Federalism and Public resources in Brazil: Federal Discretionary transfers to States

This paper analyses intergovernmental relations in Brazil based on the dynamics of the distribution of federal discretionary transfers to the states between 1997 and 2008. The theme becomes more relevant when we consider the process of fiscal recentralization that has taken place in Brazil, particularly from 1994 on, and the importance of discretionary transfers for state budgets. Our purpose is to identify which factors can account for a higher or lower state’s share in overall resources, by building on two explanatory dimensions: the partisan-political and the social-redistributive. The first one examines how the political dynamic affects resource allocation; the second, if discretionary transfers exhibit a redistributive character across states. Our findings show that a governor’s partisan alignment with the president or with the president’s government coalition are important factors in determining higher resource allocation for his/her state, and that states that are overrepresented in the lower chamber are favored. We have also found a redistributive character associated with federal discretionary transfers, favoring states with lower Human Development Index (HDI) rankings.

Transferências Voluntárias da União (TVU), to the states from 1997 to 2008.From this perspective, this work is embedded in the debate about Brazilian fiscal federalism.
A main feature of the Brazilian federal system is the tripartite configuration of its political power.The federal government, the states, and the municipalities have governments of their own, and their own financial resources and administrative authority.Hence, understanding the Brazilian public sector entails understanding these three power entities and how they relate to each other.Intergovernmental fiscal relations, the so-called fiscal federalism, are a key feature of the configuration of the Brazilian State.To further our understanding of Brazil's fiscal federalism an analysis may be conducted on the distribution of public resources and expenses across governmental levels, on what defines the level of resource decentralization, and on how vertical relations (intergovernmental) and horizontal (national Executive and Legislative) bear on the federation's sharing of revenues and expenditures.
In this work we aim to explore one of the dimensions of fiscal federalism, Brazil's TVUs, which are, as aforementioned, discretionary transfers from the federal government to the state governments.These transfers represent one of the three main sources of funding of these subnational units, the other two being the taxes they levy and mandatory federal transfers as laid out in the constitutional charter.However, unlike the taxes they levy and the constitutionally-mandated transfers, which are revenues that are guaranteed to the states, federal discretionary transfers are within the competence of the federal government, more specifically of the national Executive branch, which has full freedom to set amounts and geographic destinations for these discretionary funds, in addition to the public policies that are to be benefited.We also know that tax collection across states is very unequal, evidently benefiting the richer states.From the opposite perspective, constitutional transfers have a markedly redistributive character, for they are aimed at minimizing the effects of the country's deep economic differences.But what can be said of the federal discretionary transfers?
We do not know how they occur; nor do we know the rationale guiding the allocation of such resources to the state governments.The aim of this article is to further the understanding of the relations between the governmental bodies of the federation.This is an important theme, particularly in the present context of severe fiscal constraints imposed on the states.The recent evolution of the Brazilian federalism has curtailed the spending freedom of subnational entities, while increasingly compromising their revenues with mandatory expenditures, especially in health, education, and the payment of public debts.If payroll expenses are added to that, for both active and retired staff, plus the cost of maintaining the public machine, the conclusion is that the state budget has scarce resources and very little leeway.Hence, discretionary transfers represent important additional resources for the states toward optimizing their public-policy execution capacity.What's more, these transfers also represent an important instrument of federal power in the central government's relations with the states and the national legislature, as legislators also seek to channel part of these resources to their constituencies.Two factors must be considered if we wish to understand the resource allocation dynamics of the federal government's discretionary transfers: the partisan-political factor and the social-redistributive factor.The first reveals that the partisan-political competition logic can influence the manner whereby the distribution of resources by the Union is defined.For example it is reasonable to assume that the federal government will privilege states whose governors or a large number of federal deputies belong to the president's power base.Under the social-redistributive perspective, we may expect poorer states to be granted more federal resources, as a measure to promote greater regional equalization.Determining which of these factors matter is the investigative task we have set ourselves to undertake.In order to attain such purpose, we have sought to analyze data on the federal discretionary transfers A third relevant factor for the analysis concerns the efficiency of the states in attracting funds.We may expect that better structured and more efficient states with regard to project design and intergovernmental negotiation should raise a bigger slice of the discretionary transfers.However, a lack of good indicators to that effect does not allow us to make a more accurate analysis of such factor at the moment.
The article is organized in four sections.In the first one we define the theoretical boundaries of the research subject, federal discretionary transfers, within the debate on the Brazilian fiscal federalism.The second section presents data on the apportionment of federal discretionary monies for the 1997-2008 period, highlighting total funds transferred and comparing their allocation across states and municipalities, and showing their importance in the composition of each state government's revenues.Most importantly, we describe how these transfers have been apportioned to the states, those who have been favored and those who have been passed over.The third section presents and seeks to explain, through statistical analysis, how the federal government has allotted discretionary transfers, that is, which criteria it has used to choose their main beneficiaries, whether political or reallocative.
The findings show that a state governor's alignment with the central government, political overrepresentation in Congress, and redistributive criteria tend to increase the volume of federal discretionary funds transferred to state governments.Lastly, we close the article with some final considerations, indicating an agenda of studies for the deepening of the theme and breakthroughs on other fronts.

