From Revenue to Value Added Taxes : Welfare and Fiscal Efficiency Effects in Brazil ∗

This paper presents an evaluation of the economic impacts from the taxation reform of PIS/PASEP and COFINS that started to be collected by two regimes (cumulative and non-cumulative) associated to domestic flows and to levy imports. The evaluation made with a computable general equilibrium model adapted to new fiscal system characteristics indicates that the effects of this reform would have deteriorated macroeconomic, labor market and welfare indicators.


Introduction
From 2003 to 2004 took place a taxation reform in Brazil that affected two taxes: (1) "Contribuição do Programa de Integração Social e de Formação do Patrimônio do Servidor Público (PIS/PASEP)" and (2) "Contribuição para a Seguridade Social (COFINS)". 1 The PIS/PASEP was created in the 70's during the implementation of the labor market reforms that had finished with the employment stability, while the COFINS originated from FINSOCIAL, in the context of the 1982 Brazilian external debt crisis.
From the promulgation of the 1988 Constitution, several initiatives took place in the scope of the Union aiming at changing the Brazilian tributary system. Among the changes that had been implemented, the most significant ones were: (1) the introduction of the "Contribuição Provisória sobre Movimentação Financeira (CPMF)" in 1993, and (2) the PIS-COFINS tax reform that started in 2003 and was completed in 2004.
Prior to this reform, PIS and COFINS were cumulative taxes that charged firms' gross revenue and did not levy imports. The initial proposal of the reform was the complete conversion of these cumulative (on gross revenue) taxes in non-cumulative (on value-added) ones in order to induce economic efficiency gains by reducing the incentives to excessive vertical integration of firms. Also, the reform intended to improve the national firms´ competitiveness by apllying these taxes on imports of goods and services and maintaining the exports exemptions. By means of this reform, in 2003 the incidence of PIS on firms' gross revenue was partially changed to firms' value-added, that is, this cumulative tax was partially converted to a noncumulative one. In 2004, the COFINS was subject to a similar change and both taxes started levying imports of goods and services. Table 1, below, presents the recent evolution of some amounts of tax collections: Total, Union (broad Federal Government) and PIS-COFINS. Once the later are federal taxes, their joint collection is compared to the Union values.
It is observed an increase in the participation of PIS-COFINS collections in the Brazilian Federal public sector (Union) fiscal revenue and that from 1998 to 1999 and from 2003 to 2004 this share increased in a more significant way. In the first period: (1) COFINS rate applied on firm's total gross revenue was raised from 2.0% to 3.0%, and (2) PIS and COFINS started levying Financial Institutions (Law 9718/98). In 2004, the year in which the reform was completed, this participation achieved 22%.

