World Bank Document

FILE COPY Report No. 1267a-BR Economic Memorandum on Brazil (In Two Volumes) Volume 11: Annexes Public Disclosure Authorized

January 26, 1977 Latin America and the Caribbean Regional Office

FOR OFFICIAL USE ONLY Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Document of the World Bank

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 Worldl Bank authorization. CURRENCY EQUIVALENTS

Currency Unit: Cruzeiro

Exchange Rates Effective January 18, 1977

Selling Rate: US$1.00 = Cr$12.520 US$1 million - Cr$12,520,000 Cr$1 million = US$79,872

Buying Rate: US$1.00 = Cr$12.520

Average Exchange Rates (Selling)

1974 1975

US$1.00 Cr$6.789 Cr$8.128 US$1 m'!llL)n Cr$6,789,000 Cr$8,128,000 Cr$' allton US$147,297 US$123,031

GLOSSARY OF ACRONMIS

BEFIEX - Comissao Para Concessao de Beneficios Fiscais a Programas Especiais de Esportacao (Special Export Program)

BNDE - Banco Nacional do Desenvolvimento Economico (National Economic Development Bank)

CACEX - Carteira de Comercio Exterior (Foreign Trade Division, Bank of Brazil)

CDI - Conselho de Desenvolvimento Industrial (Industrial Development Council)

DNER - Departamento Nacional de Estradas de Rodagem (National Highway Department)

EMBRAMEC - Empresa Brasileira de Industria Mecanica (Brazilian Mechanics Enterprise)

FINAME - Agencia Especial de Financiamento Industrial (Special Agency for Industrial Financing)

FINEP - Financiadora de Estudos e Projetos (Agency for Funding Studies and Projects)

FINEX - Fundo de Financiamento de Exportacao (Export Financing Fund)

IBC - Instituto Brasileiro do Cafe (Brazilian Coffee Institute)

IPI - Imposto de Produtos Industrializados (Industrial Products Tax)

LAFTA - Latin American Free Trade Area

LIBOR - London Interbank Offer Rate

NAI - Nucleos de Articulao Industrial (Group responsible for planning and execution of federal policy to buy Brazilian machinery and equipment)

OECD - Organization for Economic Cooperation and Development

PASEP - Programa de Formacao do Patrimonio do Servidor Publico (Civil Service Compulsory Savings Program)

PETROBRAS - Petroleos Brasileiros (Brazilian Petroleum Enterprise)

PIS - Programa de Integracao Social (Social Integration Program)

PND II - Plano Nacional de Desenvolvimento (Second National Development Plan)

FEDERATIVE REPUBLIC OF BRAZIL FISCAL YEAR

January 1 to December 31 FOR OFFICIAL USE ONLY

ANNEX I

RECENT TRENDS IN EARNINGS AND THE DISTRIBUTION OF

INCOME IN BRAZIL

This document has a restrictsd distributfio snd may be used by recipsnts only in the performance oJ their ofcial duties. Its coiatents mAy Sot otherwise be disclosed without World Bank authorizaton.

ANNEX I: RECENT TRENDS IN EARNINGS AND INCOME DISTRIBUTION IN BRAZIL Page No. SUMMARY AND CONCLUSIONS ...... i I. TRENDS IN EARNINGS ...... 1 A. Average Earnings in Manufacturing Industry ...... 1 B. Urban Minimum Wage Movements ...... 6 C. Collective Wage Agreements ...... 8 II. EMPLOYMENT GROWTH ...... 14 III. TREND IN INCOME DISTRIBUTION ...... 16 IV. PIS (SOCIAL INTEGRATION PROGRAM) AND THE FOURTEENTH SALARY 23 V. THE REDUNDANCY PAYMENTS FUND (FUNDO DE GARANTIA DO TEMPO DE SERVICO) AND T'HE NATIONAL HOUSING BANK (BANCO NACIONAL PA HABITACAO) ...... 26

SUMMARY AND CONCLUSIONS

1. This discussion is concerned with the recent behavior of the distribution of personal income in Brazil. The data on recent trends in average and median earnings in manufacturing, comuerce and services are reviewed; the empirical evidence concerning the recent behavior of the urban minimum wage, the wages policy executed through "collective bargaining" agreements, trends in occupational wage differentials and the recent behavior of agricultural real wage rates are examined, The report then analyzes employment growth which is directly concerned with distributional issues. Recent trends in the personal distribution of earnings and the likely effects of recently announced changes in wages policy on the distribution of income are then Pxamined. Finally, the effects of the redundancy (unemployment) payments scheme (FGTS) and of the Social Integration Program (PIS) on the distribution of income are considered. A discussion of social welfare programs in Brazil may be found in Annex II.

Recent Trends in Earning2s 2. Trends in average earnings of employees (white and blue collar) in manufacturing indusl:ry can be seen in an index constructed from data in the annual industrial surveys undertaken by the 3razilian Institute of Geography aad Statistics (IBGE). Caution must be used in interpreting these

3. While there is :Little doubt that average wages in the formal sector have increased, changes in average earnings are not necessarily indicative of the behavior of real earnings for the mass of employees if inequality in the d:Lstribution of earnings is increasing. In such a case, it is more appropriate to consider the behavior of the median when examining trends in real earnings of the lower half of the earnings distribution. Median real earnings increased by 8.3% between 1966/67 and 1972/73, a compound amnual rate of 1.1%, and the ratio of mean to median earnings increased by 25%, showing that inequality indeed increased during the period according to these data. Furthermore, in commerce and services median real earnings were consistently lower throughout the years 1967 to 1973 than they were in 1965. There was also an increase in inequality in these sectors. There may well have been some improvement in the equality cf earnings in 1975 and 1976, but comparable data are not yet available for these years. - ii -

4. The available data, which are probably representative only of employment in the formal sector of industry, commerce and services, indicate rising real wages over 1967-73 but increasing inequality. The data do not permit analysis of the behavior of real incomes of employees in the informal sector who fall in the lower deciles of the urban distributuion.

Urban Minimum Wage Movements

5. Inferences concerning trends in real earnings of workers at the lower end of the urban income distribution are often drawn from movements in the real minimum wage. Data on trends in the real minimum wage in Guanabara (Rio) over the period 1959-1975 shows that a peak was reached in 1961 with declines every year thereafter except 1971, 1972, and 1975. The increase in 1975, following the Government's decision in November 1974 to maintain the real value of the minimum wage and even to increase it, was the largest real increase granted since February 1964. The nominal increase of 44% granted in May 1976 also implied some real growth in the minimum wage, given that the rate of inflation (April 1975 - April 1976) was 41%.

6. The crucial issue here is the extent to which the official minimum wage represents the full-time earnings of workers at the lower end of the income distriubution. That is, what proportion of the active population receives the minimum wage or earnings which are determined by reference to movements in this standard? It is extremely difficult to test this by using the evidence available on earnings. The urban minimum wage may be a poor guide to the growth of real earnings because of the possible existence of intra-enterprise wage scales (internal markets), occupational earnings ladders, and life cycle earnings profiles. At this point it can only be said that the evidence is inadequate to support from eonclusions on the relationship between the behavior of the real minimum wage and the trend ia real earnings of a substantial proportion of the labor force.

Collective Wage Agreements

7. In general, wage increases for semi-skilled and unskilled workers earning above the minimum are determined as a result of "collective bargaining" agreements (dissidios coletivos). These wage agreements which comprise official arbitration awards, cover broad groups of workers in a particular sector, for example, banking or metalworking industries. The Government's decreed wage increases for groups of unskilled and semiskilled workers in Sao Paulo have lagged behind increases in the cost of living. The apparent real decline in arbitrated wage rates should be interpreted with great caution. The impact of these wage adjustments on differentials through time is not known, nor is the extent to which these average increases are representative. During a period of rapidly growing employment, there is likely to be a substantial degree of earnings "drift" due to the divergence between - iii -

industry wide "agreed" wage rates and actual earnings arising from overtime work and shortages of employees with particular skills. Unfortunately, because of the dearth of data on earnings for specific skill grades, it is difficult to draw conclusions about recent trends in real earnings and earnings differentials for workers earning above the minimum wage.

8. Some partial data exist on trends in real earnings by skill category. Data on Sao Paulo industrial enterprises collected by a consultancy firm fromt 1967-75 clearly demonstrate that all occupational groups experienced a rise in real earnings during this period; however, this was accompanied by a tendency towards increased dispersion. The IBGE data on hourly wage rates in the construction industry in Sao Paulo, demonstrate that real. hourly wage rates have been increasing faster than the urban minimum and that the most and the least skilled construction workers have benefitted relatively.

9. Trends in wage rates for the agricultural sector can be examined on the basis of'data published by the FGV-Centro de Estudos Agricolas (for all states except Sao Paulo) and the Secretaria da Agricultura de Sao Paulo. Between 1966 and 1970, the ratio of the rural wage to the urban minimum wage remained constant. However, from 1970 onwards, agricultural wages rose relative to the urban minimum, particularly between 1973 and 1975. A possible exp.lanation is that the agricultural sector, particularly in Southern Brazil, has been undergoing a major transformation in terms of techniques of production, crop mix, and social relations of production. There has been a large increase in acreage, which has been acconpanied by the growth of new crops (soybeans) and the revival of more traditional commodities (coffee in Minas Gerais). In addition, especially in Sao Paulo, dependent types of landlord-tenant relationship are disappearing and the rural poor are being transformed into wage laborers. Meanwhile, the exodus to the cities continues with the effect that between 1960 and 1970, for example, the agricultural labor force in Brazil grew by a mere 0.7% a year. The mobility of the work force and its slow growth in relaticn to the demand for labor have served to push up wage rates, especially after 1973.

Employment Growth

10. The existing data on employment growth, must be used circumspectly, largely because of lack of year-to-year comparability. In manufacturing industry the rate of growth of enterprise employment lagged behind the growth of total sector employment, implying that there was both an absolute and a relative rise in "informal" sector employment during the 1960s. On the other hand, in commerce and in services, enterprise employment grew faster than employment for these sectors as a whole, indicating that "informal" sector employment declined as a proportion of the total. In commerce, there was even a small decline in the absolute level of "informal" sector employment, though this was not true of services. In the secondary sector as a whole, employment grew at a annual compound rate of 11.6% per year, with manufacturing growing by 9.0% and construction by -iv-

17.4% a year; in commerce and services the annual compound rate of growth was 15.0% a year. These results may well represent an upper bound estimate of employment growth since the mid-60s. Virtually nothing is known about trends In earnings and employment in the "informal" sector which absorbs approximately 50% of the urban labor force nor about inter-sectoral patterns of mobility.

Trends in the Distribution of Income

11. Few of the participants in the recent debate on Brazilian income distribution have fully recognized the extent of the serious empirical deficiencies of the data upon which most arguments have been based. Leaving such issues aside, the Gini coefficient measures of dispersion, calculated using the Ministry of Labor data, show that, from 1969 to 1973, there was a continuing trend towards increasing inequality in both the industrial sector and in commerce and services. Further difficulties arise concerning the measurement of real income treads in the bottom 40% of the income distribution. This segment contains more or less the bottom 60%OW of the rural income distribution and the bottom 20% of the urban distribution. The available evidence, however, suggests that both urban and rural poor have benefitted in absolute terms from the recent rapid growth in employment.

12. Since Novembbr 1974, the Geisel Government has introduced important modifications in the formula used to compute arbitrated wage agreements. This can be seen as part of the Government's declared policy of reversing the unfavorable trends in income distribution. The minimum wage for the Center-South decreed on May 1, 1975 was a 41.4% nominal increase over the May 1974 value. Taking into account the 10% bonus given in December 1974 (which took the form of an advance on the May 1975 rise), the real minimum in 1975 stood 6.5% higher than in 1974 (a year in which there was a very sharp decline of 8.1%). The nominal increase of 44% granted in May 1976 also implies a rise in real terms of approximately 3%.

13. It is not yet clear what the effect of continuing rises in the real minimum wages will be on the personal functional distribution of income. It appears that one objective of government policy is to reduce income differentials among wage earnings by increasing minimum wages more rapidly than other wage rates. There is reason to believe that it will not be easy to compress earnings differentials. Also, because of the diversity of work relations in the informal sector, it is not clear how the more generous minimum wage policy will affect earnings of the 5.5 million employees and the 4 mil-licn self-employed in the "informal" urban sector who are beyone the reach of compulsory minimum wage legislation. This annex deals only with wage and unemployment benefits policies, though government policies such as the POLONORDESTX program are designed to have a positive impact on the incomes of small farmers and sharecroppers in the Northeast, Brazil's poorest region. - v-

PIS (Programa de Integracao Social)

14. PIS (Social Integration Program) was designed as a mechanism for supplementing the monetary income of employees in the enterprise or "formal" sector of the urban economy. A similar fund exists for public sector employees known as PASEP, and on July 1, 1976, they were unified as PIS/PASEP. The receipts of PIS come from enterprise contributions of two kinds: the first Is a fraction of the value of total enterprise sales (0.5% since 1974) and the second is a transfer of 5% of corporate income_tax revenues. _PASEP receives_2% of the_receipts accruing to each level of governmem.t. The Fund distributes these receipts (cotas) annually to employees In the form of individual shareholdings: half of the available resources are distributed according to the employee's age and half on a strictly earnings-related basis. These shares can only be cashed in under specific circumstances such as marriage, death (when the money accrues to dependents), retirement or house purchase. The resources of PIS/PASEP are administered by the BNDE and are mostly invested in the private sector, as well as profits, if any.

15. In June 1975, the average value of an accumulated cota in PIS was Cr$1,080 (almost trice the value of the highest regional minimum wage) and Cr$2,440 in PAASEP. The yield, which is made up of monetary correction, interest arLd profits on the accumulated shareholding, will vary a great deal from year to year -- at least in current prices - but in 1975, the average yield was about 1/5 of the current monthly minimum wage. In 1974 the total number of individual employee accounts in PIS was 10.8 million and there were 3.4 million employees in PASEP. While in absolute amounts, the yields to whi,ch shareholders have access have been small, the main beneficiaries are the urban working class in the enterprise of "formal" sector. This is precisely the group which lost most from the process of income redistribution which occurred between 1960 and 1970.

16. A complete assessment of the redistributive impact of the Social Integration Program must consider possible alternative uses of the PIS/PASEP funds. About 75% of PCS receipts come from the levy on sales, but this has been compensated by reductions in the state value added tax. The rest of the resources find theLr counterpart in lower effective federal corporate taxation. In this sense, PIS/PASEP receipts can be viewed as a transfer of resources from the Government budget to an investment fund which, because imediate access to the benefits on the part of employees has been somewhat limited and because many of the lowest income groups have not been included, has mainly benefitted private sector enterprises. An alternative use of the resources with a greater redistributive impact might, for example, have been to increase the level of public expenditure on transfer payments or public consumption goods destined for the poor. Recent legislation and relaxation of the criteria governing withdrawals by low-income account holders can, however, be expected to have beneficial effects on the distribution of income within the "formal" settor. - vi -

The Redundancy Payments Fund (Fundo de Garantia do Tempo de Servico) and the National Housing Bank (Banco Nacional de Eabitacao)

17. FGTS, introduced in 1966, replaced a previous redundancy payments scheme which had several disadvantages so far as both employers and employees were concerned. Redundancy payments are especially important in Brazil because of the complete absence of an unemployment benefits system. Under FGTS, employers credit each employee's account every =cnth with a sum equal to 8% of the employee's salary. Currently, the accounts are credited annually with monetary corrections, as well as a 3% real rate of interest. Employees can withdraw the entire sum in their account upon being made redundant. Otherwise, the sum in an employee's FGTS account can only be withdrawn under special circumstances, such as retirement, house purchase and improvement, the purchase of a small business or grave sickness and other family emergencies.

13. At present, there are between seven and eight million FGTS accounts, which implies that between 40 and 50% of the employed urban labor force has a claim to redundancy payments; the self-employed and employers are, of course, automatically excluded. In general, as with the PIS scheme, FGTS only covers employees in the formal or corporate sector, thus excluding a very sizeable proportion of wage employees in urban areas whose employers do not contribute to the FGTS.