Federalism and Public resources in Brazil
Federalism is a form of organization of the political power characterized by dual territorial autonomy.This entails the existence of two autonomous territorial spheres of government: one central, which constitutes the national government, and a decentralized one, which constitutes the subnational governments.The autonomy of the subnational units, a factor distinguishing the federation from the unitary state, is premised on the independent constitution of a government, of separate legislative and administrative structures and authority, and of financial autonomy.Thereby, intergovernmental relations of a fiscal nature represent a relevant aspect in the analysis of federative political systems, characterizing themselves as a separate field of studies, which is named fiscal federalism.
Fiscal federalism presents six dimensions that lend themselves to observation and analysis which are, according to Arretche (2005, 72), "a) the setting of exclusive taxing areas; b) autonomy of government spheres to legislate on matters regarding own taxes; c) tax authority over carryovers; d) a system of fiscal transfers; e) earmarking of revenues; and f) borrowing autonomy".These dimensions show how public resources and spending are set and distributed across government spheres, thus allowing the verification of the required condition of fiscal autonomy for the actual coming into force of a federation and its level of fiscal decentralization.
As for the Brazilian case, we have observed wide fiscal federalism variations throughout the country's republican history, with periods of greater and lesser decentralization.
Part of these oscillations is associated with the instability of the political regime in the last century.We have also observed shifts in the degree of fiscal decentralization in the current democratic period, in which we have identified two distinct phases: 1980 to1994, when there was a decentralization process; and the post-1994 phase, characterized as a recentralization period.

Fiscal decentralization period: 1980-1994
The Brazilian redemocratization was followed by a process of fiscal decentralization that spanned the 1980s, as the share of the subnational spheres in the tax "cake" changed, as can be seen in Table 1.The tax reform that came into effect with the 1988 Constitution corresponded to the culmination of this process, which not only was unique in the Brazilian history, but also catapulted Brazil in terms of fiscal decentralization into the international scene (Afonso 1994, 356).
Parallel to the decentralization of revenues there was a decentralization of public expenditures.Yet, such process was conducted in an uncoordinated way, as attested by Another problematic aspect of the decentralization process was a lack of fiscal accountability rules at the state and municipal levels, entities that came to enjoy a greater volume of revenues, and fiscal freedom to incur in expenses and enter into loan contracts.The result thereof was the states' rising indebtedness, stemming from the use of state-level banks as loan agents, further compounded by inflation and the high interest rates in force.

