Document of The World Bank

Transcription

1 Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Document of The World Bank FOR OFFICIAL USE ONLY IMPLEMENTATION COMPLETION REPORT (SCL-43690) ON A LOAN IN THE AMOUNT OF US$5.0 MILLION TO THE FEDERATIVE REPUBLIC OF BRAZIL FOR A STATE PENSION SYSTEMS REFORM TECHNICAL ASSISTANCE PROJECT August 16, 2006 Poverty Reduction and Economic Management Sector Management Unit Brazil Country Management Unit Latin American and the Caribbean Region Report No: BR This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.

2 CURRENCY EQUIVALENTS (Exchange Rate Effective April 17, 2006) Currency Unit = Real R$ R$1.00 = US$ US$ 1.00 = R$ FISCAL YEAR January 1 to December 31 ABBREVIATIONS AND ACRONYMS BTOR CONAPREV CRP DF DRPSP GDP GOB ICR ICT IDB PNAFE IFI INSS LCR LIL MF M&E MPAS MPS PAD PARSEP PARSEP II PCU PDO PEU PHRD PREV-MUN PROPREV PROST PUC QAE RGPS RPPS S/SECAL SIDARP Back to Office Report Conselho Nacional Prevendenciarial Certification of RPPS Legal Compliance Federal District Department of Public Social Security Systems Gross Domestic Product Government Implementation Completion Report Information and communication technology Inter-American Development Bank IBD loan International Financial Institution National Social Security Institute Latin American and Caribbean Region Learning and Innovation Loan Minsterio de Fazenda Monitoring and Evaluation Ministerio de Prevedencia e Asistaneca Social Ministerio de Prevedencia Social Project Appraisal Document State Pension Systems Reform Technical Assistance Project State Pension Systems Reform Technical Assistance Project II Project Coordinating Unit Project Development Objective Project Executing Unit Policy and Human Resources Development Fund WB Municipal Pension Reform Project IDB Prension Project Pension Reform Options Simulation Toolkit; WB pension management software Project Coordinating Unit Quality at Entry State managed pensions for private sector workers Civil service pensions Pension Sector Adjustment Loan Current version of SIPREV is now called

3 SIPREV SPS TTL TU UEP UNDP WB Pension management software developed under PARSEP Secretaria de Prevendencia Social Task Team Leader Technical Unit Project Execution Unit United Nations Development Program World Bank Vice President: Country Director Sector Manager Task Team Leader/Task Manager: Pamela Cox John Briscoe Ronald E. Myers Linn Hammergren

4 BRAZIL BR PENSION REFORM LIL CONTENTS Page No. 1. Project Data 1 2. Principal Performance Ratings 1 3. Assessment of Development Objective and Design, and of Quality at Entry 2 4. Achievement of Objective and Outputs 5 5. Major Factors Affecting Implementation and Outcome Sustainability Bank and Borrower Performance Lessons Learned Partner Comments Additional Information 21 Annex 1. Key Performance Indicators/Log Frame Matrix 22 Annex 2. Project Costs and Financing 25 Annex 3. Economic Costs and Benefits 27 Annex 4. Bank Inputs 28 Annex 5. Ratings for Achievement of Objectives/Outputs of Components 29 Annex 6. Ratings of Bank and Borrower Performance 30 Annex 7. List of Supporting Documents 31 Annex 8. Partner Comments in Portuguese 32

5 Project ID: P Team Leader: Linn Hammergren Project Name: BR PENSION REFORM LIL TL Unit: LCSPS ICR Type: Core ICR Report Date: August 16, Project Data Name: BR PENSION REFORM LIL L/C/TF Number: SCL Country/Department: BRAZIL Region: Latin America and the Caribbean Region Sector/subsector: Compulsory pension and unemployment insurance (100%) Theme: Social safety nets (P); Other public sector governance (P); Law reform (S) KEY DATES Original Revised/Actual PCD: 04/03/1998 Effective: 11/01/ /22/1999 Appraisal: 04/03/1998 MTR: 06/30/ /30/2002 Approval: 06/30/1998 Closing: 12/31/ /01/2005 Borrower/Implementing Agency: Other Partners: GOVERNMENT OF BRAZIL/MINISTRY OF FINANCE STAFF Current At Appraisal Vice President: Pamela Cox David De Ferranti Country Director: John Briscoe Vinod Thomas Sector Manager: Ronald E. Myers Gobind Nankani Team Leader at ICR: Linn Hammergren Chris Parel ICR Primary Author: Linn Hammergren 2. Principal Performance Ratings (HS=Highly Satisfactory, S=Satisfactory, U=Unsatisfactory, HL=Highly Likely, L=Likely, UN=Unlikely, HUN=Highly Unlikely, HU=Highly Unsatisfactory, H=High, SU=Substantial, M=Modest, N=Negligible) Outcome: Sustainability: Institutional Development Impact: Bank Performance: Borrower Performance: HS HL H S S QAG (if available) Quality at Entry: HS Project at Risk at Any Time: No ICR HS