ICMS (1) IR (2) PIS-COFINS (3) SSC (4) (3)/(1) (3)/(2) (3)/(4)
It is observed that, in 1998, the PIS-COFINS amount was equivalent to 52% of the Income taxes amount and 53% of the Social Security collection. Henceforth, its amount grew faster than the other amounts, presenting, between 1998 and 2004, a growth rate of 291%, followed by IR with 130%. Following this path, in 2004 the PIS-COFINS exceeded all revenues from Social Security Contributions, becoming the third bigger tax group, surpassed only by ICMS and all forms of Income Tax. Again, as in Table 1, it deserves mention the significant increase in the ratios presented in the last three columns, in Table 2, from 2003 to 2004. 4 The data presented in tables 1 and 2 show that the PIS and the COFINS evolution changed in 2003 and 2004, when occurred the changes in their taxation regimes: (i) the introduction of a non-cumulative form of incidence for both taxes and (ii) their incidence on imports of goods and services.
Despite the analysis will focus on economic variables, some aspects of the related legislation deserves be emphasized, in order to justify some methodological issues. In order to better understand this taxation aircraft and boats (maintenance and construction), equipment for the audiovisual industry, petrochemical nafta and natural gas.
The significant amount of exceptions, mainly exemptions or suspensions, common to the laws previously mentioned, created a subsequent pressure for the extensions of these benefits, being these the main aspect of the Laws 10925, 11033 and 11051. As example of this process, the Law 10637, which created the noncumulative regime for the PIS, was later modified by the Laws 10684, 10833, 10925 and 10996, in a period of only two years from the original edition. Specifically for this work, this "mutant" characteristic caused a significant difficulty to simulate the economic effects of changes in the PIS-COFINS taxation regimes in the context of a dynamic quantitative analysis.
Considering all the previous information, the PIS-COFINS reform was basically characterized by the introduction of their: (1) incidence on firms' value-added (non-cumulative regime), (2) incidence on imports of goods and services and (3) new rates (1.65% for PIS and 7.60% for COFINS) on value-added and imports. Also, after the PIS-COFINS taxation reform these taxes started being collected by two regimes: (1) the (previous) cumulative and (2) the (new) non-cumulative. As pointed earlier in Tables 1  and 2, the PIS-COFINS collections presented significant increase in 2004 and due to their relative importance, it seems reasonable to expect that the effects of this reform on Brazilian economy were not negligible and deserves to be addressed.
Given these characteristics, the implementation of PIS-COFINS reform basically induced relative prices changes in the Brazilian economy and, consequently, the economic agents rethought their resource allocation decisions, which characterize a general equilibrium phenomenon. From the individuals' and families' point of view, modifications in the relative prices structure can significantly alter their welfare, by means of the changes in the structure of the real incomes and the expenditure of these agents. Besides, the taxation system together with the social public expenditure has been considered very ineffective in changing our income distribution 7 . Therefore, it seems very appropriate to assess the economic impacts from the PIS-COFINS taxation reform by means of a computable general equilibrium (CGE) model.
Despite the importance of this reform, the literature on its economic impacts is very scarce. Thus, it is convenient to present some details on these studies to show how our paper can contribute to address this subject. Kume (2004) employed a multissectoral fixed prices model, based on IBGE 8 input-output matrix and resources and uses tables, to estimate the change of the domestic protection due to the proposed PIS-COFINS legislation. Although the simulation results from models with price rigidities must be carefully interpreted, among the main results, we can mention that the average domestic protection would have almost doubled, from 7.50% to 14.40% on imports CIF value. However, concerning the results, it is not clear if the simulations have applied the proposed or the approved law, which has differences for sectors and the way the taxes are calculated. More important, policy analysis in an environment of fixed prices must be interpreted with restrictions, mainly for not allowing agents reaction to relative prices change, which essentially is the initial economic impact induced by the reform. SE-MFAZ (2004) reports the results of two studies that aimed at estimating the permanent (long-run) impacts of the COFINS reform on their tax burden. In the first study, assuming that, in the first months after the adoption of its new taxation regime, the COFINS collection would follow a path similar to the one present by PIS collection in the months right after its regime change it was estimated that the COFINS reform would reduce the tax burden on production, as a proportion of the firm´s gross revenue or value added, at aggregate level and in many sectors with few exceptions. However, as it is recognized in the study, the data on PIS collection used in the estimations would be expressing more transitory impacts of the PIS reform than its permanent effects. Therefore, we can say that the estimated changes in the tax burden could be temporary (short-run) effects instead of permanent (long-run) ones.
More important is the fact that the data on PIS collection used in this study to extrapolate a temporal path for COFINS do not take in account another feature of the taxation reform: PIS and COFINS started levying imports just two months after the adoption of the non-cumulative regime for the later. Thus, the data used in this study to infer the impacts of the COFINS reform do not consider the "external side" of the taxation reform that would induce more changes in the relative price system.
In the second study, by means of an input-output matrix it was calculated the impacts of the changes in the COFINS taxation regime on its legal and effective tax rates at national and sectoral levels. The results indicate that the tax burden on national production would be lowered by 7,9% and that the majority of the economic sectors would be benefited by reductions in their legal tax rates and mainly in their effective rates. However, the same remarks pointed to Kume (2004) apply to the results from this study.
Even though Kume (2004) and SE-MFAZ (2004) have contributed to address some impacts of PIS-COFINS reform, they did not evaluate its effects on the relative price system and the resource allocation in the Brazilian economy. The evaluation of these effects using CGE models was made only by Silva et al. (2004) 9 that present a descriptive section for a broader "tax reform" proposed by the Brazilian government, characterizing the participation of the main indirect taxes, before the PIS-COFINS tax reform. They simulated the implementation of this reform and also tested a possible incidence change for the social contribution of employers.
Although the taxation characteristics in Brazil are very well modeled and they provided useful results to evaluate the potential impacts of the referred reform, some characteristics and assumptions in their study deserves to be commented. First, the model's database (Social Account Matrix) is for 1998 and the use of more recent data would more accurately express the features presented by the Brazilian economy in the period in which the reform was implemented.
Second, the sectoral effective rates of COFINS on value-added used in the simulation were calculated by extrapolation procedure based on the changes presented in the PIS collection in 2003. Defining the efficiency of a tax on a determined taxation base in a given year as the ratio of its effective rate by its legal rate and the relative efficiency between COFINS and PIS on a determined taxation base in a given year as the ratio of their efficiencies (COFINS by PIS), the COFINS effective rates were calculated assuming that the relative efficiency between these taxes on value-added in 2003 was the same as in 2001, when both taxes were charging the gross revenues only. In other words, it is assumed that the sectoral levels of tax evasion and exemption were the same in the non-cumulative and in the cumulative regimes for each tax.
By this procedure they verified that some sectors 10 would present an effective rate higher that the legal rate and, thus, in the simulation, it was assumed that these sectors remained just under the cumulative regime with rate as in 2001. Due to other reasons, this assumption was applied to other sectors. 11 However, data from Secretaria da Receita Federal 12 show that PIS and COFINS were being collected by means of both taxation regimes in all the sectors present in the used classification. Therefore, the simulated taxation reform presented some characteristics that conflict with empirical data available after this paper. 9 The model employed in this study was developed from Devarajan, Lewis and Robinson (1991). 10 Mining, Electronic Equipment, Other chemical products, Plastic items and rubber, Shoes, and Other food products. 11 Dwellings and Public Administration, Automobile industry (automobiles, trucks and buses), Other vehicles and parts, Oil refinery, Pharmaceutical and perfumes, Communication, Transportation and Other services (services provided to families, services provided to firms and private non-tradeable services). 12 Brazilian Tax Collection Bureau.
Concerning the tax rates that would charge imports of goods and services, it was assumed that for each import flow the relation between the PIS and COFINS legal and effective rates would be the same as those presented between the import tariff rates. In this sense, PIS and COFINS legal rates would be subject to a system of exemptions that would make them differ from the effective rates in the same degree verified for the tariff on imports rates, whose differences are due to the trade agreements, preference systems, exemptions and exceptions. It deserves comment that is possible that this assumption can be conflictive with the empirical data available after this paper. Third, the model closure admitted full employment of labor and constant capital stocks, which nullify the impacts on GDP by assumption (GDP-neutral effects). Fourth, it is assumed in the simulations that the nominal exchange rate is endogenous while the general price index (GDP deflator) is fixed, which induce price-neutral effects by assumption (no changes in the aggregate price index). Considering the facts that the new PIS and COFINS rates were much higher than the previous ones, even though the changes in the incidence base, and that they started levying imports, it is reasonable to expect price increase and product decrease.
Finally, their focus was the impacts on macroeconomic and the sectoral indicators due to the reform, while we intend to extend the analysis towards individuals' and families' welfare effects, taking the advantage of a specific institutional modeling for these issues. Once there are some points that can be advanced, another evaluation that can to contribute to the analysis on the effects of the PIS-COFINS reform on Brazilian economy is justified.

