NICOLAU REINHARD
RONALDO ZWICKER
University of Sao Paulo, Brazil

1. Introduction--The Secretary of the Treasury

The Brazilian Federal Government has around 600,000 employees (excluding state owned companies) and is responsible for around 10% of the country's IT spending. In 1985 the country went through significant political changes: a civilian president took office under a democratic regime. At this time Brazil was renegotiating its foreign debt of US$ 80 billion, with the involvement of the IMF. Reducing government size and deficit, and controlling and giving more transparency to public spending became a political priority. For this purpose Congress created the Secretary of the Treasury (ST) in the Ministry of Economics and Finance (MEF). This new Secretary would accumulate the functions of financial management, internal and foreign debt control, internal control and auditing, and accounting for all federal government agencies. (In this text the term "Agency" is used for any organization directly controlled by the federal government.) The Minister of Economics and Finance at that time was a very prestigious and charismatic person, mainly due to the (short-lived) success of his economic stabilization program. A young, dynamic and politically well-connected economics professor became the first Secretary of the Treasury, bringing with him a number of competent and motivated executives from other government agencies to the newly formed ST.

The ST, however, had no effective control mechanisms for the prescribed functions: Ministries and other agencies had financial autonomy to execute the budget approved by Congress and could be controlled only through the aggregate cash flow, local auditing or accounting statements that would be available with a delay of 6 to 9 months, and therefore without managerial effectiveness.

The Secretary then decided to centralize the financial administration for the whole government through a centralized accounting, control and funds transfer system: Agencies would still have financial and budget autonomy, but all transactions would only be effective after being recorded in the system. This would be enforced by the fact that all payments to creditors would be made only through the system. Resources would be distributed to agencies through limited drawing rights against ST’s Central Treasury Banking Account. This system would allow the auditors to control transactions on-line, and. if necessary, stop undue payments in time. It would also allow a better cash-flow management, because of the complete knowledge of accounts payable. The system would also control Treasury-guaranteed government debt, at the federal, state and municipal levels.

The system, to be called SIAFI (an acronym for Integrated System for Financial Administration, in Portuguese), would give the ST total control over government spending, which was completely against the prevailing decentralization and liberalization tendencies of the new government. The Secretary obtained the budget to pay directly for the whole system, including equipment and installation for all other agencies. For many users this would be their first contact with a computerized on-line system. Other users would have to substitute their existing control systems for the new one. SIAFI would also require significant changes in operational procedures and strongly impact employment at the agencies, since all accounting and reporting functions would now be performed automatically by the system. It was estimated that the system would eliminate 1,000 clerical jobs, requiring also new skills for accountants, auditors and financial and planning officers.

The Data Processing Company (referred to as the Company in this case), an organization owned by the MEF, submitted a proposal to build an innovative distributed system, based on locally produced minicomputers. This system would be state-of-the-art and would take one year for project development and two more years to be implemented. The Secretary, however, considered the political timing to be unique and required the system to be operational in 7 months, that is, at the beginning of the next fiscal year. A team with members from the Secretary and the Company presented another viable alternative: a mainframe-based centralized system, using Adabas as the DBMS and a network with SNA architecture linking all users in the capital and provinces to the Company's operating centers. This proposal had many disadvantages: lower performance, high communication cost, lower availability, besides being based on an aging technology. This, however, was also its strength: since the proposed technology was already used for all other major applications in the Company, the infrastructure was already there, with trained staff at all levels. The system would run on an existing mainframe that would soon have to be upgraded.

The team had only halfhearted support from the Company's upper management, that saw the advanced solution as a unique opportunity for modernization. This conflict eventually led the Ministry to intervene later in the Company and appoint a member of the ST team to become its new president. The project was nevertheless started under the leadership of a department manager of the ST and addressed all major problems: revision of norms and regulations, system specification, infrastructure, training, user and technical support, etc. The people involved understood this project to be a unique professional opportunity and challenge; others however viewed it with disbelief.