6 3. Assessment of Development Objective and Design, and of Quality at Entry 3.1 Original Objective: The objective of the project was to prepare the Brazilian states technically to address pension reform and help them to move through the initial phases of reform. It was also to assist the Federal and State Governments to choose from among the best options and encourage them to implement sound reform. There was a further, implicit objective -- to build a constituency for reform. The project is a Learning and Innovation Loan (LIL), designed under the old format, which is to say it was intended to be flexible and to help participants explore and develop solutions to specific problems, or build the capacity to develop them. In 1999, when the PARSEP project began, the fiscal situation of most Brazilian states was precarious. As of December 1997, state and local debt was 13 percent of GDP and 38 percent of all government debt. Pension obligations constituted a large and growing component of state debt, consuming percent of revenues in some states. The gravity of the situation was also indicated by the ratio of pension expenditures to current salary bill in the public sector which in the states of Minas Gerais, Rio de Janeiro, and Rio Grande do Sul ranged from 50 to 120 percent. While the need to address these problems at all levels of government was clear, there were several complicating factors. First, during the 1990s, the number of pension programs for public workers had proliferated at the state (and municipal) level, but without adequate financing or provisions to manage contributions productively. Second, and as a result, many states managed their multiple systems on a cash flow basis, using their income from contributions to make current pension payments, and whatever remained for other current expenditures. Third, none had in operation any mechanisms (actuarial techniques or even an adequate cadastre of actual and potential beneficiaries) for estimating contingent liabilities over the longer run. Fourth, those running the programs for the most part lacked any special training in pension fund management; their duties were usually limited to collecting contributions and calculating and paying benefits. Fifth, state pensions were constitutionally a state attribution and the Federal Government thus had no obligations or rights regarding state pension reform. Moreover, the Constitution of 1988 had increased the guarantees and lowered restrictions as regards pension rights thereby augmenting the demands on the systems and reducing the room for making changes. This created enormous resistance among potential beneficiaries to anything they interpreted as threatening their acquired rights; Fernando Henrique Cardoso s unsuccessful efforts during his first term to introduce basic reforms suggested just how potent this resistance could be. The solution for what was and remains perhaps the most important macroeconomic issue facing the Brazilian Federal and state governments was not sought only in this project. As the LIL was being prepared, the government was making another attempt at reform (approved via Constitutional Amendment in December, 1998) to a) introduce principles of actuarial and fiscal balance into private and public sector workers pension systems; b) eliminate certain special benefit rules and regimes, put a benefit ceiling on RGPS (state managed pensions for private sector workers) indexed to inflation, set a minimum retirement age for the RPPS (civil service pensions), and establish a minimum vesting period of 10 years to receive higher RPPS benefits; and c) enable states and municipalities to create complementary pension funds. Taking advantage of the Brazilian initiatives to address the larger problem, the WB prepared a package of operations. In addition to PARSEP, these included a S/SECAL of $ million (approved December, 1998); a second LIL for technical assistance in effecting the constitutional changes within the INSS (the entity managing the RGPS); and a PHRD grant to prepare a second Social Security S/SECAL. Given this package of assistance, additional support from the IDB, and the Government s own efforts, attributing - 2 -

7 effects is complicated. Nonetheless, the PARSEP experience was instrumental in driving further advances. The model and software it developed were used and further refined in PREV-MUN (signed in 2003) and the IDB s PROPREV, and have spawned the follow-on project, PARSEP II. Furthermore, as discussed below, PARSEP contributed in generating the conditions for further national-level reforms and also to the drafting of additional legislation. 3.2 Revised Objective: Project objectives were not revised over the period. However, some additional PDOs (and intermediate objectives) were added and the targets for the initial PDOs were upgraded to higher levels. These developments reflected the counterpart and the Bank s evolving understanding both of the challenges and potential for change and for the most part involved setting more ambitious goals e.g. the creation of a new software, the establishment of permanent entities within the MPAS/MPS to oversee subnational programs, and the revision of national, as opposed to state, laws. 3.3 Original Components: The project was organized both in phases and components. As the project was prepared in an era when the Bank was less insistent on a uniform design format, the original documentation is somewhat inconsistent on the components, which appear in two versions in the PAD and still a third version in the loan agreement. However, whether distributed among phases, parts, or components, the basic activities are the same. The format used here is the first version appearing in the PAD. The only significant change over the life of the project was the addition of some activities, all of which are consistent with the general objective and simply reflect the Bank and borrower s improved understanding of the situation. It was expected that at least 15 of Brazil s 26 states (plus the Federal District) would complete Project Phases 1 and 2, and that Phase 3 would be completed on an exceptional basis. The phases and their components were as follows: Phase I: Diagnostic. Actually more than a diagnostic, this phase aimed at providing the states with the structural and technical tools to understand their current situations and begin the development of reform proposals. It included the following components: a. Development of cleaned sample databases of active and retired employees at the state level so as to provide actuarial data for pension modeling b. Training of two officials per state in pension reform and pension modeling making use of the Bank s PROST software c. Diagnostic of current and future pension obligations and financial implications in each state using PROST or other simulation models d. Dissemination by states of results and debates in open forums e. Strengthening of state pension units with appropriate mandates, trained personnel, budget, hardware and software to manage pension reforms f. Development of technical unit (TU) within the federal pension ministry (MPAS, later MPS) with a legal mandate and appropriate resources (including trained staff) to assist the states and the Ministry of Finance in implementing the project and monitoring through shared data the evolution of state reforms. Phase 2: Reform Proposals - 3 -