Post-1994 fiscal recentralization period
Several studies (Abrucio 1998;Kugelmas and Sola 2000;OECD 2001) contend that there has been a process of federative recentralization in Brazil since the 1990s, particularly in the post-1994 period.The federal government concentrated fiscal resources while simultaneously managing to achieve a higher degree of freedom to spend them and furthering the implementation of a stringent fiscal responsibility law for the federated entities.The analysis of the period, especially of the two presidential terms of Fernando Henrique Cardoso (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002), points to a process of recentralization of tax revenues with the central government, of rising earmarking of revenues and implementation of a national legislation focused on expenditure control and fiscal accountability by governmental entities.
The recentralization of tax revenues was done, predominantly, through the creation of or a rise in rates of the so-called social contributions (federal sources of revenue earmarked for social spending which are not subject to the constitutionally-mandated federative tax sharing provision), whose share in total tax revenues rose from 32. 1% (1985-1990) to 41.4% (1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998), according to the OECD (2001,83).Control of subnational expenditures and indebtedness through a national law was another key feature in the process of limiting the financial autonomy of state and municipal governments.The first measure approved was the "Camata Law" (Brazil 1995), which limited federal, state, and municipal payroll expenditures to 60% of net current revenues.The search for fiscal discipline at the subnational levels reached its climax in the year 2000 with the publication of Complementary Law 101 (Brazil, 2000), the so-called Fiscal Responsibility Law.This law constrained indebtedness at the three levels of government and set the following limits for payroll expenditures, in percentage of net current revenue: 50% for the federal government and 60% for states and municipalities.Penalties for noncompliance were laid down in the Fiscal Crimes Law, which was enacted in October 2000 (OECD 2001, 95-7).
The outcome of all these measures was a reshaping of the Brazilian fiscal federalism, 3) Resources from federal discretionary transfers: pursuant to the Fiscal Responsibility Law, a discretionary transfer is "the assignment of current or capital resources to another entity of the Federation, for the purpose of cooperation, aid or financial assistance, not required by any legal provision, or those earmarked for the Single Health System" (Brazil 2000).
These three sources of funds can be distinguished in terms of fiscal autonomy.Own tax resources and constitutionally mandated transfers are revenues guaranteed to the states pursuant to the law.Discretionary funds are under the authority of the national Executive, which has full control over these resources and can set the volume to be transferred, the states that are to be benefited and the public policies to be targeted.apportionment.Would the federal government be first and foremost benefiting states in which it has political allies, while also making use of such resources to reduce the country's regional inequalities, as occurs with constitutional transfers?
Volume and allocation of federal discretionary transfers to subnational units (1997-2008)   beginning of the time series and a dramatic drop in 1999, 2002 and 2003, a fact that seems to be associated with the economic crises the country went through, with poor Gross Domestic Product (GDP) growth and greater commitment of federal resources to public debt-related financial expenses.
Another highlight for the period, as illustrated in Graph 1, is an inversion of priority in terms of federal transfers to the federated entities that benefited the municipalities.The "municipalist" perspective in the distribution of these resources cannot be fully accounted for within the scope of this work, but it might be yet another expression of the "municipalism" that has pervaded the Brazilian fiscal federalism during the current democratic period, as municipalities are elected as privileged entities in the decentralization of fiscal resources, to the detriment of the states (Rezende 2010).The increasing preference for municipalities also seems to be related to the decentralization of social policies, highlighted by the actions of the Ministry of Social Development and Combat against Hunger which, over the period, is the body that transferred the most discretionary resources to governmental entities.

Federal discretionary transfers to state governments and their impact on state revenues
Total accumulated federal discretionary transfers to the states vary widely from state to state.To better assess the importance of these resources for state revenues Table 3 shows aggregate data for tax revenues collected, constitutionally mandated transfers and federal discretionary transfers per state for the entire period.The second aspect to be highlighted, which can be observed in Table 3, is the importance of federal discretionary transfers for the state revenues.In the aggregate they corresponded to 3% of total tax revenues collected by the states and 9% of all constitutional transfers to states.However, in a context of sharp regional economic disparities, the weight of federal discretionary transfers for state revenues varies widely.At the extremes, total federal discretionary funds transferred to the state of Sergipe over the period corresponded to 68% of its total tax revenues, while to São Paulo federal discretionary transfers represented a meager 0.2% of tax revenues.

State governments ranking by receipts of federal discretionary transfers
In order to rank the Brazilian states in terms of resources received from federal discretionary transfers it does not suffice to use the absolute figures presented: these figures must be weighted to take each state's population into account.(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007).

How Does the Federal Government Assign its resources?
In We seek to explain such dynamics by building on two dimensions: the partisan-political and the social-redistributive.