19. A further question concerns the rate of return on the Fund and the uses which are made of the very substantial financial resources at its disposal - a total of Cr$26,465 million (US$3 billion) at the end of 1975. The agency which administers the Fund is the Banco Nacional de Habitacao (BNH); the Bank's major aims since its inception in 1964 have been to provide finance for house purchase, particularly house purchase among low-income recipients, and to help finance water supply and sewerage investments. At the same time, the Bank is expected to administer housing finance programs in order to provide an acceptable return on the accounts which employees hold in the FGTS. There is a fundamental contradiction between these two requirements since the more ambitious the attempt to provide traditional forms of low-income housing, the greater the subsidy involved and the lower the return to accounts in FGTS. In attempting to lower the average cost of its housing loans, BNH has reduced the return paid to FGTS.

20. In order to make a complete assessment of FGTS, particularly its impact on the distribution of income , it is necessary to examine the activities of the National Housing Bank (BNH) whose main source of deposits comes from FGTS. BNH's general purpose has been to coordinate programs designed to reduce Brazil's substantial housing deficit and to keep the urban housing stock increasing in line with urban population growth. Since about 70% of urban households receive incomes less than three minimum wages, BNH has had special responsibilities for stimulating the construction of housing for owner-occupation by low-income-recipients. Up to now, there has been little public sector activity in the aran of rental accomodation in Brazil. Surveys reveal thkt owner-occupation is firmly established as most Brazilians' preferred housing situation. - vii -

21. The major way in which BNH attempted to increase the supply of housing between 1964 and 1974 was by financing the operations of the State Housing Authorit:ies (COHABs). The COHABs present plans to BNH of public housing proj;ects (conjuntos habitacionais) which are intended for purchase by low-income households. BNH then provides COHABs with finance for construction and the COHABs make mortgage agreements with the individual home purchaser. Poor performance in the area of low-income housing led the Bank to a reappraisal of its activities, resulting in the announcement of the PLANHAP program in 1973. The aims of the new program are twofold: to reduce the cost of conventional housing loans and to experiment with financing new areas of activity. PLANHAP's t:arget is 2 million new and improved dwellings by 1983. The specific measures adopted include increasing the lower and upper limits included in the low income housing scheme, lowering the rates of interest paid by low-income final borrowers, and establishing a new urban sites and services program to encourage self-help housing construction (PROFILURB).

22. The quantitative results of the low-income housing program up to 1975 are disappoin:ing. During a ten-year Deriod, BNI1 financed, wholly or partially, a total of 1,142,265 housing units, of which only 252,160 units were for households earning up to three minimum wages. With Brazil's urban population gaining 2.7 million inhabitants a year, BNH's housing program for low-income families has hardly scratched the surface of the housing problem.

23. BNH has also diversified its loan activities into the area of urban services and infrastructure: the main beneficiaries of these programs are state and municipal authorities responsible for water, sanitation, transportation and urban renewal. In 1971, the Government embarked on a new program, PLANASA, with the objective to serve 80% of the urban population living in 4/5 of Brazil's urban communities with adequate water supply by 1980 and to furnish satisfactory sewerage services to residents of the main urban centers. Up to 1975, PLANASAS's-progress was disappointingly slow and it developed some serious problems. However, recent revisions should enable PLANASA to meet its objectives readily.

24. In order to assess the redistributive effect of the combined FGTS- BNH program, the following points need to be considered:

(a) FGTS provides a strictly earnings-related pension or redundancy payment; as such, it does little to modify the distribution of income, except as between the corporate and the "'informal" sector, to the benefit of the corporate sector workers as a whole;

(b) the incidience of the employer's contribution needs to be considered: would wages be proportionately higher in the absence of FGTS, or are the employer contributions being financed through higher prices? If the latter, then taking the distribution of both costs and benefits, FGTS may actually be harmfua to the distribution of income; - viii -

(c) considering the distribution by size class of income of employees' quotas in FGTS and the distribution of housing loans by income class, low-income households (which have the highest ratio between total FGTS quotas and total housing loans) have been making substantial transfers to middle and upper income households (which have the lowest). This transfer increases with (i) any decreases in the return on FGTS quotas compared with an alternative rate of return, (ii) any increase in the differential between the cost of BNR house finance loans compared with the cost of finance from conventional sources, and (iii) increases in-the divergence between the share of lower income groups in the fund and their share in housing finance. Either owner- occupation or a public sector rental program will have to be subsidezed by the Treasury in a much bigger way or more flexible design and construction systems and standars adopted if greater progress is to be made in resolving low-income housing problems without seriously decapitalizing the Redundancy Payments Fund. I. TRENDS IN EARNINGS A. Average Earnings in Manufacturing Industry

-i.. Trends in avera,ge earnings of employees (white and blue collar) it manufacturing indu,stry can be seen in an index constructed from data in the annual industrial surveys undertaken by the Brazilian Institute of Geography and Statistics (IBGE). Caution must be used in interpreting these data becunse the surveys do not cover all manufacturing establishments. For example,those undertaken between 1962 and 1971 probably cover firms producing 90% of total manufacturing output, whereas the monthly surveys used as the source for 1971-75 cover only the thousand largest firms. In addition each source is not strictly consistent in terms of its coverage from year to year. It seems likely that since the surveys tand to cover only larger enterprises, the estimates probably give an upper-bound estimate of the rate of growth of earnings.

2. As seen in T'able 1, the data indicate that real average earnings in manufacturing increased by 27.1% between 1964 and 1975. There was a decline in average real earnings between 1964 and 1967, but, thereafter, there were positive real increases in every year up to 1975, except for 1974, when the rate c,f inflation accelerated due to the rise in the price of oil and price adjustments following a year of repressed inflation (1973). The real increase during 1975, when the aim of government policy was to raise t;he real minimum wage, was almost the largest annual increase during the period under consideration.

3. Another source of data on the trend in average manufacturing earnings is the annual survey of earnings in urhan activities undertaken by the Ministry of Labor under the "Law of Two-Thirds" which requires firms to employ a work force in which t-de-thirds of the employees are Brazilian nationals. As with IBGE data, discussed above, the survey is undertaken at the enterprise level. The Law of Two-Thirds is supposed to cover all employment protected bv the official labor legislation which grants workers certain rights such as minimum wage, paid holidays, an eight-hour day, protection against unfair dismissal and the right to redundancy payments. However, there is some evasion of this legislation-and many employees in small enterprises are not covered-.tji means that they are, in effect, in the "informal sector". Comparis^on with census figures indicates that the Ministry of Labor survey includes about 85% of all employees covered by official labor legislation and about 51% of all urban employees -2-

Table 1: AVERAGE EARNINGS IN MANUFACTURING INDUSTRY, 1959-75

Index of Average Index of Average Z Change in Annual Earnings Earnings Deflated Real Average of All Employees by Guanabara Retail Earnings in Manufacturing Price Index (1959 - 100) (1959 - 100) (1) 1/ (2) (3)

1959 100 100 - 1960 - _ 1961 1962 303.3 116.2 - 1963 583.2 130.8 +12.6 1964 1,081.3 126.7 -3.1 1965 1,637.2 115.7 -8.7 1966 2,195.5 109.8 -5.1 1967 2,779.4 106.5 -3.0 1968 3,573.9 111.9 +5.1 1969 4,824.6 123.9 +10.7 1970 6,070.0 127.0 +2.5 1971 7,808.5 136.0 +7.1 1972 9,871.6 147.6 +8.5 1973 12,238.4 149.4 2/ +1.2 1974 15,955.3 146.9 -1.7 1975 22,580.0 161.1 +9.7

1/ 1959: Censo Industrial; 1962: Registro Industrial; 1963-67: Ingueritos Economicos; 1968-71: Pesquisa Trimestral; 1971-75: Pesquisa Mensal. The index was constructed by dividing the annual salary bill by year-end employment between 1959 and 1964, and by dividing by average annual employment between 1964 and 1971.

2/ A figure of 22.5% for the rate of inflation in 1973 has been used instead of the official figure of 12.6%. - 3-

- 4. The Ministry of Labor surveys indicate that average reaJ. earnings in marmfacturing industry increased by 34.0% between 1966 and 1973. This estimate is quite close to that of the IBGE industrial surveys. The increELse in average real earnings in commerce and services over the same period (+ 25.7 is lower than for manufacturing. According to these data, real manufacturing earnings fell in only one year (1971) and in commerce and services in two years (1969 and 1971).

5. While there is little doubt that average wages in the formal sector have increased, during a period in which inequality in the distribution of earnings increases, changes in average earnings are not necessarily indi.cative of the behavior of real earnings for the mass of employees. Trends in real earnings for the lower half of the earnings distribution can be seen better from the behavior of the median. MXedian real earnings increased by 8.3% between 1966/67 and 1972/73, a compound annual raLte of 1.1%, and the ratio of mean to median earnings increased by 25% showing that inequality increased during the period. Furthermore, in comierce and services median real earnings were consistently lower throughout the years 1967 to 1973 than they were in 1965. There was alSo an increase in inequality in this sector.

i.-Cj In summary, available data which are probably representative onTy of employment i.n the formal sector of industry, commerce and services indicate rising reaL.wages over 1967-73 but increasing inequaality. The data do not permit analysis of the behavior of real incomes of employees in the informal sector who fall in the lower decile of the urban distribution.,or possible improvements in income distribution in 1975 and 1976. Table 2: AVERAGE AND MEDIAN EARNINGS IN MANUFACTURING INDUSTRY 1966-73

Average Index of % Change Median Index of % Change Index of Monthly Average in Real Earnings Median in Real Ratio of Earnings Real Earnings (Current Real Earnings Average to (Current Earnings Cr$) Earnings Median Cr$) (1966 = 100) (1966 = 100) Earnings

1966 130.8 00- 95.4 100 - 100 1967 176.9 103.6 +3.0 122.2 98.1 -1.9 105.6 1968 235.0 112.5 +8.6 159.7 104.9 +6.9 107.4 1969 287.6 112.9 +0.4 190.0 102.3 -2.5 110.4 1970 360.05 115.2 +2.0 238.7 104.7 +2.3 110.1 1971 429.2 114.3 -0.8 278.4 101.6 -3.0 112.5 1972 546.3 124.9 +9.3 340.2 106.7 +5.0 117.1 1973 718.2 134.0 +7.3 421.1 107.8 +1.0 124.4 1966/67- 72/73 1/ +27.2 +8.3

1/ We have chosen to average over the years at the beginning and end of the period because any single year's observations may be subject to considerable error.

Source: Boletim do SEPT, various years. Median estimated by simple linear interpolation between income class limits. Price deflator used was Guanabara retail price index. Table 3: AVERAGE AND MEDIAN EARNINGS IN COMMERCE AND SERVICES 1965-73

Average Index of % Change in Median Index of % Change in Index of Monthly Real Real Earnings Montlhly Real Real Ratio of Earnings Average (Deflated by Earnings Median Median Average to (CurreluL Earfulags Guanatara ReLail (CurtenL Ear-iTLhgs Eaini'L...gs'.edian Earn.LI'gs Cr$) (1965 = 100) Price Index) Cr$) (1965 = 100) (1965 = 100)

1965 117.4 100 - 80 100 - 100 1966 154.9/156.8 93.4 -6.6 100.0/107.1 88.5 -11.5 105 1967 212.5 97.0 +3.9 140.0 88.6 +0.1 109 1968 285.0 106.3 +9.6 183.5 95.0 +7.2 111 1969 347.2 106.1 -0.2 219.5 93.1 -2.0 113 1970 434.2 108.2 +2.0 265.1 91.7 -1.5 117 1971 517.4 107.3 -0.8 299.0 86.0 -6.2 123 1972 656.0 116.9 +9.0 373.2 92.2 +7.2 125 1973 807.5 117.4 +0.4 442.6 89.3 -3.1 129

Source: Boletim do SEPT, various years - 6-

B. Urban Minimum Wage Movements

7. Inferences concerning trends in the real earnings of workers at the lower end of the urban income diatribution are often drawn from movements in the real minimum wage. Table 4 contains data on trends in the real minimum wage in Guanabara (Rio ) from 1959 to 1975. The data indicate that the real minimum wage had been falling prior to the 1964 revolution, having peaked in 1961. From 1964 onwards the real value declined in every year except 1971, 1972 and 1975. The increase in 1975, following the Government's decision in November 1974 to maintain the real value of the minimum wage and even to increase it, was the largest real increase granted since February 1964. The nominal increase of 44% granted in May 1976 also implied some real growth in the minimum wage, given that the rate of inflation (April 1975-April 1976) was 41%. Despite the high rate of inflation of the first quarter of 1976, the decision to increase the real minimum wage by approximately 3% is consistent with this more liberal wages policy.

8. The crucial issue here is the extent to which the official. minimum wage represents the full-time earnings of workers at the lower end of the income_distribution. That is, what proportion of the active population receives the minimum wage orearnings which_arecIetermin-d by reference to movements in this standard? It is extremely difficult to test this using the evidence available on earnings. The source usually referred to in this context, the Ministry of Labor Law of Two- Thirds statistics, is unsatisfactory for this purpose because all workers included in this survey should, by law, be earning at a full-time rate which is no less than the minimum. Thus, the fact that there appear to be workers covered by the Law of Two-Thirds who earn less than the minimum is explained by other factors: (a) part-time employment; (b) sickness and absenteeism; (c) employees who change Jobs during the month of the survey and (d) the fact that some employees may work for several enterprises during a month may be included by each enterprise employing them at a wage which might well be less than the minimum.l/ National Household Survey (PNAD) data, the only alternative source for examining the level in earnings of the lower half of the income distribution also have serious deficiencies.

1/ To a certain extent, the estimates of median earnings above are open to similar objections; what is really required is an estimate of median full-time earnings. Table 4: TRENDS IN THEI REAL VALUE OF THE MINIMUM WAGE IN GCANABARA

Nominal Value/i Guanabara Index of the Z Change in the of the Minimum Retail Price Real Value Real Value of Wage (Cr$) 1ndex of the Mini- the Minimum :1965/67-100 mum Wage Wage (1959-100)

1959 6.00 4.98 4.98 100 1960 6.75 6.44 87.0 -13.0 1961 10.40 8.58 100.6 +15.6 1962 13.44 13.0 85.8 -14.7 1963 22.75 22.2 85.1 -0.8 1964 41.71 42.5 81.5 -4.2 1965 67.17 70.5 79.1 -2.9 1966 87.75 99.6 73.1 -7.6 1967 109.96 130 70.2 -4.0 1968 133.79 159 69.8 -0.6 1969 159.47 194 68.2 -2.3 1970 191.53 238 66.8 -2.0 1971 230.53 286 66.9 +0.2 1972 275.60 333 68.7 +2.7 1973 322.40 402/2 65.6 -4.5 1974 393.00 541 60.3 -8.1 1975 540.27 698 64.2 +6.5 1976 835.85 n.a. n.a. n.a.

/1 The minimum wage adjustments frequently occur in the middle of the year; each figure is, l:herefore, the weighted average for the whole year of ruling monthly figures. After 1963, data include effect of 13th salary (paid in December).

/2 We have used a figure Jior the rate of inflation in 1973 of 22.5%, which government spoketimen now admit to be closer to the true -clue than the official figure of 12.6%.