8 a. States develop reform proposals using PROST and other simulation models vetted by the TU b. Dissemination by states and debate of results in open forums. Phase 3: Institutional Framework a. A few states design the legislative and institutional framework necessary to implement pension reform b. States disseminate and debate results in open forums Phase 4: Government strengthening. through studies in 3 crucial areas: This phase was intended to support State and Federal reform a. Complementary legislation to implement the constitutional reforms b. Best practices for pension fund management c. Federal INSS pension related financial obligations to states and state obligations to INSS A fourth WB independent study of critical pension issues was to be done during the first year of the project. This study would use the experience of other countries and Brazil s own experience to identify relevant lessons, and would also discuss what needed to be done for state reforms to be consistent with broader reforms of the social security system. 3.4 Revised Components: Because this was a LIL, and because of the various factors impeding its rapid start-up, the project design did vary over the life of project. Most notable among these variations were the substitution of recadastramentos [1] of the entire executive branch (as opposed to the initial 15 percent samples) in seventeen states and the creation of a new pension management software as major activities. Additionally, the development of a management unit within the MPAS (now MPS) took on much greater importance and produced far more significant results than initially envisioned. This poses a challenge for the evaluators, which has been addressed by using the initial project design, but noting where additions and changes in emphasis were made. The bottom line, however, is that it was these changes that made the project so successful. Had it followed its initial design to the letter, it would have produced far less and had far fewer significant impacts. 3.5 Quality at Entry: Quality at entry was rated Excellent and the Project was recognized as one of 10 QAE distinguished loans for that period. In 2000 it also received a Social Development Award for Excellence in Project Quality in 2000 and a CMU Performance Award was given to the Social Security team for high quality analytic, technical and financial support to the Brazilian program of social security reform at both the Federal and State levels. PARSEP was showcased for its results in 2004 as a catalyst for Brazilian Pension Reform. [1] There is no good English equivalent for the term recadastre (recadastramento in Portuguese) is used to denote the process of vetting the lists of active and pensioned employees to eliminate redundancies, errors and fraud

9 4. Achievement of Objective and Outputs 4.1 Outcome/achievement of objective: Despite the obstacles and challenges faced by the project, it far surpassed its initial objectives and its rating is highly satisfactory. The objectives (and components), however, went through some further interpretation and upgrading in the course of the 7 years, as did the indicators for their success. At the time the project initiated, Brazil was only beginning its pension reforms and no one really understood the full mix of factors contributing to any solution. The initial project plan underestimated the states potential interest in the new techniques and tools it introduced as well as the importance of and possibility for effecting further changes in federal as opposed to state legislation in advancing reform. It also overestimated the appropriateness of the software program (PROST) it proposed to use for the program. Finally, unknown and thus not considered in the design stage, was the role of project activities in effecting immediate savings by detecting fraud and overpayments in existing pension systems, as well as the importance this would have in encouraging states to join the program. When one of the early samples uncovered sufficient errors and fraud to allow several million dollars of annual savings, the MPS counterparts and the TTL recognized that expanding the vetting to the entire executive branch cadastres would be important in building a constituency in the states and building interest in the project. The initial formal objective was in two parts: to prepare the states technically to address the issue of pension reform and to help them through the early stages, and to help the State and Federal Governments choose from among the best options and encourage them to implement sound reforms. As noted, there was a third, implicit objective of constituency building. On the state side, the goal was to establish adequately resourced pension units, train 2 officials in each state, and diagnose and disseminate information on the current situation and on reform proposals. The units already existed, but the project has supported their equipment and training and encouraged their more adequate staffing. In addition it has contributed to national-level legislation (not envisioned in the initial plan) to create a unified administrator for all civil service pension systems (all branches and units of state government). States did make additional changes along the way, in some cases anticipating parts of the 2003 constitutional amendment (as regards increasing contributions from employees and employing agencies) and adding provisions for investment of contributions (thus ending the cash flow management style). However, given that so much of the system was constitutionally mandated, the potential for individual state reforms proved to be more limited than initially envisioned. Training was to be in PROST, a pension management software developed by the Bank. PROST was found to be inadequate to Brazilian needs. It did not drill down to the individual account level and moreover was a proprietary software developed by Bank staff, a situation which caused some friction when the Brazilians showed an interest in making their own changes to it. The project, thus sponsored the development of (and training in) a substitute system, SIPREV, which is now being adopted nationally. SIPREV s comprehensive electronic database (modeled in part on PROST) and actuarial modules were instrumental to PARSEP s advances. A payroll and benefits module is currently under development with the assistance of other loans. Two other unanticipated developments were the creation of CRP, a website managed by the MPS and posting state legislation and actuarial analysis and evaluating its adequacy, and of CONAPREV, a national association of pension officials which has assumed an important advisory and educational role. The latter drew on a model already existing in other ministries which sponsor associations of state secretaries. On the federal side, the major aim was to create a technical unit (TU) within the MPAS (later MPS) to assist the states and the Ministry of Finance in implementing the project and monitoring through shared - 5 -