The CGE Model -main features
The CGE model used here is an extension from the model presented by Cury, Coelho and Corseuil (2005) 13 and, thus, it will be described in a shorter way, while further details can be found at this paper.

Product Supply
Foreign product supply is modeled as being totally elastic, 14 while sectoral domestic supply is represented by a three steps nested production function with three types of inputs: labor, capital and intermediate inputs. 15 First, amounts of types of labor (F l ), given by the first order firm's profit maximization conditions, are combined in a composite labor (Ld i ) for each sector i, by a Cobb-Douglas function with constant returns to scale: 16 where il β is the share of each type of labor: unskilled informal (l1), skilled informal (l2), formal with low skill (l3), formal with average skill (l4), formal with high skill (l5), public servant with low skill (l6) and public servant with high skill (l7). 17 .
Second, in each sector i, aggregated labor (Ld i ) and capital (K i ) 18 are associated by a constant elasticity of substitution (CES) function to obtain the production level (X i ): 13 This model results from a series of developments made in the model proposed by Devarajan, Lewis and Robinson (1991), as can be seen in Cury (1998), Barros et alii (2000a) and Coelho et alii (2003). 14 Thus, Brazilian demands for imported goods are fully satisfied without facing external supply constraints. 15 The model represents the 42 sectors of activities listed in the 2003 Brazilian National Accounts. 16 This means that an identical increase of every type of worker results in an identical increase of the aggregate worker. 17 Also, there are more 2 types of employers that are treated as labor and enter in the Cobb-Douglas aggregation.
a is the CES shift parameter, i α is the sector's i labor share in the production value and ip ρ is the elasticity of substitution between capital and labor.
Finally, in the third step the various intermediate inputs levels (INT i ) are obtained by a Leontief production function (e.g., fixed proportion to sector j total product, X j ): 19 Domestic producers react to the relative prices in domestic and international markets and the domestic output is divided by a constant elasticity of transformation (CET) function with imperfect substitution in products sold to these markets: where i X , i E and i D are, respectively, the domestic sector i's total output, exported volume and sales to internal market.