SIAFI system became operational in 6 months (in the form of a functional prototype). Usually long processes for the acquisition of equipment (terminals, microcomputers, communications controllers, etc.) were expedited, political contacts were used to secure the large number of data lines (in permanent short supply in the country) and other infrastructure resources. The team also promoted the required changes in regulations and legislation and obtained the support from groups that would be affected (accountants and financial and planning officers) by securing them training programs and new career perspectives, besides extensive nation-wide user training. This task was made easier because the Secretary had the normative power for most changes and used his political influence to secure the others. Agencies were instructed to relocate workers that would be displaced by the system. Special attention was given to public relations to maintain a positive image of the project in all ranks of Government and isolate resisting groups. Strong support also came from the high courts that control government activities.

On January 2nd, 1987, the system was officially installed, although with many problems: many of the 1,200 terminals were not fully operational (specially in remote parts of the country), users had many questions about procedures, and the system had low performance. These problems were eventually solved and the system stabilized around 10 months later.

The total implementation cost for the first phase of the project was US$ 30 million. This investment was returned in the 4 first months of operation, due to savings in interest paid by the government, because of the centralized cash flow operation. During the period of implementation the project employed a total of 350 people (both from the ST and the Company), that was later reduced to a permanent staff of around 100. The initial software development team had 64 members, with 27 analysts and programmers remaining for maintenance. The yearly budget for the system has grown from US$ 40 to 95 million (in 1995).

Overall the system was considered to be one of the greatest successes of a public information system in Brazil. Besides the operational savings, the system gave transparency to government action that has been instrumental in improving effectiveness and responsiveness to Congress, having played a major role in some recent political crises. Both the Secretary and his principal project manager left the ST after some time to work for the private sector and as consultants to international agencies.

SIAFI has since been expanded, both in computational power and communications network. In 1992 the Social Security Administration was included in the system by presidential decree, leading to a sharp increase in the size and usage of the system. The present dimension of SIAFI is shown in Table 1.

The evolution of SIAFI usage and performance for the last 8 years is shown in Table 2.

SIAFI presently runs on an IBM 9021/962 mainframe (310 MIPS, 768 MB and 416 GB disks). The network uses around 2,000 leased data lines (not including the connected networks).

The system has been augmented with functions for planning and budgeting, managerial control and is linked to other centralized government systems (personnel, etc.). Some of these modules compete with systems from other ministries (the conflict is most evident with the budgeting system from the Ministry of Planning). The original SIAFI emphasized only the operational transactions, that had to be used by the agencies. The added management support modules, to be used on a voluntary base, have been adopted very slowly.

SIAFI has also been used as a model for several other countries. In fact Some of the initial team members are now working for international agencies to implement these systems.

Soon after the implementation of SIAFI, the same team undertook another ambitious project to centralize the control of the largest item in government spending: Personnel, responsible at that time for 80% of the budget. Public jobs are a source of political power and their irregularities are strongly visible to public opinion. The ST is responsible for controlling these expenses, although the control of personnel policies belongs to the Ministry of Administration. This ministry was operating a reporting system that did not provide timely and complete information, but had been working for some time on a new decentralized system that was not as coercive as the one proposed by the ST, although better aligned with the prevailing political orientation.

The system proposed by the ST, to be called SIAPE (for Integrated System for Personnel Administration) was designed to centralize all payroll payments and provide users with other functions for human resources administration. Users, still responsible for their payrolls, would therefore be forced to keep the personnel databases at the Secretary permanently updated. Eliminating redundant functions would also bring operational savings. The system would be a significant improvement because of the general lack of integrated human resources systems in the ministries.

The project team of this system, similar to SIAFI. And designed to share the same infrastructure, was continually in trouble. The team with members from the ST and the EDP company had difficulties in concluding the specification process because of the large number of special conditions and constant changes in norms and procedures. Nevertheless the system was implemented by decree in a significant number of ministries, but was continually plagued by poor data quality.

SIAPE was seen by users as a control system that would allow the ST to enforce Government policies related to compensation, pay increases, promotions, etc. The lack of control had until then reduced the effectiveness of these policies, leading to constant budget overruns and administrative irregularities. Many users (understandably) resisted the implementation of the system. After 2 years the project was handed over to the Ministry of Administration, where it had never been well accepted, and only after an intense political action at the President's level, in 1993, was the use of the payroll module made mandatory in all remaining government agencies. SIAPE's other modules (human resources management, etc.), have never been implemented, although the software has been operational since 1992. The Ministry of Administration considers the design of these modules to be inadequate and has planned to redesign them.