Source: Conjuntura Economica, various issues. - 8 -

9. The urban minimum wage may be a poor guide to the growth of real earnings because of the possible existence of intra-enterprise wage scales (internal markets), occupational earnings ladders, and life cycle earnings profiles. The minimum wage, for example, might be regarded as a starting salary. Thus, it is considered unlikely that someone would remain at the minimum for more than a few years. Either the worker would be promoted within the enterprise or change jobs, thus effectively ensuring that his real earnings actually rise. There are frequent complaints in Brazil that the rate of labor turnover is high, which could mean that employees are making use of inter- firm mobility to improve their earnings. While a comprehensive analysis of labor turnover remains to be undertaken, available research results tend to support this. They indicate that turnover is higher in low skill, low income jobs and in small firms. However, before a definitive judgment can be reached on the trend in real earnings of low income workers more empirical evidence is needed on: (a) the functioning of enterprise or occupational earnings ladders which, at first sight, appear to favor higher skills to a disproportionate extent; ( b) the nature, causes and effects of high rates of labor turnover and (c) life-cycle earnings profiles. In this respect, it would be useful if the IBGE were to initiate a re-interview sample survey on incomes as part of its census enquiries. At this point,however, it can only be said that the evidence is inadequate to allow reaching conclusions on the relationship between the behavior of the real minimum wage and the trend of real earnings of a substantial proportion of the labor force.

C. Collective Wage Agreements

10. In general, wage increases for semi-skilled and unskilled workers earning above the minimum are determined as a result of "collective bargaining" agreements or dissidios coletivos. These are agreements which cover broad groups of workers in a particular sector, for example, banking or metalworking industries. As seen in Table 5 the Government's decreed wage increases for groups of unskilled and semiskilled workers in Sao Paulo have lagged behind increas,s in the cost of living. These data were prepared by a trade union research organization in Sao Paulo - DIESSE (Departamento Inter-Sindical de Estatistica e Estudos Socio-Economicos) and the cost of living index used differs somewhat (although not substantially) from the FGV index. Free collective bargaining in Brazil has been suspended since 1965/66 and most negotiations go to arbitration, where employers and employees are bound to accept the increases decreed by government wages policy. One problem .ith the aggregate figures agreed in such collective bargains is tha, there is no indication of, nor any data available on,how these increases affect earningsof employees of different skill levels. In addition with the suspension of full collective bargaining, labor unions have had no power to operate on the wage front and have been forced to concentrate on a variety of non- wage demands. There has been some progress in increasing non-wage benefits which is not captured in these statistics. -9-

11. The cause of the apparent decline in real wages conceded in collective bargaining agreements stemsfron the formula used bv the M.inistry Labor for calculating the nominal increases awarued by arbitration tribuna'ls. Before November 1974, the index was composed of two main elements: the first (or base) component consisted of a factor to restore (using a set of indices de reconstituic.o salarial) the average real wage to the level prevailing during the 24 months prior to the new agreement. This factor was augmented by a second element consisting of an estimate of one half of the expected rate of inflation during the coming 12 months plus an estimate of productivity growth. This method contributed to a decline in arbitrated wage rates because earnings were averaged over 24 months rather than taking the previous peak in real earnings. Had real wages been moving along a rising trend, averaging over 24 months, as compared with, say, 12, would also tend to reduce the adjustment. Furthermore,, the cost of living indices were lower, in most years, than either the FGV or the DIESSE cost-of-living indices. In addition, the expected rate of inflation used in the calculation diverged from the actual rate by a considerable margin prior to 1968, partly because of overoptimistic expectations concerning the results of counter-inflationary policies,and such underestimations were not corrected ex post. After 1968, estimates of the expected rate of inflation were much closer to the actual, and an ex post "'correction factor" was introduced into the formula- Finally, the estimate of productivity growth was far below the growth of per capita GDP in most years and also the productivity element was simply summed to the other components. rather than multiplied. It should be noted that alterations in wage policy announced by the Geisel Government in November 1974 have corrected the most negative features of the wage formula.

12. The apparent decline in arbitrated wage rates should be interpreted with great caution. The impact of these wage adjustments on differentials through time is not known nor is it known to what extent these average increases are representative. During a period of rapidly growing employment, there is likely to be a substantial degree of earnings "drift" due to the divergence between industry wide "agreed" wage rates and actual earnings. Overtime work and shortages of employees with particular skills will be the cause of such "drift". Since wage agreements are made individually between employer and employee, there is likely to be considerable scope for the operation of such factors during a period of rapid employment growth, when labor's bargaining position is strengthened somewhat. 'Highly skilled engineering workers, foremen, construction workers and even the least skilled construction workers are some of the groups of workers which appear to have benefitted from specific labor shortages, while the widespread existence of overtime work during the period of rapid growth boosted the pay of all workers, including those paid at the normal time rate of the minimum wage. Unfortunately, because oE the dearth of data on earnings for specific skill grades, it is difficult to draw conclusions about recent trends in real earnings and earnings differentials for workers earning above the minimum wage. -10 -

13. Some partial data exist on trends in real earnings by skill category. One set was collected by the consultancy firm of Morris and Morgan in Sao Paulo industrial enterprises during the period 1967-75. The data clearly demonstrate that all occupational groups experienced a rise in real earnings during this period; however, this was accompanied by a tendency towards increased dispersion. The trend increases in real earnings were as follows: (i) for employees earning between 1-4 minimum wages: 2.16% a year; (ii) for employees earning between 4 and 9 minimum wages: 4.40%; (iii) for employees earning between 9 and 20: 5.11%; (iv) for employees earning between 20 and 70: 7.24% and (v) for employees earning more than 70: 8.89% a year. An additional survey undertaken by Morris and Morgan confirms the tendency towards increasing inequality among executives' earnings during the period 1969-75.

14. The data on hourly wage rates in the construction industry in Sgc Paulo, published by IBGE, demonstrate that real hourly wage rates have been increasing faster than the urban minimum and that the most and the least skilled construction workers have benefitted relatively. Taking 1969=100, the Sao Paulo minimum stood at 105.8 in 1975. The index of real hourly wage rates for various categories of construction workers behaved in the following way: carpenter: 109.3; site foreman: 231.5; bricklayer: 105.4; painter: 118.8: laborer: 132.8 (the least skilled grade). These TBGE statistics on hourly wage rates are the only official series on occupational wage rates, but their main drawback is that they do not measure trends in occupational earnings. The collection and publication of more extensive data on trends in occupational earnings would be of immense value in measuring and interpreting trends in the distribution of earnings.

15. Another relevant source of data on trends in wage rates are those published for the agricultural sector by the FGV-Centro de Estudos Agricolas (for all states except Sa"o Paulo) and the Secretaria da Agricultura de Sao Paulo (see Tables 6 and 7). The data in Table 6 refer to: (i) montlly wages paid to permanent wage laborers; (ii) daily wage rates paid to casual labor, and (iii) monthly wages paid to tractor drivers. The series for all of Brazil is the unweighted average of state-wide series (excluding Sao Paulo). Given the difficulties of finding an appropriate deflator with which to deflate the nominal values and of estimating the total monthly income wage accruing to day laborers; it was decided to express the nominal values in terms of the nominal value of the urban minimum wage.2/ Between 1966 and 1970 the ratio of the rural wage to the urban minimum wage remained static. However, from 1970 onwards there was a rise in agricultural wages relative to the urban minimum, which was particularly rapid between 1973 and 1975.

2/ There is no minimum wage in rural areas. Table 6: TRENDS IN AGRICULIURAL WAGE RATES, 1960-75

Average Wage Ratio of (1) Average Daily Index of Average Monthly of Permanent to the Urban Wage Rates of the Ratio Wage Rate of Ratio of (5) Wage Laborer Minimum Casual Rural of (3) to Tractor Driver to the Urban *16Cer Cr nU &C6e kg.uL- the Urban (Current Cr$/month) Minimum Wage Per Month) rent Cr$/day) Minimum (1) (2) (3) (4) (5) (6)

1966 48.72 .56 1.63 100 90.05 103 1967 64.28 .59 2.14 105 110.27 100 1968 75.22 .56 2.58 104 131.99 99 1969 87.92 .55 3.05 103 155.85 98 1970 106.96 .56 3.65 103 188.67 99 1971 135.77 .59 4.63 108 234.55 102 1972 163.88 .60 5.61 110 290.88 106 1973 222.79 .69 7.83 131 368.52 114 1974 323.5 .82 13.00 178 513.5 131 1975 434.5 .80 17.50 174 738.0 137

/1 Unweighted average of average values for each state excluding Sao Paulo.

/2 Fundacao Getuiio Vargas, Centro de Estudos Agricolas. - 12-

c. This picture is confirmed by the data in Table 7 on trends in real wage rates for rural workers in Sao Paulo state. Although the nature of the price deflator used is unknown, the Sao Paulo series are broadly consistent with the data from the rest of Brazil.

17. A possible explanation is that the agriculture sector, particularly in Southern Brazil, has been undergoing a major transformation both in terms of techniques of production and crop mix, as well as in terms of its social relations of production. On the one hand, there has been a large increase in acreage, which has been accompanied by the growth of new crops (soya beans) and the revival of more traditional commodities (coffee in Minas Gerais). On the other hand, especially in Sao Paulo, dependent types of landlord-tenant relationship are disappearing and the rural poor are being transformed into wage laborers; meanwhile, the exodus to the cities continues with the effect that between 1960 and 1970, for example, the agricultural labor force grew by a mere 0.7% a year. The mobility of the work force and its slow growth in relation to the demand for labor have served to push up wage rates especially after 1973. - 13 -

Table 7: TRENDS IN REAL WAGE RATES IN AGRICULTURE IN SAO PAULO 1960 - 75

Resident Farm Hsznd's Casual Farm Laborer's Tractor Driver's Daily Wage at Cons- Daily Wage at Cons- Monthly Wage at tant 1971 Cr$ tant 1971 Cr$ Constant 1971 Cr$

1960 4.76 4.59 172.71 1961 4.51 5.21 159.20 1962 4.48 5.10 157.37 1963 4.14 4.56 146.74 1964 4.59 4.89 163.02 1965 5.24 5.93 238.92 1966 4.96 5.75 203.86 1967 5.39 5.49 187.80 1968 5.73 6.46 189.93 1969 5.59 5.99 218.43 1970 6.18 6.80 220.87 1971 6.44 7.04 233.86 1972 7.16 8.00 248.48 1973 8.43 8.84 279.64 1974 9.15 10.53 290.24 1975 9.23 11.46 307.55

Source: Secretaria de Agricultura do Sao Paulo; the price index used to deflate nominal values is not stated in the source. - 14 -

IL. EMPLOYMENT GROWTH

18. It is worth analyzing the existing data on employment growth, both from the point of view of market forces affecting trends in earnings and in relation to the discussion on the implications of increasing inequality. The data sources on this question are rather unreliable, largely because of the lack of year-to-year comparability in the statistics. For example, the IBGE annual surveys on industrial employment, which are frequently quoted in this context, are not at all comparable from one year to the next. The National Household Survey (PNAD) is also unreliable on this question, since it is not clear whether the 1968 survey (based on an augmented sample drawn from the 1960 census) is strictly comparable with the 1972 survey (based on an augmented sample drawn from the 1970 census). Data, used by government sources, on the growth of contributors to PIS/PASEP, FGTS (Redundancy Payments) and INPS (Health and Social Insurance) undoubtedly give an overestimate of the rate of employment growth because of continuing improvements in the coverage of these schemes.

19. Table 8 contains estimates of the long-run rate of growth of employment between 1960 and 1970 - a decade of fluctuating output growth rates and probably a similar employment experience. The data are presented in two ways: the upper portion of the Table contains estimates of employment in the enterprise sector, taken from the Economic Census in both years. In the lower portion of the Table are estimates of the economically active population, drawn from the Demographic Census. This definition goes beyond enterprise sector employment by including employees and the self-employed in the 'informal' sector. Since the Demographic Census definition of the economically active population includes the unemployed, it is not possible to draw conclusions concerning the long-run trend in open unemployment. However, in the absence of a system of unemployment benefits, the rate of open unemployment would, of necessity, be rather low. In manufacturing industry, the rate of growth of enterprise employment lagged behind the growth of total sector employment, implying that there was both an absolute and a relative rise in 'informal' sector employment during the decade of the sixties. On the other hand, in commerce and in services, enterprise employment grew faster than employment for these sectors as a whole, indicating that 'informal' sector employment declined as a proportion of the total. In commerce, there was even a small decline in the absolute level of 'informal' sector employment, hhough this was not true of services.

20. The data in Tables 9 and 10 on employment growth in manufacturing, construction, commerce and services for the period 1966-73 are taken from the Ministry of Labor earnings survey. The year-to-year comparability of the data seems to be ensured by the fact that the survey is carried out in the same month of each year, and it appears that there is little change over time in the percentage of enterprises covered. Even so, year-to-year comparisons may not be very accurate, and - 15 -

Table 8: EMPLOYMENT AND EMPLOYME1'T GROWTH, 1960-70

% Change Compound 1960 1970 1960-70 Annual Rate

In the enterprise sector

Manufacturing /1 1,861,856 2,589,842 39.1 3.1 Commerce /2 953,149 1,709,373 78.9 6.0 Personal & Commercial Services Services/2 521,259 870,907 67.1 5.3 Property Companies /3 - 29,110 - - Commercial Banks - 232,972 - - Finance Houses /3 - 17,089 - - Insurance /3 - 19,244 - -

Economically Active Population from the Demographic Census

Agriculture, Fishing, Vegetable Extraction 12,163,000 13,090,000 7.6 0.7

Mining 108,000 175,000 62.0 4.8

Manufacturing Industry 2,069,962 3,241,861 56.6 4.6

Construction 785,014/4 1,719,714 119.1 8.2

Commerce 1,520,046 2,263,539 48.9 4.1

Transport & Communication 1,088,798 1,246,395 14.2 1.3

Services 2,732,148 3,626,494 32.7 2.9

Others - Public Utilities, Public Sector, Banking, Liberal Professions 2,184,054 4,195,439 92.1 6.7

Total non-agricultural 10,488,000 16,467,000 57.0 4.5

/1 The first figure refers to 1959. Employment in both 1959 and 1970 is averaged over the whole year. End-of-year data yield a percentage increase of 54% or an annual compound rate of 4.0%. /2 Employment measured by averaging over the whole year. 73 Year end employment totals. 74 This is possibly an under-estimate since self-employed construction workers were included under 'Services' in the 1960 Demographic Census. Sources: Economic Census, 1960 and 1970. Demographic Census, 1960 and 1970. - 16 -

the trend rate of growth of employment has been calculated by averaging over initial and terminal years. In industry as a whole, employment grew at an annual compound rate of 11.6% a year, with manufacturing growing by 9.0% and construction by 17.4% a year; in commerce and services the annual compound rate of growth was 15.0% a year. These results may well be an upper bound estimate of employment growth since the mid-60's. The results should be interpreted in the light of the following factors: (i) employment growth is being measured between the trough (1966-67) and the peak (1972-73) of economic activity, and this partly accounts for the very high rates of growth observed; (ii) the data are consistent with the Census data, analyzed above, showing that between 1960 and 1970, formal sector employment growth was in the aggregate about equal to the growth of the total non-agricultural labor force. During the period of recovery in economic activity (1967- 69), employment growth was very rapid. Though at a lower rate, employment growth from 1970 onwards was still quite rapid, particularly in manufacturing and construction. This empirical evidence tends to con- firm the qualitative evidence of labor shortages so frequently mentioned in 1973-74; (iii) the data in Tables 9 and 10 measure employment growth in the 'formal' sector, which may well exceed the rate for the urban economy as a whole (including the 'informal' sector). Finally, it is worth observing that virtually nothing is known about trends in earnings and employment in the 'informal' sector nor about inter-sectoral patterns of mobility. Such data on this poorer 50% of the labor force should be collected.