Demand for products 2.2.1. Families
Families are classified according to per head household income, level of urbanization and household head characteristics: poor urban families headed by active individual (f1), poor urban families headed by nonactive individual (f2), poor rural families (f3), urban families with low average income (f4), urban families with average income (f5), rural families with average income (f6), families with high average income (f7), and families with high income (f8).
They choose commodities' consumption levels to maximize utility subject to a budget constraint 21 , according to a Cobb-Douglas functional form (similar to the production function presented earlier).
Families and firms demand domestic and imported goods as imperfect substitutes that differ according to their source (domestic or external), as proposed by Armington (1969), and their utility are measured (in product quantity) by a CES function: where M i is the imported volume of good i and D i is the consumption of the domestic good i. 18 The model closure adopted in the simulations determines that the sectoral levels of capital are fixed. 19 It is worth mentioning that Devarajan et al. (1991) makes use only the first and third steps, by combining capital with labor and value added with intermediate inputs, in this order. 20 There are no empirical estimates of Brazilian export elasticities using a CET structure for a highly disaggregated sectoral specification. Therefore, it was adopted the same procedure used in Cury (1998, pp. 112-113), which departed from the elasticities estimated by Holand-Holst, Reinert and Shiells (1994) to the American economy. 21 Actually, this utility maximization can happen along the consumers' lifetime. From the point of view of most practical applications, the maximization is on the goods and services available in a given period. 22 These elasticities values were estimated by Tourinho et al. (2002) for the same sectors considered in the model. 23 It can be interpreted as the quantity of a hypothetical mixed (imported and domestic) good that would be demanded by consumers.
The external agents demand domestic goods, reacting to changes in relative prices as well. Similarly to the import demand function, the exports demand arises from a CES utility function that represents the imperfect substitution between products from the external regions and Brazil.

Firms
Firms demand commodities to satisfy their production requirements of intermediate inputs according to the technical coefficients from the input-output matrix.
Due to the static nature of accumulation in the capital market, investments are important for product demand. Similarly to consumption, the investment is characterized as the purchases of certain goods and can be considered as a final consumption undertaken by firms. The savings represent this amount of resources and it is assumed that a share of it corresponds to investment in stocks of finished goods, while the remaining parcel represents the net investment required to expand production. The first share is defined based on a fixed proportion to the sectoral output, while the second is distributed exogenously among the sectors, reflecting information from the input-output tables (goods by sector of origin) and the matrix of sectoral composition of capital (goods by sector of destination and origin).
It is considered that investment goods are being produced but not used as increments of capital stocks. Thus, the model closure is closer to a medium-run type: constant capital stock, price flexibility and existence of involuntary unemployment in equilibrium.

Government
The Government consumption (GC) is derived from maximization of a Cobb-Douglas utility function subject to the budgetary constraint corresponding to the total expenditure that is fixed according to the total amount registered for the base year.

The Labor Market
Labor is a production factor used by firms and is classified into 7 types, according to contract status and schooling. 24 It is admitted that firms aim at maximizing profits under technological constraints conditions imposed by production function, in an environment where prices of inputs, production factors (labor and capital) and output are beyond their control. . Therefore, as a result of this maximization, for each type of workers, a specific demand curve is defined by the condition that their marginal productivities equalize their wages: 25 The labor market equilibrium (employment and wage) is determined by E / , the intersection point between the demand curve (L d ) and the wage curve (S). The wage level defined by E / does not correspond to the labor supply (L o ), and the difference L o -L is the excess of labor supply that corresponds to the involuntary unemployment level (U) in the economy. 26 24 The labor treatment that follows is applied for the 5 types of private workers. The 2 types of public servants follow the traditional labor market closure of CGE models with either wage or employment being fixed. Therefore, there is no substitution between public servants and the private kinds of workers, in the sectors where there is no public companies. In the sectors where public and private firms coexist, the changes in the public-private composition of labor are related to the changes in the public-private composition of the sectoral representative firm. 25 The derivative of the profit function in relation to the demanded quantity of each factor must be equal to the factors' price (first order condition). 26 E o would be the full employment level given by the interaction between labor supply and labor demand.
where β l reflects the firm's bargaining power in offering lower wages according to unemployment rate. 28 After defining the aggregate levels of employment, wages (w l ) and unemployment, for each type of workers, their sectoral wages (w li ) are found by means of the sectoral relative wage differentials. Using a sector and worker specific demand curve (equation 6), the sectoral employment level of each type of labor (F il ) is determined and, then, aggregated by a Cobb-Douglas function (equation 1) defining the sector i's composite labor.

The Income Transfer Mechanisms
Here it will be presented the formation process of income flows received by families and firms. The remuneration of capital is paid to firms 29 and the labor earnings to workers. In each sector, the payments to capital are distributed to the firms according to the initial share in the total earnings of capital.
The eight types (h) of families receive earnings from the seven types (l) of labor according to the initial shares (ε hl ) of these workers in these families, which also receive the remuneration of capital transferred by firms (YK) according to the family h's share in these income flows (ε hk ). Finally, the families also receive net remittances from abroad (RE h ), adjusted by the exchange rate (R), and transfers from the Government (TG), in the form of payment of benefits (direct income transfers) 30 and as other transfers (essentially domestic debt interest)that are allocated to the families according to the initial shares (θ ht ). Therefore, the family h's income is:

