Report No. 44939-UY Uruguay Poverty and Social Impact Assessment of the Tax Reform May 19, 2008 Argentina, Chile, Paraguay & Uruguay Country Management Unit Poverty Reduction and Economic Management Unit Latin America and the Caribbean Region Document of the World Bank TABLE OFCONTENTS FOREWORDAND ACKNOWLEDGMENTS .............................................................. v EXECUTIVE SUMMARY ................................................................................................ i 1 1 2.. INTRODUCTION.................................................................................................... 1 2.1. Description o f the Tax Reform ................................................................................... 2 URUGUAY'S TAX REFORM................................................................................ 2.2. ExistingStudies on Tax Incidence andThe Tax Reform inUruguay........................ 6 3. METHODOLOGY................................................................................................... 6 3.1. Simulation o f Impact o f IRP and IRPF ...................................................................... 8 3.2. Simulation o f Impact o f Indirect Taxes ...................................................................... 8 10 3.4. Scenarios Simulated inthe Analysis ........................................................................ 3.3. Income Capture through the Household Survey....................................................... 4. 12 4.1. Impact of Direct Taxes ............................................................................................. IMPACT OFTHE TAX REFORMON EQUITY AND POVERTY................12 13 4.3.1 Workers ..................................................................................................... 13 4.3.2 Pensioners.................................................................................................. 17 4.2. Impact o f Indirect Taxes........................................................................................... 19 4.3. Impact o f Combined Total Effect o fthe Tax Reform .............................................. 20 20 5. 4.3.1 Change inTax Burden............................................................................... 28 References......................................................................................................................... CONCLUSION....................................................................................................... 29 Annex 1. Summary o f the Principal Modifications Under the Tax Reform: New taxes and Eliminated taxes ................................................................................................................ 31 Annex 4. Sensitivity Analysis To Assumptions OnCOFIS And IncomeData ................41 Annex 3.Poverty, Inequality andProgressiveness Indicators........................................... Annex 2.Uruguay Tax Incidence Studies ......................................................................... 39 47 Annex 5. Concentration Curves And Progressiveness Indices.......................................... 51 55 Annex 6. Statistical Annex ................................................................................................ Horizontal Inequality, RerankingandVertical Inequality ................................. 57 LISTOFTABLES Table 2.Uruguay: Estimates o fthe revenue-neutral impact o fthe tax reform ............................... Table 1.Uruguay. LatinAmerica and OECD: Comparison o ftax structure (% o f GDP)..............2 5 Table 3.Treatment of various incomes to estimate IRPFdeductions ............................................. 8 Table 4.Comparison o fthe average householdincome. EGMvis-a-vis ECH. per geographical 11 Table 5.Consistency o f IRPestimates (UR$thousands. unless otherwise stated) ....................... area (inpercent) ............................................................................................................................. 11 Table 6. Capture o frental income inthe ECH.............................................................................. 11 Table 7.Meantax rate o fIRPand IRPFby decile (%)................................................................. 13 Table 9. Situation o fworkers under the post-reform tax system. by worker characteristic ..........15 Table 8.Workers paying more under the post-reform system...................................................... 14 Table 10. Situation o fworkers under the post-reform tax system. by industry ............................ 17 Table 11.Pensioners: Mean tax rate of IRP andIRPFby decile (%) ........................................... 18 Table 12.Uruguaytax reform: Summary of impacts on workers and pensioners ........................ 19 Table 13.Mean indirect tax rate (IVA + COFIS) onhouseholds before and after the reform. by20 decile (%)....................................................................................................................................... Table 14. Disposable income andaverage tax burdenbefore and after the tax reform (in December 2006 prices) .................................................................................................................. 21 Table 15.Inequalitymeasures before and after taxes ................................................................... 24 Table 16. Confidence intervals for inequality measuresbefore and after the taxes ...................... 24 Table 17.Uruguay. LatinAmerica and OECD: Comparison o f inequality before andafter taxes ....................................................................................................................................................... 25 Table 18.Impact o fthe tax reform on poverty and extreme poverty............................................ 26 Table 20.Effect o fthe tax reform onpoverty for those aged under 18 and above 64..................26 Table 19.Confidence intervals for poverty measuresbefore and after the reform ....................... 27 LISTOF TABLES December 2006 prices) .................................................................................................................. Table A 1. Disposable income and average tax burdenbefore and after the tax reform (in 48 49 Table A 3.Impact o fthe tax reform on progressiveness (scenarios 1.1and 11.2) ......................... Table A 2. Inequalitymeasuresbefore and after taxes.................................................................. 55 Table A 4.Breakdownof the redistributive effect o ftaxation...................................................... 56 Table A 6.Labor Income: Minimum, mean andmaximum income, per decile (inUR$)............57 Table A 5.Pre-reform and post-reform tax system....................................................................... 58 Table A 7.Variation inlabor income (disposable) by decile (UR$)............................................. Table A 8. Situation o fworkers under the post-reform tax system, by occupation (one digit) ....59 58 Table A 9.Situation o fworkers under the post-reform tax system, byworker characteristic ......60 Table A 10. Situation o fworkers under the post-reform tax system, by industry......................... 62 Table A 12.Pensions: Minimum, mean andmaximum value, by decile ...................................... Table A 11.Situation o fworkers accordingto type o f occupation (selected occupations) ..........61 62 Table A 13.Variation o f average pensions by decile (UR$)......................................................... 63 LISTOFFIGURES Figure 1.Meantax rate on labor income IRPand IRPF............................................................... 14 Figure2.Variation inaverage incomeby decile due to the tax reform (%) ................................. 14 Figure 4.Variation inaverage pensions due to the tax reform. by decile (%) .............................. Figure3.Mean tax rate onpensions underthe IRP and IRPF. byvingtile (%)............................ 18 19 + vingtile (%) .................................................................................................................................... Figure5.Mean indirect tax rate (IVA COFIS) on households before and after the reform. by20 regime and the post-reform situation............................................................................................. 23 Figure 6.Tax burden as share o fper capita pre-tax income: Comparisonbetween the pre-reform Figure 9.Differencesbetweenthe Lorenz after tax income curve and the concentration curves: 53 Figure 8.Pre-reform concentration and post-reform curves: Indirect and direct income taxes ....42 Figure 7.An example o f a concentration curve for a progressive tax ........................................... 52 Tax burdenredistnbution approach............................................................................................... . . . curves: Incomeredistribution approach ........................................................................................ Figure 10.Differencesbetween the Lorenz after tax income curve compared to the concentration 53 Figure 11.Non-Parametric estimate o fthe differentiation o ftaxes with respect to income level 54 FOREWORDAND ACKNOWLEDGMENTS The objective o f the poverty and social impact (PSIA) analysis o f the Uruguay tax reform is to assess whether the tax reform i s likely to have significant effects on equity and poverty. The tax reform came into effect inJuly 2008. Here, an assessment o f the impact o f the reform on equity andpovertyis presented. Rather thanbeing a discrete piece o fwork, the PSIA is intendedto be a programmatic activity. The PSIA analysis commenced as an activity to support the first Programmatic Reform Implementation Development Policy (PRIDPL I) operation, approved by the World Bank Executive Board in May 2007, which supports the implementation o f the tax reform. The study is consistent with the FY2005-2010 Country Assistance Strategy (CAS) objectives for the Bank's non-lending program, which aimed at mainstreaming the use o f PSIAs. Whenpost-tax reform data becomes available, the team aims to follow-up with a further analysis on the impact of the reform as requested by the Government. It is intended that the second stage study will take into account modifications inindividualbehavior inresponse to the tax reform. This report was prepared by a team o f World Bank staff and consultants under the guidance o f Mauricio Carrizosa (former Sector Manager, LCSPE), Rodrigo Chaves (Sector Manager, LCSPE), James Parks (Lead Economist and Sector Leader, LCSPR) and David Yuravlivker (Country Representative, LCCUY). The study team included Veronica Amarante, Rodngo Arim and Gonzalo Salas, researchers at the Universidad de la Republica, who prepareda background paper containing the micro-simulation analysis presented inthe study. The report was written by Emily Sinnott (Country Economist, Task Team Leader, LCSPE), with excellent research assistance provided by Paula Cobas Barque (Junior Professional Associate, LCSPE) and Bemardo Diaz de Astarloa (Consultant, LCSPE). The team benefited greatly from comments from Fernando Lorenzo, Director o f the Macroeconomic Advisory Office, Ministryo f Economy and Finance, Uruguay. The finalization o f the report profited from the comments given by Carlos Grau (CINVE), Hugo Vallarino (DGI) and participants at a seminar at the Universidad de la Republica in Montevideo held in December 2007. Finally, the team i s grateful to the peer reviewers for comments: David Coady (PSIA Group, Fiscal Affairs Department, IMF), Eduardo Ley (Lead Economist, PRMED), and Maria Beatriz Orlando (Senior Economist, LCSPP). EXECUTIVESUMMARY The Poverty and Social Impact Assessment (PSIA) analyzes the impact o f the tax reform, which came into effect in July 2007, on tax incidence and poverty in Uruguay. The essence o f the reform i s the introduction o f a dual personal income tax, which taxes labor income at progressive rates and capital income at lower, proportional rates. A further modification is the reduction inthe revenue share o f indirect taxes. The study aims to provide information to inform policy discussion on distributional implications o f tax reform. In addition, it gives impetus for furthermore sophisticated analysis o fcurrent andproposed tax reforms. Indesigningatax system, atrade-offexistsbetweenefficiency, equityandadministrative simplicity. The paper focuses on one aspect o f this trade-off by evaluating the equity impact o f the tax reform inUruguay. Neither the efficiency o f the post-reform tax system nor the effect on tax administration i s examined. Assessing the distributional impact o f a tax reform is important, firstly, as there is a potential to mitigate the equity-efficiency trade-off in the design o f tax structures, and secondly, as the expenditure side o f the budget can then be employed to diminish any adverse distributional impacts. The aim o f the tax reform in Uruguay i s to create a more efficient and equitable tax structure, and to promote productive investment. The reform i s designed to be broadly revenue neutral. The report concentrates on the equity dimension o f the reform. Inthis regard, the pre- reform tax system was somewhat inequitable due to the importance o f consumption taxes and the partial taxation o f income. The pre-reform tax system did not meet the basic criteria for horizontal equity. Direct taxes on individual income were derived almost exclusively from labor income from employees and pensions, while personal income from capital sources, such as financial assets and property, was tax exempt, irrespective o f whether it was generated domestically or abroad. With regardto the impact o f the reform on equity, the study focuses on the impact o f eliminating the personal receipts tax or wage tax (Impuesto a las Retribuciones Personales, IRP) and the implementation of the dual personal income tax (Impuesto a la Renta de la Personas Fisicas, IRPF), as well as the effects o f reducing the VAT rate and eliminating the social security financing contribution tax (Contribucibn a1 Financiamiento de la Seguridad Social, COFIS). The analysis employs a static micro-simulation model using household survey data to calculate for each individualhousehold the total amount o f direct and indirect taxes due before and after the tax reform. This makes it possible to assess how each individualhousehold i s affected, and thus calculate the poverty and income inequality indicators before and after the reform. These static models, unlike behavioral models, do not incorporate changes in individuals' behavior inresponse to changes in the tax system. However, using a simple micro- simulation approach for the analysis entails certain benefits: it allows comparison with previous studies done on Uruguay and other countries; and it is a transparent method that generates results, which are easy to communicate to a general audience. The welfare measure used i s individualhousehold income taken from the 2006 Expanded National Household Survey (Encuesta Nacional de Hogares Ampliada, ENHA). Income rather than a consumption measure is used as at the time when the analysis was done the latest i household expenditure survey was over a decade old. Nevertheless, the Household Income and Expenditure Survey is employed in the analysis in order to provide detail on consumption patterns (Encuesta de Gastos e Ingresos de 10s Hogares, EGIH 1994-1995). The average income tax rate falls for the bottom 80 percent of income earners under the post-reform tax structure 14 , (inpercent) 1 12 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10 Deciles o flabor income Pre-reformwage tax (IRP) W Dualpersonalincome tax(IRPF) Source; Based on the ENHA. To begin, the study examines the impact o f the replacement o f the wage tax (IRP) with the dual personal income tax (IRPF). Results indicate that the replacement o f the IRP with the IRPF results in an improved situation for the bottom 80 percent o f labor income earners, for whom the mean tax rate falls. The mean direct tax rate increases for the top 20 percent o f taxpayers. The mean tax rate here refers to the average IRP and IRPF payments by decile as a share o f labor income, including any labor income from secondary occupations. Workers in the tenth decile are calculated to pay, on average, 7 percentage points more tax on their labor income, which implies an increase in their tax burdenby 150 percent. For the ninth decile, the increase inthe meantax rate i s much lower at 0.3 percentage points. For pensioners, the tax burden i s increased from the eighth decile onwards due to the introduction o f the IRPF. Pensioners paid little wage tax on pension income prior to the tax reform. The top 20 percent o f pension earners faced a tax rate o f 2 percent prior to the tax reform. Following the reform, for the top 30 percent o f pension earners the difference between the average tax rates before and after the reform is inexcess o f 8 percentage points. The bottom 70 percent o fpension earners continue to pay no tax. .. 