III. TRENDS IN THE INCOME DISTRIBUTICN

21. Few of the participants in the recent debate on Brazilian income distribution have fully recognized the extent of the serious empirical problems involved in using the data upon which most arguments have been based. These problems are of several types. Brazil does not possess data on the distribution of earnings of full-time employees. The data from the Ministry of Labor survey which have been utilized in the controversy, in fact suffer from a variety of deficiencies. The data include part-time employees who have left employment during the enquiry month and those that have been employed in more than one enterprise. Furthermore, these data only refer to employees in the "formal" sector. The bias imparted by the first set of characteristics is to increase the proportion of low earnings recipients in the distribution. The exclu- sion of employees in the "informal" sector (domestic outworkers and casual construction workers, for example) probably has the reverse effect. When measuring inter-temporal trends in the distribution, such sources of bias would be unimportant if there was some degree of cen- stancy in the incidence of part-time employment, in the rate of labor turnover, in the share of "informal" sector employees in total wage labor employment and in the ratio between mean earnings in the "formal" and "informal" sectors. This is a demanding set of assumptions and conclusions based on the imperfect Ministry of Labor data need to be judged in the light of these deficiencies. Leaving such issues aside, the Gini coefficient measures of dispersion, calculated using the - 17 -

Table 9: EMPLOYEENT GROWTH IN MANUFACTURING, CIVIL CONSTRUCTION, PUBLIC UTILITIES AND EXTRACTIVE INDUSTRIES, 1966-73

Total Industry Extractive Industries Manufacturing Public Uti- Construction lities and Others

1966 2,153,721 47,527 1,676,582 135,900 293,712 1967 2,365,569 71,022 1,755,775 178,032 360,740 1968 2,628,252 89,742 1,969,626 126,061 442,823 1969 3,116,039 95,872 2,325,174 117,011 577,982 1970 3,280,716 1.03,845 2,381,690 148,691 646,490 1971 3,391,486 110,578 2,466,946 160,878 653,084 1972 3,770,472 104,443 2,669,951 176,807 819,271 1973 4,291,556 1.24,096 3,087,618 191,217 888,625

% Change in Employment

1966-67 +9.8 +49.4 +4.7 +31.0 +22.8 1967-68 +11.1 +26.4 +12.2 -29.2 +22.8 1968-69 +18.6 +6.7 +18.1 -7.2 +30.5 1969-70 +5.3 +8.3 +2.4 +27.1 +11.9 1970-71 +3.4 +6.5 +3.6 +8.2 +1.0 1971-72 +11.2 -5.5 +8.2 +9.9 +25.5 1972-73 +13.8 +18.8 +15.6 +8.2 +8.5 1966/67- 1972/73 +78.4 +92.8 +67.7 +17.2 +161.0

Source: Boletim do SEPT. Table 10: EMPLOYMENT GROWTH IN COMMERCE AND SERVICES, 1966-73

Finance, Total Commerce Banking & and Services Commerce Insurance Transport Communications Services Public Services

1966 1,297,763 531,544 199,831 189,618 42,188 173,916 160,669 1967 1,485,598 586,376 228,225 202,452 46,454 235,088 167,079 1968 1,788,571 652,760 257,507 270,327 54,651 299,803 253,523 1969 2,190,155 794,875 271,542 316,109 70,927 343,648 393,05 1970 2,449,192 858,166 277,200 344,569 76,320 396,792 496,145 1971 2,677,870 917,566 282,631 362,589 66,072 459,020 589,992 1972 3,105,346 1,038,806 331,031 401,752 87,426 543,325 703,006 1973 3,322,631 1,053,610 342,726 422,298 143,765 781,315 578,91

CD % Change in Employment

1966-67 + 14.5 +10.3 +14.2 + 17.3 + 10.1 + 35.2 + 4.9 1967-68 + 20.4 +11.3 +12.8 + 21.5 + 17.7 + 27.5 + 51.7 1968-69 + 22.5 +21.8 + 5.5 + 16.9 + 29.8 + 14.6 + 55.0 1969-70 + 11.8 + 8.0 + 2.1 + 9.0 + 7.6 + 15.5 + 26.2 1970-71 + 9.3 + 6.9 + 2.0 + 5.2 - 13.4 + 15.7 + 18.9 1971-72 + 16.0 +13.2 +17.1 + 10.8 + 32.3 + 18.4 + 19.2 1972-73 + 7.0 + 1.4 + 3.5 + 5.1 + 64.4 + 43.8 - 17.0

1966/67- 1972/73 +130.9 +87.2 +57.4 +100.0 +160.8 +223.9 +291.9

Source: Boletim do SEPT. - 19 -

Ministry of Labor data, are presented in Tables 11 and 12. They show that, from 1969 to 1973, there was a continuing trend towards increasing inequality in both the industrial sector and in commerce and services.

22. A second problem concerns the measurement of trends in real income in the bottom 40% of the income distribution. This segment contains more or less the bottom 60% of the rural income distribution and the bottom 20% of the urban distribution. As has been seen, the Ministry of Labor survey excludes the urban poor. The Family Expenditure Survey (PNAD), while it presents data on money income in the whole of the distribution and includes information on the existence of non-monetary income, makes no attempt to quantify or put a monetary value to the non- monetary component. In addition, the series for the period 1967-69 is probably not strictly comparable with that for 1972-73. The data on trends in agricultural wages certainly show a rise in real wages in recent years, but it is questionable how far this can be extrapolated to all members of the rural labor force. Nevertheless, it can in general, be concluded that both urban and rural poor have benefitted in absolute terms from the recent rapid growth in employment. The Demographic Census of 1960 -ad-197G-are- prone t htW~e- criticism as the other sources available. The census odnly measures m'one income, and the fact that the monetary income of the bottom 40% increased faster than that of the next 30% in the distribution (urban employees in the "formal" sector) does not necessarily imply a relative improvement in the conditions of the poorest during the decade. This measured improvement in the position of the bottom 40%,who are largely in the agricultural sector,relative to the urban working class (though not relative to the mean of the distribution) may only be indicative of the increasing importance of money income. In conclusion, at least until the early 1970's, it is unlikely that the real incomes of the rural poor and of employees and the self- employed in the "informal" urban sector diverged significantly from trends in the real uirban minimum wage.

23. The essentially "static" measures of inequality which are used, entirely fail to piclc up the possible life-cycle trends in income. In the context of the recent rapid growth of employment, this point is of particular importnnce , since life-cycle earnings may be less unequally distributed than earnings measured at a point in time since the ranks of the low paid would be swollen by rapid employment growth.

24. As of November 1974, the Geisel Government introduced important modifications in the formula used to compute arbitrated wage agreements. This must be seen as part of the Government's declared policy of reversing the unfavorable trends in the distribution of income. The two major modifications to the formula are: (i) the base component of the formula is to be computed by restoring the average value of the real wage in the previous 12 months, attenuating the depressive tendencies - 20 -

Table 11: GINI COEFFICIENT MEASURE OF TEE DISPERSION OF EARNINGS IN THE INDUSTRIAL SECTOR, 1965-73

(1) (2) (3)

1965 .305 1966 .361 .364 1967 - .365 1968 - .370 1969 _ .407 1970 - .421 1971 .422 1972 .438 1973 .459

Sources and Notes:

(1) 1965 and 1966, data from Boletim do SEPT. For comparison, sectors entitled urban industry, extractive industries and agriculture were excluded for 1965, while the civil construction sector was included in 1966 and petroleum excluded.

(2) and (3) 1966-67, data from Boletim do SEPT. The industrial sector includes all the sectors in the manufacturing sector plus civil construction, electric energy, and vegetable and mineral extraction. Columns (2) and (3) are not comparable. - 21 -

Table 12: GINI COEFFICIENT MEASURE OF THE DISPERSION OF EARNINGS IN COMMERCE AND SERVICES IN URBAN AREAS (1965-73)

(tL) (2) (3)

1965 .349 1966 .380. .387 1967 -- .394 1968 - .409 1969 -- .450 1970 -- .469 1971 - .475 1972 -- .513 1973 -- .502

Sources and Notes:

(1) 1965 and 1966, data from Boletim do SEPT, covers 22 sectors in commerce and services.

(2) and (3), data from Boletim do SEPT, covers data from 29 sectors in each year. Columns (2) and (3) are not comparable due to change in the timing of minimum wage increases in 1969. imparted by the previous practice of averaging over 24 months; (ii) the productivity growth component or to quote, as stated by Law 6147 November 20, 1974), "the coefficient corresponding to workers' participation in the increase in productivity", is to be made multipliea- tive rather than additive. While the phrasing is ambiguous as to whether the aim is to increase, lower or maintain constant the share of worker's wages in value added, the aim of the policy appears to be to keep the money wages of manual workers rising at least as fast as the cost-of-living. This is consistent with the Government's declared goal of reversing the trend towards increasing inequality. This strategy also was evident in the announcement of a 10% wage bonus payable in December 1974, and in subsequent wage awards. Although there has been a change in the wage formula to be applied to arbitrated wage settlements, the individual components which go to make it up are no longer to be published; instead the authorities are to publish a "salary readjustment factor"

25. The minimum wage increase for the Center-South decreed on May 1, 1975 was a 41.4% rise in nominal terms on the May 1974 value, which when compared with the FGV cost-of-living index for May 1974-May 1975 (+25.4%) and that of DIESSE (+26.5%), was a clear demonstration of the Government's Policy. Taking into account the 10% bonus given in December 1974 (which took the form of an advance on the May 1975 rise), the real minimum in 1975 stood 6.5% higher than in 1974 (a year in which there was a very sharp decline of 8.1%). The nominal increase of 44% granted in May 1976 also implies a rise in real terms.

26. It is not yet clear what the effect of continuing rises in the real minimum wages will be on the personal and functional distribution of income and on other economic variables such as the rate of inflation. It appears that one objective of government policy is to reduce income differentials among wage earners by increasing minimum wages more rapidly than other wage rates. This can be seen from the fact that the readjustment coefficient for arbitrated wage rates (i.e., for workers receiving between 2 and 5 minimum wages) was lower during 1975, at between 35 and 39%, than the minimum wage adjustment. The Government also attempted to limit the wage increases of top executives during 1975 to flat rate 30% of the value of 30 minimum wages. However, no statutory authority was established to monitor the increases in either monetary or non-monetary income accruing to top wage-earners.

27. There is reason to believe that it will not be easy to compress earnings differentials. First, as explained above, the average figures decreed for dissidios coletivos may well not be operative, since individual workers with particular skills in short supply may be in a position to maintain and even increase their earnings differentials as compared with the minimum wage. Second, such evidence as is available on the recent evolution of top salaries tends to suggest either that there has been little implementation of government policy or that more generous fringe benefits have been the response to restrictions on increases in - 23 -

monetary income. The only available quantitative evidence on the effect of the change in minimu3n wage policy indicates that between 1974 and 1975, salaries in the lowest range increased by 5.45% in real terms; the percentage increases for the next four salary groups in ascending order were: 2.27%, 6.67% and 22.5% (excluding "fringe benefits"). Thus, the more generous wages policy at lower levels was accompanied, to a large extent, by a tendency towards increasing differentiation throughout the earning structure, part:cularly at the highest levels.

28. Because of the diversity of work relations in the informal sector it is not clear how the more generous minimum wage policy will affect the 5.5 million employees and the 4 million self-employed in the "informal" urban sector who are beyond the reach of compulsory minimum wage legislation. There wou:Ld certainly be some "trickle down" effect, but any attempt to push the minimum wage substantially ahead of cost-of- living increases is unliLkely to be fully implemented in this sector. The self-employed category is also extremely diverse, but a majority of all self-employed workers earn less than the minimum wage. Even though this group is not direct:ly affected by official wages policy, their incomes are probably fairly closely related to movements in the urban minimum.

29. It has been suggested that the squeeze in the urban minimum wage between 1964 and 1974 had a beneficial effect on employment in both rural and urban sectors and that a rise in the urbnn minimum might reduce employment growth and be detrimental to the interests of the "poorest" sections of the community. However, it seems equally reasonable to argue that a rise in wages and in the share of labor in urban value added is not only consistent rith further rapid employment growth in the "formal" sector (provided the rate of accumulation can be maintained), but will tend to raise incomes and employment in the "informal" sector as well.

IV. PIS (SQCIAL INTEGRATION PROGRAM) AND THE FOURTEENTH SALARY

30. PIS (Social Integration Program) was designed as a mechanism for supplementing the monetary income of urban employees in the enterprise or "fotmal" sector. A similar fund exists for public sector employees known as PASEP, and on July 1, 1976, they were unified as PIS/PASEP. The receipts of PIS come from enterprise contributions of two kinds: the first is a fraction of the value of total enterprise sales (fixed at 0.50% from 1974 onwards) and the second comes from a transfer of 5% of the corporate income tax burden. The resources for PASEP come from a transfer of 2% of the receipts accruing to each level of government, avoiding the double counting problem involved in intra-government transfers. The Fund distributes these receipts (cotas) annually to employees in the form of individual shareholdings: half of the available resources are distributed according to the employee's age and half on a strictly earnings related basis. These shares can only be cashed in one of the - 24 -

following situations: marriage, death (when the money accrues to dependents), retirement or house purchase. In the first and last cases, where the employee remains a member of the labor force, new shares are issued in his name at the next year's distribution of shareholdings.

31. The resources of PIS/PASEP are administered by the BNDE and are invested in a variety of ways. For example, a recent regulation permits the Bank to use PIS funds to acquire equity participation in nationally owned enterprises. The PIS/PASEP resources yield interest of 3% a year plus monetary correction, as well as profits, if any. The total annual yield can be withdrawn in cash or reinvested, in which case the employee receives extra shareholdings. Thus, an individual's account comprises the annual cota and the annual yield arising from the investment of PIS resources. The latter payment is related directly to the capital value of individual accounts.

32. In June 1975, the average value of an accumulated cota in PIS was Cr$1,080 (almost twice the value of the highest regional wage) and Cr$3,440 in PASEP. In June 1974, the average new cota distributed to shareholders in the form of shareholdings in PIS was equal to slightly more than one monthly minimum wage, while the minimum addition to shareholdings was about 2/3 of one minimum wage. The yield, which is made up of monetary correction, interest and profits on the accumulated shareholding, will vary a great deal from year to year - at least in current prices - but in 1975, the average yield was about 1/5 of the current monthly minimum wage. Such a sum represents a fairly minor addition to average income and, since the value of shareholdings and their yields are strictly earnings related, the redistributive impact of the payout is, at least amongst corporate sector employees, fairly negligible. 33. In 1973, the total number of individual employee accounts in PIS was 8.1 million (1974: 10.8 million) and there were 2.7 million (1974: 3.4 million) employees in PASEP. This figure excludes multiple incomes following a change in employment. Since the total size of the urban labor force in 1973 was 21.8 million (including 4.9 million employers and self-employed workers), about 50% of the total urban labor force (just 64% of employees) has access to PIS/PASEP shareholdings and their yields. This implies that the income of the bottom three or four deciles of the urban income distribution is unmodified by PIS/PASEP yields. Similarly, a lower proportion of the population in upper urban income deciles have access to PIS/PASEP because of their status an non-employees and therefore, while PIS/PASEP yields are earnings related, average incomes in the top deciles are increased by less than incomes in the deciles in the middle range of the urban income distribution. Thus, while in absolute amounts, the yields to which shareholders have access have been small, the main beneficiaries are the urban working class in the enterprise or "formal" sector. This is precisely the group which lost most from the process of income redistribution which occurred between 1960 and 1970. - 25 -

34.. A complete assessment of the redistribution impact of the Social Integration Program must consider possible alternative uses of the PIS/PASEP funds. About 75X of PCS receipts come from the levy on sales, but thts has been compensated by reduction in the state value added taxes. The rest of the resources find the:Lr counterpart in lower effective federal corporate taxatiou. In this senne, PIS/PASEP receipts can be viewed as a transfer of resources from the Government budget to an investment fund which, because imediate access to thet benefits on the part of employees has been somewhat limited and because manty of the lowest income groups have not been included, has mainly benefited private sector enterprises. An alternative use of the resources with a greateir redistirbution impact micht, for example, have been to increase the level of public expenditure on transfer payments or public consumption goods destined for the poor. The lost potential for income redistribution through thie Government budget frequently is referred to as the "forced savings" aspect: of PIS/PASEP.