The EDP company that operates the system has had a difficult time in receiving payment for the operation of SIAPE (not solely due to budget shortages), but cannot stop providing the services because of the strategic importance of the payroll.

SIAPE, in its present state is a simple payroll system, although it can identify irregularities and stop undue payments. It has a limited use for managerial control and planning, because of its inability to support payroll projections, analysis of compensation policies, manpower and career planning, performance evaluation, etc.

Recently, some ministries, including the MEF, have developed their own human resources systems, with interfaces to SIAPE payroll. There are constant pressures to discontinue the whole SIAPE system.

A few years later, the same team of the ST started a new project in another area of strategic importance to the control of government spending: purchases of goods and services and materials administration (SIASG, for Integrated System for Management of General Services). This area, despite heavy regulations, is a constant source of problems that regularly makes the news. The system would be of great benefit to the government due to standardization, economies of scale and its usefulness in promoting industrial policies, besides the possibility of control of operational procedures. Some ministries have their own systems, which they even have shared with other agencies. The project, however, never went beyond the preproject phase in the ST and was, after some time, also handed over to the Ministry of Administration, which has the responsibility for control and normatization in this area. Although the project has not yet been implemented, there have been some advances: the development (without implementation) of a codification and a comprehensive catalog of materials and services, a centralized suppliers directory and a central exchange for materials among agencies. This system, too, is not high on the Ministry priority list, because of budget shortages and lack of human resources to manage the project.

Both projects, SIAPE and SIASG, are designed to be integrated operationally with SIAFI and would, together, constitute the backbone of the integrated control system that would help in achieving the effectiveness and efficiency of government action. The existing technical infrastructure has also bccn used for some other support systems, like a successful electronic mail system that is used for administrative transactions and fast broadcast of new norms and regulations to agencies and the public interfacing with government. ST is also considering to give the public limited access rights to aggregated data of SIAFI.

In January 1995 a new president took office and invited the former Secretary of the Treasury and his principal project manager at that time, to participate in the new government as, respectively, Vice-Minister of Planning and Vice-Minister of Finance. Balancing the budget, reducing the size of government are still top priorities to control inflation (for which the structuring systems are highly strategic instruments). The political process resulting from governing with a coalition of political parties, regional disparities and local interests add to the complexity of the problem. Possible alternatives include the unification of the Ministry of Planning Budget system with SIAFI, providing resources and support for SIAPE and SIASG, etc. The strongest constraints are not financial, but political and managerial.

For the systems themselves, there remain the challenges of technical actualization, integration with other systems, and having them become widely used planning and management control instruments. One project, for which government is seeking World Bank funding is the development of a new client/server architecture for SIAFI with a centralized server for operational transactions and clients in the ministries with copies of the relevant databases and appropriate tools for management applications, to be developed centrally or locally by individual users. In a later phase the servers for transaction processing and data storage should also be distributed. This infrastructure would also be shared by the other systems. Although SIAFI itself is beyond question, the ST has had difficulty in justifying the proposed changes and the investments in the other centralized control systems.

1. Discuss the importance of scheduling the deliverables (systems functions) to the success of a project.
2. Discuss the importance of identifying the resources required for the project (both tangibles and intangibles) and their "owners".
3. Discuss the importance of having measurable goals for a project. How should one establish the (politically) most meaningful measures?
4. Discuss the risk of using innovative technologies in large inter-organizational systems projects.
5. Are there differences in Information Systems success factors between public organizations and private companies? Are there differences in project management strategies between turbulent environments and stable environments? Between strategic applications and support functions?
6. What can be done to increase the survival rate of operational transaction-processing systems in turbulent environments? Are they different from those of decision-support systems?
7. Discuss the relationship between the reorganization of the government and supporting IT.
8. What are the political and economic implications of the success of the government changes supported by IT?
9. Read some material about the economic, political, and cultural environment in Brazil. Then point to issues relevant to this case.
10. Identify issues in this case that are similar to those of TradeNet.