11 Indirect tax rates fall by over 2 percentage points on average for under the post-reform tax structure (in percent) 17 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10 Deciles o f household income Pre-reform indirect tax rate Post-reform indirect tax rate Source: Basedon the ENHA. The analysis next examines the impact o f the reform o f indirect taxes on household income. The pre-tax reform situation is estimated to have been regressive, inthat the percentage decrease inincome due to indirect taxes is higher for the bottom 40 percent compared to the top 60 percent o f income-earninghouseholds. All households benefit from a reduction inthe indirect tax rate due to the tax reform, which falls by on average over 2 percentage points over the different income groupings. In terms o f average indirect tax rates, the post-reform indirect tax structure remains regressive. To restate-as i s regular inthis type o f analysis-it is assumed that indirect taxes on goods are shifted completely to consumers. An analysis o f the total impact o f the reform shows it to be progressive. The grey and black smooth lines represent total taxes before and after the reform, respectively. Under the pre- reform system, the tax burdenas a share o f pre-tax income remains relatively constant (close to 12percent) throughout the distribution curve. The pre-reform system was neutral from the point o f view o f inequality. This neutrality arose from the combination o f regressive consumption taxes and progressive personal income taxes. The reform alters the relative neutrality o f the pre- reform tax system. There i s a monotone increase o f the tax burden from the median onwards. As a result, the global tax burdeno fthe households below the 16thpercentile falls, while it becomes markedlyhigherinthe last three vingtiles (five-percent bands) o fthe distribution. If the total impact on households is considered, the reform represents a progressive change. Taking into account changes to VAT, COFIS and the introduction o f the IRPF, along every measure considered the tax reform has a small, but positive impact on equity. The Gini coefficient o f after tax income i s estimated to fall due to the reform: The Gini coefficient o f before tax income i s calculated as 0.45 and the Gini coefficient o f after tax income following the reform i s calculated as 0.44 (the pre-reform Gini o f after tax income is 0.45). ... 111 The combinedimpactof the tax changeis to reducethe tax burden for the bottom80 percent and to increasethat of the top 20 percentof incomeearners (Tax burden as a share ofper capita pre-tax income) 4% 2% 0% 8% 6% 4% 2% 0% I 2 3 4 5 6 7 8 9 I O 11 12 13 14 15 16 17 18 19 20 - - Householdincomevingtile VATtaxes beforetax reform 4 'IRP - Total taxes before tax reform *IRPF VATaftertaxreform -Total taxes after tax reform Source: Based on the ENHA. Disposableincomeincreasesfor the bottom80 percent of households after the reform, but falls for the top 20 percent of householdsdue to the direct tax change - - Variationdue to change indirecttaxes Variation(total) 1-8%' I Source: Based on the ENHA. iv The reform also has a small, but beneficial, impact on the poverty headcount. Poverty incidence is estimated to fall by 0.7 percent due to the change in the tax structure. The slight reduction in poverty i s fimdamentally due to the price reduction effects generated by the modified consumption tax regime on the basic food basket and the poverty line. Ineaualitv measures are estimatedto fall following the tax reform Before tax reform After tax reform After After After After After After Before taxes indirect direct taxes indirect direct v u k d h taxes (1) taxes taxes (2) taxes taxes ( 2 H1) Gini coefticient 0.454 0.453 0.457 0.448 0.442 0.456 0.440 -0.011 Entropy0 0.351 0.350 0.357 0.341 0.338 0.361 0.334 -0.012 Entropy 1 0.376 0.376 0.382 0.366 0.353 0.380 0.349 -0.023 Source: Basedonthe EGIH and the ENHA. V Vi 1, INTRODUCTION 1. The Poverty and Social Impact Assessment (PSIA) analyzes the impact of the reform of the tax system on tax incidence and poverty in Uruguay. The tax reform, which came into effect inJuly 2007, has been a centerpiece o f the government's reform program.' The essenceo f the reform i s the introduction o f a dual personal income tax, which taxes labor income at progressive rates and capital income at lower, proportional rates. The aim i s to create a more efficient tax structure, while promoting greater horizontal and vertical equity, and stimulating productive investment and employment. 2. The PSIA study providesan analysisof the impact of the tax reform on equity and povertybasedon micro-simulationsusinghouseholddata. Itbuilds on existing studies on the impact o f the tax reform inUruguay. The analysis adds value in a number o f ways. It is the first studyto look at the impact o fthe actual tax reform implemented. Previous studies were based on the 2005 government proposal for tax reform. The PSIA is the first analysis to quantify with precision the magnitude o f the impact on equity, as well as its incidence for different groups o f households and individuals. The study is more comprehensive than those preceding it, both in terms o f the equality and tax incidence indicators analyzed and in that it corrects the household income data for underestimation. An additional subject covered, deserved o f attention but not treated to date, is the impact o fthe reform on poverty. 3. The study proceeds as follows. Insection 2, a description o f the pre- and post-reform tax system is given. A summary o f existing studies on tax incidence and the impact o f the tax reform on equity inUruguay is presented. Section 3 provides a description o f the data and methodology used inthe analysis. This is followed by a presentation o f the main results regarding the impact o fthe tax reform on equity andpoverty insection 4. Conclusions are presented in section 5. 2. URUGUAY'S TAX REFORM 4. In reforming the tax system, the economic costs of collecting and administering taxes need to be considered.Not all taxes have the same costs. Taxes may be associated with economic distortions if they cause changes in market prices. The incidence o f taxes may differ for different groups. There may be ease o f tax collection reasons to favor one tax over another. In designing a tax system, a trade-off then exists between efficiency, equity and administrative simplicity. The paper focuses on one aspect o f this trade-off by evaluating the equity impact o f the tax reform inUruguay. Neither the efficiency o f the post-reform tax system nor its effect on tax administration is examined. Assessing the distributional impact o f a tax reform is important, firstly, as there is a potential to mitigate the equity-efficiency trade-off in the design o f tax structures, and secondly, as the expenditure side o f the budget can be employed to diminish any adverse distributional impacts. 1. The tax reformlegislation was approvedby the UruguayanParliament inDecember 2006. 1 2.1. DESCRIPTION THE TAX REFORM OF 5. The aim of the tax reform in Uruguay is to create a more efficient and equitable tax structure, and to promote productive investment. The reform i s designed to be broadly revenue neutral. The reform introduces a dual personal income tax (IRPF), which combines a progressive tax schedule for labor income with a low flat tax rate on capital income. It reduces VAT and corporate tax rates, streamlines and eliminates minor taxes, and broadens the VAT base. In addition, the government has made substantial efforts to strengthen tax administration. Details o f the tax reform are provided in Box 1. A more detailed description o f the post-reform tax regime is given inAnnex 1. 6. Personal taxation has constituted a small share of overall tax effort in Uruguay (Table 1). In contrast, consumption taxes have been a substantially more important source o f revenues than on average for Latin America or the OECD. Tax revenue is the main source o f income for the government in Uruguay. Excluding social security, taxes represented 71 percent o f total revenue on average over 2004-06. The tax burden i s unlikely to be reduced in the medium term. Following the 2002 crisis-with public sector debt reaching a higho f 97 percent o f GDP in2004'-the government has sought to strengthen the fiscal position in order to reduce macroeconomic vulnerability. The level o f disposable fiscal income, i.e. resources available to the state after servicing debt and meeting the commitments o f the social security system, i s about 40 percent o fthe total revenues (Barreix and Roca, 2007). Table 1.Uruguay, Latin America and OECD: Comparison of tax structure (YOof GDP) Uruguay LatinAmerica OECD (2005) (2004) (2005) Tax revenuea 28.0 20.2 36.2 Value added tax (VAT)b 12.8 5.8 6.5 Income tax 4.0 3.8 13.0 Corporation tax 3.1 2.6 3.8 Personal income tax 0.9 1.2 9.2 Social security contributions 7.0 2.8 9.2 Source; IMF (Uruguay), Barreix andRoca (2007) (Latin America) and OECD. a Includes social security contributions (pensions). For Uruguay, value added taxes include specific regimes and other indirect taxes. 7. Prior to the tax reform, direct taxes on individual income were derived almost exclusively from labor income from employees and pensions, while personal income from capital sources, such as financial assets and property, were tax exempt, irrespective of whether it is generated domestically or abroad. The system was based on the principle o f only taxing labor and capital when they came together in a business enterprise, when labor income would be subjected to the wage tax and capital income to one o f the corporate income tax regimes. This tax structure led to capital income from financial investments being exempt from taxation, while income for independent professionals paid fixed taxes o f a small amount. Indirect tax rates inthe pre-reform system were relatively high. The country had the highest consumption tax rate o f Latin America (26 percent), when the joint effect of V A T and COFIS is taken into account (23 percent and 3 percent, respectively) (Barreix and Roca, 2003). 2 Box 1.Uruguay: Main Components of the Tax Reform The tax reformbillbecameeffective on July 1,2007. The mainreforms to the tax system are as follows: Introduction of a personal income tax, the Impuesto a las Rentas de Personas Fisicas (IWF), on all domestic sources o f income. This tax replaces the wage tax (IRP). The tax has a progressive tax schedule with six income tax brackets and rates ranging from 0 to 25%. Deductions are permitted for social security contributions, fixed health allowances for pensioners and fixed deductions for children. Capital income is taxed at 12%, a lower rate than for labor income. Nonresidents' income generated in Uruguay is taxed at 12%. Interest income from public debt i s to remain tax exempt. Rationalization of the corporate income tax with the introduction o f a tax-Impuesto a las Rentas Empresariales (1RAE)-replacing four business taxes (Impuesto a las Rentas de Industria y Comercio (IRK), Impuesto a las Rentas Agropecuarias (IRA), Impuesto a las Comisiones (ICOM), Impuesto a las Pequefas Empresas (IMPEQUE)). These are be consolidated into a single 25% corporate income tax, a reduction from the former rate o f 30%. There is a surtax o f 7% on dividend income. Reduction of the rate of the value-added tax (VAT). The basic VAT rate i s reduced from 23% to 22% and the reduced rate falls from 14% to 10%. The VAT base i s broadened and various IVA exemptions eliminated. The base expands by incorporating the formerly exempt tobacco products, and financial services at the 22% rate, and h i t s and vegetables, health services, public transportation, and the first sale o f real estate properties at the 10%rate. VAT tax exemptions remain inplace for certain agricultural products such as milk. Unification of the employer's social security contribution rate and elimination of sectoral exemptions. The contibution rates paid are unified into a single 7.5% general rate. The exemptions for employers' contributions inthe rural, industrial, and transportation sectors are eliminated; inthe case o f the rural sector through the adoption o f a special regime. Elimination of several taxes (equal to an estimated revenue reduction o f about 5% o f GDP), including the taxes on corporate income and wages; the consumption tax on industrial goods used to finance social security contributions (Impuesto de Contribucidn a1 Financiamiento de la Seguridad Social, COFIS), and the taxes onbanking assets, health services. 8. The key change to the system is the introduction of a dual personal income tax, combining a progressive tax schedule for labor income with a low flat tax rate on capital income. The progressive labor income tax-the dual personal income (Impuesto a la Renta de las Personas Fisicas, IRPF)-has six marginal income tax rates ranging from zero in the first bracket to 25 percent inthe 6thbracket. The flat corporate tax, the tax on income from corporate activities (Impuesto a la Renta de las Actividades Empresariales, IRAE), varies from 0 percent inthe case of assets heldabroad to 12percent inthe case offoreign currency domestic deposits. The personal income tax to be levied on all labor and capital income sources results in a broadening o f the tax base and improves horizontal equity inthe system. The introduction o f the dual system was based on the need to balance fundamental equity objectives with the need to provide conditions that favor investment and make the tax structure compatible with current revenue collection capacities. The dual tax system inUruguay follows a similar "small country" strategy as that pioneered by the Nordic countries in the 1990s. It responds the plight o f small open economies that are unable to trace non-domestic sourced income in the face o f increased capital mobility across borders. For these countries, a low flat tax on capital income reduces the risko ftax evasion from residents with capital investments abroad. 9. The pre-reform system was judged not to provide adequate incentives to promote investment. The corporate income tax was relatively high (although lower than in the regional 3 economies) and distortive, whereby some activities contributed a substantially lower share o f their income than was the average for the economy. Manufacturing and construction firms, for example, were exempt from paying social security contributions; instead, retail and services paid a 12.5 percent rate. In addition, the pre-reform tax system included certain taxes, such as the IMABA, which increased the cost o f credit, and therefore deterred investment. A thorough overview o f the shortcomings o f the pre-reform tax system is presented in Barreix and Roca (2003). 10. The reform increases the neutrality of the tax system across factors of production and the reduction in the corporate income tax from 30 to 25 percent is expected to foster private investment. The unification o f the employer's social security contribution at a 7.5 percent rate creates a more level playing field between manufacturing and construction on one hand (which were exempt from social security contributions) and retail and services (which paid 12.5 percent contributions) on the other. Similarly, the elimination o f various VAT exonerations creates a more level playing field across sectors. By taxing income from all factors (labor, capital, or any combination o f the two), the reform creates less perverse incentives in terms o f the business structure adopted. For example, professionals which were not taxed under the previous regime faced disincentives to be incorporated. Finally, the elimination o f COFIS implies that there i s no longer an asymmetry between firms that were allowed to collect this tax on their sales (thereby taking credit for what they paid on their inputs) and those that were not. Under the pre-reform regime, the latter still hadto pay COFIS on inputs and so faced a larger net tax burden. 