,927.9 0.5%Fondo para la Infraestructura SocialMunicipal 14,051.7 13978.7 0.5%Fondo de Aportaciones para elFortalecimiento de los Municipios 15,030.3 14,952.1 0.5%Fondo de Aportaciones Múltiples, quese distribuye para erogaciones de: 5,206.2 5,179.2 0.5%Fondo de Aportaciones para laSeguridad Pública de los Estados y delDistrito Federal 5,170.0 5,170.0 0.0%

The 2001 budget was discussed in the midst of cautious behavior on the part of

legislators, since most politicians were eager, even if the PRI lost the election, to prevent

a crisis like the one that occurred at the end of 1994. The budget witnessed the creation of

a fund to strengthen states, a concession that was quickly adopted by the federal

government, notwithstanding that the rationale for that fund was rather unconvincing. In

2002 the state fund was retained while the provision for equal shares of 1 percent of FIS

were eliminated. However, the most influential factor in determining the actual

allocations for the aportaciones in the 2002 budget was the shortfalls of the federalbudget. Since congress severely reduced the revenue expectation of the tax reform

submitted by the Executive, it became necessary to downsize the aportaciones tomaintain the fiscal targets (see Table C below). Once again, the aportaciones appeared to

be a critical residual factor for budget (and, in this case, tax reform) negotiations.

55

Table C. Modifications of selected decentralized budgetary items (millions) 2002

Approved BudgetProposedBill Amendment

AportacionesFAEB 145,424.8 146,182.0 -0.5%FAS 25,313.9 25,758.8 -1.7%FIS 19,064.1 19,729.5 -3.4%FAFMDF 19,539.1 20,221.2 -3.4%FAM 6,231.1 6,423.9 -3.0%FAET 2,781.5 2,822.8 -1.5%FASP 5,594.8 3,000.0 86.5%PAFEF 1,280.7 0 100.0%

Participaciones 196,914.4 207,087.6 -4.9%

The above summary of the recent evolution of the aportaciones reflects how thetransformation from largely discretionary fund management to transparent, formula-

based transfers has followed two contradictory forces, each one of them trying to give

final shape to the Ramo 33 transfers. On the one hand, the administration seeks totransform an increasing share of those funds into transparent formula driven transfers; on

the other, year after year there appears to be a need to “invent” another fund which can be

assigned to states or municipalities following loosely defined criteria --regardless of their

poverty conditions or infrastructure needs.

A common perception in Mexico is that the current balance of political forces

(including traditional political culture) appears to prevent the full transformation of the

various social investment funds transferred to states and municipalities into incentive

compatible instruments that reward effort. In a nutshell, opposing political forces

working within the federal legislature have both partially transformed and prevented full

transformation of the funds that finance social infrastructure in states and municipalities -

-inherited from the macroeconomic stabilization period-- into budgetary items subject to

allocation and disbursement rules capable of guaranteeing effectiveness, efficiency,

accountability, evaluation and control. As discussed in Section 2 of this paper, the reason

behind the apparent schizophrenic behavior of the federal congress is that it is in the

congressmen’s interest to negotiate unconditional resources to their states on an

56

individual basis rather than seeking to establish rules that would allocate resources based

on fiscal effort or expenditure efficiency. Since the legislature has the power to hold the

central government hostage during budget negotiations and approval, the central

government uses those funds to “buy” the votes of different congressmen. The problem

is compounded because, since congressmen cannot be re-elected, the partisan coalitions

that have passed the budget in that last five years have been little concerned with their

constituent jurisdictions.

It is true that the transformation of social investment funds took place with

surprising speed and smoothness in the case of Mexico. Other Latin American countries

(including Bolivia, Colombia, the Dominican Republic, Ecuador, Peru and most Central

American countries) have been forced to either postpone or entirely forget the

transformation of the funds by the funds own built-in interest and apparent delivery

capacity. However, resistance to substitute decentralization transparency for centralized

patronage has not withered away in the case of Mexico. Year after year, at the time of

budget negotiations within and between the Executive and the Legislature, there appears

and reappears the claim for new, broadly defined funds that add complication to the

already weak picture of subnational accountability and control.

The implication of this hypothesis is that “economic solutions” to the

transformation of FIS and other social funds will not work in the medium or long term

until the political forces (including political incentives) underlying current legislative

decisions are properly dealt with. This is easier said than done. Each country has an

idiosyncratic budget process with legislatures that have different incentives. In Mexico,

the link between the budget process, the social investment funds, and the incentives of the

legislature have been largely missed by studies on decentralization.