35. Recent legislation has modified the "forced savings" aspect of PIS/PASEP by increasinig the level of current yields from the fund. From 1977, an individual who has held a shareholding for fine years and is earning less than five minimum wages, will be able to withdraw annially a sum equal to oane regional minimum wage from his account. The extent of the withdrawal will be limited by the availability of funds in individual PIS/PASEP accoutnts. Those earning above five minimm wages would continue to have accesa to the annual yield on their accounts. Since low-incme employees already have access to the yields on their shareholding, the tendency of the measure is to equalize in absolute terms the current yields from PIS/PASEP shareholdings. This permitted withdrawal is known as the "fourteenth salary", although it is neither paid by employers nor an earnings related sum (as is the "thirteenth salary"). The drawback to the ner regulation is that recipients of the fourteenth salary will no longer be able to withdraw that portion of the yield on their accounts corresponding to monetary correction. The purpose of this measure is to prevent the decapitalization of the accumulated cota in real terms. Since ;It the moment, annual distribution of new cotas are approximately equivalent to one monthly minimum wage, the withdrawal of one miniimm wage-Yi- thC b f&iiteentFfialarywli not -prejtudice the real value of accumulated cotas. On the positive side, an additional salary, paid at the ral:e of one monthly minimum wage, is a fairly substantial addition to gross income for low-income households. Moreover, the increment will be inversely related to earnings and will, therefore, have beneficial effects on the distribution of income within the "formal" sector. Furthermore, the measure certainly modifies the "forced savings" aspect of PIS/PILSEP as it operates at present. A further modifi- cation is that the fus:Lon of PIS/PASEP will have the effect of equalizing the cotas distributed to participants in each fund since the same distributional criteria will be adopted. Hitherto, the volume of resources in PASEP placed public sector employees in a relatively privileged position. - 26 -

V. TYE REDUNDAICY PAYMENTS FUND (FUNDO DE GARANTIA DO TEMPO DE SERVICO) AND THE NATIONAL HOUSING BANK (BANICO NACIONAL A,_HBITACAO)

36. FGTS, introduced in 1966, replaced a previous redundancy payments scheme which had several disadvantages so far as both employers and employees were concerned. Redundancy payments are especially important in Brazil because of the complete absence of a system of unemployment benefits. The essential feature of the previous system was that any worker who had been employed in one job for more than a year and who had been subject to unfair dismissal (sem justa causa) was entitled to receive to the number of years worked in that job mult4plied by the highest monthly wage so far received. After ten years in one job, a worker would actually hold the job as a right and was entitled to a lump sum payment on dismissal or retirement equal to twice the value fo the number of years in that employment times the highest monthly wage. The disadvantage of this scheme to employers was that since they were not making regular contributions to a redundancy fund, the cost of large scale redundancies represented a heavy burden. It was argued that this tended to inhibit technical progress, as well as the process of industrial concentration, since even the inefficient firms could not afford to go into liquidation. Firms also found themselves with an aging labor force and the right to a stable Job, granted after 10 years, reduced labor mobility overall. From the point of view of employees, there was effectively no access to redundancy payments during the first year of employment. There was also some possiblity of losing one's job after nine years of continuous employment.

37. Under FGTS, employers credit each employee's account every month with a sum equal to 8% of the employee's salary. In theory, employees were given the choice between the old system and FGTS, although there appears to have been some pressure on them to opt for FGTS. In cases where employees prefer to remain in the former system, 8% of the wage is credited to the employer's account. Currently, the accounts are credited annually with monetary correction, as well as a 3% real rate of interest. Employees can withdraw the entire sum in their account if they are made redundant sem justa causa. In addition, they may claim a further sum from the employer equal to 10% of the total outstanding balance.l/ The sum in an employee's FGTS account can be withdrawn under other circumstances, such as retirement, house purchase and improve-mpnt, the purchase of a small business or grave sickness and other family emergencies. In the case of withdrawals for house purchases, amortization of mortgages and house improvement, there are some fairly severe restrictions: the employee is required to have worked five years, be up to

1/ Even the employee who has been dismissed com lusta cause has access to his account in conditions of distress. - 27 -

date with his mortgage payments and the financial transaction must be with the official housing finance system. Given the absence of a system of unemployment bexiefits, the average redundancy payment is supposed to provide between three and four months of income support, during which the unemployed worker seeks a new job. Once this has been used up, however, there are no other sources of income maintenance and one re-employed, the extent of future claims on the Fund are limited by the size of the new account.

38. As compared with the old system, there is no doubt that FGTS does provide employees with some advantages. First, they can claim redundancy payments if dismissed during their first year of employment. Second, the FGTS accounts can be drawn on in cases other than a redundancy or retirement. There has been a steady rise since 1967 in the ratio of withdrawals to employer contributions. During the first months of 1975, this ratio fluctuated between 45% and 55%, with total withdrawals equal to about 1% of national income.

39. At present, there are between seven and eight million FGTS accounts which implies t'hat between 40 and 50% of the employed urban labor force has a claim to redundancy payments; the self-employed and employers are, of course, automatically excluded. In general, as with the PIS scheme, FGTS o,nly covers employees in the formal or corporate sector, and thus excludes a very sizeable proportion of wage employees who are not in a strong enough bargaining position to compel their employer to make his contribution to FGTS.

40. A further question concerns the rate of return on the Fund and the uses which are made of the very substantial financial resources at its disoosal - a total of Cr$26,465 million at the end of 1975. The agency wh__:h administers the Fund is the Banco Nacional de Habitacao; the Bank's major aims since its iniception in 1964 have been to provide finance for house purchase, paritcularly house purchase among low-income recipients, and to help finance wai:er suppy and sewerage investments. At the same time, the Bank is expected to use the revenues on-its operations in the area of housing finance to provide an acceptable return on the accounts which emDlovees hold in the FGTS. There is a fundamental contradiction between these two requirements since the more ambitious is the attempt to provide traditional forms of low-income housing, the greater necessarily has to be the _

41. In attempting to lower the average cost of its housing loans, BNH has reduced the reiturn paid to FGTS. Between 1966 and 1971, FGTS accounts were corrected on a quarterly basis according to the following scheme: monetary correction plus 3t on the outstanding balances for employees with up to twio years in the same job; monetary correction plus 4% on the balances of employees with between three and five years of continuous service; monetary correction plus 5% on the balances of employees with between six and ten years of continuous service and monetary correction plus 6% on the baLances of employees who had worked for more than - 28 -

ten years in the same job. From September 22, 1971 onwards, however, this band of interest rates was reduced to a single real rate of 3% for all new accounts, although the existing rights of employees to higher interest rates were preserved, so long as they did not change employment. The accounts of all employees who had not opted for FGTS were subsequently paid interest at a uniform rate of 3%. In 1972, the system for capitalizing FGTS account was again changed and monetary correction and interest subsequently were paid on an annual basis using the rate of inflation of the following year to correct the previous year's balance. The implication of this was that new inflows would lose on average six months of monetary correction plus interest, whereas under the system of quarterly correction, this average loss would be equal to 1-1/2 months. In addition to this loss, which accrued to every de- positor, employees who withdrew their funds before the end of the year would lose the monetary correction and interest appropriate to that year. While these changes allowed the Bank to reduce its loan charges to borrowers, holders of FGTS accounts lost out in the process. However, on December 5, 1975, the Geisel administration announced that it intended to reintroduce the system of quarterly correction of the accounts. 42. The Government has acted in the following way: deposits as of De- cember 1974 are to receive monetary correction of 23.76% (to compensate for inflation during 1975) plus interest. The end-of-year balance of new deposits made during the course of 1975 is to be corrected at a rate which represents what would have happened if the accumulating balances had been corrected on a quarterly basis; the index to be used for this purpose is 11.3% which is slightly less than one half of the rate of inflation plus interest (this is equivalent to correcting the average balance by the annual rate of inflation plus interest). The duly corrected 1974 balances, together with the duly corrected 1975 deposits are to receive monetary correction plus interest for the first quarter of 1976. However, the monetary correction factor of 6.7% adopted for this purpose is only one-half of the actual rate of inflation observed during the first quarter of 1976. Under the new system, deposits will lose an average of 1-1/2 months of monetary correction but the more substantial losses incurred under the old system will have been corrected. Even so, the rather arbitrary nature of the number assigned as the monetary correction factor means that there will still be room to squeeze FGTS accounts in favor of cheaper home loans.

43. In order to make a complete assessment of FGTS, particularly its impact on the distribution of income, it is necessary to examine the activities of the National Housing Bank (BNH) whose main source of deposits comes from FGTS. BNH's general purpose has been to coordinate programs designed to reduce Brazil's substantial housing deficit and to keep the urban housing stock increasing in line with urban population growth. Since about 70% of urban households receive incomes less than three minimum wages, BNH has had special responsibilities for stimulating the construction of housing for owner-occupation by low-income recipients. Up to now, there has been little public sector activity in the - 29 -

area of rental accommodation in Brazil. Sarveys reveal that owner-occupation is firmly established as most Brazilians preferred housing situation.

44. The major way in which BNE attempted to increase the supply of housing between 1964 and 1974 was by financing the operations of the State Housing Authorities; (COHABs). The COHABs present plans to BNE of fairly large scale public housing projects (coniuntos habitacionais) which are designed for purchase by low-income households. BNH then provides COHABs with finance for construction and the COHABs make mortgage agreements with the individual home purchaser. BMH thus has acted only as a second line financial agent, and has had no direct responsibility for executing the housing program. On the supply side, it appears that one of the weaknesses of the program has always been the fact that its execution depends on state level agencies who are often not suffi- ciently well qualified technically and who are subject to interruptions in their rhythm of work due to changes in political administration at the state level. The other major constraint on the low-income housing program has come from the side of demand. Universal owner-occupation is an extremely difficult housing objective so far as households with low and variable incomes are concerned, unless there is a substantial level of subsidy from the state. While there was ample demand for housing finance the mid-1960s, demand gradually fell for the following reasons. (a) Loan contracts specify positive real rates of interest. (It is worth remembering that owner-occupiers in developed countries have recently been paying strongly negative real rates of interest.) However, real interest rates charged to EMNI's borrowers are inversely proportional to the sum tiorrowed, which implies a considerable degree of cross-subsidization as between high and low-income borrowers. The degree of cross-subsidizations has increased and at present the range is between 1% for loans up to US$3,000 and 10% for loans between US$16,400 and US$52,200 (BNE's upper limit), with the average cost of funds to BNH being 4.5% a year. It is also the case that the term of loans is inversely proportional to the size of loan, with the maxmtsum repayment period having been extended to 25 years in 1972; for low income borrowers this is both a maximum and a minimum period. (b) During the early years of the scheme, monetary correction on the outstanding debt was at a higher rate than the nominal increase in the minimm wage, so that the monthly repayment sum increased faster than wages. An alternative scheme was devised (Plano de Equivalencia Salarial) whereby householders could opt for repayments which increased at the same rate as the minimum wage. The problem with this scheme was that there could still be a positive outstanding debt at the end of the mortgage term, since the debt was corrected using the indices for correcting ORTNI/ which might exceed those for the minimum wage. Recently, the Government has promised to find a way of liquidating such outstanding debts, although the need to do so has not yet arisen. The effect of (a) and (b) above, combined with the low and variable incomes of householders, has resulted in many being in arrears in their mortgage payments. Of- ficial sources indicate that over 30% of households in the low-income group of up to three minimum wages are more than three months behind schedule with their debt repayments. Initially, the main burden of non-payment of debt falls on the COHABs. Though they are refinanced by BNH, their basic financial viability appears to affect their ability to plan new projects. Such widespread inability to maintain repayment schedules acts as a disincentive to new borrowers as well as retarding new housing construction activity for low-income groups. (c) Because of high urban land prices the conjuntos often are located on the periphery, distant from places of work, and residents incur onerous transport costs. Conjuntos tend to lack community centers or recreational facilities and resident's organizations are too weak to provide a cohesive social identity. Abandoned apartments are susceptible to vandalism and attract squatters-, which diminish the desira- bility of the conluntos. These conditions tend to depress the demand for new units and discourage the emergence of an active second-hand market for apartments in the conjuntos.

45. The poor performance in the area of low-income housing finance led the Bank to a reappraisal of its activities, resulting in the announcement of PLANHAP in 1973. The aims of the new program are twofold: to reduce the cost of conventional housing loans and to experiment with financing new areas of activity. PLANHAP's target is 2 million new and improved dwellings by 1983. The specific measures adopted are as follows.

(a) The range of household to be included in the low-income housing scheme is increased from those earning between one and three minimum wages to those between 0.5 and five minimum wages. The upper limit for a housing loan in this scheme was increased from US$4,700 to US$7,500; although this latter charge hardly seems consistent with a policy of directing resources at those most in need. (b) An attempt has been made to improve the financial viability of the COHABs by changing them lower rates of interest on their loans and reducing their share in the financing

1/ Monetary correction on FGTS accounts is at this rate as well. - 31 -

of any projtect to a maximum of 15%. As an incentive to constructioni for low-income household, COHAB financial participation is zero on units whose average cost is US$4,771 or less. COHABs will also be given loans to build up stocks of land at pre-betterment values. (c) There has been a reduction in the real interest rate charged to final borrowers: the maximum loan under the previous low-income finance program of US$4,800 bore 6% and it now bears only 3%. This reduction seems to have been achievred through increased interest rate discrimination and cross-subsidization: the corollary to this is that BNH wi.ll have to pursue its lending program to high income hous;eholds. A measure announced in July 1975 (Decree Law 1358) attempted further to reduce the cost of housing loans for low-income households. Previously, households not paying income tax had never been able to benefit freom the fiscal incentive which allows house pur- chasers to deduct 20% of their annual installment payments from taxable income. Decree Law 1358 gives a credit, worth 12% cif the annual value of installments (subject to a minimtm in 1976 of Cr$480.00 and a maximum of Cr$3,960.00) against insitallments due next year. To give an example, for householders earning up to 1-1/2 minimum wages (there are 136,000 mortgage holders in this income group), a monthly installment of Cr$122.14 would be reduced by Cr$40.00 to Cr$82.14 - a reduction of 33%. In fiscal year 1975, this transfer to low-income households, from the federal budget, came to Cr$300 million representing the only federal subsidy to BNH's low-income housing program. (d) PLANHAP has established two funds, FINC and FINEC, to give loans to state and municipal authorities for infrastructural services and essential community facilities. The interest cost of such programs previously ranged from 4% to 8% and now lies between 2% and 7%. The airm of the funds is to avoid the situation where the cost of providing such infrastructure services directly falls upon the low-income house purchaser; the cost of such loans will be borae through general tax revenue. (e) PLANHAP has introduced PROFILURB, a site and serv4 ces program. The aim of PROFILURB is to finance the act_vities of local state authorities in the provision of vacant lots, complete with water, sewage, electricity and adequate access roads as well as slum improvement programs. Poor households iwill be able to acquire these urbanized lots with a view to building their own houses. At present, a large proportion of the mass of urban workers lives in self-construicted housing: for most households, in fact, own capital formation represents their only means of ac- quiring adequate housing. The main deficiency of these houses is their lack of adequate urban services, and PRO- FILURB represents, therefore, an intelligent response to existing social realities. The maximum cost of an improved lot will be US$1,789, with BNH financing the total cost of the investment and the maximum level of family income commitment set at 9%. The target number of loans under this program is 476,000 by 1979. However, pilot programs have shown up some deficiencies in PROFILURB: a speculative market in improved lots has grown up, and while there has to be some incentive to ensure that construction takes place, it is difficult to distinguish between the pure speculator and the man who is waiting to find the time and resources to start construction. BNH programs such as RECON and RECON SOCIAL, which make loans for the purchase of building materials will be of use here though the interest rate on RECON loans (10%) could be a strong disincentive. (f) There is a new program of home improvement grants direct to existing owner-occupiers: the average value of such loans is expected to be US$1,298 and they will be entirely financed by BNH. This sum implies a commitment of a- bout 5% of family income by a family earning the median urban income of 2.08 minimum wages.