The Government
The Government spends by consuming ( ∑ i i CG ) and transferring resources to the economic agents. It plays a very important role in the process of determination of secondary income, once it also directs a 27 A brief description of wage curves can be found in Cury, Coelho and Corseuil (2005). Broader explanation about them can be found in Oswald (1990 and1994). 28 These parameter values were taken from Reis (2002), who estimated then for the Brazilian case. 29 Small (self-employed people) and large (other firms). 30 These transfers include the social security benefits as well as other programs such as unemployment benefits, income transfer social programs and other cash benefits.
share of its transfers to firms 31 as interests on the domestic debt and demands products. Similar to families, the sharing of government transfers to the types of firms follows the proportions observed in the base year (θ k ). Finally, it also transfers resources to abroad (GE) and its total expenditure is: To face all expenditures, the Government relies on three types of collections: (1) direct taxes levied on firms' and families' income (φ h and φ k , respectively), and (2) indirect taxes on domestic and imported goods (proportional to production (X), domestic sales (D), imports (M) and value added (VA) amounts). Besides these sources, it also receives transfers from abroad (gfbor) and, finally, there is the balance of the social security system (SOCBAL). 32 Thus, the Government total revenue is: where η i are the tax rates on production, ξ i and π i are, respectively, the sector i's PIS-COFINS rates on domestic sales value (cumulative regime) and on value-added (non-cumulative regime), i σ and κ i are, respectively, the ICMS-IPI tax rates on value-added and imports, µ i is the tariff on imports, while γ i are the PIS-COFINS rates on imports of commodity type i.
An eventual lack of government resources is defined as a government deficit that, together with domestic private (firms and families) and foreign savings, defines the amount of resources spent as investments.
The implementation of the PIS-COFINS reform changed the way by which the Government collects indirect taxes that levy domestic and imported commodities. Thus, the indirect tax revenue (INDTAX) from domestically produced goods is given by: where PX i .X i is the production value, PD i .D i is the gross revenue value from domestic sales and VA i , η j , ξ i , σ i and π i were presented in equation 10.
The presentation of this equation is very important to understand the way the implementation of the fiscal reform will be simulated. According to PIS-COFINS tax revenue data from "Receita Federal", all sectors are being levied in both cumulative and non-cumulative regimes. Then, the domestic part of the simulation will consist in applying the ξ j and π j tax rates that were verified in 2004 at sectoral level.
The other equation that contributes to the Government revenue and deserves mention is the indirect taxes on imports revenue, which is given by: where pwm i is the external price of imports (in US$), exr is the exchange rate, µ i is the tariff on imports, κ i is ICMS-IPI rates on Imports and γ i are the PIS-COFINS rates on imports.
Again, this equation is important to understand the way that the fiscal reform will be simulated, once another feature of this reform was that the imports started being levied by PIS and COFINS taxes. Thus, the implementation of this part of the reform will consist in applying γ i tax rates that were collected from import flows of commodity type i in 2004. 31 The same applies for labor supply. 32 In fact, social security is treated as an agent apart from the Government in the model, not only because of the considerable amount of resources that it handles in Brazil, but also because of the contributions that it applies on either the company's income (here again in a different form), or on the installments of the added value of labor.

The Model Data Base
Almost all data used in the model and simulations were derived from a Social Account Matrix (MSC2003) that was specifically made to be used in this research and contains all the quantities and prices information in 2003 (the model's base year). Besides, all the model's coefficients and parameters obtained by means of a model calibration process are calculated from this data matrix. 33 It deserves mention that it was made based on information from the latest officially published Brazilian National Accounts by the Instituto Brasileiro de Geografia e Estatística (IBGE). Another set of data used to calculate the economic shocks that will be simulated and evaluated will be presented in the next section.