57

6. The fourth phase of Mexico’s recent decentralization and federalization policies:policy options towards strengthening monitoring, control, and evaluation ofsubnational spending responsibilities.

Previous sections have illustrated, i) the phased-in way Mexico developed substantial

vertical and horizontal imbalances that are compounded by confusing division of

responsibilities, ii) why it appears to be politically unlikely that Mexico can undertake a

comprehensive reform of decentralization or fiscal federalism in the short term, and iii)

current, timid and patchy measures towards re-establishing vertical and horizontal

balances in Mexico.

Aware of some of the above problems with decentralization, the country has for sometime entertained the idea of a comprehensive overhaul of fiscal federalism. However,political and technical difficulties have deterred any sustained effort to this effect. Thetechnical obstacles for introducing a technically sound distribution of revenue sources,transfers and responsibilities are formidable. There is only limited tradition of accountingbudgeting and reporting standards at subnational levels. Besides, the country does notappear to be politically prepared for either a comprehensive or an overnight reform of thecountry’s fiscal federalism. Before pushing further with the design, enactment andimplementation of a sophisticated federalist model tailored made to Mexico, the countryneeds to continue getting rid of a culture of centralization that is rooted in political andfiscal dependency mechanisms, many of which are still in operation.

Although Mexico has never followed a planned path towards decentralization19,

pragmatic decisions over time have lead to frequent course adjustments. When analyzed

ex-post, four main turns can be clearly identified as the main phases of the Mexican

process of decentralization during the last 20 years --each phase characterized by the use

of a particular set of technical instruments that are more or less adequate to reaching the

specific goals of Mexico’s decentralization at a particular point in time. Each one of the

four phases of Mexico’s decentralization has broadly followed changes in the broad

political context at each point in time. When necessary, governments have sacrificed

technical purity in adjusting decentralization goals and tools to political pressures that

appear to be of overall concern at a given time. It has only been recent, during the last

19 As a matter of fact, it is only ex-post that the four phases described above can be reconstructed in asequential, manner.

58

two phases of Mexico’s pursuit of decentralization, that sound fiscal and management

guidelines --like elimination of discretionary transfers or enhanced accountability—

overlap with government styles and political agendas20. Besides the broad political

context, increasing technical refinement in the design of decentralization framework has

also been stimulated by governments’ own evaluation of the process of decentralization

as well as lessons learned from countries going through similar waves of decentralization.

This appears to be the case, for instance, with Mexico’s prudent subnational borrowing

policies enacted in 2000-2001. Still, political barriers often demand enactment of policies

or instruments –as in the case of the reform of social investment funds-- that go in

opposite direction to sound decentralization policies.

Since goals of and approaches to decentralization, and selected technical

instruments, were ordinarily conditioned by the political context, it is no surprise that

instruments often missed the proposed decentralization target. However, it is the

cumulative effect of the four phases of decentralization reform, --more so than the

particular set of measures and instruments adopted at each phase in time-- that makes the

current decentralization framework inconsistent and hard to reform on a piecemeal basis.

As many decentralization experts have indicated in the past, the Mexican system is so

confusing and it sends so many inconsistent messages to decentralization stakeholders

that the best the country can do is to, first build consensus towards a Mexican model of

federalism and decentralization and begin taken steps in that direction, then enact an

overhaul reform of fiscal federalism and overall distribution of revenues and

responsibilities among levels of government.

The problem with the proposed comprehensive reform is twofold: i) the political

mood of the country does not seem prone to such a major political consensus, and ii)

20 It must be recognized that many Latin American countries preferred general revenue sharingarrangements similar to those adopted by the Mexican Fiscal Pact of the 1980s. This was, for instance, thecase with Brazil, Colombia and Venezuela during the late eighties. However, none of those countriesdeprived so drastically the intermediate level of government of their own tax bases. On the contrary, theBrazilian constitution of 1988, the Colombian tax reform of 1983 and the Venezuelan Decentralization Lawof 1988, each one of them expanded the regional tax base.It must also be admitted that some Latin American countries initially pursued sector decentralization in asimilar vein to the way Mexico did in the nineties. This was, for instance, the case with Brazil’s return todemocracy in the late eighties and the Colombian first phase of decentralization reform of 1986-87.However, Colombia soon changed course towards a more gradual, negotiated transfer of sectorresponsibilities under the Constitution of 1991 and law reform of 1993. Venezuela enacted a negotiatedapproach to sector decentralization since the country first enacted decentralization in 1988.