Partisan-political dimension
In the new constitutional order, control over the Union's budgetary process is centralized in the national Executive branch, which is responsible for drafting three laws: the Multi-year Plan (Plano Plurianual (PPA)), the Budget Guidelines Law (Lei de Diretrizes Orçamentárias (LDO)) and the Annual Budget Law (Lei Orçamentária Anual (LOA)), all of them submitted to Congress for discussion and amending, and thereafter to the president of the Republic to be sanctioned or vetoed.
The Annual Budget Law is an authorization for the federal government to incur in expenses, not an obligation, which means that certain budgeted expenditures may not come to be executed, including federal discretionary transfers.In practice, this is further evidence of the power of the federal Executive to set volumes and destinations of discretionary transfers.
Moreover, budgeted execution of these resources has been recurrently frozen by the federal government, which also applies to budget-related amendments passed by the legislature.
Thus, the bottom line is that the distribution of federal discretionary transfers involves several institutions, actors and moments.A state may succeed in approving the inclusion of provisions of its interest in the Annual Budget Law at the moment the proposal is drafted in the ministries or during the legislative process, through amendment bills. 2 In the execution of the budget one may even seek to bargain for discretionary resources while at the same time trying to release funds agreed upon during the budget formulation process.
There is an important discussion in the literature concerning the criteria adopted by the Executive to transfer budget amendment bills, criteria which would be used as pork to ensure party discipline and support in the Legislature.For instance, Pereira and Mueller (2001;2003) argue that the federal government executes a lawmaker's own budget amendment bill in exchange for support for the passage of legislation of the interest of the Executive.Figueiredo and Limongi (2008), in turn, contend that the lawmakers' individual amendment bills have no centrality in Executive-Legislative relations, as they are the ones with the lowest execution rates.According to these authors, the greatest concern regarding budget freezes, as well as with regard to amendment cutting, is the convergence of proposed expenditures with fiscal targets and the federal government's established public policies priorities.In the view of Figueiredo and Limongi, the overarching principle in amendment execution is partisan, more specifically whether the lawmaker's party belongs or not to the government's power base in Congress, rather than related to individual lawmakers.
Thus, the political dimension of national-level Executive-Legislative relations constitutes a key variable in the search for understanding discretionary transfers to the states.Hence the assumption that the logic underlying the partisan-political competition does matter when it comes to the federal government's allocation of resources, and that the president's party or the parties that make up the president's power base, as held by Figueiredo and Limongi (2008), are favored in the allocation of discretionary transfers.
Therefore, states can benefit more if they possess larger government-supporting delegations in the Chamber of Deputies.
Nevertheless, over and beyond the horizontal dimension of the Executive-Legislative political relation, other factors related to the federative dynamics can explain federal discretionary transfers to state governments.Mostly, it must be stressed, because federal discretionary transfers are significantly much higher than all of the individual budget amending bills, plus the fact that the execution dynamics of individual budget amending bills is not captured accurately in the allocation of discretionary resources.From the standpoint of intergovernmental relations strictly we can assume that governors can attract more resources to their states when they belong to the same party as the president or the government coalition.
The support lent by the states toward electing a president may also be a relevant aspect in explaining how much the states receive.For one, as stated by Arretche and Rodden (2004, 554), based on Cox and McCubbins (1986), it is likely "that risk-averse rulers will tend to invest in regions from which they received support in the more recent elections".
However, still according to Arretche and Rodden (2004), rulers can also favor those districts with more undecided constituencies, with lower ideological alignment, and, thereby, more susceptible to respond positively to government spending-related incentives.

Social-redistributive dimension
In the debate about the advantages and disadvantages of the federation, May (1969) contends that the federative system reproduces and tends to increase regional inequalities, favoring the richer regions to the detriment of the smaller and poorer.This would occur for two reasons.First because the larger regions have larger electorates, thus greater influence over representatives.Second because their larger wealth is leveraged politically to pressure the central government.To attenuate the problem, May (1969) describes how the government acts toward establishing the system's balance through fiscal transfers: The function of fiscal transfers therefore is to allocate revenue, given the distribution of governmental functions, so as to achieve the greatest satisfaction of all governments.This implies a redistribution of income between units such that the benefits to small and poor units are sufficient to maintain their interest in federation, while the sacrifice of revenue by large and rich units is not considered by them to be greater than the benefits accruing from federation.(May 1969, 56) (Brazil 1989) set forth that 85% of FPE resources are to be allocated to the Northeast (53.5%),North (25.3%) and Central-West (7.17%), while each state's quota share is set on the basis of population size and an inverse income per capita factor.
We can assume that federal discretionary transfers may have a similar perspective and this seems to be more reasonable when we consider that, in order for the states to receive federal funds, the Budget Guidelines Law has set, over the last years, offsetting funds on the basis of state location, favoring the poorest regions. 3Yet it is clear that the differentiated offsetting funds guideline only creates an incentive for the states situated in the North, Northeast and Central-West regions to access these resources, while the federal government holds the discretionary power to favor these regions or not.
Therefore, the explanation for governmental transfers to states is not simple.It entails elements of the two dimensions proposed, while an adequate analysis must consider them jointly.The purpose herein is that of verifying to which extent variables of a political nature -specifically party alignment and the president's electoral strength in the states -influence these transfers, by controlling variables of a social, economic and redistributive character.