10 data, the evolution of state pensions. Again the initial goal was far surpassed. The implementation team (UEP) has metamorphosed into a fully staffed ministerial department (Department of Public Social Security Systems, DRPSP) with a (new) constitutional mandate to monitor and evaluate sub-national RPPS systems. (World Bank, 2006, p. 34). The Secretaria de Prevedência Social, within which it sits, has also evolved into a much stronger body, responsible for overseeing all RPPS programs. Over the course of the project, and helped by inputs it and its Brazilian team have provided, the principle that the federal government oversees all sub-national programs is now firmly established, in law and in the organizational structure. On both the state and federal side, the aim had been to study and develop recommendations for system reform. The results have gone far beyond this to help produce a second constitutional reform (in 2003) which made several across the board changes to the RPPS, as shown in box 1 below: Box 1: Main Components of the December 2003 Reform to the RPPS World Bank (2006), p ) Stricter Rules for Retirement in the Public Sector. The introduction of penalties ranging from 3.5 to 5 percent of the wage base for each year of retirement prior to the normal retirement age of 60 for men and 55 for women; the abolition of proportional retirement provisions for those who did not have the right to retire from this program before the reform; the increase in the required years of service in the public sector from 10 to 20 years, and in the last position from 5 to 10 years. 2) New Benefit Formula. The change of the pension base, from the last wage to the average of 80 percent of the highest real wages after 1994; 3) New Indexation Rule. The switch from wage to price indexation of benefits; 4) Rules on Acquired Rights. Measures (1)-(3) do not apply to workers who had already acquired the right to retire, under the old rules. Measures (2)-(3) do not apply to workers who meet the new stricter conditions for retirement described in (1). 5) Survivor Pensions. A reduction in survivor s pensions from 100 percent of the original pension to 70 percent above a threshold of 2,400 Reais per month; 6) Imposition of Contributions/Tax on RPPS Pensioners. The introduction of a contribution rate of 11 percent for pensioners, on the benefit portion above the threshold of 2,400 Reais per month; 7) Imposition of Benefit Ceiling for RPPS combined with Creation of Complementary Pension Funds. The introduction of a benefit/contribution ceiling, equal to the respective ceiling in the RGPS, provided that the employer (the federal, state, or municipal Government) creates a complementary defined contribution pension fund; 8) Overall Wage and Benefit Caps. Maximum wage and benefit levels were introduced, and defined by the salary of the respective leaders of the executive, judicial and legislative branches of the federal and subnational governments; 9) Minimum Subnational Contribution. The introduction of a minimum contribution of 11 percent for all active state and municipal civil servants; 10) RGPS Ceiling. The RGPS (and thus RPPS) contribution/benefit ceiling was raised to 2,400 Reais per month. Other legal reforms of particular importance include setting the mandate of the federal MPS to oversee the RPPS and enforce compliance with new RPPS legislation, including legal frameworks, actuarial analyses, reporting, and most recently the clean up of pension data bases and the adoption of SIDARP (as the current version of SIPREV is now called). While the results cannot be attributed entirely to PARSEP, and affect more than the states employees, the information amassed through the project was instrumental in promoting the reform and its eventual passage. SPS staff working with PARSEP also helped draft legislation leading to and building on these amendments. The consequences of the 2003 constitutional amendment should have an important impact on state pension pay-outs, and pave the way for further - 6 -

11 reforms of the systems. 4.2 Outputs by components: Phase/Component 1 a) States develop a sample data base of 15 percent of active and retired state employees sufficient to extract actuarial data for pension modeling Highly Satisfactory. This became one of the most successful and most critical elements of the project. It was originally believed that 15 states would create cleaned partial samples as a basis for future modeling. Instead 17 states have completed or contracted full recadastres and 24 in total undertook some kind of exercise. Because of legal limitations and their own resistance, the databases of other branches of government could not be included in most cases (judiciary and legislature, in total not comprising a significant percentage of all state employees, although because of their high pensions, representing significant costs and possibly significant savings). However, the 2003 constitutional amendment through its creation of a single administrator has changed that situation, and the addition of the other branches will be covered under PARSEP II. Nonetheless, the recadastres produced an additional enormous benefit and one contributing to the popularity of the program the savings generated by identifying fraud, overpayment, or other intentional or unintentional error. This was not a result anticipated in the project documentation, although interviews with project staff suggest they saw this as a further consequence of the sampling exercise. Table 1: Potential Savings through August, 2005, from Nine States [6] State Branch & Employee Status Obits [8] Amapá Goiás Maranhão Mato Grosso do Sul Pernambuco Executive (active, inactive, pensioners) Executive (active, inactive, pensioners) Executive (inactive and pensioners) All branches (active, inactive, pensioners) Ghost Employees [7] Accúmula-tion of Jobs [9] Total employees after cadastre upgrade Annual Savings [10] (R$) Executive (active, inactive, pensioners) Piauí Executive & Tribunal de Contas (active, inactive, pensioners) Rio Grande Executive (active, do Norte inactive, pensioners) Sergipe Executive (active, inactive, pensioners); Legislative and Judiciary (actives, inactive) Tocantins Executive (active, pensioners) Total