11. The pre-reform tax system was based on a large number of taxes; yet, the bulk of revenues came from just a few. During the last twenty years, a customary practice to sustain revenue collection inUruguay under adverse conditions has been to incorporate new taxes or to increase the rates o f some already existing taxes. This generated a sizable number o f taxes, many o f which turned out to be difficult to administer and to generate low levels o f revenue. Pre- reform, there were 12 different direct taxes and 18 indirect taxes, with 20 taxes each individually accounting for less than 1 percent o f revenue. Just five taxes, three consumption taxes (VAT, IMES12and COFIS3) and two income taxes-the personal receipts tax/wage tax (Impuesto a la Renta de la Personas Fisicas, IRP) and the income tax rate for corporations and individuals (Impuesto a las Rentas de Industria y Comercio, IRIC)-provided over 85 percent o f tax revenues (excluding social contributions). The proliferation o f small taxes was inefficient as it led to a large number o f distorting and low-revenue generating taxes, and substantial tax exemptions. 12. The elimination of various low-yield distortive taxes and a number of tax exemptions brought about a substantial simplification of the tax system with beneficial efficiency implications for tax collection resource allocation. The tax reform repealed fourteen taxes and empowered the Executive to annul another three, some o f which are incorporated in the new taxes, while others are eliminated with the aim o f reducing the indirect tax weight, and some merely for reasons o f administrative simplification. The tax reform also unified the 2. IMESI (Impuesto Especifico Interno or Specific Internal Tax) is an excise tax levied on the first sale of goods including alcoholic beverages, soft drinks, cosmetics, tobacco products, and motor vehicles. 3. COFIS (Impuesto de Contribucion a1 Financiamiento de la Seguridad Social) was a type of wholesaler's VAT used to finance social security. 4 contribution rate o f employers to social security, with the objective o f eliminating the sector discriminations with regard to labor inputs. 13. The reformaims at an improvementin vertical equity through the reductionof the revenueshare of indirecttaxes and the increasein the revenueshare of direct taxationThe elimination of the COFIS tax (3 percent) and a small reduction inthe VAT rates (from 23 to 22 percent) along with the introduction o f the personal income tax reduces the share o f indirect taxes in total revenue. The proportionally higher reduction in the VAT rate for basic consumer goods (from 17 to 10percent) also contributes to equity. 14. One of the aims of the reformis to increasehorizontalequity (treatingequallythose of equal income) and vertical equity (treating unequally those of unequal income). Horizontal equity should improve as a system that only taxed dependent labor income i s replaced by a comprehensive personal income tax. It is additionally likely to lead to an improvement in vertical equity as the labor income tax has progressive marginal rates. In spite o f the overall beneficial impact on equity o f the reform, the implication o f capital income being taxed at a lower rate than labor income is that neither horizontal nor vertical equity are being maximized by the reform. The justification for the adoption o f a lower tax rate on capital compared to labor income sources i s that as capital is a highly mobile factor the dual tax treatment mitigates arbitrage toward (untaxed) foreign capital instruments. Table 2. Urueuav:Estimatesof the revenue-neutralimDactof the tax reform Gains Concept Losses Losses ("$ ml) GDP2005) (Yo of Gains (%GDP) of 2005 Eliminated taxes 825 4.9 Modification to VAT taxes 155 0.9 163 1.o o/w Reduction inVAT rate 155 0.8 o/w Eliminationof COFIS 25 0.2 o/w Broadening of V A T base 163 1.o New taxes 743 4.4 o/w Corporate income tax (IRAE) 394 2.3 o/w Personal income tax (IRPF) 349 2.1 Total 980 5.8 907 5.4 Deficit 73 0.4 Source: Ministry of Economyand Finance, Uruguay, 2006. Note: Calculatedby the Ministry of Economyand the Financebasedon 2005 revenuecollection. 15. The tax reform is designed to be broadly revenue neutral, with a slight fall in revenueof 0.4 percentof GDP projectedby the Governmentduringthe preparationof the tax reformlegislation(Table 2). The losses from the eliminated taxes and the reduction inVAT rates are due to be made up by gains from the broadening o f the VAT base, and the introduction o f the personal income tax (IRPF) and the corporate income tax (IRAE). Therefore, it is not expected to impact negatively on government expenditure. 16. It is assumed that there is full compliance with the post-reform tax system. If tax compliance was instead to vary with the income o f an individual, then this would impact on the 5 progressiveness o f the post-tax regime. For example, in increasing the marginal tax rate, it may be the case that tax compliance falls for higher-income individuals. The Government has reduced the probability o f a fall in tax compliance ensuing due to the tax reform by substantially strengthening tax administrationand collection efforts inUruguay. 2.2. EXISTINGSTUDIESONTAX INCIDENCE AND THETAX REFORM INURUGUAY 17. Studies of the pre-reform tax system find it to have been slightly regressive or neutral interms of its distributive impact (Grau and Lagomarsino, 2002; Perazzo, Robaina and Vigna, 2002). All the existing studies use a direct tax incidence approach and apply a partial equilibrium approach. Findings are that the aggregate impact o f the taxes on which the pre- reform revenue collection mostly relied upon-VAT, IMESI, and IRP-was regressive, if analyzed by the average tax rate imposedby decile andthrough indicators o f income inequality. 18. The IRPhad a neutral impact on the income distribution according to the Grau and Lagomarsino (2002) analysis. The average IRP tax rate rose with income until the seventh decile. This result is due to the lower income share o f wages and pensions (as opposed to other income sources) inthe income o f the highest and lowest deciles compared to the middle deciles, and also due to the a higher propensity for self-employment and informal work. Thereafter, for these higher income strata, the increase inthe IRP tax rate is not adequate to compensate for the reduction inthe share o f overall income taxed by IW. 19. The effective VAT rate by decile shows its weight to be greater for the five lowest income deciles (in proportion of overall income, not in absolute terms). The inequality indices are somewhat lower before compared to after the imposition o f VAT tax. With regard to IMESI, the effective imposition rate does not show a clear pattern by income decile, while the variation o f inequality indicators i s almost zero, indicating neutrality. 20. Existing studies agree that the impact of the tax reform is likely to be relatively progressive, if the post-tax reform system is compared with the pre-reform system. Prior to the presentation o f the tax reform to parliament a number o f studies as to its likely impact on equity had been carried out and some o f the analysis was used as an input for the preparation o f the reform (Grau and Lagomarsino 2002; Perazzo et a12002, Barreix and Roca 2003,2005). The general conclusion o f these studies was that the tax system could be modified to have a positive impact on inequality. A summary o f the various studies on the equity o f the tax system in Uruguay is given inAnnex 2. 3. METHODOLOGY 21. The study evaluates the impact on equity and poverty of eliminating the IFW tax and the implementation of the IRPF, as well as the effects of reducing the VAT rate and eliminating the social security financing contribution tax (Contribucidn al Financiamiento de la Seguridad Social, COFIS). An assessment o f the impact o f the tax reform on economic efficiency or tax administration i s not attempted. The welfare measure used is individualhousehold income taken from the Expanded National Household Survey (Encuestu Nucionul de Hogures Ampliudu, ENHA) 2006. Income rather than a consumption measure is used as at the time when the analysis was done the latest household expenditure survey was over 6 10 years old. Nevertheless, the Household Income and Expenditure Survey is employed in the analysis where detail on consumption patterns was needed(Encuesta de Gastos e Ingresos de 10s Hogares, EGIH 1994-1995). 22. The analysis uses a static micro-simulation approach to assess the impact of the tax reform. Using household survey data, the direct and indirect tax burden is calculated for each individualhousehold. This makes it possible to assess how each household is affected, and thus calculate the poverty and income inequality indicators before and after the reform. The approach used for the analysis is static in that it does not take into account second order or general equilibrium effects. Inthe study, the direct impact o f the change inthe tax system parameters on income is assessed. Using a simple micro-simulation approach for the analysis entails certain benefits: a recent CGE model is not available; it allows comparison with previous studies done on Uruguay and other countries; and it is a transparent method that generates results, which are easy to communicate to a general audience. 23. Changes to the tax system are analyzed using different indicators of income inequality, poverty and tax incidence before and after the reform. A tax is defined as progressive when the average rate (i.e. proportion o f income paid intax) increases with income. Measures o f tax progressivity, inequality and poverty are presented before and after the reform o f the tax system. For the poverty analysis, a simulation exercise is carried out for the change in the value o f the poverty line due to changes in the prices o f the goods, on the basis o f different assumptions on the price elasticity o f supply and demand. A full account o f the indicators used is given inAnnex 3. 24. There are certain limitations associated in applying a static approach. These static models, unlike behavioral models, do not take into account modifications in an individual's behavior inresponse to changes inthe tax structure. For instance, people o f a working age could change their labor participation decisions with regards to working hours or number o fjobs, as a fhction o f how they are affected by the changes in indirect taxes. Something similar happens with consumption expenditure: it is reasonable to assume that household decision-making is impacted by changes in relative prices caused by the reform. However, these types o f modifications in individual behavior are not included in this exercise, since to do so it would be necessary to have information on demand functions for various goods, price elasticities and elasticities o f substitution. A complete modeling o f these aspects would require using a CGE (computational general equilibrium) model, or specific studies for subsets o f goods that merit it because o f their relevance. In short, the results o f the micro-simulation presented in this report are an estimation o f the first order changes in tax incidence, since behavioral changes are not captured. 25. Regarding assumptions for the transfer of the tax incidence, markets are assumed to be competitive and the total indirect tax burden is assumed to fall on the buyers. Inthe case o f consumption, it i s assumed that there i s no tax evasion. For direct taxes this assumption is not made, since the household survey makes it possible to identifythose workers that pay into social security and those who do not (information that is included inthe analysis). 7 3.1. SIMULATIONOFIMPACT OFIRPAND IRPF 26. The analysis of the impact of eliminating the IRP and implementing the IRPF is conducted based on the 2006 ENHA. The survey collects information on disposable income received in the previous month. To calculate the IRP and IRPF payments corresponding to each individual, the before-tax nominal income must be estimated by adding tax contributions to disposable income. In calculating tax contributions, the specific tax conditions facing each occupational group o fworkers are considered, both for the main and secondary occupations. For example, the case o f members o f the military and policemen (contributors to the Military and Police Funds) i s distinguishedto take into account health insurance and pension contributions. Table 3. Treatmentofvariousincomesto estimateIRPFdeductions Concept Deductions allowed Annual tax rate Marginal tax rates Health coverage for children under 18 years: 6.5Upto 60BPC: No payment Labor income Salaries, commission, BPC ina year From60 to 120BPC: distribution o fprofits Aportes a la seguridad From 120to 180BPC: 20% 15% social From 180to 600 BPC: 22% From600 to 1200BPC: Retirement Retirementandpensions Health coverage: Over 1200BPC: 25% andpensions 12 BPC ina year Flat tax rates Capital Rents, leases 12% income Profits None 7% Interest 12% Source: Amarante, Arim and Salas (2007). Notes: The value ofBase de Prestaciones Contributivas (BPC) in February 2008 was UR$1,775. 27. At the household level, income also includesincomefor rentals, leases, interestsand other capital incomes, which are subject to the IRPF. These capital incomes have their own rates and are taxed separately (dual system) (Table 3). In the case o f rental and lease income, income above UR$5,000 per month is taxed at 12 percent. All deposit interest is assumed to be taxed at 12 percent-the tax rate for foreign currency deposits. While tax rates on interest income differ depending on the type of deposit, approximately 86 percent of the deposits in Uruguay's financial system are foreign currency deposits. Therefore, it is assumed that all interest income i s for foreign currency deposits, as the household survey does not indicate the type o f deposit from which interest income originates. Deductions on capital income such as real property tax and the primary tax on leases are not taken into account. This capital income is underestimated in the household survey, so the value o f making this type o f adjustment is doubtful. 3.2. SIMULATION OF IMPACT OFINDIRECT TAXES 28. The analysis considers the reduction in the two indirect taxes, namely the elimination of the 3 percent consumption tax on industrialgoods that was used to finance social security contributions (Impuesto de Contribucih a1 Financiamiento de la Seguridad Social, COFIS), and the reductioninthe VAT maximumratefrom 23 percentto 22 percent and the VAT minimum rate from 14 percent to 10 percent. With regards to the other 8 consumption tax, the excise tax levied on specific goods such as certain alcoholic beverages (IMESI), gradual modifications are being planned to adjust its structure, but no specific changes have been specified. For that reason, the IMESI is not considered in this analysis. Information from the EGM 1994-95 i s combined with data from the ENHA 2006 to calculate the effect o f the reduction in indirect taxes. Previous household income surveys also are used for comparison with other available sources for income data to calculate the underestimation o f household incomes associated with differenttypes o f income. For these calculations the Encuesta Continua deHogares (ECH), which was replaced by the ENHA in2006, is used. 29. The first issue to be addressed is the treatment of the defacto VAT surcharge, the social security financing contribution tax (COFIS), which is eliminated under the post- reform tax regime. The tax is legally applied to all taxpayers registered with the Revenue Authority (Direccidn General Impositiva), andit is not possible to know with certainty the extent to which consumer prices incorporate the COFIS. One alternative i s to assume that consumer prices do not include the COFIS, and that this tax is fully borne by the producers. Another alternative i s to assume that COFIS is fully passed through to the end consumer. This is the assumption used inthe previous analysis reviewed (Barreix and Roca, 2003) and inbackground analysis undertaken on the reform by the Ministry o f the Economy, which stated that the effective tax burden on goods drops from 26 to 22 percent for those that pay VAT at the maximum rate (consisting o f the joint effect o f reducing VAT by 1percent and eliminating the COFIS). 