59

current management capacity at each level of government does not appear to match the

necessary requirements of a new inter-government system in terms of roles and

responsibilities, enhanced autonomy and efficient coordination among levels of

government. Since no political consensus on comprehensive decentralization reform

would be feasible unless levels of government and inter-government relations are ready

to implement it, the rest of this section will deal with the second set of obstacles to

streamlining decentralization reform in Mexico.

By the beginning of this millennium, Mexico’s decentralization is only partial,

incomplete at best. Relations between the federal level and subnational governments are

plagued by substantial vertical imbalances, confusing division of responsibilities and lack

of clear accountability, monitor and control mechanisms. From the point of view of

distribution of revenues, the fiscal federalist pact that began in 1980 and is still going on

–though reformed several times-- severely limited tax capacity and crippled potential tax

effort at the state level. Sector decentralization initiated in the early nineties still lacks a

well defined incentive framework for subnational efficiency in key sectors such as

education, health, water or roads. Formula driven transfers that are broadly targeted to

earmarked purposes, and elimination of most federal discretionary transfers –as adopted

in 1998-2000, although a step in the right direction, fell short of effectively transforming

broadly defined transfers or enhancing subnational accountability and responsiveness. In

sum, Mexico is still far away from a decentralization framework that meets the most

essential principles of decentralization, including vertical balances, clear division of

responsibilities and accountability, and inter-government coordination for efficiency in

service delivery. This section discusses the federal government current approach towards

re-establishing vertical and horizontal balances and enhancing accountability and hard

budget constraint.

Seen in a medium to long term perspective, Mexico appears to be entering still one

more phase of the federalism/decentralization process since the beginning of the present

decade. Although not quite explicitly formulated, Mexico seems to pursue these days

two, mutually reinforcing purposes: i) transparency, monitoring and control of

decentralized spending, and ii) fiscal discipline and hard budget constraint at state and

local level. For the latter purpose, the Fox’s administration is struggling to enforce the

60

new market-oriented framework for subnational borrowing inherited from the Zedillo

administration. For transparency and control, the preferred instrument of the new phase

appears to be performance-driven (conditioned) transfers that pursue strategies and

results identified by subnational governments within a wide menu of sector options.21

Subnational governments have to meet those conditions either as a pre-requisite

(eligibility requirements) for receiving a grant or a loan (eligibility criteria) or during

execution of grants or loans. Different from the general revenue sharing agreements of

the early 80s, the confusing deconcentration/decentralization sector transfers of the early

nineties, or the --largely uncontrolled-- formula-driven transfers of the late nineties, the

federal government is now trying to balance revenues and responsibilities by

emphasizing result-conditioned transfers.

During the last three years the federation and some of the states have beenactively pursuing expenditure coordination schemes that combine federalresources with federal transfers to subnational governments, includingparticipaciones and aportaciones. Matching grants programs (commonly knownas pari passus) have been expanded –particularly for poverty alleviation purposesthrough subsidized programs in the areas of agriculture, nutrition, health andemployment creation. The previous Zedillo administration initiated theFORTEM program geared to capacity building at subnational levels via acombination of conditioned loans and subsidies. The Zedillo administration alsoexperimented with inter-government performance agreements for health or theenvironment.The Fox administration has supported innovative ways to funnel participacionesor aportaciones to guarantee inter-government payments for water. It has alsofurther expanded the search for a more balanced package of fiscal transfers –onethat gives more weight to transparency, monitoring, evaluation and controls andless weight to transfers that are either freely disposable or hard to evaluate interms of allocation or production efficiency. To mention but a few examples ofthe Fox’s administration move in this direction: i) current delegation and ortransfer of agriculture responsibilities (including human resource management)through inter-government negotiated agreements, as in the case of the federal-Coahuila agreement; ii) ongoing reform efforts to strengthen budget-resultindicators for subsidized programs jointly administered between the federation,the states and municipalities; iii) SEDESOL-coordinated matching grants programfor water, electricity and rural roads targeted to isolated indigenous communities,

21 Enhanced transparency and accountability is also being pursued at the federal level. The National Planfor Transparency and Anti-Corruption as well as myriad of individual financial restructuring, processsimplification and information technology strategies are being planned or implemented throughout thefederal level. As a matter of fact, the National Plan for Transparency and Anti-Corruption is also beingextended to subnational governments via inter-government performance agreements.