12 [6] Source: MPS reports [7] Employees that do not appear in the 2 stages of recadastramento [8] Former employees whose death was communicated during recadastramento [9] Employees who accumulate 3 or more jobs, which is against the law. [10] Calculations assumed gross costs of R $ 600/employee and 13 salaries a year. The results above are only for nine states and for the most part only for the executive branch, but they still total over US $100 million of potential annual savings. While fraud and overpayment are hardly the greatest problem posed by the current pension system, the opportunity for making these immediate savings has enhanced the program s popularity among other states and at the municipal level. The recadastres have allowed cross-checking with other databases to capture still more fraud, and their results now feed into a central system managed by the SPS. They also pave the way for component C below, the actuarial studies to determine future liabilities without change to the current system. b) Training of 2 officials per state in pension reform and modeling via PROST Highly Satisfactory. Here again the project took an unexpected turn, with the substitution of a new, project-funded pension system management software, SIPREV which is being expanded into a multi-module system for managing the all pension programs. A decree is expected to be enacted which will mandate SIPREV s (now called SIDARP) use in all subnational governments. Aside from the development of SIPREV and its adoption in 20 states, there have been 19 training programs in its use, and 5 multi-state workshops, the latter with a total of 590 participants. The project also supported multiple training activities in the principles of pension system management and the publication of related texts. These activities began as early as 1999, although they have picked up speed in the last few years. c) States produce a diagnostic of current and future pension obligations via PROST or other simulation models Highly Satisfactory. As a consequence of the 1998 constitutional amendment, actuarial analyses were legally required at the time of project effectiveness. However, virtually no state had the capacity to do one and there was no means of forcing compliance (because of the states independence). The few that tried produced very poor results. The focus on recadastramentos (as opposed to the sample surveys) doubtless slowed the process, but, once it was begun, generated more faith in the results. The strategy was to help the states conduct a first study, and then on that basis have them continue on their own later. So far studies have been completed in 20 states with a few more pending. Most were done in 2004 and The creation of a website managed by the MPS on which the results of the actuarial studies are posted provides an additional impetus for continuing the practice, as do the MPS s new legal authority to oversee, vet and provide technical assistance to the subnational programs and the provision for a cut-off in federal funding for state noncompliance. d) States disseminate and debate results in open forums Satisfactory. As documented in the reports from the PCU, the states have held numerous forums on pension management both for system managers and affected workers. Through the project they have also been provided published materials to back up the discussions. However, wider discussions of the actuarial results have been delayed because of the slow start of this component. Nonetheless, following the completion of each study, the MPS auditors met with state officials to debate findings, leading in some - 8 -

13 cases to the states decision to increase taxes and contribution rates. The savings from the recadastres have been publicized, creating other sources of support for the undertaking. If the purpose was the creation of recognition of the need for change, it was fully met. However, given the expectation that the discussions would draw more fully on the actuarial results, the rating is only satisfactory. e) MPAS develops a technical unit with a legal mandate and appropriate resources to assist states and MF in implementing the project. Highly Satisfactory. This is another of the dramatic successes of the program, far surpassing the initial expectations. Once the project coordination responsibility was removed from Finance Ministry, and in some sense even before, those units in the MPS responsible for the oversight of pension projects were transformed into the Secretaria de Prevedência Social (SPS) and the Department of Public Social Security Systems (DRPSP) and given the legal mandate to oversee all RPPS (including those of subnational entities). The SPS, considered by Bank specialists to be a unique institution in the region, received project support for consultants, logistics, training, IT, publications, and related expenses. Both the Project Coordinating Unit (PCU), which reports to the MPS Executive Secretary and manages all its IFI-sponsored projects, and the Project Executing Unit (PEU) located in the SPS and responsible for the technical execution of PARSEP, have also received support, and the latter in particular has been responsible for such innovations as the development of SIPREV, the CRP system, and CONAPREV, as well as input into the drafting of new legislation. With PARSEP support, the MPS also created the National Council of State RPPS Directors (CONAPREV) which has assumed an increasingly important consultative role representing state views on proposed reforms. In short, rather than limiting itself to the development of a project-dedicated technical unit, PARSEP has contributed to the formation of a series of permanent organizations within the MPS legally responsible for overseeing the entire RPPS and thus destined to last beyond the project s life. Phase/Component 2 States Develop Reform Proposals Using PROST and any Other Simulation Models Vetted by the TU Satisfactory. As noted, the potential for reform at the state level was limited by the constitutional status of the programs. State reform has instead gone in the direction of strengthening their pension management units, creation of arrangements to invest contributions, and use of the recadastres and cross-checking with other data bases to detect fraud. Still state participation in the project has increased openness to the further constitutional and national level legal reforms which in the end will reshape their own programs. In the end, this phase or component is really a part of the following item, which does get a higher grade. Phase/Component 3 A few States on an Exceptional Basis, Design the Legislation and Institutional Framework Necessary to Implement Pension Reforms Highly Satisfactory. This can be considered a legitimately modified component. Once it became apparent that state systems were limited as to the changes they could introduce on their own and that further legal and constitutional modifications were instead required at the national level, project efforts shifted in this direction. The results included a constitutional amendment, as outlined in Box 1 above, which changed some of the basic rules governing pension systems in all states. The results have been far more important than a state-by-state approach would ever have produced. However, work with the states in the preceding activities clearly fed into the development of the final national-level changes