30. Two alternative scenarios are considered in the analysis. The first assumes that the COFIS i s not passed through to the end consumer, i.e. that the reduction inindirect taxes consists solely o f the reduction o f the VAT rate. The second considers that COFIS is passed through in full to the end consumer andthus that the effective tax burden on goods is reduced from 26 to 22 percent in the case o f goods subject to the maximum VAT rate and from 17 to 10 percent for goods subject to the minimumVAT rate. Inboth cases, the assumption i s that VAT is paid 100 percent by households, an assumption which i s usual inthis type o f estimates. Inthe case o f the first scenario, households pay only VAT before and after the reform, while in scenario two before the reform the households pay VAT and COFIS and after the reform they only pay VAT. The actual transfer o f COFIS i s probably an intermediatepoint between these two scenario^.^ 31. The assumption that the full burden of VAT or COFIS falls on the consumer implies that the pre-reformtax system is more regressive than the situation in which part of the burden is taken on by the producer. The assumption o f full pass-through o f consumption taxes to the consumer impacts on equity by passing on to the consumer the full reduction in indirect taxes. Simulations are presented in Annex 4 showing an opposing assumption for COFIS, i.e. assuming that the final burden o f COFIS i s not passed through to consumers. 32. A special treatment is given to health care expenditures. These were exempt from V A T prior to the reform, but now are taxed at 10 percent. However, a tax that affects the sector directly is eliminated: the specific tax on health care services (Impuesto Espec$co a 10s 4. It should be noted that if the evolution o f consumer prices i s analyzed, and inparticular for goods subject to VAT, no major increase is detected at the time ofintroductionofthe COFIS. 9 Sewicios de Salud, IMESSA). It is difficult to forecast how the final price o f these services is affected, and to the extent that the dues paid to the mutual (HMO) are a major expense for households, this aspect o f the tax reform could be relevant. Since this is a regulated price, the degree to which these changes are passed on to consumers will depend on the final policy decision. Based on consultations with a specialist on the subject, the assumption used in the analysis is that the price o fhealth care expenseswould increase by 3 percent. 33. The micro-simulationanalysis assumes that households do not change consumption decisions in responseto a shift in relativeprices,Le. that the demandis completelyinelastic. While this is a strong assumption, the difficulty inbuilding inbehavioral modifications means that it is used inmost studies on tax incidence, both in developed and developing countries. An additional assumption i s that the household's consumptiordincome ratio i s kept constant when the shares o f indirect taxes resulting from the EGIH 1994-95 are allocated to the ENHA 2006. This implies assuming that the household's savings rate is the same inthose two years. The ratio between final private consumption expenditure and gross national disposable income from the national accounts estimated by the Central Bank o fUruguay amounts to 74.1 percent in 1994 and to 76.8 percent in2004 (the last year available). The ratio shows a slight increasing trend inthe period under analysis (with the exception o f 2002). This would indicate that indirect tax payments may be underestimated in the study as a result o f the methodology adopted. However, it is difficult to infer the potential effect on the distributional findings, since no information is available on the evolution o fthe savings rates for the various income strata. 3.3. INCOME CAPTURETHROUGHTHE HOUSEHOLD SURVEY 34. Household surveys, in all countries, often present underestimationsof household incomes, fundamentally due to the difficulties individuals have in reporting variable incomes and also due to intentional under-reporting of income. This is an important limitation for this work, which is completely based on the income captured by household surveys. Attempts at estimating the underestimation in household surveys are often based on comparisons with the income reported in income and expenditure surveys or with the estimates o f the System o f National Accounts (SNA or Sistema de Cuentas Nacionales, SCN). CEPAL (2001) reports on efforts at investigating the pattern o f misreporting inUruguay by matching the incomes captured by the ECH with those o f the National Accounts (SNA). For example, Amarante and Carella (1997) find that the average under-reporting rate for 1989 and 1990 was around 23 percent and 22 percent respectively). Regarding the comparison between household surveys and surveys o f income and expenditures, an existing study for Uruguay found a considerable level o f income sub-capture in the household income survey (ECH) (Mendive and Fuentes 1996). The study finds that the under-capture i s concentrated in the higher income quintiles. The income and expenditures survey (EGM) captures income better since in the process o f data gathering it develops a financial balance sheet at the individual household level and goes deeper into controlling the consistency o f the micro data. The data collection methodology o f the EGIH also implies longer-term contact o f the surveyor with the households, which favors a better capture ofthe income. 35. A recent paper by Arim and Vigorito (2006) states that the 2006 householdsurvey captures around 95 percent of the pensioners based on the administrative records. This 10 result is consistent with Mendive and Fuentes (1996) for pensions, andreaffirms the idea that for that source o f income, the household survey gives a very good approximation (Table 4). 36. Simulations suggestthat the ENHA 2006 is an adequate source to analyze wages and salaries. Regarding salaries, an indirect way o f considering the quality o f their capture by means o f the household survey is comparing the tax revenues for IRP that result from the expansion o f the household survey with the effective tax receipts. Such a comparison was made for 2006. The results suggest that the survey captures 9 percent more thanthe actual receipts (Table 5). Table 4. Comparison of the average household income, EGIHvis-&vis ECH, per geographical area (in percent) Household Salaries Retirements Rents and Transfers and Imputed incomes and wages employed ensions and interests subsidies Rent Montevideo 111.2 103.3 114.4 147.0 99.3 146.0 164.3 110.7 Interior 111.9 105.1 112.1 129.4 95.2 181.0 131.3 125.6 Source: Mendive and Fuentes (1996). Table 5. Consistency of IRP estimates (UR$ thousands, unless otherwise stated) IRPTax Collection Household survey (ENHA, 2006) 5,566,420 Accounting General (Contaduria General de la Nacidn, CGN) 5,092,120 Capture of IRP2006 (ENHNCGN) 9% Source: Amarante, Arimand Salas (2007). 37. Rental income would seem to be significantly underestimated in household surveys. An attempt was made to quantify the underestimation problems inthe 1997 ECH, by comparing the rental data that emerged from this survey with those o fthe SNA in1997.5Comparing the two sources, the ECH only be captures 32 percent o f the total rental income. The E C H does collects information about the implicit rent from own-housing by asking each householdhead to estimate how much it would cost to rent the dwellinghe owns. However, as the SNA figure refers to the gross production value o f the real estate services activity, in theory it does no correspond to household income reported by the ECH. The SNA category includes not only rental o f own property, but also real estate intermediation activities for the rentingand the management o f real estate on a fee or contract basis. Ifit is assumedthat the rental income that accrues to households i s approximately 80 percent o f the gross rental income (excluding these other activities), the result i s that the ECH underestimates rental income by about 60 percent (Table 6). Table 6. Captureof rentalincomeinthe ECH Estimate in National Accounts (SNA) and Household Survey (ECH) Estimation of gross rentalproduction SNA (UR$ million) (1) 8,055 Estimation o fnet rental production SNA (UR$ million) (2) 6,444 Estimation rental income ECH (UR$ million) 2,555 ECH/SNA (taking gross rental measure) 1997 (1) 31.7% ECH/SNA (taking net rental measure) 1997 (2) 39.7% Source; Amarante, Arim and Salas (2007). 5. That total includes rentals collected by the households and also by companies, thereby overestimating the component corresponding to households. 11 38. Simulationssuggest that interestincome is underestimatedin household surveys. An attempt was made to assess the performance of the household survey incapturing interest income derived from deposits. For the calculation o f aggregate interest accruing to households, an estimate o f the stock o f household deposits inthe banking system in2005 was used (Miraballes, 2006) and the corresponding interest was calculated based on the average annual interest rates for 30-day deposits. The total interest resulting from this calculation was compared to that resulting from the aggregation o f deposit interest recorded in the ECH in 2005. This estimation indicates that the ECH captures a mere 23 percent o f the total interest received by resident households from bank deposits. Interestingly, interest income captured by the ECH i s strongly concentrated inthe higher quintiles (8 percent inthe fourth quintile and 86 percent inthe fifthin 2006).6 39. The household survey would then seem to be a good source to analyze wages and salaries and pensions, but one with strong limitationswith respect to capturing rental and interest income. Based on these considerations, the decision was made to work with two scenarios with regardto household income. The first considers the income reportedinthe ENHA without introducing corrections. The second adjusts income from each source to correct for the average sub-capture estimated by Mendive and Fuentes (1996) inMontevideo and inthe Interior, since this i s the only source that analyzes the global capture o f the ECH and gives a breakdown o f sub-capture by source. If no adjustment is made to the income measure, pre-tax income and therefore, the equity-enhancing impact o fthe tax reform are underestimated. 3.4. SCENARIOS SIMULATED INTHE ANALYSIS 40. The results presented in the study, unless otherwise stated, correspond to the following scenario: (i) ENHA household income is not corrected (scenario 1.1); and (ii) COFISis fully passedon to the end consumers (scenario11.2).The sensitivity o fthe results to a change inthese two assumptions is investigated in annex 4. The alternative assumptions are as follows: (i) With respect to income, ENHA income is adjusted with the Mendive and Fuentes (1996) sub-capture coefficients (scenario 1.2); and (ii) COFIS is not passed through to the end consumer (scenario 11.1). Inall scenarios, VAT is assumed to be fully shiftedforward to the final consumer. 4. IMPACTOFTHETAXREFORMONEQUITYAND POVERTY 41. The section analyzes the impact of the tax reform on poverty and income distribution. The results are presented inthree sections. First, the impact o f the elimination o f the wage tax (IRP) and introduction of the dual personal income tax (IRPF) for individuals is considered in section 4.1. The section looks at the impact on an individual's labor income. Next, the impact o f the reduction in the indirect tax rate on household income i s analyzed in section 4.2. Finally, the overall impact o f tax changes, including a decomposition o f the effects o f direct and indirect taxes, on household income i s presented in section 4.3. Here, the impact o f the introduction o fthe dual personal income tax (IRPF) on non-labor income is taken into account. 6. Part o f the underestimation o f bank deposits could be due to the households having bank accounts to support their productive activity when self-employed. Inthis case, interest income is not considered as "household" income, butrather as "business" income. 12 4.1. IMPACT OFDIRECTTAXES 42. The analysis here is focused on the impact o f the reform on the labor income o f active individuals and pensioners. To construct income by decile, total nominal income is used for active individuals, including those who do not contribute to social security, and total pension income for pensioners. 4.3.1 Workers 43. The bottom 80 percent of workers face a lower average tax rate following the tax reform. It is only for the individuals in the ninth and tenth deciles that the post-reform tax system implies a higher mean tax rate compared to the pre-reform system.' The mean tax rate before and after the tax reform, by decile, is given in Table 7. The mean tax rate here refers to the average IRP and IRPF payments by decile as a share o f labor income, including any labor income from secondary occupations. Workers in the tenth decile are calculated to pay, on average, 7 percentage points more tax o f their labor income, which implies an increase in their tax burden by 150 percent. For the ninthdecile, the increase inthe mean tax rate is much lower at 0.3 percentage points. Figure 1illustrates the differences inthe mean tax rate on labor income per vingtile (or twentieth), illustrating the significant increase in the burden at the upper end o f the distribution. The increase inthe average income o f deciles one to eight due to the tax reform i s minor, while the drop inthe average income o f the tenth decile is o f a considerable magnitude (Figure2). 44. Due to the reform on average a greater tax burden results from the ninth decile onwards. Although there are workers who pay more from decile six onwards due to the introduction o f the income tax, there i s a highconcentration o f the workers who pay more inthe last two deciles. O f the all workers, 17.5 percent pay more when the IRP is removed and the IRPFimplemented(Table 8). Ifformal workers only are considered-that i s those making social security contributions-27 percent o f them pay more under the post-reform tax system. Table 7. Mean tax rate of IRP and IRPFby decile (YO) Difference Difference Decile IRP(a) IRPF(b) (b-a) ((b-a)/a)* 100 1 0.66 0.00 -0.66 -100.0 2 0.04 0.00 -0.04 -100.0 3 0.26 0.00 -0.26 -100.0 4 0.33 0.00 -0.33 -100.0 5 1.16 0.00 -1.16 -100.0 6 1.93 0.01 -1.93 -99.7 7 2.34 1.01 -1.33 -56.8 8 4.21 2.82 -1.39 -33.1 9 4.88 5.15 0.27 5.4 10 4.62 11.59 6.97 150.8 Source: Based on the ENHA. 7. The minimum, average and maximumincomes per decile are presented inTable A 6. 