61

and iv) SECODAM is extending transparency and anti-corruption tools andstandards to subnational governments via inter-government performanceagreements.

The present Fox administration appears to be convinced that Mexico should not

keep expanding fiscal transfers while no level of government is clearly accountable for

service delivery. More than additional fiscal transfers or decentralization of new

responsibilities –as frequently demanded by state governments or opposition parties—

the federal government is struggling to reorganize the patchy, often inconsistent structure

the country has built for over two decades. Effective implementation of this fragmented

decentralization framework appears to be more realistic than immediately adding still one

more layer to the pile of distribution of revenues and responsibilities. At this point in time

in the evolution of Mexico’s decentralization, ensuring decentralization meets the basic

principles of clear accountability and efficiency enhancement appear to be more

important than pursuing the politically and fiscally uncertain path of the reform of fiscal

federalism.

The current phase can be seen as imposing stricter federal government controls

and –in that regard—reinforcing a pyramidal, center-based public sector. However, it is

more appropriate to interpret this phase as an inevitable step towards capacity building

for efficient service delivery at subnational levels and enhanced inter-government

coordination. There is no way Mexico can keep adding fiscal transfers to subnational

governments --as frequently demanded on primarily ideological grounds-- unless citizens

perceive substantial progress in subnational fiscal responsibility, clear division of

responsibilities and sector performance and results.

Short of a comprehensive reform of federalist relations, what Mexico appears to need themost, at this juncture of the country’s process towards decentralization, is a transitionphase that helps prepare the ground for the fiscal federalism reform. The focus of such atransition phase is to enhance government accountability, transparency and expenditureassessment at all levels of government. However, the country does not appear to beprepared to adequately implement the instruments selected for the transition phase. Inter-government performance agreements are a politically delicate instrument that requirereliable and precise information and inter-government reporting channels the countrydoes not appear to have at this point in time. Unless the country adjusts instruments towhat is doable in the present context, the current search for meeting the most basic

62

premises of decentralization may end up by adding frustration and further weakening thecredibility of the decentralization process. Need to be more specific and a bit less cryptic

The priority presently given to the purpose of strengthening mechanisms for

checks and balances can hardly be challenged from a technical point of view. It appears

to be the only way to guarantee Mexico’s original intention of elevating expenditure

efficiency via sector decentralization. It is also a way to prevent disillusion and

frustration that may eventually arise re-centralization forces –such as currently seen in

other Latin American countries. Rather than its purpose, the problem with the current

approach lies in the primary tool selected for enhanced transparency and accountability.

The main instrument is negotiated transfers that incorporate specific performance, output

or investment related benchmarks. Government financial agencies (development banks

such as Banobras) should play a critical role in promoting, signing, monitoring and

enforcing contracts with interested states or municipalities. This “incomplete contract

approach” (as those performance and result-oriented contracts are known in the fiscal

decentralization literature) would help move Mexico into the realm of disbursement

based on compliance with the agreed benchmarks. However, federal financial agencies

do not appear to be equipped to effectively play their new role. Furthermore, it is hard to

think of any other federal agency that counts on the necessary information and leverage

power to adequately identify appropriate base-line or performance or result indicators for

each individual contract. Lack of the necessary information will probably weaken the

credibility of federal agencies as capable monitoring and enforcing agencies. It must be

taken into account that inter-government performance agreements –even when smoothly

developed— are politically sensitive by themselves.

An additional problem with the current choice of instruments for transparency

enhancement is that those instruments aim at solving two different decentralization

challenges that can hardly be combined into one single set of tools. One is the challenge

of minimum fiscal and institutional standards subnational governments must meet in

order to ensure efficiency and accountability in service delivery. The other one is the

challenge of reaching sector specific performance and/or results. Relying on a single set

of instruments (inter-government performance contracts) for reaching the two purposes

is extremely difficult to do. International experience tells us that whenever those two

63

targets have been intended with one single shot, the result is either weak monitoring and

enforcement or a heavy burden of conditions that debilitates the incentives for

subnational governments to sign those complex incomplete contracts. In either case,

Mexico would not be building the necessary trust and transparency to elevate inter-

government relations to a higher plateau of efficiency in decentralized service delivery.