14 Phase/Component 4 Government Strengthening Satisfactory. Governance strengthening was highly satisfactory as indicated in the above ratings. However this component anticipated something else the execution and discussion of a series of studies as an educational tool. The studies were done and numerous workshops and publications produced as a result. However, their contribution to the overall progress of the project may have been less than the lessons learned from the other activities. This is not a design or implementation fault. Had the counterparts adopted the activities less enthusiastically, the studies might have played a larger role. As it turned out, their impact was overtaken by events and the events produced far more than had been anticipated. 4.3 Net Present Value/Economic rate of return: This was a remarkably cheap project with a remarkably large impact. However, many other actions were occurring simultaneously so that attribution of all potential impacts is difficult. Like the financial returns (see 4.4 below) these are not immediately calculable, because they will accrue from the project s contributions to further reforms in Brazil s pension system. So far most attention has focused on on-going fraud. The next step is a more serious consideration of the actuarial studies, their implications for contingent debt, and the introduction of measures to curb the effects. This is intrinsically a longer term and highly politicized process. However, changes introduced in the 2003 Constitutional Amendment are an important first step as is earlier legislation that established the MPS s right to oversee and vet state pension systems. Better information on actual and future pay-outs, assuming a continuation of the current system, may not be sufficient to overcome the many vested interests, but they are an important tool in doing this. As the new software (SIPREV) also allows drilling down to individual cases, a second and more potent form of analysis might focus on the intrinsic inequalities of the current system how much of the public budget is paying the pension of what percentage of the population and how this is distributed among the beneficiaries. None of this would have been possible without the changes introduced by PARSEP and thus the project has contributed to advancing a still uncalculable improvement in public finances and the opening of fiscal space. 4.4 Financial rate of return: The project has offered immediate financial gains to the participating provinces. The most obvious and least contestable results are the savings made available through purging of the pension cadastres and the identification and elimination of fraudulent and erroneous payments. Early results from 9 states showed potential savings (assuming agencies act on the findings) of US $100 million annually. Once all recadastres are completed it is anticipated that the savings should be at least double, and once the other branches of government are added (fewer pensions, but among the largest) savings should be even greater. Moreover, the linkage of the state databases in the SPS will also allow identification of fraud and illegal payments that extend across states individuals collecting excessive amounts or excessive numbers of pensions in several jurisdictions. Against an initial investment of $5 million, this is already an enormous return. 4.5 Institutional development impact: As detailed throughout the text, the project s institutional development goals have been more than fully met. State pension units have had their internal capacity strengthened, their staff trained in pension management, and their technical resources upgraded. The new units in the MPS are now fully charged with overseeing development of all government pensions included in the RPPS and have the capacity to do this. The creation of the CRP and CONAPREV support these institutional changes and their sustainability over time

15 Moreover the institutional environment; as regards a broader understanding of the workings and challenges of pension management has also been altered for the better. All of this was incorporated in the project s initial goals, but the results have far surpassed those initially predicted. MAIN PARSEP ACHIEVEMENTS BASELINE TO PRESENT* Loan Objective Situation in June, 1998 Situation Today & PARSEP Contribution Develop strong Government/MPS role in sub-national government pension regulation and supervision. Develop an organization in MPS capable of assuming an important role in sub-national pension M&E and capable of providing technical assistance Develop in MPS and the States comprehensive, homogeneous systems to assist pension management and assure data adequacy. MPS had no constitutional role or right permitting them access to information, nor right to review or validate pension fund management. There was reticence to assume responsibility because of the sub-national pension debt. There was minimal regulation and supervision of sub-national government pensions which resulted in an untenable situation. Note that at the time of loan preparation there was no way to oblige sub-national governments to participate in PARSEP and take advantage of the funding and TA offered through MPS nor was there an expectation that the role of MPS would be much altered over the 3 year project life. New legislation (Law no /98 and Decree no /99, Constitutional Amendment No. 20/98, Law no , Complementary Law no 101, Law no , Decrees 3.112/99, 3.217/99, 6.209/99, among, others established Broad-based pension reforms Sub-national pension regulation MPS s key role in monitoring, evaluating and regulating performance and providing technical assistance SPS now monitors, validates and provides systematic technical assistance in legal, actuarial, administrative, systems, and legislative areas. It has assumed a key role in sub-national pension administration. The drafting of much of this legislation, establishment of CRP, and monitoring, validation and provision of technical assistance were supported through the establishment of DEPSP and PARSEP funding. In fact, much that the Government has achieved in the area of sub-national and to some extent, national --pension reform has directly or indirectly been supported through PARSEP financing of SPS and DEPSP which has exercised a central role in implementing virtually all of the reforms (see below). MPS had no unit responsible for DRPSP was created in SPS with professional and support sub-national pensions, no data bases staff, many funded by PARSEP. The unit implemented the with information on state pension PARSEP technical assistance loan and prepared organizations, administrators, phone PREV-MUN. Today DRPSP maintains a complete data base numbers, etc., and no data on the covering virtually every aspect of State pension programs and pension systems for the 26 state and includes technicians who have worked with almost all states over 2,000 sub-national pension and know the programs and their administrators well (only programs. During loan preparation a MT opted out of PARSEP). SPS developed the CRP and small implementation unit was monitors legislative compliance as it will do with actuarial envisioned that would grow into a compliance. It provided, with PARSEP support, funding and permanent organizational unit but it TA for cadastre rationalization, actuarial analysis, and was unclear how this would be done systems work. It has been legally empowered and has since there was no legal mandate assumed a major role in provision of TA and dissemination of empowering MPS to oversee state & pension laws, regulations, good practices. (Discussed municipal pensions below). Each state (and municipal) pension program had its own system subsequently found to be mostly incomplete and inadequate for pension management and they were unknown to SPS. PARSEP PROST was adapted to the Brazil reality, courses were given and Federal and state officials were trained. It is still used occasionally in Brazil research but SPS determined that PROST did not satisfy its needs and developed, with PARSEP support, SIPREV--a superior product for Brazil and a possible international best practice. SIPREV includes