Simulations -modeling issue
Pursuing to reduce the cumulativeness of the PIS and the COFINS, the taxation reform has changed part of these taxes incidence from the firms' gross revenue from domestic sales to their value-added and also established a legal (nominal) rate on the later magnitude. Even though a tax on value-added is noncumulative, the effective tax rate differs from the legal rate because it is established, by law, that to calculate the former rates that must be applied on the value-added amounts, it is necessary to consider the incidence of these taxes on the ICMS rates, which is sector specific. Besides, the legal and the applied rates can be different due to fiscal exemptions.
BY the PIS-COFINS implemented reform all sectors in the classification used here started to be levied by a taxation regime that is a mix of the cumulative (previous) and the non-cumulative (new) regimes Thus, as exposed earlier, the domestic part of the simulation will consist in applying the tax rates on domestic sales (production less exports) and value-added amounts, calculated based on 2004 flows, at sectoral level.
More specifically, the economic shocks (tax rates changes) were calculated based on data of PIS and COFINS collections by taxation base (firms' gross revenue, value-added and imports) obtained from the Secretaria da Receita Federal (SRF). Once these data are classified according to sectoral classification CNAE that differs from the IBGE codification used in the MCS-2003, the sectors in both codes were matched assuring that the aggregated amounts of collections were equal.
It is important to mention that the legal compound PIS-COFINS rates are not directly imposed on imports because the legislation establishes that to determine the rate to be applied it is necessary to use a specific formula by which these rates must interact with the ICMS and tariff rates on imports, besides themselves.
Also, the existence of special taxation regimes benefiting some import flows by exempting them from tariffs and/or indirect taxes, as for example the imports under the drawback regime, can make the applied rates differ from the legal ones. Then, given these possible differences, the implementation of the incidence of PIS and COFINS on imports will consist in applying the tax rates that were verified from these flows in 2004.
Therefore, the impacts of the PIS-COFINS taxation reform will be simulated by implementing a mixed taxation regime, which consists of the following features: (1) the taxes levy sector's revenue from domestic sales and value added, (2) application of the new tax rates on domestic flows verified in 2004, (3) the taxes start levying imports and (4) application of the new tax rates on import flows verified in 2004. Therefore, we expect the impacts calculated here will differ from those presented by Silva et alii (2004) because while we will simulate the impacts of the changes in these taxes rates calculated based on the verified collections they simulated the effects of the changes in these taxes rates calculated based on extrapolation procedure as presented in the introduction (first section).
The taxation reform will be simulated in two steps. In the first step, the mixed taxation regime will be implemented (features (1) and (2)). Then, in the second step, the PIS-COFINS rates on imports are also implemented (features (3) and (4)), taking as database the resulting scenario from the first step. Thus, the results from the second step capture all the impacts of the taxation reform. Henceforth, these simulations will be referred as PCVA (domestic reform) and PCVAM (complete reform).

Model Closure
As previously mentioned, the model closure is closer to a medium-run type, since it is being assumed that: (1) sectoral capital stocks are constant, (2) prices are flexible, (3) involuntary unemployment exists in equilibrium and (4) trade balance is exogenous. The constancy of capital stocks is due to the fact that, in the model, the investment goods are being produced but not considered as increments of capital stock. The existence of involuntary unemployment in equilibrium is a consequence of the labor market modeling (section 2.3).
By admitting trade balance as exogenous, the exports adjust not only due to the price responsiveness of external demand but also to adjust the changes in imports, maintaining the trade balance unchanged.
Once one of the main purposes of the simulation is to evaluate the potential distributive impacts of the PIS-COFINS fiscal reform, the analysis will also focus on the impacts on employment, wages and household income. However, the impacts on selected aggregated variables will be presented in order to present the direction and the magnitude of the macroeconomic effects.

Macroeconomic impacts
The simulated macroeconomic impacts of the PIS-COFINS reform are reported in the first column (PCVAM) in Table 4. The effects of just the domestic part of this reform are also reported in the second column (PCVA), even they are not the focus of analysis.