The other purpose of the current phase of decentralization is to effectively enforce

fiscal discipline and hard budget constraint. In this case, the country appears to have

selected an adequate, powerful instrument capable of reaching the intended objective.

Indeed, during 2000-2001 Mexico pioneered a market-oriented reform that minimizes

(federal) moral hazard, stimulates market enforcement of fiscal sustainability and

provides transparency and disclosure mechanisms. The 2000-2001 reform eliminates the

participation of the federal government in guaranteeing subnational debt. The risk is now

borne by creditors that are required to make proportionate capital provisions according

the indebtedness capacity of the borrowing subnational entity. Indebtedness capacity is

measured by independent credit rating agencies. As expected, the rate of interest will be

related to the borrower’s and/or the loan risk. Rather than the federal government,

subnational governments will provide guarantees of their own.

As international experience has demonstrated time and again, market-oriented

subnational borrowing regulatory frameworks similar to the one recently adopted by

Mexico requires a strong political will. When subnational governments find that

accessing capital markets is more demanding than in the past, they tend to put the

pressure back on the federal government to intervene with development banks for

relaxing the new regulations. Although the new system is still too young to predict the

likely reactions of the federal government to growing subnational pressures, already

available evidence indicates that the government is firmly behind the new regulatory

framework –even at the cost of feeding additional political opposition from over-indebted

subnational governments. This is not an insignificant political cost at a time government

and congress or political parties appear to be at odds over a number of fiscal and non-

fiscal issues.

In sum, Mexico’s current quest for transparency, efficiency and accountability in

decentralized service provision is unlikely to lead to completion of the most basic

64

decentralization pre-requisites or streamlining decentralization. The instruments being

used are just too weak or inadequate to re-invigorate decentralization and federalism

reform. The remainder of this section will be devoted to discussing what are the possible

options for further advancing fiscal federalism and decentralization reform in today’s

Mexico. Given today’s political and technical constraints in Mexico, this question can be

properly rephrased as how can Mexico reinforce decentralization and minimize the risks

of future fiscal federalism reform by resorting to finely-tuned incentives that have the

capacity to begin building blocks towards more comprehensive reform of fiscal

federalism?

As stated above, the federal government is presently seeking to enhance

transparency, accountability and efficiency in subnational government expenditure. In its

pursuit of accountability and efficiency, the federal government has had to accept two

significant constraints: i) a radical redistribution of tax revenue sources may have to wait

for some more years to come, and ii) some key sector responsibilities (such as water,

education or health) have already been partially transferred. Given those two constraints,

the federal government’s best chance to achieve decentralization efficiency and

accountability is to resort to inter-government fiscal transfers and borrowing regulations

as incentives capable of stimulating the required minimum standards as well as fiscally

sustainable service delivery.

Fiscal transfers and borrowing requirements are indeed being used by the federal

government to reach three purposes: i) common minimum accounting, budgeting, and

reporting standards, ii) sound subnational fiscal and financial management, and iii) more

clearly defined sector responsibilities. However, the tools being used by the federal

government have only limited leverage. On the one hand, there is a problem with the size

of the incentive tools the federal government is currently managing for inter-government

performance agreements. Since the two biggest types of fiscal transfers –the

participaciones and the aportaciones—are not being conditioned to enhanced subnationalperformance, additional conditionality is being introduced through ad hoc inter-

government arrangements regarding matching grant or subsidized (pari passu) programsand special credit lines managed by federal development banks. Up to now, the federal

government has not yet intended to add accountability and performance requirements at

65

the time of annual budget negotiations regarding the –potentially powerful-- earmarked

aportaciones.On the other hand, minimum standards and performance conditionality are being

required in a combined way that adds a heavy burden –and debilitates the incentive

effect— to lending and grants instruments. Were Mexico to insist on adding a heavy load

of requirements on relatively weak instruments, the country may learn the hard way some

of the lessons already learned by other Latin American countries. In fact, Latin American

countries that have tried to combine both minimum standards and sector-conditionality

(performance or results) have failed to reach either goal. As demonstrated, for instance,

by the recent Colombia’s “certification process”, the political and fiscal pressure to

bypass or somehow accept less-than-minimum eligibility requirements is simply too

much for federal sector secretaries to resist when deciding on transfer of resources or

performance-required loans. This has also been the experience with multi-sector

minimum institutional standards for emergency investment funds. As a matter of fact, this

is precisely why sector investment funds are broadly deemed incompatible with multi-

sector decentralization.