16 Upgrade pension cadastres Building reform constituencies through systematic dissemination and training envisioned implanting PROST in modules to validate data and perform actuarial analyses. It Federal and state pension agencies, a should soon have a payroll module and work needs to Bank software used around the recommence on a planned benefits module. Note also that world by Bank analysts, including in SIPREV communicates electronically with MPS which Brazil. addresses the need for regular reporting. It also provides states with an electronic system for unifying and managing its executive, legislative, judicial and other systems. MPS had no knowledge of state cadastres other than that with few exceptions they were incomplete, maintained on inadequate systems, and resulted in significant unnecessary or fraudulent payments, based on results from the Parana on a 15% cadastre sample conducted during project preparation. It was hoped that PARSEP could help build reform initiatives in the states which would also support Federal reform efforts. At the time of the loan, state pension management was autonomous from Federal intervention and there were strong vested interests. These interests prevented the inclusion of the judicial and legislative systems in most states, a problem apparently now resolved with the creation of a single pension administrator. There was also virtually nothing written on Federal and state pension legislation and management. Finally, there was little public understanding of the fiscal burden imposed by pensions --considered then as now by the Bank to be Brazil s principal macroeconomic problem. 20 states have already implemented SIPREV with more expected. Training continues. Actuarial and payroll modules have been successfully tested but more work is required to develop the benefits module. There is now a need to consolidate and disseminate SIPREV and available modules throughout remaining states and especially municipalities where the modules are expected to be of even greater usefulness. This effort is funded by PARSEP and will be continued by PREV-MUN and PARSEP II. PARSEP offered funds to each state to upgrade cadastres. By close of project, 20 states had completed or contracted full revisions of their cadastres for the executive branch employees, 24 had participated in some part of the exercise and a few had been able to add judicial and legislative employees. Reports from 9 states indicate potential annual savings of R$$255.8 million from this effort R$74 million in PE alone. The expectation is that this example, once disseminated, will provide all states and municipalities with sufficient incentive to carry out comprehensive cadastre upgrades. This work is part of PREV-MUN and PARSEP II. The reform constituency has evidently grown enormously over the intervening years as has knowledge regarding pensions at the Federal and sub-national levels. It is impossible to measure PARSEP s contribution as the entire political environment has shifted toward reform. However, at the time of the mid-term review, state and MPS participants termed PARSEP as a critical catalyst in supporting the reform movement. To this end PARSEP has funded the following activities each with participation normally exceeding 100 stakeholders up to state run stakeholder workshops in 15 states 11 workshops, courses, and seminars for Federal and state pension and other officials 9 training seminars for state officials Dissemination and training activities have further been supported by printed materials including 9 state pension publications publication and broad dissemination of Informe do PARSEP publication of the Colecao Previdencia Social