13
The overall impacts from fiscal reform were adverse since it would induce a real GDP fall of 0.70%, an aggregate employment decrease of 1.60% and inflation of 2.51%. The effect on real GDP can be reflecting the fact that the reduction of the cumulativeness of these taxes had relatively strong negative effects on output at sectoral level and, therefore, on aggregate product.
The taxation of value-added (VA) induces an increase in its price, which is equivalent to a rise in marginal costs. To achieve the equilibrium, in perfect competition, the representative firm needs earn higher marginal revenue or reduce marginal costs, which can be done by reducing the VA components usage. Considering the way that the labor market operates and the model's closure features, this implies in a lower labor demand, inducing a decrease in wages, and so, reducing the available income and, consequently, consumption expenditure.
Also, by taxing imports, that is, increasing their prices in domestic market (+6.80%), the reform induces another adverse effect on aggregate consumption. Once domestically produced and imported commodities are not perfect substitutes, even changing the relative prices in favor of the domestic ones, 34 this price increase raise the composite commodities prices in internal market (+2.51%). This relative increase of prices in the internal market induces the households to consume less, but substituting imported commodities by larger amounts of domestic ones. In a similar way, the firms tend to substitute in some extent import inputs by domestic ones. However, this positive effect is not strong enough to offset the negative effect on consumption induced by the taxation of imports, and so, there is a second adverse impact on consumption demand.
The macroeconomic closure considers that the investment is determined by the savings behavior and that the Government consumption is fixed. This implies that the changes in the tax revenue affect the Government savings and, therefore, the public deficit and investment. Thus, the investment increases by 0.28% due to the rise of 0.54% in the public sector revenue, which induces a decrease of 10.30% in the public sector deficit.
Exports fell due to the price-responsiveness behavior of external agents and the model external closure characteristics. First, the reform induced an increase in domestically produced commodities prices, which, by turn, caused a decrease in external demand by Brazilian commodities. Second, the rise of import prices and the reduction of internal absorption (activity) induced a fall in demands for imported commodities, and in order to not affect the trade balance equilibrium, exports decreased.
The fall in aggregate consumption and exports more than offset the increase in investment and the fall of imports, thus, the PIS-COFINS reform induced a decrease in the national real GDP (-0.70%) and in employment (-1.60%).
Although the implementation of mixed taxation regimes only (column PCVA) is a more stylized scenario, its effects can reveal some interesting results when compared to the effects from the complete reform (column PCVAM). We see that the effect on GDP would still be negative but less strong (-0.52%) but, in an opposite way, the total government revenue would fall (-0.34 %), inducing decrease of public savings and, therefore, decrease in investment, which would be the opposite result from the complete reform and shows the importance of PIS-COFINS taxation on imports.
The differences between our results and those reported by Silva and alii (2004) are due the differences in the simulated impacts and in the models´ closure, since in this paper it was assumed full employment of labor and constant capital stocks, which nullify the impacts on GDP by assumption (GDP-neutral effects), and that the nominal exchange rate is endogenous while the general price index (GDP deflator) is fixed, which induce price-neutral effects by assumption (no changes in the aggregate price index). As we were expecting, the reform would induce price increase and product decrease, mainly because of the incidence of PIS and COFINS on imports.
To better understand the relationship between public sector fiscal revenue and PIS-COFINS taxes according to their three different sources, the amounts for the model base year (2003) and the two simulations scenarios are shown in table 5, below. Despite not being a really formal method for comparison, the total collection showed in table 5 is 0.96% higher than the amount verified in 2004, when it is deflated to 2003 value, R$ 91,140. Also, from the above data, we can verify that the total value collected in the partial reform (column PCVA) is very similar to the base value with an increase of just 3.84%. These results partially confirm the hypothesis that only changing the taxation regime would not significantly alter the total PIS-COFINS collection. However, when the taxation on imports are also simulated, the total value collected significantly increases by R$ 17,991 millions (+24.30%) with PIS-COFINS collection on imports (R$ 13,616 billions) representing almost 80% of that growth. 36 This means that the imposition of PIS and COFINS on imports were the major determinants of fiscal revenue rise.

Sectoral impacts
In this section we will analyze the impacts at sectoral level, once each sector performance can vary in a significant way comparing to others. The analysis will be based on a combination of prices and quantities indicators represented by real gross revenues in the table below. The least benefited sector is Leather and shoes and its performance can be explained by the fall of quantity produced, basically due to decline in exports and consumption. The effects on Coffee and Sugar industries are understood by a similar explanation. The effects on Steel industry and Retail/Wholesale trade are explained in a different way. These two sectors are benefited in the taxation reform by paying a lower indirect taxes amount that reduces their production prices. Besides this price decrease, the Steel industry also presents fall of output due to the reduction of its domestic sales and exports. In the case of the Retail/Wholesale trade the increase in output is not strong enough to offset the effects on prices. 35 The base year total value of this table has a small difference from the data reported in table 1 and 2 above because the later has some deductions due to total taxes GDP participation methodology adopted by "Receita Federal". 36 Despite not being a good method for comparison, the total value showed in table 4 agree with the amount verified in 2004, when it is deflated, R$ 91,140 (see Receita Federal, 2006), since it is just 0.96% higher than this value.
On the other hand, the fifth most benefited sector are the Services provided to firms that would benefit from a price effect due to a difficulty in finding substitution for its products. The two most benefited sectors, Chemical elements and Non-iron metallurgy share the property of increases in production prices and quantities. These combinations of effects are possible in the context of a strong sectoral import substitution in the market of intermediate inputs. The increase in the real gross revenue of the Rubber and the Plastic industries are due to increase in their prices, above the general price index, that offset the fall in their output levels. Table 4 revealed that the reform would reduce aggregate employment (-1.60%). Table 7, below, reports the fiscal reform impacts on the employment by labor type (line PCVAM). The results show that employment would fall for all categories of workers in the private sector only. The public servants employment does not change because public sector does not follow the behavior of private sector concerning hiring/firing people and so, by assumption, their employment levels are fixed and their labor market adjust only by means of wages.