Hypotheses and variables
Building on the information and discussions above we may establish as working hypotheses that the federal government transfers more discretionary funds to states: H1: where the governor's party is the same as the president's; H2: where the governor's party is a member of the national government coalition; H3: which have fewer inhabitants per federal deputy (greater overrepre-sentation); H4: which have a higher proportion of federal deputies in parties that are members of the government coalition, considering the state's total number of deputies; H5: which exhibited a higher percentage of votes for the president in the most recent election and H6: which have lower Human Development Index (HDI).
Given the strong asymmetry in the allocation of funds across states, we have decided to logarithmize the dependent variable -per capita volume of federal discretionary funds transferred to state governments for each year of the period selected.As independent variables we have: 1) a dichotomous variable, a dummy, to measure if the governor's party is the same as the president's, by assigning a value of "one" to such condition and "zero" if otherwise; 4 2) another dummy to measure if the governor belongs to one of the federal government coalition parties, codified as "one" for "yes" and "zero" for "no"; 3) a variable informing about the number of inhabitants per federal deputy in the state; 4) the proportion of federal deputies, in the state, that are part of the national government coalition. 5Other independent variables are: 5) the percentage of votes received by the president in a given state at the time of the president's election to that post in the first round; 6) a dummy for FHC or Lula, with "one" for Lula and "zero" for FHC and 7) the state's HDI.
Arretche and Rodden ( 2004) is a main reference for the analysis we are undertaking, as the authors presented a similar model to explain federal discretionary transfers to the states for the 1996-2000 period.However, some findings were distinct, a fact we may attribute to the different time frames analyzed, certain independent variables used, for example, they used GDP while we used HDI, and the regression technique applied.