13 Figure 1. Mean tax rate on labor incomeIRP and IRPF 14 I 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Vmgtile oflabor mcome Source: Basedon the ENHA. Figure 2. Variation in average incomeby decile due to the tax reform (%) l o , I 1 2 3 h -2 Deciles o f labor mcome -4 -6 -8 Source: Based on the ENHA. Table 8. Workers paying more under the post-reform system Distribution o f those who Decile pay more Percentage who pay more 6 0.1 0.2 7 5.2 9.0 8 16.5 28.8 9 24.6 43.1 10 53.6 93.8 100.0 17.5 Source: Based on the ENHA. 14 45. Next, the impact of the reform on different categories of workers is examined (Table 9). It is important to look at the effect o f the reform on different groups o f workers as there has been concern that those some maybe overly penalized by the reform. For instance, pre-reform those with more than one job were not assessed according to their aggregate labor income, but rather separately according to the amount o f each wage contract. Post-reform, marginal tax rates are imposed based on aggregate labor income. Workers in the health and education sector are among those negatively affected by the reform as it is common inthese sectors to hold more than one position. In terms o f the category o f the workers' main occupation, the most affected are public sector employees and entrepreneur/employers, with 32 and 35 percent o f workers inthese categories paying more due to the tax reform, respectively. Given the distribution o f employees across occupational categories, it in the private sector that the majority o f those who lose out fiom the tax reform are found: over half o f the workers negatively affected are in the private sector. Table 9. Situation of workers under the post-reform tax system, by worker characteristic % Losers Distribution All by Distribution of occupied IRP IRPF Difference Difference category oflosers workers (a) (b) (b-a) ((b-a)/a)*lOO All 17.5 ... ... 3.58 5.8 2.2 61.8 Private 15.8 48.9 54.2 3.7 5.8 2.1 55.4 Public 31.9 28.5 15.6 4.5 6.6 2.0 44.8 Entrepreneur/employer 35.3 9.6 4.7 4.3 8.4 4.0 92.4 Self-employed(w/o site) 1.9 0.7 6.5 0.2 0.6 0.5 306.9 Self-employed(w site) 12.6 11.9 16.5 1 3.2 2.2 222.7 Other 0 0 2.2 0.4 0 -0.3 -93.1 Non-Professional 15.1 81.9 94.7 3.4 5 1.6 46.4 Professional 59.3 18.1 5.3 4.6 10.6 6.0 129.7 One job 14.7 75.4 89.6 3.6 5.3 1.7 46.5 More than onejob 41.3 24.6 10.4 3.5 8 4.5 128.3 Women 14.1 35.1 43.5 3.4 4.9 1.5 42.6 Men 20.1 64.9 56.5 3.7 6.3 2.6 71.6 Interior 12.6 41 57.1 3.3 4.3 1.1 32.7 Montevideo 24.1 59 42.9 3.9 7.1 3.2 83.2 Mayor urban 18.7 92.9 86.7 3.7 6 2.4 65 Minor urban 9.1 3.0 5.9 2.8 2.9 0.1 3.3 Rural 9.5 4.0 7.4 2.8 3.8 1.o 35.2 Source: Basedon the ENHA. 46. The greatest differences between pre- and post-reform tax payments are found in the case of entrepreneur/employers and the self-employed. The mean tax rate on income has doubled for entrepreneur/employers. For the self-employed, there are large differences in the effect o f the tax reform on individuals due to their different locations in the income distribution. Among professionals, 60 percent pay more taxes due to the reform, represents 18 percent o f the total workers who pay more under the reform. The average tax rate increases by six percentage 15 points for this group that, on average, pays 130percent more intaxes.* Overall, for workers who are negatively affected by the reform, the mean tax rate increases by almost six percentage points, which implies that these taxpayers pay an additional 132 percent. For those that benefit, the taxes bill decreases by 46 percent. 47. Regarding multiple jobs, 41 percent of workers with more than one job pay more with the reform, as compared to 14.7 percent of those with one job. Prior to the reform, individuals paid a single tax rate applied to each work contract. With the introduction o f a marginal income tax based on total income from all labor contracts, multi-jobholders with higher incomes pay more. Inspite o fthe greater burden for workers with more than onejob, it should be noted that 75 percent o f the workers that pay more under the post-reform tax system have only one job. Salary differences and differences in labor market participation rates between men and women mean that 65 percent o f the workers with an increased tax burden are men. On average, the tax increases by 2.6 percentage points for men, while for women the variation is 1.4 percentage points. 48. Due to the salary differentials between Montevideo and the urban interior, the post- reform tax system impacts more negatively on workers in the capital. Salaries are in general higher inMontevideo compared to the rest o f the country and therefore, the average tax rate for workers increases by 3.2 percentage points, as compared to 1.1percentage points for the workers inthe interior o f the country. Thus, almost 60 percent o fthe workers affected negativelybelong to the country's capital. The workers who negatively affected by the changes in the tax system are concentrated in urban localities with more than 5000 inhabitants. There is little difference in the impact o f the reform on minor urban zones and rural ones, although the mean tax rate increases on average by one percentage point inthe latter. 49. Construction workers and domestic servants in private households are those who gain most from the tax reform, with these low-salaried workers paying on average 40 percent and 20 percent less in taxes, respectively (Table 10). Manufacturing, trade, public administration and defense, health care and social services and education are the sectors with almost 60 percent o f the workers affected negatively by the reform. Financial intermediation is the industry with the largest share o f workers who pay an increased tax burden due to the tax reform (63 percent), followed by the electricity, gas and water sectors (40 percent). Social services and health workers, some o f whom have more than one employment contract, on average pay 3.5 percentage points more due to the tax reform). It i s those in professional occupations that lose out interms o f greater taxes paid due to the reform. 8. All those who are employed and declare having completed university education are classified as professionals. 16 Table 10. Situatioi Ifworkers underthe post- formtax system, by industry % All Losers Distrib. by Distrib. occupied IRPF Diff. Diff. category o flosers workers IRP(a) (b) (b-a) ((b-a)/a)* 100 ~~ All 17.5 ... ... 3.58 5.8 2.22 61.8 A. Agriculture, livestock, game and forestry 9.9 6.0 10.7 2.7 4.2 1.5 54.0 B.Fishing 18.4 0.3 0.2 3.7 7.7 4.0 106.8 C. Exploitationo f mines and quarries 24.9 0.2 0.1 4.0 5.7 1.7 41.8 D.Manufacturing industry 17.1 13.4 13.7 3.7 6.0 2.3 62.0 E. Supply o felectricity, gas and water 39.6 2.4 1.o 5.3 8.5 3.2 60.3 F.Construction 3.4 1.2 6.2 3.8 2.3 -1.5 -40.4 G.Major andminor commerce, vehicle repair 11.9 13.1 19.2 2.7 4.4 1.8 66.3 H.Hotelandrestaurants 10.2 1.5 2.6 2.8 3.3 0.5 16.7 I.Transport, storage and communications 30.3 9.1 5.3 4.3 6.6 2.4 56.0 J. Financial intermediation 63.4 5.5 1.5 5.2 12.2 7.0 135.9 K.Realestate activity 27.1 8.8 5.7 3.6 7.6 4.0 112.1 L.Public administration and defense 27.8 11.7 7.3 4.5 6.3 1.8 40.3 M.Teaching 27.6 9.0 5.7 3.8 5.1 1.4 36.3 N.Healthandsocial services 32.9 12.8 6.8 4.1 7.6 3.5 87.2 0.Other activitiesin provision o f communal, social and personal services 14.4 4.0 4.9 3.1 4.8 1.7 56.0 P. Private household domestic service 2.1 1.1 9.O 1.o 0.8 -0.2 -24.1 Q.Organizations and extraterritorial organs 2.2 0.0 0.3 0.3 0.0 -20.3 Source: Basedon the ENHA. 4.3.2 Pensioners 50. For pensioners, the tax burdenincreases from the eighth decile onwardsdue to the reform.The impact o fthe reform onpensioners is o finterest as groups o fpensioners inUruguay have launched a particularly strong challenge against the application o f the personal income tax (IRPF) to pension income. Pensioners paid little wage tax on pension income prior to the tax reform (Table 11 and Figure 3). The top 20 percent o f pension earners faced a tax rate o f 2 percent prior to the tax reform.' For the ninthand tenth decile o fpension earners, the difference between the average tax rates before and after the reform i s in excess o f eight percentage points 9. The minimum,average and maximumpensionvalues per decile are presented inTable A-10. 17 (Figure 4). Pensioners negatively affected by the tax reform represent 27 percent o f the total pensioners. The bottom 70percent o fpension earners continue to pay no tax. Table 11. Pensioners: Mean tax rate of IRP and IRPFby decile (YO) Decile IRP (a) IRPF(b) Difference (b-a) Difference ((b-a)/a)*lOO 1 0.0 0.0 0.0 2 0.0 0.0 0.0 3 0.0 0.0 0.0 4 0.0 0.0 0.0 5 0.0 0.0 0.0 6 0.0 0.0 0.0 7 0.0 0.0 0.0 8 0.4 8.7 8.2 1994 9 2.0 10.0 8.0 390 10 2.0 12.5 10.4 510 Source: Based on the ENHA. Figure 3. Mean tax rate on pensionsunder the IRP and IRPF, by vingtile (%) 16 I I 14 12 n 2e,e 10 k0i? , 8 6 4 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Vmgtile o fpension income Source: Based on the ENHA. 18 Figure4. Variation in averagepensionsdue to the tax reform,by decile (%) 0 -2 -4 8 h -6 -8 -10 -12 Source: Based on the ENHA. Table 12.Uruguaytax reform: Summary of impactson workers andpensioners % o fLosers IRP (a) IRPF(b) Difference Reduction in Increase in (b-a) disposable income disposable income for losers (uR$) for winners (uR$) Labor income 17.5 3.58 5.80 2.22 1516 75 Pension income 27.1 1.08 6.86 5.78 1375 0 Source: Based on the ENHA. 4.2. IMPACT OFINDIRECT TAXES 51. The analysis here is focused on the impact of the reform of indirect taxes on householdincome. The pre-tax reform situation is estimated to havebeen regressive, inthat the percentage decrease inincome due to indirect taxes is higher for the bottom 40 percent compared to the top 60 percent o f income-earning households (Table 13 and Figure 5). All households benefit from the reductioninthe indirect tax rate due to the tax reform, which falls by on average over 2 percentage points over the different income groupings. In terms o f average indirect tax rates, the post-reform the tax structure remains regressive. To restate, it i s assumed in the analysis that indirect taxes on goods are shiftedcompletely to consumers. 4 19 Table 13.Meanindirecttax rate(IVA + COFIS) on householdsbeforeand after the reform,by decile(%) Pre-reform Difference Difference Decil (a) Post-reform(b) (b-a) ((b-a)/a)*lOO 1 11.4 8.6 -2.8 -24.8 2 11.3 8.6 -2.7 -23.8 3 11.3 8.7 -2.6 -23.0 4 11.1 8.6 -2.5 -22.4 5 10.9 8.6 -2.4 -21.5 6 10.4 8.1 -2.3 -22.2 7 10.6 8.3 -2.3 -21.6 8 10.4 8.2 -2.2 -21.3 9 10.0 7.9 -2.1 -20.9 10 9.6 7.7 -2.0 -20.5 Source: Basedon the ENHA. Figure5. Meanindirecttax rate(IVA + COFIS) on householdsbeforeand after the reform,byvingtile(YO) 14 I --- - 1 2 - - - - - - - - +------ - h &?lo- - - - a 5 *r v 8 - % 8$ 6 - 4 - --pre reform -post reform 2 - 0 I I I , , , , , ~ ~ ~ ~ ~ , , , ~ ~ , Source: Based on the ENHA. 4.3. IMPACT OFCOMBINEDTOTAL EFFECTOFTHE TAX REFORM 52. Next, the results o f an analysis o f the impact o f direct and indirect taxes on households are presented. Household income includes all non-labor capital income. 4.3.1 ChangeinTax Burden 53. The average pre-reform tax burden for households was 12.3 percent of pre-tax income.For the average household, the total impact o fthe tax reform on the tax burdeno f direct and indirect tax changes i s given inTable 14. Inthe case o f direct taxes, the reform increases the average burden by 68 percent. In the case o f indirect taxes, the reform reduces the average household tax burdenby 21 percent. Overall, the reform causes the household tax burden to fall 20 by 4.8 percent. Falling indirect taxes compensate for the increase indirect taxation. As intended by the reform, the relative weight o f direct taxes and consumption taxes changes on average. Under the pre-reform regime, indirect taxes represent 81percent o f the taxes paid by households. The reform causes the share of consumption taxes to fall, leading to VAT representing 67 percent o f the household tax burden and the IRPF becoming equal to a third o f the total taxes payable by households. Incomes (without adjustment) Disposable income COFIS (fully passed on) Averaee income before taxes " 6646 Before tmx reform a) Total taxes on households 816 664 152 12.3% Ib)Indirect taxes 10.0% c) Direct taxes 2.3% After tax reform 1I a) Total taxes on households 777 b)Indirect taxes 522 c) Direct taxes 255 Tax burden (as % of income before taxes) a) Total taxes 11.7% b)Indirect taxes 7.9% c) Direct taxes 3.8% Change in tax burden due to tax reform (%) a) Total taxes -4.8% b) Indirect taxes -21% c) Direct taxes 68% 54. An analysis of the total impact of the reform shows it to clearly be progressive ( 21 Figure 6). The grey and black smooth lines represent the situation before and after the reform, respectively." Under the pre-reform system, the tax burden as a share of pre-tax income remains relatively constant (close to 12 percent) throughout the distribution curve. The pre-reform system was neutral from the point o f view of inequality. This neutrality arose from the combination o f regressive consumptiontaxes and progressive personal income taxes. 10. The minimum, average and maximum values of the household per capital income vingtiles (weighed per individual) are presented inTables A-4 (uncorrected income). 22 Figure6. Tax burdenas shareof per capitapre-taxincome:Comparisonbetweenthe pre- reformregimeandthe post-reformsituation 4%, / i 0% 2% 1 2 3 4 5 6 7 8 9 I O 11 12 13 14 15 16 17 18 19 20 - - Householdincomevingtile VAT taxes before tax reform --31( ' IRF' -Total taxes before tax reform *IRPF VATaftertaxreform -Total taxes after tax reform Source: Basedonthe EGIHandthe ENHA. 55. The reform alters the relative neutrality of the pre-reform tax system. Up to the distribution median (approximately vingtile lo), the total tax burden shows no systematic variation in terms o f the distribution; however, there is a monotone increase o f the tax burden from the median onwards. As a result, the global tax burden o f the households below the 16th percentile falls, while it becomes markedlyhigher inthe last three vingtiles o f the distribution. 56. The progressive impact of the reform stems fundamentally from the process of replacingthe IRPwith the IRPF. The change inindirect taxes does not generate a substantial distributional impact, although it does generate an increase in the income o f all households. Although both the IRPand the IRPFare progressive, that is the tax rate on the household income increases with the level o f income, the IRPF shows a sharper progressiveness pattern. The tax burden is lower than that o f the IRP approximately up to the sixteenth vingtile, and then increases exponentially. Nevertheless, it should be noted that the average direct tax rate remains at relatively moderate levels, below 9 percent for the twentieth vingtile. The net impact o f replacing the IRP with the IRPF turns negative from vingtile fifteen, but the reduction o f consumption taxes more than compensates for that effect up to vingtile eighteen. As a result, only the wealthiest 20 percent o f the population are negatively affected by the reform. The rest o f the households see their disposable income increase by an amount inthe range o f 2 to 3 per cent. The results presented correspond to the scenario where household income is not corrected and it is assumed that COFIS is fully passed through to the end consumer. However, the result also holds under the alternative scenarios analyzed. 23 57. The progressiveimpact of the reform is shown by the improvement in inequality that results from the change in the tax system (Table 15 and Table 16). The results illustrate that the pre-reform system is characterized as neutral in terms o f the impact on the distribution. Pre-reform, the synthetic inequality indices remain practically unchanged before and after taxes. Post-reform, the last column o f Table 15 shows that all the synthetic inequality indicators decrease after taxes. The variation in the inequality indicators due to the tax reform is in the range o f 1 to 2 percentage points." The reform i s estimated to have achieved a modest, but significant reduction in the Gini: An examination o f the upper bound limit o f the confidence interval o f the Gini coefficient after the tax reform reveals that is smaller than even the lower bound o fthe pre-tax Gini (Table 16). Table 15. Ineaualitvmeasuresbeforeandafter taxes Before tax reform After tax reform After After After After After After Before taxes indirect direct taxes indirect direct variation taxes (1) taxes taxes (2) taxes taxes (2)-( 1) Gini coefficient 0.454 0.453 0.457 0.448 0.442 0.456 0.440 -0.011 Entropy 0 0.351 0.350 0.357 0.341 0.338 0.361 0.334 -0.012 Entropy 1 0.376 0.376 0.382 0.366 0.353 0.380 0.349 -0.023 Source: Based on the EGIHand the ENHA. Table 16. Confidenceintervalsfor ineaualitvmeasuresbeforeand after the taxes Gini Entropy 0 Entropy 1 Lower Average Upper Lower Average Upper Lower Upper bound bound bound bound bound Average bound Before taxes 0.452 0.454 0.456 0.349 0.351 0.357 0.369 0.376 0.381 After taxes (pre-reform) 0.452 0.453 0.456 0.350 0.349 0.358 0.370 0.375 0.381 After taxes (post-reform) 0.443 0.445 0.447 0.335 0.338 0.343 0.349 0.355 0.362 58. While progressive, the impact of taxes on the Gini under the post-reform tax structure for Uruguay is much lower in magnitudethan seen in many Europeancountries. The change in the Gini due to taxation in Uruguay i s more broadly in line with that found for Latin America than for Europe (Goiii et al, 2008): Taxation has a close to neutral or modestly progressive impact on the Gini (Table 17). Uruguay joins Mexico as one o f the only two Latin American countries where the Gini falls after taxation. The tax structure i s limited as a tool for redistribution in Latin America due to low level o f personal income taxation. By contrast, in Europe personal income taxes are higher, better-targeted and more progressive allowing the tax system to have a greater impact on equity. 