Impacts on employment and wages
Among workers in the private sector, the effects would be more pronounced among the less skilled ones, regardless their labor contract status (L1 and L3). The second higher impacts would be on employment of more skilled workers, also regardless contract status (L2 and L5). The less affected category would be the formal with average skill worker (L4).
In our interpretation, with lower imports there will be a pressure to overvalue the exchange rate that will tend to make exports more expensive, which will be reinforced by an increase in input prices used to produce exported goods. The sectors in which exports are more sensible to price changes are the most traditional ones. Thus, by exporting less, there would be a tendency for these sectors to produce less and, therefore, to employ less workers, especially the less skilled ones.
The decrease in employment of more skilled workers is due to the fall in the output of sectors that produce goods with higher technological content and demand this kind of worker in a more intensive way (automobiles, auto parts, electronic, electrical, and pharmaceutical).
Now, it will be presented the effects on wages by labor type. It is worth remembering that, it is assumed that the sectoral wage differentials are rigid. Thus, the wage structure can only react to the type of labor. As a consequence, we report, in table 8, below, the changes in real wages for each type of worker without any sector desegregation. Note that the general effect is a real wage fall. The real wage of informal workers (L1 and L2) would fall relatively less comparing to the other categories. Among private sector workers, the decrease in wages would be lower among the less skilled workers (L3 and L4) and the fall strength is proportional to the increase in qualification. The higher reduction of public servants' earnings is due to the assumption that the equilibrium in their labor market is almost exclusively achieved by means of wages adjustments.
In general terms, it does not seem that there was a labor category that had benefited more or less from the reform in a very significant way. Classifying the workers from the most to the less benefited worker group, according to the impacts on employment, we would have the ordering: public servants, formal with average skill, highly skilled and low skilled. However, according to the impacts on average real wage, we would have almost the opposite ordering. The classification of workers into informal, formal private and formal public categories also does not show any pattern.

Impacts on household income
The effects of the fiscal reform on household income are presented in table 9, below. The results show that all types of households have their real incomes negatively affected by the fiscal reform. It is notorious that only the poor urban families headed by non-active individuals (F2) would present a much lower decrease in their real income (-0.16%) comparing to the fall experienced by the other types of families (stronger than -1.18%). This happens because the total income of this family (F2) presents the lowest dependency on labor earnings.
In Brazil, the labor income has large weight in the total the household income. Nevertheless, income transfers have an important participation, mainly for the poor households by means of the social security retirement pensions and benefits, besides the direct transfers from social programs (as "Bolsa Família").
Considering the distribution of impacts on families' real income, it is not clear that the reform had affected the inequality in income distribution in Brazil. However, the reform had almost linearly reduced the average real income in the country and had induced a decrease in employment and consumption, then, it is possible to interpret the general impact from the reform as a welfare reducing one.  Therefore, taking in account the income levels, the main losers of PIS-COFINS tax reform would be the poorest households. Even it is not so clear to infer that the reform would unequivocally worse the income inequality, it is reasonable to expect that it had increased the poverty gap once the poor families real income levels would fall, reducing more their, already low, consumption levels.

Conclusion
In this paper, we have investigated the effects of a change in the Brazilian Fiscal System on some economic indicators, also trying to infer the impacts on social welfare. More precisely, together with macro indicators we have estimated how wages and employment structures, as well as the household income distribution would react to the tax reform characterized by the introduction of a mixed (cumulative and non-cumulative) taxation system for PIS/PASEP and COFINS, and by their incidence on imports of goods and services.
Despite the increase of Government fiscal revenues and the reduction in the public deficit, that induces increase in investment, the results show that the reform would have adverse effects on macroeconomic aggregates, as GDP, general price level, employment, consumption and external trade flows (exports and imports). Also, the intensity of changes induced by the reform would depend on the taxation level of PIS-COFINS on imports. As imports are taxed, the public sector indicators would become better but the adverse effects on other macroeconomic would be enhanced, except for investment.
For the labor market, we can notice a clear general deterioration, but the effects differ among categories of workers. The negative effects on the employment structure are concentrated among less skilled workers in the private sector (L1 and L3), regardless their labor contract status. The second higher impacts would be on employment of more skilled workers (L2 and L5), also regardless contract status. These effects are due to the decrease in exports of sectors that represent high shares in these workers distribution along productive activities. Again, the intensity of results would be enhanced with taxation of imports.
There would be a general welfare loss for all families. Only the poor urban families headed by non-active individuals would present a relatively lower fall in income due to their lower dependency on labor earnings with relation to the other types of families. All the other families would present percentage changes in income very close to the others. Therefore, considering the income levels, the main losers of PIS-COFINS tax reform in the simulation are the poorest households. Even it is not so clear to infer that the reform would unequivocally worse the income inequality, it is reasonable to expect that the reform had increased the poverty, once the poor families' real income levels would fall, reducing more their, already low, consumption levels.
Finally, it deserves be emphasized that the imposition of PIS and COFINS taxes on imports induced a stronger and much more relevant impacts than the effects related to domestic flows, which consisted in partially changing the tax base from firms' gross revenue to firms' value-added, especially because of the induced increase in the general prices level, whose control was, and still is, the main objective of the Brazilian economic policy in a broad sense.