Analysis of the findings
In order to test the hypotheses we used regression models whose dependent variable is the volume of federal discretionary funds per capita transferred to state governments.
As we used time-series cross-section data, these models are subject to problems of heteroscedasticity, cross-correlated residuals and contemporary correlation of errors, that is, an error related to one state may be correlated with errors in other states for the same year.To solve these problems we will use the technique known as the "panel-corrected standard error", which is based on the traditional ordinary least squares (OLS) method, to correct the structure of the errors.The OLS method was proposed by Beck and Katz (1995), having since then become an important methodological innovation in political science.
In the models that follow, assessments were carried out of the impact that variables of a political and social nature have on funds' transfers to the states, such as whether the governor belongs to the same party as the president or to the president's government coalition; disproportionate state representation, as a function of the number of inhabitants per deputy; the proportion of federal deputies, elected in the state, that are part of the federal government's coalition; the percentage of votes received by the president of the Republic in the state in the most recent election; whether the state government is pro Lula, and the HDI.
It is important to bear in mind that the HDI is related to disproportionate state representation in the Chamber of Deputies (Soares 1973;Soares and Lourenço 2004).
Accordingly, there remains the problem of finding out if greater resource outlays to overrepresented regions can be accounted for only in terms of political motivations, mostly aimed at Legislative support, or if it is also related to these federative entities' needs.By using the two variables in the same model, HDI and disproportionateness, we are controlling for reciprocal effects.
In Table 5 we present two similar models.The only difference between them regards changing the variable that informs if the governor is affiliated with a party that belongs to the national government coalition for a variable informing if the governor is a member of the president of the Republic's party or not.As the latter is subsumed in the former, including both in the same model would generate multicollinearity problems.The figures suggest that the political relationship between the federal government and the governor does indeed matter with regard to the volume of discretionary transfers to the states.In the first place we can observe that the fact that they belong to the same party showed a high statistical significance in explaining the dependent variable.In the second place comes the fact of whether the governor is a member of one of the parties of the central government's coalition, which also showed a significance level of 1%.Both behaviors were confirmed as had been anticipated: the central government favors those governors who belong to the president's party in the allocation of resources and, to a lesser extent, governors of the government's power base.It is worth recalling that these parties hold important ministries, bodies playing a key role in resource allocation.
The proportion between the number of deputies and the size of the populationindicative of the size of a state's disproportionate representation -also proved to be an important variable in explaining resource transfers, in both models.The bigger the number of inhabitants per deputy, the bigger the number of votes required to elect a deputy, and the lower the outlays tend to be in proportional terms.In other words, the president seems to favor overrepresented states to the detriment of the underrepresented.The hypothesis is that the president seeks to "cheapen" the cost of Legislative support by transferring more resources per capita to states where the cost is relatively lower.These results are convergent with the findings of Gibson, Calvo and Falleti (2004), who identified overrepresentation in the Chamber of Deputies as the most important variable in accounting for the distribution of government expenditures in Brazil, Argentina, Mexico, and the United States.
The variable related to the president's electoral performance in a state also presented statistical significance, a 5-percent level, in the first model.The data show that the votes the president received in the most recent election help explain, with a 5% significance level, the quantity of resources that were transferred.This result is in line with the result found by Arretche and Rodden (2004, 568) that "the president rewarded with federal discretionary transfers the states that most supported him in the 1994 and 1998 elections".This suggests that the president is more concerned with consolidating the support received than with improving his electoral performance in states where he was poorly voted.This interpretive possibility goes counter the argument by Lindbeck and Weibull (1987) that candidates (or parties) tend to concentrate efforts on marginal districts, that is, on those where they failed to get a majority of votes, though by a narrow margin.
The variable controlling information on the presidential term shows that, proportionally, Lula made more transfers than Fernando Henrique.For the sake of speculation, we could say that Lula's behavior would have been a way of making it up to the relatively smaller power base he had in the legislature.It could also have been a way Lula found to reinforce the strategy of increasing his party's embedding in less developed regions, which received more resources particularly during his administration.
Another variable that proved highly significant was the one controlling each state's level of development, herein measured by the HDI.The negative sign shows that the more developed the state is, the lesser the federal government tends to allocate funds of a discretionary nature to it.The result suggests a redistributive perspective in the action of the central government, transferring resources to the neediest states, a perspective similar to that of the mandatory transfers.We may again assume that this is also a strategy adopted by the president to seek "cheaper" political, electoral, and governmental support, that is, in states that are more dependent on federal funds, where the impact of and the return on the resources invested may be proportionally bigger.
The proportion of deputies belonging to the government coalition did not prove significant, in either of the two models.A suggestive explanation of such result is that in a system as polarized and fragmented as the Brazilian is the government cannot simply limit itself to meeting only the demands of its consolidated power base; it must also make use of the flexible resources it has available to grab additional votes from the opposition.Figueiredo and Limongi (2008) showed that the central government executes both budget amendments submitted by deputies from its power base, though to a higher degree, and by those of the opposition, and that, in executing individual lawmakers' amendments it takes into account its own priorities in terms of public policies.With this we are suggesting that a similar rationale underlies discretionary transfers of resources: both the states with many deputies in the government's power base and the states with a greater number of political adversaries receive discretionary resources.We should also highlight that amendment bills represent a smaller share of the discretionary transfers.This means that the government might be prioritizing its allies' amendment bills, but not exactly these allies' states; even more so because, as our models indicate, the central government is also looking at the political allies at the head of subnational governments.
These results also contribute to the debate by characterizing federal discretionary transfers as yet another tool available for the Executive branch to set the Legislative agenda, in addition to those already described by the literature on the field, such as the president's legislative powers and rules conducive to the centralization of the decisionmaking process (Figueiredo and Limongi 2001), power of agenda and patronage (Santos 1997), appointment of ministerial posts (Neto 2000), and execution of budget amendments (Pereira and Mueller 2001).
In short, though we were unable to establish a causal relation, the data suggest that the amount of resources that are arbitrarily allocated by the federal government to the states depend on the political relation with the governor, the state's level of development, and disproportionate representation in the lower chamber.The last two are related and can be traced back to an old discussion in the Brazilian political science that suggested a "promiscuous" relation between federal government and the less developed states (Soares 1973).The idea is that the discretionary distribution of resources to the poorest, and overrepresented, states would be an additional tool in the hands of the president of the Republic toward strengthening the Executive's political capital and reducing filibustering.
That would be so because the effect of a given quantity of resources in needy regions, with a relatively large number of representatives, generates a much greater impact than in richer regions exhibiting a representation "deficit".This is, however, a highly speculative hypothesis we do not intend to test herein.The concern in this article was merely to identify the important variables in explaining the central government's motivation when it transfers resources to this or that state.To have a more accurate view of its consequences for the legislative arena, new research must be conducted that should take into account the deputies' role in the budget-making process and the lawmakers' behavior in roll-call votes, in addition to other aspects of the lawmaking action.