11. Synthetic inequality indices weigh income transfers differently in the various parts o f the distribution curve, which could produce diverse results. However, inthis case the conclusions are robust for the index used. 24 Table 17. Uruguay, Latin America and OECD: Comparison of inequality before and after taxes Inequality Inequality Contribution before taxes after taxes oftaxes Latin America Uruguay (Pre-reform) 0.454 0.453 0.00 Uruguay (Post-reform) 0.454 0.445 0.01 Argentina 0.49 0.48 0.01 Brazil 0.54 0.55 -0:Ol Chile 0.46 0.46 0.00 Colombia 0.53 0.53 0.00 Mexico 0.50 0.49 0.01 Peru 0.49 0.49 0.00 LatinAmerican average 0.50 0.50 0.00 OECDBenchmarks (Selectedcountries) Denmark 0.49 0.42 0.06 Finland 0.49 0.45 0.05 France 0.42 0.38 0.04 Germany 0.43 0.38 0.05 Greece 0.47 0.44 0.04 Ireland 0.53 0.48 0.05 Italy 0.48 0.44 0.04 Portugal 0.49 0.44 0.05 Spain 0.47 0.42 0.05 Sweden 0.45 0.41 0.04 UnitedKingdom 0.52 0.48 0.04 Europe average 0.46 0.41 0.05 Source: Goiii et al. (2008) and OECD. 59. Next, the impact of the reform on poverty and extreme poverty is analyzed. Again, it should be recalled that this exercise is subject to limitations, fundamentally because it does not take into account the potential household behavioral changes that may occur due to the modification in the tax system or other general equilibrium effects such as a potential impact o f the reform on investment. The poverty line used is that calculated by the National Institute o f Statistics (Instituto Nacional de Estadistica, NE) in 2002. It i s based on information from the Household Income and Expenditure Survey o f 1994, which is used to design the basic food basket or food poverty basket (Canasta Busica Alimentaria) and non-food consumption, which are the components o f the poverty line. A household is considered to be inextreme poverty ifits per capita income is below that o f the food poverty basket, while it i s characterized as poor ifits income does not reach the value o f the poverty line. The indices FGTo, FGTl and FGT2 are used to measure the effects on the incidence, intensity and severity o f poverty and extreme poverty.'* First, the FGT indices for the year 2006 were calculated using current income reported by the 12. These indices are described inthe methodological annex. 25 households. Second, the impact o f the reform on the prices o f goods included in the poverty basket due to changes inthe indirect tax structure was taken into account. Table 18. Impact of the tax reform on poverty and extreme poverty Indicators o f Poverty FGT(0) FGT(1) FGT(2) Indigence Indicator before the reform 0.019 0.004 0.001 Indicatorpost reform (indirect taxes) 0.017 0.004 0.001 Effect of indirect taxes -0.002 0.000 0.000 Indicator post reform (indirect + direct taxes) 0.017 0.004 0.001 Effect of direct taxes u.000 0.000 0.000 Poverty Indicator before the reform 0.270 0.088 0.040 Indicator post reform (indirect taxes) 0.265 0.085 0.038 Effect of indirect taxes -0.005 -0.003 -0.002 Indicator post reform (indirect + direct taxes) 0.263 0.084 0.036 Effect o.,fdirect taxes -0.002 -0.001 0.000 _1_1 r'iggregatc effect ofthe reform -0.007 -0.004 -0.002 Source: Basedon the EGIHand the ENHA. Table 19. Confidence intervals for poverty measures before and after the reform FGT(0) FGT(1) FGT(2) Lower Upper Lower Average Upper Lower Average Upper bound Average bound bound bound bound bound Before thereform 0.269 0.270 0.271 0.088 0.088 0.089 0.040 0.040 0.040 After the reform (direct taxes) 0.264 0.265 0.266 0.085 0.085 0.086 0.038 0.038 0.038 After the reform (direct and indirect taxes) 0.262 0.262 0.263 0.084 0.084 0.085 0.036 0.036 0.036 Source: Based on the EGIHand the ENHA. 60. There is a small, but beneficial, reduction inpoverty due to the reform (Table 18 and Table 19). The simulated reduction in poverty represents a reduction o f about 0.7 percentage points in poverty incidence. The small variation in poverty observed stems from the changes in VAT and the elimination o f COFIS, rather than being due to the variations indirect taxes. This result i s in contrast to that obtained in the analysis o f the redistributive impact, where the main force that made it possible to characterize the reform as progressive i s linked to direct taxes. Individuals in the poorer segments o f the income distribution were in general not taxed by the IRP.Around the third decile-the location o f the poverty line-the amount o f IRPpaid was less than 1percent o f pre-tax income, so that a reduction o f this amount does not substantially alter the availability o f resources for households. The impact o f the tax reform on extreme poverty i s 26 practically ~ e r 0 . This result i s to be expected as the share o f individuals below the extreme l ~ poverty line i s low, at about 2 percent o f the total population. Table 20. Effect of the tax reform on poverty for those aged under 18 and above 64 FGT(0) FGT(1) FGT(2) Less than 18 years Indicator before the reform 0.461 0.160 0.074 Indicator post reform (indirect taxes) 0.447 0.153 0.071 Effect of indirect taxes -0.014 -0.006 -0.004 Indicator post reform (indirect + direct taxes) 0.444 0.152 0.070 -0.003 -0.001 -0.001 -JJ Effect ofdirect taxes - - -.! Aggregateeffect ofthe reform -0.0I7 -0.008 -0.004 Greater than 64 years Indicatorbefore the reform 0.085 0.021 0.008 Indicatorpost reform (indirect taxes) 0.079 0.019 0.007 Effect of indirect taxes -0.005 -0.001 -0.001 Indicator post reform (indirect + direct taxes) 0.080 0.019 0.007 Ejfect of direct taws 0.000 0.000 0.000 Aybpgdte effect o f the reform -0.005 -0.001 -0.001 Source: Based on the EGIHand the ENHA. 61. One of the characteristics of poverty evolution in Uruguay since the beginning of the 1990s is the dynamics according to age groupings. While poverty among the elderly (age 65 and over) fell during most o f the period-with the exception o f the period between 2001 and 2004, when, as a consequence o f the deep economic crisis, the poverty headcount increased for all age segments-poverty among children and young adults increased markedly (Amarante et al, 2005; PNUD, 2005). Therefore, it is important to analyze whether the tax reforrn has a differential impact depending on age group. The results o f this analysis reflect that the impacts per age group are analogous to those obtained for the total population. Poverty indicators are slightly reduced with the implementation o f the reform, with the drop somewhat higher among younger individuals for whom the absolute value o f the change is around 1 percent. The low impact on the elderly i s not surprising, since the pensioners who tend to be close to the poverty linewere not subject to the LRP and do not pay the IRPF(Table 20). 13. The poverty calculations were only made for urban areas (localities with 5000 inhabitants and more), as there is no official poverty line available for rural areas. 27 5. CONCLUSION 62. The focus of the PSIA is on the equity impact of the changes in the tax system on households, specificallythe changes in VAT and COFIS and those relatedto the IRP and the IRPF. Results indicate that the replacement o fthe IRPwith the IRPFresults inan improved situation for the bottom 80 percent o f labor income earners, for whom the mean tax rate falls. The mean direct tax rate increases for the top 20 percent o f taxpayers. The mean tax rate here refers to the average IRP and IRPFpayments by decile as a share o f labor income, including any labor income from secondary occupations. Workers in the tenth decile are calculated to pay, on average, 7 percentage points more tax on their labor income. For the ninthdecile, the increase in the meantax rate is much lower at 0.3 percentage points. 63. For pensioners, the tax burden is increased from the eighth decile onwards due to the introductionof the IRPF. Pensioners paid little wage tax on pensionincomeprior to the tax reform. The top 20 percent o f pension earners faced a tax rate o f 2 percent prior to the tax reform. Following the reform, for the top 30 percent o f pension earners the difference between the average tax rates before and after the reform is inexcess o f 8 percentage points. The bottom 70 percent o fpensionearners continue to payno tax. 64. With regard to indirect taxes on household income, the pre-tax reform situation is estimatedto havebeen regressive,inthat the percentagedecrease in incomedue to indirect taxes is higherfor the bottom40 percentcomparedto the top 60 percentof income-earning households. All households benefit from a reduction in the indirect tax rate due to the tax reform, which falls by on average over 2 percentage points over the different income groupings. Interms of average indirect tax rates, the post-reform indirect tax structure remains regressive. To restateas is regular in this type o f analysis-it is assumed that indirect taxes on goods are shifted fully forward to consumers. 65. An analysis of the total impact of the reformon households shows it to represent a progressivechange. Under the pre-reform system, the tax burden as a share o f pre-tax income remains relatively constant (close to 12 percent) throughout the distribution curve. The pre- reform system was neutral from the point o f view o f inequality. This neutrality arose from the combination o f regressive consumption taxes and progressive personal income taxes. The reform alters the relative neutrality o f the pre-reform tax system. There i s a monotone increase o f the tax burden from the median onwards. As a result, the global tax burden o f the households below the 16th percentile falls, while it becomes markedly higher in the last three vingtiles (five-percent bands) o fthe distribution. Taking into account changes to VAT, COFIS and the introduction of the IRPF, along every measure considered the tax reformhas a small, but positive impact on equity. The Gini coefficient o f after tax income is estimated to fall due to the reform: The Gini coefficient o f before tax income is calculated as 0.45 and the Gini coefficient o f after tax income following the reform is calculated as 0.44 (the pre-reform Gini o f after tax income i s 0.45). 28 References Amarante, V., Arim, R. and Salas, R. (2007). Impact0 Distributivo de la Reforma Impositiva. Mimeo. Asociacion Rural del Uruguay (2006). Informe de coyuntura. Direccion de Estudios Agroeconomicos. Disponible en: htt~://~~~.aru.org;.uy/Documentos/Info%2007-06.pdf Barreix, A. and Roca, J. (2003). Sistema Tributario: condiciones actuales and propuesta. Universidad Catolica del Uruguay, Montevideo. Barreix, A. and Roca, J. (2005). Propuestaspara la reforma tributaria de Uruguay. Documento elaboradopara el Ministerio de Economia y Finanzas, Montevideo. Barreix, A. and Roca, J. (2006). Arquitectura de una propuesta de reforma tributaria. Universidad Catolica del Uruguay, Montevideo. Barreix, A. and Roca, J. (2007). Strengthening a fiscal pillar: the Uruguayan dual income tax. CEPAL Review 92. CEPAL (2001). Analisis estadistico de 10s efectos sobre 10s principales indicadores socio- econdmicos resultantes de 10s cambios metodoldgicos introducidos en la encuesta continua de hogares. By Bucheli, andM.Furtado, M.Montevideo, Uruguay. Cinve (2005). Implicancia de la reforma de 10s aportes patronales sobre la demanda laboral. Montevideo. Duclos, J.Y. and Araar, A. (2006). Poverty and Equity. Measurement, Policy, and Estimation withDAD. Springer/IDRC. Ottawa. Gasparini, L. Horenstein, M, and Olivieri, S. (2006). Economic Polarisation in Latin America and the Caribbean: What do Household Surveys Tell Us? CEDLAS Working Papers 0038, CEDLAS, Universidad Nacional de LaPlata. Goiii, E., Lopez, H. and L. Serven L. (2008), Fiscal Redistribution and Income Inequality in LatinAmerica. World Bank Policy ResearchPaper 4487, World Bank, Washington, D.C. Grau C. and Lagomarsino G. (2002). La estructura tributaria de Uruguay y su incidencia en la distribucidn del ingreso de 10s hogares. Fundacionde Cultura Universitaria. Grau C., Lorenzo F. and Oddone G. (2004). Ideas y lineamientos para la reforma tributaria. Centro de Investigaciones Economicas(CTNVE). Instituto de Economia (2006) Impactos de la reforma tributaria sobre 10s ingresos de 10s hogares. Anexo del Informe de Coyuntura del Instituto de Economia, Montevideo. Mendive C. and Fuentes A. (1996) "Evaluation de la captacion del ingreso de 10s hogares" in Aspectos metodoldgicos sobre la medicidn de la linea de pobreza: el cas0 uruguayo. CEPAL, Oficina de Montevideo, Instituto Nacional de Estadistica. Ministerio de Economia y Finanzas. (2006). Proyecto de Ley de modijkacidn del sistema tributario. 29 Miraballes G. (2006). Sewicios de Intermediacidn Financiera Medidos Indirectamente. Una propuesta alternativa para Uruguay. Tesis para la obtencion del titulo de Magister en Economia, Facultad de Ciencias Economicas y Administracion, Universidad de la Republica. Perazzo I.,Robino C. and Vigna A. (2002). Sistema impositivo y distribucidn del ingreso en Uruguay. Trabajo monografico presentado en la Facultad de Ciencias Economicas y de Administracion, Universidadde la Republica. World Bank. (2007). Income Transfer Policies in Uruguay: Closing the Gaps to Increase Welfare. ReportNo. 40084-UY, October 18,2007. 30 k cw 0 2 r 4 m 5 c W W w 4 ci P e, a C .I FigureAl. Uruguay: Tax RevenueCompositionBeforeandAfter Tax Reform Before reform: Tax as a YOof revenue After reform: Tax as a YOof revenue Source: Ministry o fEconomyand Finance, Uruguay. Note: Elaboratedby the Ministry ofEconomy andthe Financebasedon 2005 revenue collection 38 0 d J s 3 3 2 2 m 0 P 8 x W 2 9 Annex 3. Poverty, Inequality and Progressiveness Indicators 66. To analyze the changes in poverty, the class o f indicators proposed by Foster, Geer and Thorbecke (1984) (FGT (a))is used. FGT (a)i s a poverty index that includes a family o f indices based on different values o f the parameter, a,which represents the degree of poverty aversion in society. The FGT (a)index i s defined as: FGT(a) =-C[(Z-Y,)/Z] 1 4 a n i=l 67. where n is the total sample size; yi i s the standard o f living indicator for person i, for which typically a measure o f income i s used. Let Y = (y ~y2, ,...,y,, ) be a vector o f incomes in increasing order. z > 0 is the pre-determined poverty line, with z - yi is the income shortfall o f the ith individual; and q = q(y; z) is the number o f poor individuals having income no greater than z. The measures are defined for a 2 0, and a i s a measure o f the sensitivity o f the index to poverty. If a = 0, the FGT index i s transformed into the equivalent o f the poverty headcount, since it quantifies the percentage o f households below the poverty line. Ifa = 1, the FGT index i s transformed into the poverty gap, measuringthe gap between the income o f the poor and the poverty line. If a = 2, the FGT index includes the relative variance o f income across the poor. 68. To analyze changes in inequality, the Gini coefficient and entropy indices are used. The Gini coefficient o f incoem, G,, i s an indicator that increases with the degree o f inequality, with 0 representing equidistribution and 1 representing the extreme case when only one person receives the whole income o f ~0ciety.I~ This index may be defined as the average difference between all possiblepairs o f incomes o fthe population (yJ, expressed as a share o ftotal income, i.e.: where is the mean income. 