17 Training officials and staff working in state social security agencies Conducting actuarial and financial analyses Provision of hardware and software There was/is no public service career for technicians working in previdencia. Pension specific courses were almost nonexistent and pension officials were mostly self-taught and migrated from other career paths. Training venues was perceived to be a major issue given anecdotal evidence of lack of preparation of pension staff in states and municipalities. While states were legally required to do actuarial analyses, there was no means of enforcement. Few had done so (e.g São Paulo, Bahia), no programs were believed to be in actuarial balance, knowledge of the importance of such work was minimal, and there was a reticence in some cases to verify the seriousness of the pension problem which was a main contributor to the states fiscal crises. *Prepared on the basis of a table provided by the TTL One outcome of the workshops was the formation of CONAPREV, an association of officials at State and Municipal levels working in previdencia similar to associations of sector secretaries at the state and municipal levels. CONAPREV is expected to grow into an important role in advancing reform and training and as a forum for debate and dissemination. These efforts have been crucial in preparing federal and state officials to participate in the reform dialogue PARSEP sponsored workshops and training seminars are enumerated above. The search for cost effective, accessible pension training programs for large numbers of staff extended from pilots of weeklong training seminars for a few officials from each state to the development of university degree programs (SC). PARSEPII & PREV-MUN will continue to look for efficient training vehicles including training for officials in state and distance learning programs possibly making use of INSS capacity. Actuarial analysis is now monitored by SPS as is that for sub-national governments pension legislative frameworks. Results of both are posted on the CRP. An actuarial model has been included in SIPREV to facilitate the work. Actuarial instruction has been provided in workshops and training venues. This has all been supported by PARSEP. Furthermore, consultant companies were contracted to work with states to produce baseline actuarial analyses and financial simulations of various alternatives for addressing fiscal imbalances. Work has been completed in 20 states with a few more pending. Little was known regarding the Small packages of less than $30,000/state were provided to hardware and software being used in all states to ensure that PROST and other tasks could be the states but with the exception of accommodated and as an institutional strengthening/ added the larger states there were questions incentive to fully participate in PARSEP. Total equipment as to whether it was adequate to costs is about R$1.2m. PARSEP is investigating equipment accommodate pension management upgrades for a number of mid-sized states who find that more needs. Research in PR found that capacity is needed but it is unclear that this can be done given the state needed more powerful impending project closure and available resources. computers to run PROST. 5. Major Factors Affecting Implementation and Outcome 5.1 Factors outside the control of government or implementing agency: Two factors were critical in affecting implementation. The first was the limited knowledge, on the part of the Bank or the borrower, as to the situation of the pension systems at the state level. Had this project only developed that information it might have been worth the money invested. It is possible, but not verifiable from the project documentation, that the changes initially envisioned were more of a policy than organizational nature. However, over the early years it became apparent that basic organizational strengthening, at both the federal and state level, would be a necessary prelude to policy change. The strides made in the former area are enormous, but in some sense not predicted given the lack of clarity as to the starting point. This is not the fault of either participant, and it is to their credit that they took up the

18 better defined challenge and pursued it as far as they did. If Brazil today is in a better position to evaluate more basic policy changes, and has already enacted a second constitutional amendment to impose more order on the system, then PARSEP (I), the rest of the donor package, and the Brazilian reformers can take the collective credit. The second factor has to do with various aspects of implementation arrangements and changes in the responsible authorities. As first designed, PARSEP was to be managed out of the Ministry of Finance, through a PCU already implementing the huge IDB PNAFE loans. It is not clear how this decision emerged and whether it was promoted by the Bank, the government, or both. These arrangements prevailed for the first years despite the fact that it was the MPAS/MPS that designed the loan and did the operational work. The results were not positive; the MPS did all the procurement, but then encountered delays in Finance, due in part to a greater interest in the IDB project. One year the PCU even budgeted incorrectly, attributing PARSEP to IDB. It cost about 8 months to straighten that out and get the money flowing again. Finally, in 2001 the MF handed the loan over to MPS, but the transfer took another 6-8 months. Once it was affected, the project implementation speeded up considerably as shown in the reports on disbursements and activities undertaken. 5.2 Factors generally subject to government control: There was a further problem posed by changes in ministerial and lower level leadership, both in Finance and in the MPS. The team in Finance was constantly changing and their attention was inevitably directed to the larger IDB project. The MPS went through at least four changes in Minister, five Executive Secretaries, and 5 PCU heads during the life of project. Other key staff remained longer, but turnover at other levels contributed to delays. 5.3 Factors generally subject to implementing agency control: These were only positive. Technical leaders who remained within the implementing agency despite the frequent changes of other personnel began with a vision of what they wanted to achieve and were able to push it forward successfully. To a certain extent, additional objectives not included in the initial project documentation were identified by, these leaders. One example is the shift from cleansing of partial samples to entire cadastres. An initial sample done in Parana (when the project was under preparation) indicated the enormous savings possible through identifying and eliminating fraudulent and erroneous payments. While it was believed this would be a strong selling point for attracting state participation as well as involving states in full purges of their registries, it was recognized that it could simultaneously increase resistance from those benefiting from the errors. 5.4 Costs and financing: These did not affect implementation. The project achieved an enormous amount for the financing available and despite the addition of new activities, was able to do this within the amounts allocated. 6. Sustainability 6.1 Rationale for sustainability rating: The changes introduced by the project are now well entrenched by constitutional and legal provisions and by new organizational structures and capabilities. Activities undertaken (e.g. the actuarial studies) have also increased understanding of the risks posed by the existing pension system and thus the need for further policy change. Some of those changes are already included in the 2003 Constitutional Amendment and others are under study. It would be difficult to imagine any reason for reversals the remaining question is simply how fast additional changes, leveraged by the project and related undertakings, will proceed. The improvements in pension management introduced under PARSEP are solid, and represent part of what the TTL calls a revolution in Brazil s approach to its pension systems. This is arguably one of the most