46. The quantitative results of the low-income housing program up to 1975 were disappointing. During a ten-year period in which BNH financed, whether wholly or partially, a total of 1,142-,265 housing units, only 252,160 units were financed for households earning up to three minimum wages. With Brazil's urban population gaining 2.7 million inhabitants a year, BNH's housing program for low-income families has been making only marginal contribution to the solution of the housing problem. Data from Conjuntura Economica show that lending to low-income families virtually ceased during 1974. It remains to be seen whether the new incentives will reactivate the demand for conventional housing loans, and whether the COHABs will present sufficient new plans. The results for 1975 were encouraging and a strong recovery occurred in the number of loans to finance popular housing. In December 1975, the number of low-income units financed was 20,000, easily surpassing the 6,225 upper income anits financed by BNE. The total of low-income housing loans agreed during 1975 was 60,000 which compares favorable with an annual average of 20,000 a year between 1964 and 1974. However, an even more sizeable federal budget subsidy than the one outlined above may be required to keep this program going.

47. BNH has also diversified its loan activities into the area of urban services and infrastructure: the main beneficiaries of these programs are state and municipal authorities responsible for water, sanitation, transportation and urban renewal. In 1971, the Government embarked on a new program, PLAN&SA, with the objective to serve 80% of the urban population living in four-fifths of Brazil's urban communities with adequate water by 1980 and satisfactory sewerage services to residents of the main urban centers. Investments of US$2.5 billion were projected in the sector by the end of the decade. The massive PLANASA program is financed - 33 -

half by BNH and half by the states, through individual State Revolving Funds for Sanitation (FAE's). Along with interest on and repayment of former loans, BNH's prirLcipal financial resources are the deposits of retirement funds. The ]AE funds are obtained from allocations of up to 5Z of the states' anniual revenues. In poorer states' 50% financing commitment, BNH extends supplementary loans to the states. 48. The PLANASA program focuses on statewide, rather than individual city water and sewerage needs. In order to enter the program, states must sign agreements with BNH which establish single statewide water and sewerage compamies, draw up statewide investment plans and generally provide for sitate guarantees of BNH loans (Minas Gerais is an exception and the Banco de Desenvolvimento de Minas Gerais guarantees BNE loans for a fee). All of Brazil's twenty-two states have formally signed such agreements and entered the program. The statewide system of project preparation and financing has two novel features that help achieve a more equitable distribution of income:

(a) Varying interest rates. BNH funds were initially loaned to states at interest rates varying from 4% to the poorest states in the Northeast to 8% to the relatively more prosperous states of the South and Center of Brazil, Principal of loans is corrected for changes in the general price index; and

(b) Statewide or regional tariffs. The PLANASA program requires the establishment of statewide or regional tariff structures. Such a policy permits the generation of cash in wealthier cities, which may then be made available to finance projects in poorer and smaller communities.

49. Up to 1975, P'L!'NASA's progress was disappointingly slDv and it had developed some serious problems. Investment plans, of uneven quality and slow to materialize, proved to be much costlier than originally envisaged. The state water companies experienced difficulties in obtaining concessions for water distribution systems and in maintaining adequate tariffs. Many companies were having organizational problems as well, and some only maintained an adequate financial position by financing a proportion of their worlIs with state grants. The National Sanitation Training Program was cne year behind schedule and it appeared that PLANASA's original targets would not be met. This led to a realistic readjustment of PLANASA along the following lines.

(a) The PLANASA national targets were modified. The present objective is to service 80% of the population in 80% of the communities with water and to provide the population of the main urbani centers with sewerage by 1980.

(b) A technical assistance program to the state water companies was launc'hed with the Pan American Health Organization (PAHO) as managemnent consultant. - 34 -

(d) BNE's real lending rates (i.e., after monetary correction) were lowered to a range of 2% to 7%.

(e) Consultants were hired to prepare a National Tariff Law for the sector, treating various aspects of tariff problems and the ownership of municipal water resources. (f) Where the tariffs necessary to sustain the very high finan-' cial standards prove excessive for the population's ability to pay, limited grant financing is envisaged.

50. The Gover=ment's major goals for the sector are to establish a viable national financing scheme which will serve the water and sanitation needs of the majority of the urban population by the end of the decade, within the frame-work of a mechanism that will equally serve rich and poor states, communities and individuals, with the incidence of the financial burden falling more heavily on the rich than the poor. In spite of its problems, PLANASA has made useful progress towards achieving its goals. The recent revisions in the program should enable it to meet its objectives readily.

51. In order to assess the redistributive effect of the combined FGTS- BNH program, the following points need to be considered.

(a) FGTS provides a strictly earnings-related pension or redundancy payment; as such, it does little to modify the distribution of income, except as between the corporate and the "informal" sector, to the benefit of corporate sector workers as a whole;

(b) The incidence of the employer's contribution needs to be considered: would wages be proportionately higher in the absence of FGTS, or are the employer contributions being financed through higher prices? If the latter, then taking the distribution of both costs and benefits, FGTS may actually be harmful to the distribution of income;

(c) Considering the distribution by size class of income of employees' quotas in FGTS and the distribution of housing loans by income class, low-income households (which have the highest ratio between total FGTS quotas and total housing loans) have been making substantial transfers to middle and upper income households (which have the lowest). This transfer increases with (i) any decrease in the return on FGTS quotas compared with an alternative rate of return, (ii) any increase in - 35 -

the differential between the cost of BNH house finance loans compared with the cost of finance from conventional sources, and (iii) increases in the divergence between the share of lower income groups in the fund and their shares in housing finaance. One feasible way of reversing this tendency is to increase the level of the federal government's transfer of "zero" cost resources to BNH. Either owner-occupation or a public sector rental program will have to be subsidized by the Treasury in a much bigger way or more flexible design and and construiction systems and standards adopted if greater progress is to be made in resolving low-income housing problems, 4ithout seriously decapitalizing the Redundancy Payments Fund.

ANNEX II

SOCIAL WELFARE PROGRAMS IN BRAZIL

ANNEX II: SOCIAL WEIFARE PROGRAMS IN BRAZIL

Page No.

SUMMARY AND CONCLUSIONS ...... i-vi

Introduction ...... I

I. The State's Activities ...... 2

A. Federal Spending ...... 2

Social Assistance ...... 2 Health and Sanitation ...... 3 Conclusions ...... 5

B. State Government Expenditures ...... 5

II. INPS (National Institute for Social Welfare) . . 6

A. Origins ...... 6

B. Receipts ...... 7

C. Coverage ...... 12

D. Benefits ...... 13

General ...... 13 Cash Benefits ...... 14 Health Care ...... 22 Free Pharmaceuticals Program ...... 24

III. FUNRURAL (Rural Worker Assistance Fund) . . 26

IV. PRONAN (National Food and Nutrition Program) ...... 30

A. Food Distribution ...... 30

B. Food Production ...... 31

V. CSU (Urban Social Centers) ...... 33

VI. FAS (Social Support Fund) ...... 33

SUMMARY AND CONCLUSIONS

1. The advent of the Geisel Administration in November, 1974 has brought a substantial increase in the importance attached to social welfare programs. This change is illustrated by the higher level of real expenditure devoted to social welfare, the introduction of new programs and efforts to improve existing schemes in order to extend the coverage of the social welfare system. Furthermore, the Geisel Administration exhibits readiness to alleviate absolute poverty and its causes by direct programs to benefit designated target groups. Although the following review draws attention to certain shortcomings and deficiencies of current social welfare programs, the new approaches are encouraging. This report does not consider any of the neut government production programs such as Polonordeste or Sertanejo wich are intended to benefit income groups.

2. The major health and social welfare programs in both urban and rural areas are administered by semi-autonomous public agencies whose funding is virtually ind,ependent of either the federal, state or municipal budgets: about 70% of the urban labor force has access to the cash benefits and health programs administered by INPS (Instituto Nacional de Previdencia Social) while FUNRURAL provides similar benefits to the rural labor force, although FUNRURAL's health program is still in an embryonic state. The role of the state's activities, financed out of budgetary revenue, is consequently rather limited.

Federal Expenditures

3. Federal expenditures on social programs can be divided into two categories: those on "health and sanitation" and those in the general area of "social assistance". "Social assistance" absorbs the largest share, a and in 1975 it amounted to Cr$11.6 billion out of a total Federal budget of about Cr$95 billion. Federal Government spending in the area of "social assistance" is for the most part strictly related to the state's activities as an employer. At the present time, the Federal Government has no general responsibilities in the area of social assistance to the mass of the population. Total federal spending on "health and sanitation" in 1975 was Cr$2.6 billion or less than 3% of total federal spending, although the Geisel Government has substantially increased the resources committed in this area.

4. The share of expenditures actually devoted to health and welfare provision to the community at large can be put at Cr$4.8 billion in 1975 or about 5% of federal spending. In countries with fairly progressive tax systems, health and welfare expenditures financed out of the federal budget constitute a faily powerful device for redistributing income. The clear challenge facing Brazilian social policy at the present time is to finance health and welfare programs in a manner which in fact improves conditions for larae members of thp poor nnd hence has a greater redistributive effect. State and Municipal Expenditures

5. State government budgetary estimates for 1975 suggest that state expenditures in the area of social welfare are as important as those made by the Federal Government. State public health services are more widely available in poor states, and in the metropolitan areas of richer states. The hospitals and clinics of the state health system tend to be used mainly by the poorest households - those who are not insured with INPS. While total expenditure by the states is 80 to 90% of total federal spending, municipal expenditure is less than 30% of total expenditure at the state level.

INPS (Institute Nacional de Previdencia Social)

6. The INPS is the centerpiece in the Brazilian urban health and social security system. The institute was established in 1967, as a semi-autonomous public agency (currently it is subordinated to the Ministerio de Previdencia e Asistencia Social: MPAS), as the result of a fusion of several smaller institutes serving the social security and health requirements of specific groups of urban workers. At its inception, INPS inherited from its predecessor agencies the principle that only those who contribute to the Tnstitute can have access to its benefits. There is a contradiction between this principle and the extension of access to welfare benefits on the basis of need. In order to circumvent this contradiction, it has been the aim of government policy since 1967 to increase the percentage of the urban labor force (plus dependents) who are "insured" with the system. Thus, a greater "universalization" of access to the Institute's benefits has been achieved, while retaining the fundamental principle of "he who benefits must pay". Very recently, there has been a weakening of this principle, but its continued retention means that certain groups of the urban poor are excluded from the -Institute's programs.

7. The major source of INPS revenue (73% of the estimated total for 1976) comes from contributions made by employees, employers and the self-employed. The most important type of contribution (67% of the estimated total for 1976) comes from 16% of earnings (including the thirteenth wage) paid by both employee and employer (8% each) in the corporate sector. The cost of INPS is largely financed by its beneficiaries: either directly through their contributions or indirectly, to the extent that employees' contributions are shifted forward into the prices of goods purchased by the beneficiaries. General tax revenue plays a minor role. Total INPS receipts (both real and nominal) have grown extremely rapidly in recent years to Cr$48.8 billion in 1975 as compared with total federal expenditure of Cr$95 billion. The resources at the disposal of INPS represent about 7% of GDP.

8. The number of "insured " members in INPS has virtually trebled between 1967 and 1975. This expansion can be explained by the rapid growth of urban employment, the extension of coverage amongst previously ineligible groups (especially among "employees" and the self-employed in the "informal" sector) and by the inclusion of newly-eligible groups (domestic servants). The percentage of the labor force covered by INPS is much larger than the proportion which participates in either the Redundancy Payments Scheme (FGTS) or the Social Integration Program (PIS). The latter two systems are confined laLrgely to the "formal" or corporate sector of urban activity. Despit:e recent expansion, however, fully 30% of the urban labor force is not yet included in INPS, and this group is likely to consist of the poorest members of urban society.

9. Benefits under INPS are basically of three kinds: cash benefits, health care and social assistance. On the benefits side, there is an implicit duality between, on the one hand, cash grants (largely old age, widows' and dependents-' pensions) and, on the other hand, health care and social assistance. While the administration of cash grants embodies an insurance principle, the existence of health care and social assistance tends to imply an open--end commitment, without there being any direct ralationship between cc,ntributions and benefits. The basic problem here is that a single source of funds finances both aspects. Furthermore, in the absence of more substantial transfers from the Federal Budget, the INPS may not be able both to expand health care facilities and pay out pensions on the generous scale promised when, in the future, the growth of receipts is deceleratirng and the population is becoming more aged. There are grounds for recommending the separation of the two functions on the benefits side and the introduction of separate sources of revenue for each activity. A second argument favoring this course of action lies in the nature of the externalities associated with benefits due to health expenditures. The existence and importance of such externalities means there is a strong economic case for including everyone in the health care system and for financing its costs from general tax-revenue.

10. All workers insured with INPS and their dependents have access to the health services financed by the Institute. These, together with the public sector services provided by the State and Municipal Governments and the public health program sponsored by the Ministry of Health, are the major public initiatives in this field. In recent years, there has been a rapid growth in the share of INPS expenditures devoted to medical assistance. However, it is clear that, at least between 1970 and 1974, INPS was not equal to the task of providing health care for the vast numbers of those who were becoming newly eligible to receive these benefits.

11. Although the INPS is a semi-autonomous public agency, the provision of medical assistance is not modelled on a National Health Service of the British or Scandinavian. type. In general, the Institute's funds are used to increase the coverage of the system of private medical care. Never- theless, there are several ways in which publichealth care is available under INPS. First, the INPS operates its own network of outpatient clinics and hospitals. The present government policy is to freeze the INPS's own hospital network at its present size. There is substantial excess demand for the services of outpatient clinics largely stemming from the great mass of the population who cannot pay private medical fees (even when INPS pays some proportion of them). Excess demand results in inevitable queuing and low quality of treatment. Secondly, INPS finances medical assistance through contracts (convenios) signed with a variety of entities: iv-

trade unions, university medical schools, enterprises and municipal and state governments. Thirdly, INPS finances medical assistance through its financial support for private sector medicine. There is clearly a very strong lobby, both within the medical profession and MPAS, favoring the continuing "privatization" of INPS. The private medical facilities account for only 18% of outpatient consultations, but 90% of all hospital cases financed by INPS.

12. The free pharmaceuticals program (Projeto de Assistencia Farmaceutica) begun in 1975 as a pilot project in Greater Rio de Janeiro, is undoubtedly one of the most audacious innovations in Brazilian social policy in recent years. The program is financed entirely from the resources of INPS and is being administered by INPS in conjunction with CEME (Central de Medicamentos). Fifty-four of the most frequently prescribed drugs will be available under the program.

FUNRURAL and New Rural Programs

13. FUNRURAL, an agency of the Ministry of Social Welfare, administers social welfare programs in the rural sector. Its activities began in the mid-sixties, but the program was given a massive injection of financial resources in 1971, with the establishment of PRORURAL. In terms of the volume of resources available, Cr$5-6 billion in 1975 or about 0.7% of GDP, FUNRURAL must be adjudged a serious attempt to replicate in rural areas the kinds of benefits available in urban areas through INPS. FUNRURAL has two main sources of receipts. First, a levy of 2.4% on the urban wage bill which is paid by employers to INPS, which transfers the funds to FUNRURAL. In 1974, this source provided 66% of FUNRURAL's income. The second source is a 2.5% levy on sales of rural output, which accounted for 30% of FUNRURAL's receipts in 1974. The beneficiaries of the program are "rural workers" and their dependents, but this category includes rural wage laborers, as well as small proprietors, subsistence farmers, sharecroppers, tenants, etc. Seasonal wage labourers and workers in factories processing rural produce are also entitled to FUNRURAL's benefits. The rural population was estimated at 43.1 million in 1975, and this represents the social group which FUNRURAL is designed to reach.