69. The Entropy indices o f grade 0 (Eo)and 1 (Theil Index, (E,)) o f income distribution are defined as: i = I,...,n (3) 14. The Gini coefficient (G)is a measure that may be visualized geometrically interms of the Lorenz curve as the equivalent to a quotient. The numerator is the area between the equidistribution straight line and the Lorenz curve, and the denominator is the area below the diagonal. 41 where yi represents the per capita income o f individual i, p is the mean income o f the population, and Xj i s the share in the income of individual i. The index Eo has greater sensitivity in the lower strata (tending to infinite when the income approaches zero). In the case o f El the weight of a transfer i s greater the lower the income o f the receiver, and the greater the distance between the individual who receives and the individual who i s deprived from the income. Figure 7. An example of a concentration curve for a progressive tax 70. The concentration curve i s an important tool for examining the impact o f taxation transfer policies. This may be used to capture the equity of tax systems when it i s compared to the before-tax income Lorenz curve. The Lorenz curve, Lx(p), i s defined as the proportion o f total income, x, received by the lowest pth fraction of the population, arranged in ascending order o f income. The Gini coefficient i s the area between the line o f equality (the 45-degree line) and the observed Lorenz curve (the observed deviation from the line o f perfect equality) divided by the total area under the diagonal (the maximum possible deviation from perfect equality). This i s equal to two times the area between the diagonal and the observed Lorenz curve: 71. The concentration curve o f a tax CT(P) i s defined similarly to the Lorenz curve, replacingthe cumulative proportion o f income received by each fraction p, by the cumulative proportion o f taxes or other payments contributed by each fraction o f the population ordered by income. The tax concentration index CT is analogous to the Gini coefficient and corresponds to one minus twice the area between the concentration curve and the line o f equality divided by the total area under the diagonal. The index takes a negative (positive) value when the tax concentration curve lies above (below) the line o f equality. A negative value indicates a disproportionate concentration o f taxes among the poor. The index has a range o f [-1,1]. 42 72. There are two leading approaches to look at the distribution o f a tax (Duclos and Araar, 2006). Under the tax redistribution (TR) approach, a tax structure i s progressive ifthe tax concentration curve lies below the Lorenz curve for pre-tax income (Figure 7). When tax payments are regressive, the concentrationcurve o f taxes lies above the Lorenz curve, i.e. the bottom pth fraction o f the population pay more than p% o f tax. In addition, a tax A i s more TR-progressive than a tax B if the tax concentrationcurve o f the tax A lies below that o f tax B. where the subscript T refers to taxes and the subscript Y refers to pre-tax income. 73. The second approach i s that focuses on the income redistribution (IR) impact o f the tax. A tax is IR-progressive if the after-tax income distribution i s more progressive than the pre-tax income distribution. Let Ndenote post-tax income and Y continues to denote pre-tax income, then IR-progressiveness requires the after tax income concentration curve CN(P)(= CY-T(P)) to be greater than the pre-tax income Lorenz curve: ` N (P) LY (P) ti'P E [OJI ' (9) 74. The evaluation criteria derived from conditions (7) and (9) make it possible to characterize taxes as globally progressive provided that the conditions stipulated are met for all percentiles. If the concentration curves, however, cross the Lorenz gross income curve, these criteria are not sufficient to characterize the p01icies.I~In turn, if the concentration curves cross, even when a classification o f the taxes as globally progressive or regressive may be possible, the policies could not be ranked according to their relative degree o f progressiveness. In other terms, conditions (7) and (9) provide a qualitative and partial ranking. 75. A comparison between the tax concentration curves and the pre-tax income Lorenz curve may lead to erroneous conclusions regarding the distributional impact o f taxes should taxation cause strong household reranking processes across the distribution, which could potentially cause their redistributive effect to be sharply reduced. Reranking may reduce the redistributive effect because o f a difference in pre-tax and post-tax ranking. The underlying income distribution could alter substantially, but have no impact on the concentrationindex if income ranks are preserved. Therefore, the distribution o f taxation needs to be related to the income distribution and the degree o fpre-tax income inequality. 15. This problem i s similar to that stemming from usingthe Lorenz curve as a rankingcriterion. Ifthe Lorenz curve corresponding to pre-tax income intersectswith the concentration curves, it is not possible to determine the global effect ofthe tax on inequality. 43 76. The Kakwani index, IIK, progressivity measures the divergence from o f tax proportionality. It is given by the difference between the concentration index o f taxes (CT) and the Gini coefficient o fpre-tax income: 77. For a progressive tax, the tax concentrationindex i s higher than the Gini o f the initial income distribution, i.e. the tax i s less equally distributed than initial income. Therefore, the higher the positive value o f IIK the more globally progressive is the tax. For a given pre-tax income Gini, Gy, the value o f IIK varies from (-Gy) to (l-Gy), a negative IIK representing a regressive tax. Ifthe incometax i s proportionalCT= Gy and the IIK equal zero. would 78. The Reynolds-Smolensky index, IIRs, often referred to as the redistributive/vertical equity effect (VE), i s a measure of the extent to which the tax system redistributes income. The after tax Gini i s denoted by GN (=GY-T).The IIRs i s given by: The nkindex has a range o f [-1,1]. For nRSgreater are than zero, taxes redistribute income to the poor and are characterized as progressive. A negative value for the index represents a regressive tax. 79. As inthe case o f the Gini coefficient, these indicators provide the same weight to the distance between both curves across the whole o f the distribution, which causes the segments where the difference i s greater to have more weight inthe final result. However, it i s possible to generalize these progressiveness indicators by incorporating a weighing function K indexed by aparameter p which captures the priority given to the various percentiles: Using this family o f linear weights, the generalized Kakwani and Reynolds-Smolensky indices adopt the following form: Increasing the parameter p increases the priority attached to the lower percentiles o f the distribution. Note that p = 2 results inthe traditional Gini and Concentration indices. 80. Finally, it is possible to break down the net redistributive effect o f taxes into a factor measuring the vertical equity (VE) and another one that captures the reranking inducedby the tax system (RR). The vertical equity effect measures the tendency o f taxes to reduce the 44 dispersion o f after tax income with respect to the before tax income distribution, i.e. the progressiveness o f the tax system. If the tax induces rerankings, the RR effect will always have a negative impact, reducing the effectiveness of the tax system as a mechanism to achieve a more egalitariandistribution o f income. 81. The concentration curve CN@) for net incomes is defined as and typically estimated as: 1 C,(p=i/n)=- ~ P NN.i C' j=l where the Nj have been ordered inincreasingvalues o f the associatedpre-tax incomes rj. Note that CN@)i s different from the Lorenz curve o fnet incomes, L&), which is definedas: 82. Empirically, the Lorenz curve for net income i s typically estimated as butwhere the observations havebeenre-ordered inincreasingvalues ofpost-tax incomes, withNl 5N2 1..1 Thus, CN@)sums upthe expectedvalue o fpost-tax incomesupto pre- N,. tax incomepercentilep. LN@) however, sums uppost-tax incomes upto apost-tax income percentilep. 83. The income concentrationcurve after taxes CN(P)(= CY-T(P))will always be above the Lorenz curve o f income after taxes LN(P)(= Ly-~(p)),except if there is no reranking o f any type, in which case both curves may coincide (Lambert, 1993). Thus, a tax will cause rerankings if and only if CN(P) > LN(P)for some p. The distance CN(P)- LN(P)can therefore be usedas an as an indicator o freranking. A reranking index is based on the distance between both functions using a similar procedure to before, the reranking index can be defined as (Duclos y Araar, 2006): where IN(p) i s the weighted inequality index o f after tax income and 1C~(p)is the index o f concentration o f after tax income. Thus, by using the progressiveness indices defined above, 45 it is possible to breakdown the difference between the Lorenz curves of income before and after taxes as follows: The greater the difference between the vertical equity effect and the reranking effect, the more redistributive the tax. These results may be used to form a synthetic index (indexed to the parameter p): 84. The differences betweenthe indices of pre-tax and post-tax inequality measure the net redistributional effect. The first term on the right hand side of equation (20) captures the redistributional effect and the second term, the reranking effect. Note that if p=2 is assumed, the previous expression is simply the difference between the pre-tax and post-tax Gini coefficients, the vertical equity (VE) effect is captured by the traditional Reynolds- Smolensky index and the redistribution effect is the reranking index (RR) known as Atkinson-Plotnick. G, -G, =nz-RJ YE RR 46 ANNEX 4. SENSITIVITYANALYSIS TO ASSUMPTIONS ON COFISAND INCOMEDATA 85. The main results o f the analysis are presented showing a variation in the assumptions in income and COFIS pass-through as a sensitivity analysis. To recap, with respect to income, two scenarios are defined: 0 ENHAincomewithoutadjustment(scenario1.1); and 0 ENHAincomeadjustedwiththe MendiveandFuentes(1996)sub-capture coefficients(scenario 1.2). Regardingthe COFIS, two scenarios are defined: 0 COFISis notpassedthroughto the endconsumer (scenario 11.1); and 0 COFISis fully passedthroughto the endconsumer(scenario11.2). 86. The pre-reform tax burdenfor householdswas around 10 percent to 12 percent, depending on the hypothesis made on the transfer of the COFIS. The adjustment for under-reporting o f income results in an average per capita income that i s approximately 11 percent higher than the uncorrected amount captured by the ECH. The income correction hypothesis impacts more on the nominal amount o f income before and after taxes, while the proportion o f income paidintax remainsroughly the same. 87. The totalimpacton the tax burdenof direct and indirecttax changesdependson the assumption made regarding income (adjustedhot adjusted) and the pass-on of COFIS to the final consumer. Table A 1 presents the impact o f the reform on the mean income as a function o f the four scenarios described. The income assumption affects the outcome o f the reform in terms o f average household tax burden. Inthe case o f direct taxes, the reform increases the average burden by 68 or 95 percent depending on the income hypothesis used. Inthe case o f indirect taxes, the reform reduces the average household tax burdenby 5 or 21percent, depending on the hypothesis usedwith regardto the transfer ofthe COFIS. Overall, the reform causes the household tax burden to rise, inthe range o f 10.8 and 15.6 percent, depending on the income scenario used. This increase i s fundamentally due to the rise indirect taxes. IfCOFIS i s assumedto have been included inretail prices, the reform causes a small decrease in the tax burden, which varies between 0.8 and 4.8 percent depending on the income scenario considered. This reduction i s due to the joint effect o f the increase indirect taxation and decrease inindirect taxes. 47 Table A 1. Disposable income and average tax burden before and after the tax reform 10.6% 12.3% 10.4% 12.2% 8.3% 10.0% 8.3% 10.0% b) Indirecttaxes 522 581 581 c) Directtaxes I 255 255 522 1 310 310 Tax burden(as YOof incomebeforetaxes) a) Total taxes 11.7% 11.7% 12.1% 12.1% b) Indirecttaxes 7.9% 7.9% 7.9% 7.9% c) Directtaxes 3.8% 3.8% 4.2% 4.2% Change intax burdendueto tax reform (%) a) Total taxes IO.8% -4.8% 15.6% -0.8% b) Indirecttaxes -5% -21% -5% -21% c) Directtaxes 68% 68% 95% 95% Source: Basedonthe EGIH and the ENHA. 88. The reform has a moderate redistributional impact. The last column o f Table A 2 shows that all the synthetic inequality indicators decrease, whichever income and COFIS pass-through scenarios considered. As expected, the reduction in inequality is greater when the elimination o f COFIS i s assumed to be transferred in full to the final consumer (scenario 11.2). The reform i s progressive: An examination o f the upper bound limit of the confidence interval o f the Gini coefficient after the tax reformreveals that i s smaller than even the lower bound o f the pre-tax Gini. 48 Table A 2. Inequality measures before and after taxes Beforetax reform After tax reform After After After After After Before After indirect direct taxes indirect direct Variation taxes taxes (1) taxes taxes (2) taxes taxes (1)-(2) 11.1 COFIS is not passedthrough A- Income: Uncorrected(1.1) Gini coefficient 0.454 0.453 0.457 0.448 0.442 0.456 0.440 -0.024 Entropy0 0.351 0.349 0.356 0.341 0.338 0.361 0.334 -0.032 Entropy 1 0.376 0.375 0.381 0.366 0.353 0.380 0.349 -0.059 B- Income: Corrected (1.2) Gini coefficient 0.456 0.459 0.461 0.454 0.448 0.461 0.446 -0.024 Entropy0 0.356 0.356 0.361 0.348 0.340 0.361 0.337 -0.045 Entropy 1 0.380 0.395 0.398 0.386 0.370 0.395 0.367 -0.063 11.2 COFIS is passedthrough A- Income: Uncorrected(1.1) Gini coefficient 9.454 0.453 0.457 0.448 0.442 0.456 0.440 -0.024 Entropy0 0.351 0.350 0.357 0.341 0.338 0.361 0.334 -0.034 Entropy 1 0.376 0.376 0.382 0.366 0.353 0.380 0.349 -0.061 B- Income:Corrected (1.2) Gini coefficient 0.456 0.459 0.462 0.454 0.448 0.461 0.446 -0.024 Entropy0 0.356 0.357 0.362 0.348 0.340 0.361 0.337 -0.048 Entropy 1 0.380 0.396 0.399 0.386 0.370 0.395 0.367 -0.066 Source; Basedonthe EGIH andthe ENHA. 49 50 Annex 5. ConcentrationCurvesAnd ProgressivenessIndices 89. An analysis of the distributional impact of the reform can be made based on tax concentration curves. A first approximation to the relative progressiveness o f the taxes i s a comparison o f their concentration curves with the Lorenz curve. The concentration curve shows the relative share in the total tax receipts o f the tax corresponding to different individuals, ranked by gross income. Figure 8 gives a comparison o f the income concentration curves before and after the reform considering the total taxes paid by households, and distinguishingbetween direct and indirect taxes. The blue lines identify the pre-reform taxes while the red lines show the situation after the reform. In turn, the smooth lines reflect the total effect, the thick dotted lines represent the concentration o f direct taxes and the thinner dotted lines represent the indirect taxes.16 90. The Lorenz income curve before and after taxes and the concentration curve of the pre-reform tax system practically overlap. That is, the pre-reform system tends to generate a tax burden proportional to the share o f pre-tax per capita income; the tax structure does not significantly alter the distribution o f income before and after taxes. This situation i s modified by the reform. The concentration curve o f the taxes paid directly by households after the implementation o f the reform (full red curve) i s systematically below the Lorenz curve before taxes, which indicates that the tax burden i s reallocated, generating a more progressive distribution. These results are consistent with the synthetic inequality indices presented previously, which showed that the pre-reform system did not generate changes in the indicators related to the gross income distribution, while the reform resultedina moderate reductionininequality. 