Mexico should not include sector lending for responsibilities that are partially in thehands of the federal level. Most of the federally-subsidized programs, for instance,interfere with actual or potential subnational responsibilities --to the point that a givensector-result cannot be attributed to subnational sector borrowing. Therefore, currentmeasures need to incorporate mechanisms that clearly identify what are those results thatcan be effectively (not only in the law but in practice) attributed to subnationalmanagement --even at the cost of reducing the menu of options or otherwise slowingdown project preparation or borrowing negotiations between subnational governmentsand sector secretaries or development banks. By so doing, the federal level will indeedhelp further clarify the distribution of responsibilities among levels of government.Failing to do so will further confuse accountability in Mexico’s entire public sector.

In its quest for result-conditioned grants or loans, Mexico needs to separate

instruments for sector performance (such as investment requirements) from general

decentralization pre-requisites. The menu of sector investments that can be financed must

be limited to those responsibilities that are effectively managed by states or

municipalities. As demonstrated time and again, it would be very difficult, for instance,

to provide education loans to states and municipalities when there is so much federal

66

interference in the production of education services. As widely demonstrated by the Latin

American experience, a state may always argue, for instance, that a certain result or

performance cannot be met because collective bargaining at the national level forced the

state to reallocate education resources in a certain way. Selecting those responsibilities

that are effectively in the hands of subnational governments (and they can be held

accountable for) would probably demand fine, case by case sector identification work

Fine judgment of what can/cannot be effectively conditioned is well beyond the capacity

of federal development banks. Inter-sector and inter-agency coordination will be needed

to fine-tune incentives built-in in conditioned transfers or loans.

In order to meet the above needs, the federal government will need to align three

different instruments: the aportaciones, matching grant (pari passu) programs and specialcredit lines for subnational strengthening and inter-government performance agreements.

As negotiated and approved in next year’s budget, the aportaciones could graduallyincorporate eligibility requirements. The pari-passu, subsidized programs, could in turn

incorporate sector specific, negotiated goals. Special credit line would in turn reinforce

minimum accounting, budgeting and reporting standards –as eligibility requirements—

while further stimulating sector specific goals. Such a package of mutually reinforcing

instruments should gradually build the capacity at each level of government --and basic

inter-sector coordination— as it will be needed when the political climate of the country

allows for comprehensive reform of fiscal federalism.

67

7. Conclusions (even more preliminary)Mexico is in a deep transition in the decentralization process. The current

decentralization framework can best be understood in terms of four reforms

placed on top of each other in the last 20-25 years: (i) the centralization of taxing

powers in exchange for unconditional revenues with the introduction of the VAT

tax in 1980, (ii) the deconcentration of key sectors in the early 1990s the most

important of which were education, health and water; (iii) the creation of social

investment funds in the late 1980s and their subsequent transformation into

formula driven transfers in the late 1990s; and (iv) finally the fourth phase in

which the current administration is dealing with the current framework and

decentralization pressures through two instruments which appear too weak to deal

with all the objectives.

These key reforms created a confusing picture of responsibilities with no

clear responsibilities at any level of government fro key public services. In

addition, there are deep imbalances of revenue and expenditures at the subnational

level which are alleviated through transfers which are not providing the best

incentives for efficient tax collection, expenditures, or service provisions.

The conclusions of the paper are not all that optimistic. It appears that the

increased democratization in Mexico is not the most auspicious atmosphere for

improving the fiscal relations framework. The paper argues that the current

political institutions are unlikely to pass a far reaching reform of the

decentralization framework that would place México on a path to a more efficient

and equitable fiscal federalisms framework. Rather the best route appear to be

incremental changes which point in the right direction and are solid steps towards

improved efficiency and equity of revenue collection and expenditures at the

subnational level as well as incentive compatible transfers from the federal

government. The incremental path in México is common to other countries which

are also going through increased democratization. The important issues are to

have a clear view of what is the best decentralization framework for Mexico given

its political institutions as they are or as they have the possibility of changing.

68

References (particularly incomplete)

Casar, Maria Amparo (2000) “La LVII legislatura de la Camara de Diputados en Mexico.Coaliciones y Cohesion Partidaria” documento de Trabajo 113 CIDE

------- (1999) Coaliciones Parlamentarias: Una nueva forma de hacer politica”Documento de Trabajo 28 Fundacion Rafael Preciado Hernandez

Courchene, Thomas and Diaz-Cayeros, Alberto (2000) “Transfers and the Nature of theMexican Federation” in Giugale and Webb.