14. Benefits are of two kinds: transfer payments and medical assistance. Transfer payments are of five types: old age pensions, invalid pensions, widows' and dependents' pensions, funeral aid, and sickness benefits. Transfer payments are by far the most important item of FUNRURAL expenditures, absorbing 75.3% of its total resources. Taking the distribution of both benefits and costs into account, the program of transfer payments has a positive effect on the distribution of income. A comparison between the regional distribution of FUNRURAL's receipts and of its transfer payments demonstrates that the program also brings about a fairly significant inter-regional transfer of income. The program of medical assistance is much less well-established and absorbed only 16.4% of FUNRURAL's budget in 1974 (Cr$631 million). In terms of its coverage, this program encounters - v -

the well-known problems of attempting to introduce urban medical systems, with their bias in favor of advanced technology and their relative lack of mobility, into rural areas. The cost-effectiveness of programs addressed to the real causes of rural ill-health (contagious and debilitating diseases and malnutrition) is lLkely to be much greater than that of conventional medical programs. In this respect, it is worth mentioning the PIASS (Programa de Interiorizacao das Areas de Saude e Saneamento) along with its Proj eto de Medicina Simplificada (or "barefoot doctor" program). The aim of the pilot-program would be to select areas in the North-East and establish small medical posts in villages of between 200 and 500 inhabitants. Nutrition Programs

15. One consequence of Brazil's unequal distribution of income and of the absolutely low levels of income in both urban and rural areas is poor nutrition. In 1974, the Geisel Government took an important step to alleviate nutritional deficiencies by establishing PRONAN. PRONAN is at one and the same time a basic nutrition program and a program for increasing food production. On the side of food distribution, the aims of PRONAN in the pericd 1976-79 are: (1) to provide 1.9 million pregnant and breast-feeding women with 30% of their basic nutritional requirements, as defined by international standards; (2) to provide 4.8 million children in the age group 0-6 with 30% of their basic nutritional requiremests and (3) to expand the existing school meals program to encompass 9.1 million children . The income distribution impact should be favorable, since the program is open tci every family, regardless of income-level, and the program is financed from general Federal budget revenues. The welfare foods program is ambitious, especially given the absence of prs- vious administrative experience of action in this area.

16. The aim of the food production part of PRONAN (PAPP - Programa de Apoio do Pegueno Produtor) is to orient this substantial demand fc r basic foodstuffs towards a particular classof small, independent rural producers; specifically, those operating units of land between 5 and 50 hectares. Using add:Ltional resources (Cr$3.2 billion, at 1975 prices, between 1976 and 197'3), the plan aims to provide the credit, transportation, marketing facilLties, and modern technological inputs with which to equip the medium-to-small sized producer to be the major beneficiary of the plan. It remains to be seen whether the PAPP scheme as present .7 envisaged can be effectively and fully implemented without complemen. ry policies to modify land tenure patterns.

Urban Social Centers

17. In a recent decree, the Government has introduced a plan to set up 600 social centers in urban areas. The aim of these centers will be to compensate for the virtual absence of cultural and leisure amenities on the peripheries of large and medium-sized cities. The establishment of these centers depends on the initiative of state and municipal authorities and a good part of the finance both for construction and the running cost vi - of the centers has to be met from state and municipal resources although each respective ministry (Education, Health, etc.) also make some contribution. In light of the fiscal weakness of state and municipal governments and the known deficiencies of their administration and planning personnel, it is possible that implementation of these interesting plans may be delayed.

Social Projects Fund

18. The Social Support Fund (FAS) is a recent innovation in Government policy and was set up with the aim of creating a line of credit, at lower cost than was available in the financial system, to finance projects of a social nature, both in the public and private sectors. For administrative purposes, FAS is subordinated to the new Conselho de Desenvolvimento Social. In general FAS has a tendency to charge more to private sector borrowers, to favour the interior as against state capitals, charitable as against profit-making institutions, and poor as against rich regions. The present level of applications for the Fund's resources indicates that the demand is far in excess of available resources. SOCIAL WELFARE PROGRAiAS IN BRAZIL

(a) the monthly income supplement (renda mensal vitalicia), equal to half of the highest regional official minimum rate, which can be claimed by any person over 70 years of age; this age group also has recently been given free access to the health achemes administered by the National Institute of Social Welfare (INPS: Instituto Nacional de Previdencia Social);

(b) the Plano Para Pronta Acao recently established by the Ministry of Health, extends emergency health treatment to the entire urban population;

(c) the PIASS program (Programa de Interiorizacao das Areas de Saude e Saneamento) introduced in 1976 to establish a system of health posts and improve sanitary facilities in rural areas and small urban centers in the Northeast;

(d) PRONAN, the national nutrition program to alleviate the nutritional deficiencies of low-income families and especially of pregnant women, nursing mothers and children under 15 years of age. A subsidiary program, PAPP (Pro- grama de A:..io ao Pegueno Produtor) is intended to orien- tate the cdeand for basic foodstuffs generated by PRONAN towards relatively poor rural areas in order to raise the output and incomes of small peasant farmers particularly in the Northeast region. PRONAN has been allocated budgetary resources of US$1.3 billion at 1975 prices for the period 1976-79.

2. These various prci rams reveal the present Government's concern to reduce absolute poverty and its willingness to innovate in order to improve the social conditions of low income groups in both rural and urban areas. Although the following review draws attention to certain shortcomings and deficiencies of current social welfare programs, the evidence of this concern and the new approaches it has prompted are encouraging. The State's Activities

3. The major bealth and social welfare programs in both urban and rural areas are administered by semi-autonomous public agencies whose funding is virtually independent of either the federal, state or municipal budgets: about 70% of the urban labor force has access to the cash benefits and health programs administered by INPS (Instituto Nacional de Previdencia Social) while FUNRURAL provides similar benefits to the rural labor force, although FUNRURAL's health program is still in an embryonic state. The role of the state's activities, financed out of general budgetary revenue, is consequently rather limited. The aim of this section is to-Qutline brieflv the main areas of state action and to quantify the financial cost of these programs.

A. Federal Spending

Social Assistance

4. Federal expenditureson social programs can be divided into two categories: those on "health and sanitation" and those in the general area of "social assistance". "Social assistance" absorbs the largest share and in 1975 it amounted to Cr$11.8 billion out of a total federal budget of about Cr$95.0 billion. Roughly speaking, this total can be subdivided into three elements:

(1) Cr$2.2 billion was transferred to INPS - a sum which is supposed to cover the Institute's administrative costs; this in fact only covers about one-half of administrative expenditure. It was originally planned that INPS should be funded by equal tripartite contributions from employers, employees and the Federal Government, but this arrangement was never put into effect.

(2) Cr$8.2 billion under the heading of "social assistance" consisted of pensions and dependentd allowances paid to retired public servants, both civil and military; the ratio of the total of civilian to military pensions is of the order of 2 to 3. These expenditures represent transfer payments which the Government is obliged to make as an employer to a specific section of the community. The Federal Government has no pension program directed to the community in general;

(3) Cr$1.4 billion under the heading of "social assistance" consisted of government payments to PASEP. As explained above, these are essentially the contributions of the government qua employer to the national "forced savings" plan: since these funds do not consist of current transfers, they can hardly be classed as social welfare expenditure. There are several other programs of social assistance such as Legiao Brasileiro de Assistencia (LBA), Fundacao Abrigo Cristo Redentor (FACR), Fundo Nacional do Bem-Estar do Menor (FUNABEM) and Servico de Assistencia e Seguro Social dos Economarios (SASSE); but total expenditure here is less than Cr$100 million.

To conclude,Federal Gdvernment spending in the area of "social assistance" is for the most part strictly related to the state's activities as an employer; even those transfers made to INPS are substantially less than the level originally envisaged and represented just 4% of INPS's total receipts in 1975. Thus, at the present time, the Federal Governmenit has no general responsibilites in the area of social assistance to the mass of the population.

Health aad Sanitation

5. Total federal spending on "health and sanitation" in 1975 was Cr$2.6 billion of which Cr$2.2 billion went to the Ministry of Health and most of the rest to the Ministry of the Interior in sanitation programs (Cr$388.1 million). "Health and sanitation" expenditure represented therefore less than 3% of total federal spending, although the Geisel Government has substantially increased the resources committed in this area (the Ministry's budget almost doubled between 1974 and 1975). The activities of the Ministry of Health are basically concentrated in the areas of preventive medicine, nutrition and the extension of public sector health programs in rural areas. Federal expenditure in these areas declined in real terms during the late 1960s and early 1970s (See Table 1) leaving the administration of public health in a perilous state. Evidence of a substantial deterioration in public health standards, such as the rising infant mortality rate in Sao Paulo City, the resurgence of transmissible diseases previously thought under control (such as bu- bonic plague and malaria) and the meningitis epidemic in 1974 led to pressures on the Federal Government to put more resources into this area.

6. The ministry has four main programs currently (i) Programa Nacional de Vigilancia Epidemiologica: the aim is to establish a network of laboratories and a central reference laboratory in order to monitor the evolution of epidemic diseases; (ii) Programa Nacional de Controle de Doencas Transmissiveis (Contagious Diseases): recently this program has concentrated on the national meningitis vaccination campaign, involving 82 million people; (iii) Alimentacao e Nutricao: this area will receive a massive injection of funds as a result of the implementation of the national nutrition program, PRONAN; and (iv) Servicos Med- ico-Assistenciais: since the major aim of government policy is to sti- mulate private medical practice, the expansion of public sector health services is mainly to be limited to more remote rural areas which hold out little attraction for private medical practitioners. Table 1. HEALTH EXPENDITURE: THE RELATIVE POSITION OF INPS AND STATE EXPENDITURES IN CURATIVE AND PREVENTIVE MEDICINE, 1965-69

(In current Cr$) million

% changes Federal 2/ in constant Year INPS Government States IPASE-' FUNRURAL- Total Prices

Curative Medicine

1965 179 16 109 24 - 328 - 1966 184 51 193 32 - 460 +3% 1967 649 55 272 45 9 1,030 - 1968 860 82 416 58 3 1,419 +10% 1969 1,494 191 509 72 67 2,333 +42%

Preventive Medicine

1965 - 131 107 - - 238 - 1966 - 181 101 - - 282 -20% 1967 - 213 198 - - 411 - 1968 - 219 215 - - 434 -22% 1969 - 166 256 - - 422 -25%

Total

1965 179 147 216 24 - 566 - 1966 184 232 294 32 - 742 -8% 1967 649 268 470 45 9 1,441 - 1968 860 301 631 58 3 1,853 +0.8% 1969 1,494 357 765 72 67 2,755 +26%

1/ IPASE

2/ FUNRURAL

Source: F. Rezende and D. Mahar, Saude e Previdencia Social (IPEA, Rio, 1974). The activities of the! Fundacao de Servicos de Saude Publica as well as the "barefoot doctor" program are important here. There is a strong argument to be made c,: efficiency grounds for a national structure which integrates the administration of both preventive and curative medicine (this is already foreseen in the plans for the Sistema Nacional de Saude) and an extension of public sector health provision (which runs counter to the Government's present policy). As compared with the health budgett of INPS (Cr$11.3 billion in 1975), the Ministry of Health which has the major responsibility for health programs in rural areas is still obviously under financed.

7. Because of changes which have occurred over time in the clas- sification of expenditures, it is difficult to measure trends in the share of government resources devoted to social welfare programs. However, it can be tentatively suggested that the share of total federal expenditures which went to health, sanitation and social assistance fell between the mid-1960s and 1970 from about 18% to 12%; the greater emphasis on this area by-the Geisel Government can be seen from the fact that the share rose to 15% in 1975.

Conclusions 8. As important as the low value plac d on social expenditures in quantitative terms is the qualitative nature of these programs: the major component consists of transfers to a privileged class of retired public servants. The share of expenditures actually devoted to health and welfare provision to the comunity at large can be put at Cr$4,799 million in 1975 or about 5% of federal spending. In countries with fa±rly progressive tax systems, health and welfare expenditures financed out of the federal budget conastitute a fairly powerful device for redistributing income. The clear challange facing Brazilian social policy at the present time is to finance health and welfare programs in a manner which in fact improves conditions for large numbers of the poor and hence has a greater redistributive effect,.

B. State GovernmeLt ]-xPenditures 9. The budgetary estimates for state government expenditures for 1975 suggest that state expenditures in the area of social welfare are as important as those made by the Federal Government. Expenditure on "health and sanitation" was estimated to be Cr$3.8 billion and on "social security" Cr$6.5 billion, which together represented about 12% of total state expenditures. The planned budget expenditures for 1976 represent an increase of a very high magnitude, "health and sanitation" expenditures riaing bt Cr$9.1 billion,, and those on "social security" to Cr$8.8 billion. State government expenditure in the social welfare field will rise to about 15% of total state expenditures. This dramatic increase can largely iae explained by a large increise-in expenditure In the State of Sgo Paulo. -6°

10. State public health services are more widely available in poor states, and in the metropolitan areas of richer states. Since these services are financed out of general taxation, azy section of the commuity can obtain free treatment through them. Effectively, however, the bospitals and clinics of the state health systi--t_en t-o_buiised mainly by the poorest households - those who are not insured with INPS. Though the services are used by some DIPS members, U1PS has consistently failed to reimburse the state authorities, arguing that since state facilities are financed out of general taxation, everyone in the community has the right of access to them; recently, INPS has changed its policy on this issue. Many state hospitals provide excelent medical care, but there are reports indicating that some lack the necessary funds to provide an acceptable level of service.

El. As at the federal level, the states' expenditures on social assistance largely take the form of pensions paid to retired state public servants; the states' activities here, therefore, are largely limited to the fulfillment of their obligations as employers.

12. While total expenditure by the states is 80-90% of total spending at the federal level, municipal expenditure is less than 30% of total expenditure by the states. The latest available disaggregate data on municipal spending are for the years 1968-71. These data are of little helo because the most important category (general urban services) which may well contain much of the expenditure of interest h here is not disaggregated. Qualitative evidence suggests once again that municipal expenditure on health services is an important element in the provision of health services in rural areas, as weli as in large metropolitan areas.

II. INPS (National Institute for Social Welfare)

13. The INPS (Instituto Nacional Para Previdencia Social) is the centerpiece in the Brazilian urban health and social security system, and an analysis of its origins, growth and current performance is necessary for understanding the entire social services network.

A. Origins

14. The institute was established in 1967, as a semi-autonomous public agency (currently it is subordinated to the Ministerio de Previdencia e Assist4ncia Social: MPAS), as the result of a fusion of several smaller institutes serving the social security and health requirements of specific groups of urban workers. At its inception, OIPS inherited from the individual component institutes the principle that only those who had contributed to the institute could have access to its benefits. There is a fairly obvious contradiction between this principle and the extension of access to social welfare benefits on the basis of need. Since the state's contribution, from its general tax revenue, to the institute's income has always been rather small, as noted above, there was no statutory requirement for the institute to provide social welfare benefita to the urban population as whole; nor, a course, - 7 -

did INPS have the re:sources to do this. In order to get around this contradiction, it has been the aim of government policy since 1967 to increase the percentage of the urban labor force (plus dependents), who are "insured" with the system. Thus, a greater "universalization" of access to the institute's benefits has been achieved, while retaining the fundamental principle of "he who benefits must pay". Very recently, there has been a weakening of this principle, but its continued retention means that certain groups of the urban poor are excluded from the institute's programs.

B. ReceiVts_

15. The major source of INPS revenue (73% of the estimated total for 1976) comes from contributions made by employees, employers and the self-employed. The most important type of contribution (67% of the estimated total for 1976) comes from 16% of earnings (including the thirteenth wage) paid by both employee and employer (8% each) in the corporate sector. Self-employed adherents to the institute pay a "double" contribution-(16% o:f income) and this source represents 3.1% of estimated INPS income for 1976. Enterprises purchasing the services of self-employed workers musf pay 8% of the earnings bill (19% of total INPS income). Certain kinds of institutions are exempt from paying the 8% employer contribution, and so their employees pay just 8% of earnings, the so-called "contriLbuig4o unitaria de segurados", which amounts to 1 .0% of total INPS income. Following a recent decree-law (Law 5859/1972) domestic servants and their employers also each pay 8% of the minimum wage. In many cases, the monetary component of the income of domestic servants is considerably less than the minimum wage, so that 8% of the minimum represents a considerable burden to them; even so, there has been a rapid adherence by domestic servants to the institute and in some cases employers pay both contributions (1.0% of INPS income).