91. Inthe case of direct taxes, both tax concentrationcurves (for the IRP and IRPF) are below the Lorenz curve, although the IRPF is clearly more redistributive. For example, almost 5 percent o f the IRP i s accumulated by the distribution median (p=0.5), while in the case o f the IRPF the figure falls to 2 percent. The concentration curves for indirect taxes reflect the regressive nature o f this tax (as per the definition used inthe study). Both the before and after reform curves are above the Lorenz curve, indicating that the lower income percentiles account for a greater share o f total revenues than their share o f pre-tax income. This regressive pattern i s not altered by the reform, but indirect taxes are slightly more progressive compared to the previous tax system. However, the relative share o f the various percentiles in total indirect tax revenues does not alter substantially, unlike the case o f direct taxes. Figure 9 and Figure 10 show that the reform may be characterized as redistributive both based on the tax burden redistribution approach and in terms o f income redistribution, which i s not the case for the pre-reform tax system." 16. In the previous section, it was noted that the qualitative results do not vary systematically with the assumption on the effects of the eliminationof the COFIS. The importanceofthe assumptionregarding COFIS relies, rather, on estimating the magnitude of the redistributional effect. The same applies to the adjustment of income.To the extent that the resultspresentedbelow are qualitative, it was decidedto present in the study only the results under scenarios 1.1and 11.2 (uncorrectedincome and full transfer ofthe COFISto retail prices). 17. The figures correspondingto the other scenarios are includedinthe statistical Annex (Figures A-10 to A- 18) 51 Figure 8. Pre-reform concentration and post-reform curves: Indirect and direct income taxes I-- .......Pre-reformindirecttaxes - - - IIRP -Pre-reform total taxes 'IRPF .......Post-reformindirecttaxes-Post-reform total taxes Lorenzcum(aftertaxes) Source: Basedon the EGIHandthe ENHA. 92. The tax burden redistribution approach posits that a tax is progressive if its concentration curve is systematically below the Lorenz curve. That i s undoubtedly the case with the post-reform tax system. However, the pre-reform system does not comply with this global property, since the proportion o f the fiscal burden borne by the higher strata (in particular for the last decile) is lower than their share o f the total income. InFigure 9, the full thick blue and red curves represent the difference between the concentration and Lorenz curves for the tax system ex ante and ex post reform, respectively. The curve for the pre- reform regime has a section below zero, towards the end o f the distribution, so it i s not possible to characterize it as globally progressive. 93. The income redistribution approach produces similar results. While the after tax income concentration curve is systematically above the pre-tax income Lorenz curve in the case o f the post-reform tax system, this characteristic i s not met in the last centiles o f the distribution for the pre-refom regime. Inother words, the cumulative share inpre-tax income i s greater than in gross income for all percentiles in the case o f the reform, while under the pre-reform regime, the cumulative share in net income i s higher than in gross income when reaching the last centiles, so that it cannot be stated globally that the pre-reform tax system redistributes resources from the higher to the lower strata, as i s the case with the post-refom system. 52 Figure 9. Differencesbetween the Lorenz after tax income curve and the concentration curves: Tax burden redistribution approach Difference between Lorenz and concentration curves: ProgressivitylTax:TR Approach _______ Source: Basedon the EGIHandthe ENHA. Figure 10. Differencesbetween the Lorenz after tax income curve compared to the concentration curves: Income redistribution approach Difference between concentration and Lorenz curves: ProgressivitylTax:IR Approach oo,s-I _ _ _ _ .- ----__- . - - __ - -...-- __ __ __ - _ - Source: Basedon the EGIHandthe ENHA. 53 94. Both approaches indicate that both the IRP and the IRPF are broadly progressive, since the differences shown in the figures are always positive. However, the IRPF is systematically above the IRP, reflecting its greater concentration inthe higherstrata. The curves o f consumption taxes take on negative values along their entire trajectory, which once again reflects their regressive character. However, the curve corresponding to the post- reform situation i s slightly above the pre-reform curve, confirming that the changes in the consumptiontax regime are slightlyprogressive. 95. In turn, the position of the concentration curves shows that the post-reform tax system globally dominates the pre-reform regime in terms of distributional impact. It should be recalled that given two tax regimes, A and Byregime A dominates regime B globally if the post-tax income concentration curve o f regime A is systematically above the curve of regime B (income redistributive approach), or alternatively, if the tax concentration curve i s higher in the case o f regime B. Figure 8 to Figure 10 clearly show that both conditions are met. Figure 11. Non-Parametric estimate of the differentiationof taxes with respect to income level Non parametric derivate regression of Y on X 0 131 I 0 121 , 0 I l t I 0 l o t h 0 03- Source: Basedonthe EGIH andthe ENHA Note: Non-parametncestimation o fthe derivativeo ftaxes with respect of income. 96. Unlike in the post-reform case, the pre-reform tax system is not globally progressive, i.e the level of taxation does not increase systematically with the income level. However, it i s possible that even without being globally progressive, the system could be locally progressive in some o f the distribution segments. This i s confirmed by Figure 11, which shows the estimate o f a non-parametric regression o f the differentiation o f the taxes vis-a-vis per capita income. The increase intax receipts as a function o f the per capita income i s observed both under the pre-reform and post-reform regimes, approximately up to the 70th 54 percentile. However, while the marginal change in the tax level per additional income unit remains at positive but constant levels under the pre-reform regime, it continues to increase markedly in the case o f the post-reform tax system. This means that the pre-reform regime meets the condition o f local progressiveness only in the distribution segments that are below the 70thpercentile-the tax differentiation with regard to the level o f income i s increasing in income-while that condition i s met in the case o f the post-reform tax system for the whole o f the distribution. These differences inthe localprogressiveness pattern explain why the pre- reformregime cannot be characterized as globally progressive. 97. The post-reform tax system can be characterized as globally progressive, butthis does not imply that it is possible to assert, based on that characteristic only, that the new regime "dominates" the pre-reform regime in terms of the impact on inequality. This judgment depends on an assessment to be made on income transfers across the distribution. As can be seen in Figure 11, there are segments o f the distribution where the marginal tax burden increasesmore under the pre-reform regime. Thejudgment on the equity effect depends on the relative weight assigned to income variations in different parts o f the distribution curve. Table A 3 shows the generalized indices o f Kakwani and Reynolds- Smolensky synthetic progressiveness indices. A positive (negative) value o f the index being an indication that the tax is progressive (regressive). The parameter p gives the priority assigned to lower percentiles o f the distribution. Increasing the parameter p increases the priority attached to the lower percentiles o fthe distribution. p=l.5 0.010 0.080 -0.028 -0.024 0.009 0.335 p=2 0.011 0.095 -0.033 -0.028 0.010 0.382 p=3 0.011 0.093 -0.032 -0.027 0.010 0.359 98. The results in Table A 3 show the post-reform tax regime to be more progressive than the pre-reform regime. The results are not sensitive to changes inthe parameter p. The index values are slightly positive under the pre-reform regime and increase in a systematic and considerable way with the reform. The indices for indirect taxes, however, are negative, illustrating their regressive character. After the reform this pattern i s slightly attenuated (the absolute value o f the indicators i s lower). Progressiveness indices confirm that the IRPF i s more progressive than the IRPwage tax. Horizontal Inequality, Reranking and Vertical Inequality 99. This section analyzes an aspect that receives less attention in the empirical literature on the distributional impact of public policies, the issue of horizontal inequality. Within the framework o f redistributive justice theory, the most classical formulation o f the horizontal equity principle posits that individuals who enjoy a similar welfare level ex ante government intervention should receive equal treatment from the policy under analysis. An alternative approach conceptualizes horizontal equity based on the absence of reranking in welfare distribution induced by public policies. For a tax to be horizontally equitable, the ranking o f individuals in accordance with their welfare level shouldnot be alteredby the tax system (Duclos and Araar, 2006). 100. There is a potential problem of rerankings generated by the implementation of a tax system aiming at greater vertical equity. At one extreme, ifthe tax system generates a perfect exchange in ranking between individuals without altering the mean income, vertical inequality-measured by means o f the Lorenz curve-would not change. Inthat sense, it i s possible to disaggregate the net redistributive effect o f taxes into one factor measuring vertical equity and another that captures the rerankindreordering induced by the tax system. The first component measures the proclivity o f taxes to reduce the dispersion o f after-tax income compared to pre-tax income. If the tax system induces rerankings, the reranking effect will always be negative, reducing the effectiveness o f the tax system as a tool to achieve a more equitable distribution of income. Table A 4 summarizes the results obtained from the calculation o f the net redistribution index and its breakdown into a vertical equity effect and reranking effect. The parameter p i s varied according to the degree o f aversion to inequality. Table A 4. Breakdown of the redistributive effect of taxation ComDonents Changein Vertical equity inequality effect (VE) Re-ranking (Redistributionof Effect(RR) (2)/(1) (3141) GN(P)-GY(P) incomes) (1) (2) (3) Before tax reform p=1.5 0.003 0.003 0.000 1.040 0.040 p=2 0.004 0.004 0.000 1.030 0.030 p=3 0.004 0.004 0.000 1.030 0.030 After tax reform p=1.5 0.009 0.009 0.000 1.020 0.020 p=2 0.010 0.011 0.000 1.010 0.010 p=3 0.010 0.010 0.000 1.010 0.010 Source: Basedonthe EGIHandthe ENHA. 101. Reinforcing the previous results, the net redistributive effect of the tax system i s positive in both cases, but its magnitude is greater after the tax reform. This increase in progressiveity of the post-reform tax system is due to the greater progressiveness o f the IRPF. This i s captured in Table A 4 by the Reynolds-Smolensky family o f indices, which measure the vertical equity effect (VE). For both pre- and post- reform tax systems, the progressiveness is not negatively impacted by the reranking effect. In other words, the pre- and post- tax reform does not cause major reranking problems for households in the distribution. 56 Annex 6. StatisticalAnnex Table A 5. Pre-reform and post-reform tax system Tax Pre-reform Post-reform 1 Impuestoa las RentasIndustriay Comercio(IRIC) 30% 2 Pequeiiasempresas Fixedbasic 3 Impuestoa las RentasAgropecuarias(IRA) 30% 4 Impuestoa las Rentasde Actividades empresariales (IRAE) 25% Impuestoala Enajenaci6nde BienesAgropecuarios (IMEBA) Multiple Multiple 6 Impuestoalas RetribucionesPersonales(IRP) 2%- 6% 7 Impuesto Renta de lasPersonasFisicas(IRPF) 10%-25% 8 Impuestoa las SociedadesFinancierasde Inversi6n(SAFI) 0.30% 0.30% 9 Impuestoa1Patrimoniode lasPersonasJuridicas 1.5%- 2.8% 1.5%- 2.8% 10 Impuestode Control de las SociedadesAnhimas Fixedbase Fixedbase 11 Impuestoa 10s Activos de las EmpresasBancarias(IMABA) Max. 2% Min. 0.01% 12 Impuestoa1 Controldel SistemaFinancier0 Max. 0.36% Min. 0.01% 13 Impuestoal Patrimoniode las PersonasFisicas Progresive0.7%-3% 0.10% 14 CompraVenta BienesMueblesen RematePublico 0.20% 15 ImpuestoCesiones o PermutasDerechos Deportistas 5% 16 Impuestoalas TrasmisionesPatrimoniales High 4% High4% Low 3% LOW 3% 17 Impuestoal Valor Agregado (IVA) Minimum 14% Minimum 10% Basic 23% Basic 22% Contribuci6nal Financiamientode la Seguridad Social l8 (COFIS) 3% 19 impuesio EspecificoInterno(IMESI) Multiple Multiple Adicional Impuestoala Enajenaci6nde Bienes 2o Agropecuarios (adic. IMEBA) 0.2% - 0.4% 0.2%- 0.4% 21 ImpuestoCompraMonedaExtranjera(ICOME) 2% 2% 22 FondoInspecci6nSanitaria(FIS) 1% 1% 23 Impuestoa 10s Sorteos(Fondo Nac. de Recursos) 20% 24 ImpuestoEspecificoa10s Serviciosde Salud (IMESSA) 3yo 25 Impuestoa 10s IngresosCompaiiiasde Seguros Multiple Multiple 26 Impuestoa las Comisiones 9% 27 Impuestoa las Ventas Forzadas 2% 28 Impuestoalas Telecomunicaciones Multiple 29 Impuestoalas Tarjetasde Credit0 Fixedsum 30 Detraccionesa la Exportacion Multiple Multiple Source: Basedon MEF (2006). 57 TableA 6. LaborIncome: Minimum,meanand maximumincome,per decile(in UR$) Minimum Average Maximum 1 0 354 1,040 2 990 1,637 2,300 3 2,179 2,887 3,574 4 3,370 4,061 4,800 5 4,538 5,325 6,105 6 5,763 6,720 7,696 7 7,253 8,428 9,716 8 9,162 11,075 13,211 9 12,467 15,834 20,192 I O 19,040 35,457 402,578 Source: Basedonthe ENHA TableA 7. Variationinlaborincome(disposable) by decile (UR$) Beforethe tax Variation inaverage reform After the tax reform income 1 352 352 0 2 1,636 1,637 0 3 2,879 2,887 7 4 4,048 4,061 14 5 5,263 5,325 62 6 6,590 6,720 130 7 8,23 1 8,343 112 8 10,608 10,762 154 9 15,061 15,019 -42 10 33319 31,349 -2,470 58 N 9 % 09 - N 2 0 w 4 I G se, E 0 B m 3 s2 h; 3 .. :; 7C 1 f U $ P P f 6a 3 E fa 7ac E 4c U c I b ..cec ii cE 7 e 5 i 3c a d r n U Q W r i . . . . . , Table A 11. Situationof workers accordingto type of occupation(selected occupations) IRPF(a) Difference YOLosers by IRp(b) (a-b) occupation Losers Directors and company senior managers 14.6 3.9 10.7 79.7 Other department managers 15.2 5.5 9.7 88.0 Managers ofproduction and operation departments 14.0 5.3 8.7 83.3 Specialists inbusiness organization and administration and related 12.6 4.2 8.5 78.7 Civil service managers 13.6 5.3 8.3 81.4 Member o fthe executive branch and o f legislative bodies 13.0 5.0 8.0 46.3 Physicians and related 11.5 4.7 6.8 73.7 Architects, engineers and related 10.9 4.7 6.1 70.9 Technical personnel inmaritime and aeronautic navigation 10.9 4.8 6.1 62.7 Professionals inbiological sciences and other disciplines 10.4 4.4 5.9 66.7 ITProfessionals 10.1 4.6 5.5 68.3 Physics, chemistry and related 9.9 4.8 5.1 50.7 Law professionals 10.7 5.6 5.0 46.7 Winners Glass, ceramics and related 1.4 2.7 -1.3 52.5 Machine operators for the fabrication o f wood products 1.4 2.8 -1.4 50.9 Drivers o f vehicles and heavy machinery 0.5 2.0 -1.5 25.9 Officials and operators inconstruction and related 2.0 3.5 -1.5 57.3 Day laborers inminingand the construction 1.7 3.6 -1.9 47.9 Source; Based on the ENHA. Table A 12. Pensions: Minimum,mean and maximumvalue, by decile Minimum Average Maximum 1 0 1,141 1,800 2 1,804 2,257 2,500 3 2,501 2,667 2,800 4 2,801 3,103 3,400 5 3,403 3,799 4,200 6 4,201 4,798 5,300 7 5,302 6,189 7,000 8 7,006 8,115 9,000 9 9,001 10,842 13,228 10 13,239 22,131 92,600 Total 0 6,438 92,600 Source: Based on the ENHA. 62 Table A 13.Variationof average pensionsby decile(UR$) Variation inaverage Before the reform After the reform income 1 1,141 1,141 0 2 2,257 2,257 0 3 2,667 2,667 0 4 3,103 3,103 0 5 3,799 3,799 0 6 4,798 4,798 0 7 6,189 6,189 0 8 8,081 7,4 13 -669 9 10,621 9,758 -863 10 21,680 19,375 -2,305 Total 6,368 5,996 -372 Source: Based on the ENHA. 63