Final Considerations
This article sought to analyze federal discretionary transfers to state governments.
First it characterized, through concepts and values, the meaning of these transfers.Next, it ranked states according to the different allocations of such resources, demonstrating how some states were much more benefited than others in terms of resources received from 1997 and 2008.In the third section we sought to mobilize some factors that might prove instrumental in accounting for differences in federal discretionary transfers across states and we performed a statistical test with some variables.We found out that the alignment of a governors' party with the president or with the president's government coalition impacted positively on the revenues received annually by the state.States with greater disproportionate representation in the Chamber of Deputies were also benefited.Lastly, less developed states, in HDI terms, were favored in the allocation of federal discretionary funds, which points to a redistributive perspective that may be associated with a central government strategy of seeking "cheaper" political support.
Nonetheless, this is a new agenda of studies that may gain much larger contours than the ones outlined in this article.Deepening the issue requires adopting a similar perspective toward the municipalities and seeking to describe and understand their position toward such resources.In parallel, it is also critical to examine the types of public policies favored in the states and municipalities, which have been given priority in the allocation of resources.Coincidently, it provides us with data on a 12-year period, six for each of the two last governments.
2 Amendment bills are submitted by federal deputies and senators for the purpose of nullifying, introducing or reallocating appropriations in the Executive's budget proposal.With regard to their introduction, amendments can be submitted by a single lawmaker, a state's delegation, a regional delegation, and Senate and Chamber permanent committees.
3 To receive federal discretionary funds, the states must contribute with a required minimum percentage of the amount received, the offsetting funds.The 2010 Budget Guidelines Law set forth in Article 39 that such percentages shall be: a) 10% and 20% for priority states as established by the National Policy for Regional Development (Política Nacional de Desenvolvimento Regional (PNDR)), in the areas supervised by regional development agencies (Superintendência de Desenvolvimento do Nordeste -(SUDENE), Superintendência de Desenvolvimento da Amazônia (SUDAM) and Superintendência de Desenvolvimento do Centro-Oeste (SUDECO)); and b) 20% and 40% for the all other states (Brazil 2010).
during the 1997-2008 period, thus comprising the two last years of the first administration of president Fernando Henrique Cardoso (Partido da Social Democracia Brasileira (PSDB), in the Portuguese acronym) and his entire second term of office, the first administration of Luiz Inácio Lula da Silva (Partido dos Trabalhadores (PT), in the Portuguese acronym), and the two first years of his second term of office. 1 The privatization of state-level government-owned banks through the Program to Stimulate the Reduction of the Government Sector in the Financial Sector (Proes), created by Provisional Measure 1,514/96 and implemented in February 1997, deprived the states of an important source of public-spending funding and indebtedness.In addition, state debt restructuring agreements were reached, a process which took place between September 1996 and January 1997, when the federal government negotiated the debt of 22 states of the Federation.Terms ranged from 15 to 30 years and, as guarantees for compliance with the agreed upon accords, the states put up their revenues of Tax on the Circulation of Goods and Services (ICMS) and State Revenue Sharing Fund (FPE)(OECD 2001, 91).
greater volume of non-earmarked tax resources concentrated with the central government and rising accountability and control of state and municipal expenditures.In this context, discretionary transfers gained in attractiveness, as they allowed subnational governments to enhance their capabilities to implement public policies, especially investment expenditures.They also represented for the national Executive a powerful political coordination tool in intergovernmental relations with states and municipalities (vertical relations) and in negotiations with Congress (horizontal relations), given the legislators' interest in grabbing a bigger slice of the discretionary transfers to their constituencies.the Distribution of Federal Discretionary Funds to the States: Who Wins and Who Loses Federal discretionary transfers account for one of the three major sources of state revenues, which are: 1) Own tax revenues: taxes, fees and improvement charges collected by the states, such as the Tax on the Circulation of Goods and Services (ICMS) and the Tax on the Ownership of Automotive Vehicles (IPVA).2) Resources from constitutional or mandatory transfers: constitutionally-mandated transfers from one federative entity to another.In the Brazilian case the most important federal transfers to the states are the State and Federal District Revenue-Sharing Fund (FPE); the Fund for the Maintenance and Development of Basic Education (Fundeb); and the Export Compensation Fund (FPEx) (National Treasury Secretariat 2008b).