Diario Oficial de la Federacion (1998) Presupuesto de Egresos de la Federacion para elEjercicio Fiscal de 1998

Diario Oficial de la Federacion (1999) Presupuesto de Egresos de la Federacion para elEjercicio Fiscal de 1999

Diario Oficial de la Federacion (2000) Presupuesto de Egresos de la Federacion para elEjercicio Fiscal de 2000

Diario Oficial de la Federacion (2001) Presupuesto de Egresos de la Federacion para elEjercicio Fiscal de 2001

Diario Oficial de la Federacion (2002) Presupuesto de Egresos de la Federacion para elEjercicio Fiscal de 2002

Diaz Cayeros, Alberto, Rafael Gamboa & Fausto Hernández Trillo (2002) "FiscalDescentralization in Mexico: The Bailout Problem" Working Paper InteramericanDevelopment Bank.

Hernandez Chavez, Alicia (1998) La politica Economica de Mexico en el Congreso de laUnion (1982-1997) Mexico: FCE

Levy, Santiago (1998) “Análisis Metodológico de la Distribución de los Recursos en elRamo 33” in Comisión de Desarrollo Social Instrumentos de Distribución de losRecursos del Ramo 33 (México: Cámara de Diputados).

Linz, Juan (1990) “The Perils of Presidentialism” Journal of Democracy 1(fall).Linz, Juan and Valenzuela, Arturo (1994) The Failure of Presidential Democracy Vol. 1

Baltimore and London: Johns Hopkins University Press.Mainwairing, Scott (1993) “Presidentialism, Multipartism and Democracy: the Difficult

Combination” Comparative Political Studies 26(2).Mogollón, Olivia (1997) "Pobreza y Distribución de Recursos Descentralizados del

Fondo de Desarrollo Social Municipal" Memorias del Congreso Nacional deCiencia Política: Federalism, México: CNCPYAP.

Moulin, Hervé (1995) Cooperative Microeconomics. A game theoretic introduction.Princeton: Princeton University Press.

Nacif, Benito (2000) “Legislative Parties in the Mexican Chamber of Deputies” inMorgenster, Scott and Nacif, Benito (eds.) Legislative Politics in Latin AmericaCambridge: Cambridge University Press.

Scott Andretta, John (2002) “La otra cara de la reforma fiscal: la equidad del gastopublico” CIDE

Secretaria de Hacienda y Credito Publico (1998) Proyecto de Dictamen del Presupuestode Egresos de la Federacion para el Ejercicio Fiscal de 1998

Secretaria de Hacienda y Credito Publico (2000) Exposicion de Motivos del Proyecto dePresupuesto de Egresos de la Federacion para el Ejercicio Fiscal de 2000

69

Secretaria de Hacienda y Credito Publico (2002) Proyecto de Dictamen del Presupuestode Egresos de la Federacion para el Ejercicio Fiscal de 1998

Tsebelis, George (1995) “Decision Making in Political Systems: Veto Players inPresidentialism, Parliamentarism, Multicameralism and Multipartyism” BritishJournal of Political Science 289-325.

Ugalde, Luis Carlos (2000) The Mexican Congress: Old Player, New Power Washington:CSIS Press

Weingast, Barry R. (1979) “A Rational Choice Perspective on Congressional Norms”

Weldon, Jeffrey (2000) Voting in Mexico’s Chamber of Deputies, 1998-1999 Paperpresented at the Latin American Studies Association Meeting, Miami, March 16-18.

--------- (1996) “The Mexican Congress and the Presidency 1917-1940” Paper presentedat the Annual Meeting of the American Political Science Association, SanFrancisco, August.

--------- (1998) “Legislative Delegation of the Budget in Mexico: Assessing Legal andPartisan Constraints” Paper presented at the Boston, September 3-6.

American Journal of Political Science 23:245-62.

Willis, Eliza, Garman, Cristopher, Haggard and Stephan (1999) “The Politics ofDecentralization in Latin America” Latin American Research Review 34(1).

World Bank (2000) Achievements and Challenges of Fiscal Decentralization. Lessonsfrom Mexico edited by Giguale, Marcelo and Webb, Steve, Washington.