Federalism
Thus, it is incumbent upon us to ask what these resources represent for the state governments and what has been the behavior of the federal government in their

Furthermore, a1
cross-section study of discretionary transfers, focused on the lawmakers' amendment bills, may help to clarify the connection between budgetary resources and the Executive-Legislative relation at the national level.New steps in the research will advance the field of study of fiscal federalism, plus constituting a source of information for greater transparency and control of public resources by society.Time period choice was dictated by data availability at the National Treasury Secretariat.

Federalism and PubliC resources in Brazil: voluntary transfers from de Union to the states bpsr
Over the 12-year period analyzed, federal discretionary transfers to states and municipalities amounted to 103 billion reals, inflation adjusted, i.e., 8.6 billion reals a year on average.This amount represented 1.7% of the 2008 budget's total expenditures, which totaled 509.7 billion, excluding all intergovernmental transfers.This is of a magnitude similar to the expenses incurred by the federal government with the cash-transfer Family Allowance Program, which accounted for 2.1% of total expenditures for fiscal2008 (National   Treasury Secretariat, 2008d).This amount is, therefore, capable of generating impact on the political relation between the federal Executive branch and the other entities of the federation, especially on the less developed states.

Table 2 .
Federal discretionary transfers to state and municipal governments 1997-2008 (in thousands of reais)

Table 2
presents discretionary transfers by subnational level.Total transfers to states and municipalities exhibited a random behavior, with higher amounts transferred in the

Table 3 .
Own tax revenues, constitutional transfers and federal discretionary transfers to state governments, 1997-2008 (in thousands of reais)The first finding is that the amounts of own resources collected by each state vary widely.At the extremities we have Roraima, with 0.38% of what the state of São Paulo collected.Even within the "wealthy" group, Minas Gerais or Rio de Janeiro did not reach a third of what the state of São Paulo collected while, together, the three states collected more than 50% of the total collected by all the other federative entities.The share of the nine states

Márcia Miranda Soares and Pedro robson Pereira Neiva bpsr
of the Northeast region in total tax collection was 14%.Constitutional transfers, however, exhibited a different pattern.São Paulo participated with 16%, Minas Gerais with 8% and Rio de Janeiro with 3%, totaling 27% of all constitutional transfers.By contrast, the Northeastern states' share was 34% of total constitutional transfers.Lastly, federal discretionary transfers also exhibited a distinct share per state, something we shall detail below.

PubliC resources in Brazil: voluntary transfers from de Union to the states bpsr
Still, if the data seem to assign a redistributive character to the discretionary transfers, they are far from signaling an unquestionable convergence between federal discretionary transfers and income.For example, the state of Acre was better ranked than nine other states with lower per capita income.São Paulo, ranking second in per capita income, received more resources than other four states ranked lower in terms of income.The assumption is that other factors have influenced the differences observed in resource allocation.At any rate, it seems reasonable to assume that social factors, combined with political factors, Table4presents federal discretionary transfers per capita for each state government over the period.As can be seen, ranking top as major beneficiaries are the entities of the North region, followed by the Northeast and Central-West regions.What could account for the central government's preference?Perhaps the answer lies in the economic situation of these regions: the assumption is that the central government would be allocating its discretionary resources to the poorer states.Such redistributive perspective seems to be corroborated when we compare each state's federal discretionary transfers per capita ranking with the state's GDP per capita ranking.operate in determining success or failure of subnational entities in receiving discretionary resources from the federal government.Federalism and

Table 4 .
State ranking -Federal discretionary transfers and income per capita(1997-2008 average) In Brazil, a country characterized by sharp regional inequality, constitutional transfers have worked as a mechanism for the promotion of greater equalization and balance across states and regions.The poorest receive more resources than the richer.The main form of redistributive transfer is the State and Federal District Revenue-Sharing Fund (FPE), which amounted to more than R$ 38 billion, in 2008.Pursuant to Article 159 of the Constitution, 21.5% of the net collection (gross revenue minus fiscal incentives and rebates) of taxes levied on income and miscellaneous earnings (income tax) and on industrial products (IPI) is to be earmarked for the State and Federal District Revenue-Sharing Fund.Complementary Law 62