16. The other major sources of receipts come entirely from employer contributions: 4.0% of the wage-bill goes to pay "family allowances" (sa.1rio familia), and forms 12.7% of total INPS income. Then there is a further 3% of the wage-bill which takes the form of insurance for accidents at work (5.9% of INPS receipts), and another 0.3% of the wage-bill which contributes to the payment of a "maternity wage" (0.9% of INPS receipts).

17. A further source of income comes from atransfer from the Federal Government's buidget. As the Brazilian social security system was originally conceived, the state was supposed to bear an equal share of the costs along with employers and employees; the state's budgetary contribution was to comne from certain specially designated taxes. In the early sixties, the Federal Government's role was modified, so that the state was only responsible for covering personnel and administrative costs in the institute and for any deficit in its operations. In fact, however, the Federal Government has consistently failed even to cover fully the personnel and administrative costs of INPS. In 1976, for example, the Federal Government's contribution is estimated at Cr$2.2 billion, whereas administrative costs will be of the order of Cr$5.6 billion. The Federal Government has a nominal debt to INPS, estimated - 8-

in 1971 at Cr$1,7 billion but this sum is not subject to monetary correction and it is doubtfulif the institute could ever call on it. It is important to realize, therefore, that the "subsidy" element from the Federal Budget to INPS is rather small, and has actually been a declining share of INPS receipts (8.7% in 1972 and an estimated 3.4% in 1976). The initial conclusion is that the cost of INPS is largely financed by its beneficiaries: either directly through their contributions or indirectly, to the extent'that employees'-contributions are shifted forward into the prices of goods purchased by the beneficiaries. General tax revenue plays a minor role.

18. A further source of INPS receipts comes from returns on its real and financial assets, but in real terms this has actually been a declining source of receipts in recent years. This has implications for the institute's long-run viability in an actuarial sense, which shall be explored below.

19. Total receipts (both real and nominal) have grown extremely rapidly in recent years (see Table 2), showing greater elasticity even than the Federal Budget's receipts. This has been due to three main factors: (a) the rapid growth of employment in the formal sector since 1970; (b) steady increases in real average earnings and (c) the steady increase in coverage of the urban labor-force (see Table 3). This rapid growth of receipts has meant that, at least in a short-term sense, the Institute has not been short of funds. In fact, the Institute has transformed itself from being a dissaver in the mid-sixties to being a source of net savings in the earlr-seventies. 2/ The existence of these surpluses highlighted a fundamental dilemma facing the Institute between its role as an insurance institute for a specific section of the labor force and its responsibilities, especially in the field of curative medicine, both to its insured members and to the population at large. Should it accumulate financial assets in order to be able to secure worthwhile old-age pensions for its memebers, or should it rely on a constant flow of new resources and spend to the limit of its rapidly growing revenue? Since 1974, INPS seems to have adopted the latter strategy.

20. Finally, it is worth considering the size of INPS receipts in relation to total federal expenditure. Total receipts in 1975 were Cr$48.8 billion, in relation to total federal expenditure in the order of Cr$95 billion (including a transfer of Cr$2.2 billion to INPS from MPAS). Another way of looking at the magnitude of the resources at the disposal of INPS is to see that its receipts represent about 7% of GDP. IMPS, therefore, has considerable potential in terms of the total resources it can devote to transfer payments and provision of health care.

j The major source of savings is the surplus in the "family allowance" account. - 9 -

Table 2: INPS: RECEIPTS" AND THE DISTRIBUTION OF EXPENDITURE BY ACTIVITY

Total % Change Total Transfer Health Receipts in receipts Current Payments Expenditure (Current in Constant Expenditure in Current % Share of in Current % Share of Cr$) Cr$ 2/ 3/ Cr$ 3/ Expenditure Cr$ Expenditure

1965 1.0

1966 1.8 +41.2

1967 2.7 +22.2 2.6 1.5 (52.7) 0.7 (26.9)

1968 4.1 +24.2 4.2 2.6 (62.0) 0.9 (21.2)

1969 5.7 +16.7 5.7 3.3 (58.0) 1.5 (26.2)

1970 7.4 +10.2 8.0 4.6 (57.5) 2.2 (27.5)

1971 10.0 +14.6 10.2 6.0 (58.8) 2.7 (26.5)

1972 13.5 +18.0 13.9 8.5 (61.2) 3.4 (25.8)

1973 21.2 +35.0 19.7 12.0 (61.0) 4.7 (23.8)

1974 31.9 +15.0 28.3 19.3 (68.2) 6.6 (23.3)

1975 50.0 +17.0 47.9 29.7 (62.0) 11.4 (23.8)

1976(est.)64.7 60.7 36.1 (55.8) 18.0 (29.6)

1/ Excluding employers' contribution to family allowances. 2/ Using implicit GDP deflator 1965-72. 3/ Excluding family allowances. Mainly old age, widows and dependants pensions.

Source: Mensario Estatistico do INPS. Relatorio de Atividade do INPS, 1975. - 10 -

Table 3: COVERAGE OF URBAN HEALTH AND SOCIAL SECURITY SYSTEHS

(1) (2) (3) (4) No. of Insured (2) as % of Total No. in INPS, IPASE Including Total Urban Insured Under and SASSE Dependents Population INPS

1960 4,422,470 13,709,657 43.3 n.a.

1967 n.a. n.a. n.a. 5,695,687

1968 n.a. n.a. n.a. 7,763,058

1969 n.a. n.a. n.a. 8,701,157

1970 9,178,855 28,449,490 54.6 8,745,422

1971 10,065,163 31,198,480 57.3 9,640,695

1972 11,328,535 35,115,291 61.6 10,307,834

1973 13,001,618 40,299,006 67.7 n.a.

1974 14,485,059 44,896,021 72.3 14,065,817

1975 n.a. n.a. n.a. 14,600,000 - 11 -

Table 4: INPS: CHARACTERISTICS OF THE INSURED MEMBERS OF INPS

Public Employees in Self- Un- Sector Philanthropic Voluntary Employees Employers Employed Employed Employees Institutions Contributor Total

1967 ------5i695 5,695,687

1968 6,308,615 966,542 266,352 31,059 91,790 55,538 43,162 7,763,058

1969 7,057,895 961,019 390,058 37,659 117,870 92,684 43,972 8,701,157

1970 7,071,745 971,723 407,397 30,894 124,926 87,351 51,386 8,745,422

1971 7,649,848 1,236,675 420,468 32,075 134,135 114,638 52,856 9,640,695

1972 8,001,879 1,306,160 537,435 87,764 151,522 147,110 75,966 10,307,834

1973 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.

1974 11,218,500 n.a. 994,900 n.a. n.a. n.a. n.a. 14,065,817

1975 12,239,500 n.a. 1,074,300 n.a. n.a. n.a. n.a. 14,600,000

Source: INPS: Mensario Estatistico. _ 12 _

C. Coverage __ 21. Between 1967 and 1975, the number of "insured" members in INPS has virtually trebled (see Table 4. Col. 8). This truly phenomenal growth can be explaindd by the rapid growth of urban employment, by the extension of coverage amongst previously ineligible groups (especially amongst "employees" and the self-employed in the "informal" sector) and by the inclusion of newly-eligible groups (domestic servants) The most recent estimate of the size of the employed non-agricultural labor force -- for the last quarter of 1973 -- is put at 21,784,328. It is probably true that by 1975 between 65 and 70% of the urban labor force were insured with INPS. If we consider the total number of insured workers in both INPS, IPASE (Instituto de Previdencia e Assistencia aos Servidores Estaduais) and SASSE (Servico de Assistencia e Servigo Social dos Economiarios) (welfare schemes for civil servants and the very poor) and their dependents, then the percentage of the urban population covered by these systems increased from 55% in 1970 to 72% in 1974 (see Table 3, Col. 2). 22. The important point about coverage is that the percentage of the labor force covered by INPS is much larger than the proportion which has rights either under the Redundancy Payments Scheme (FGTS) or the Social Integration Program (PIS). The latter two systems are confined largely to the "formal" or corporate sector of urban activity. INPS has succeeded in reaching out to include employees in the "informal" sector (such as female domestic outworkers, employees in small firms, high turnover construction workers) and self-employed workers in both "formal" and "informal" sectors. Since INPS offers such employees the only possible access to health services, it seems that they put pressure on employers to make: contributions to INPS, even when the employees do not have an account with FGTS and PIS; alternatively, they pay the contribution of the self-employed person. 23. Even so, fully 30% of the urban labor force is not yet included INPS, and this group is likely to consist of the poorest members of urban society. In addition, households without a regular income earner (e.g., single-parent families) are excluded. Both of these groups will be excluded from the system of cash benefits and health care available under INPS. However, as health care is concerned, such groups can have access to the public sector clinics, funded by the state and municipal authorities. The old and the infirm, if they have not contributed to INPS during their working life, are also excluded from its provi- sions, although INPS is gradually diluting the contributory principle so far as this group of the urban poor is concerned. - 13 -

D. Benefits

General

24. Benefits undier INPS are basically of three kinds: cash benefits, health care and social assistance. On the benefits side, there is an implicit duality between, on the one hand, cash grants (largely old age, widows' and dependents' pensions) and, on the other hand, health care and social assistanco. I hile the administration of cash grants embodies an insurance principle, the existence of health care and social assistance tends to imply an open-ended commitment, without there being any direct relationship between contributions and benefits. The basic problem here is that when a single source of funds finances both aspects, it is not at all clear what; proportion of resources should be devoted to the pensions fund and whsLt proportion to health care. Recently, INPS has been devoting an even smaller share of resources to transfer payments (about 80% in the micl-sixties down to 63% in 1976), while health expenditure has grown very rapidly as a share of total expenditure (17.3% in 1964 to 28i; in 1976) (see Table 2). While it is true that INPS does operate some rules-of-thumb with respect to expenditure shares, it seems that these are purely administrative devices without any long-run operational significance. During a period when resources are growing fast and the "insured" population has a very young age structure, such an allocation does not impose financial constraints on INPS. However, it is not at all clear whether, in the absence of more substantial transfers from the Federal Budget, the INPS will be able both to expand health care facilities ad pay out pensions on the generous scale promised when, in the future, the growth of receipts is decelerating and the population is becoming more aged. There is, then, a very good argument for separating the two functions on the benefit side and introducing separate sources of revenue for each activity.

25. A second argument favoring this course of action lies in the nature of the externalities associated with benefits due to health expenditure. The exi3tence and importance of such externalities means there is a very strong economic case for including everyone in the health care system and for financing its costs from general tax revenue. In Brazil, the very rapid expansion of private curative health care facilities financed by INPS has been accompanied by a stagnation or even a reduction in real terms in expenditure by federal, municipal and state governments on curative and, especially, on preventive medicine; (the main beneficiaries of the curative programs in the public sector tend to be the very poor). The fact that INPS artificially segments the population into those that have access to INPS and those that heve access only to deteriorating public sector facilities reduces the social efficiency of INPS health care expenditures. Further, the lack of any central administration of resources in this area and the fact that the Federal Government hagi left INPS with the major responsibility for urban health, has led to a serious misallocation of expenditure between curative and preventive medicine, as argued in greater detail below. A final point is that a centrally-administered health service, financed - 1 -

from general tax revenue, is likely to be more effective in terms of its redistributive impact than the present system, which,-taking both the burden of costs and the distribution on benefits, is broadly neutral in its distributional impact. 26.- The Government appears to have accepted the logic of these ideas and in policy statements accompanying PND II (1974), two programs were suggested, each covering social insurance and health. For health, the Government has suggested establishing a Fundag7 de Sau'de Nacional to preside over the Sistema Nacional de Medicina Social. The idea appears to be unify the sources of finance from the institutes and to integrate the public sector programs at the federal, state and municipal levels. It seems that the contributory principle in the present system will be retained. In the area of social insurance, the Uovernment plans to establish the Sistema Nacional de Previdencia Social. In both cases, it remains to be seen how the Government will tackle the problem of those presently excluded from existing social welfare schemes.

Cash Benefits 27. Details of the cash benefits available to those insured with INPS are set out in Table 5, along with the value of each benefit and the minimum requirements concerning contributions. If the pensions category is considered, then the most important numerically is the widows' and dependents' allowance, of which more than one million will have been grant5ed by the end of 1976 (see Table 6). The average value of these pensions in 1974 was about 60$% of the maximum regional minimum wage. The nominal value of all outstanding pensions is corrected by the same coefficient as the minimum wage, and it has been suggested that, as a result, pensioners suffered substantial real income losses in the period 1964-74. However, probably as a result of the continued entry of people with new entitlements to pensions, the average real value of dependents' pensions rose by 51% between 1971 and 1974. 28. The total number of all other pensions (old-age, retiremer± invalid and special) is estimated to be 1,578,400 in 1976 (see Table 6). On the whole, these pensions are much more generous than the widows' and dependents' allowance. Retirement pensions in 1974, on the average, were worth 2.8 as much as the minimum wage; old-age pensions were worth 1.3 of the value and invalid pensions (the most common) 0.85 the value of the minimum wage. The average real increase in these pensions was 32.4% between 1971 and 1974.

29. It is possible to make a rough estimate of the age-distribution of pension recipients. If it is assumed that half of the recipients of invalid pensions are over the age of 60, as well as all of these in receipt of old-age or retirement pensions, then we have an upper bound estimate of 815,700 pensions recipients over the age of 60. The 1973 National House- hold Survey estimated the male urban population over the age of 60 at 1,493,965, implying that, using our estimate, about 50% at most of the group receive pensions. -15 -

Table 5: CASH BENEFITS AVAILABLE UNDER INPS

Name Beneficiary Value of Benefit Contribution Requiraemnta

1. Retirement Pension Workers with a Monthly cash payment. 60 months (Aposentadoria por minimum of 30 Women: 95% of average wage (minimum) tempo de Servico) years employment. during last 3 years in employment. Men: 80% of average wage A..r4n& -l nst- I Vparn In employment plus 3i for each year over 30 up to maZ. of 95Z of terminal earnings. Minimum: 90% of local minimum wage. Maximum: 18 x local minimum wage

2. Old-Age Insured workers who Monthly cash payment. 60 months Pension have reached the age 70% of average earnings (minimum (Aposentadoria of 65 for men and 60 during last 3 years in por Velhice) for women. employment plus 1% up to a maximum of 25% for each year covered by insurance contributions or in receipt of some benefit due to incapacity. Minimum: 90% of local minimum wage.

3. Widows' and Dependents (in a Monthly cash payment. 50% of 12 months Dependents' certain order of the pension to which the in- (minimum) Allowance priority) who have sured person would have had a (Pensao por been supported by right or which he was enjoying Morte) the insured person. at time of death plus 10% according to no. of dependents up to max. of 5. Minimum: 60% of local minimum wage. - 16 -

Name Beneficiary Value of Benefit Contribution Requirements

4. Invalid Pension Person considered unable Monthly ca,h payment. 70% of average 12 months min. - can be waived (Aposentadoria to take up any form of earnings during last year in employ- in very serious cases. Not por invalidez) economic activity. ment plus 1% up to a maximum of 30% given if impediment occurred for each year insured. before the insured joined Min. 90% of local minimum wage. social insurance program.

5. Special Pension Insured worker who has 70% of average earnings during last 60 months. (Aposentadoria worked in difficult in- 4 years of employment plus 10% for Insured must have ceased especial) dustrial conditions for each year of activity in which w working to be a beneficiary 15, 20 or 25 years de-- worker was insured up to max. of under this program. pending on sector of 30%. activity. Min. 90% of local minimum wage.

6. In-work Pension For those who are al- 20% of average earnings during last 60 months. (Abono de Per- ready entitled to re- 3 years of employmuent if employed monencia em tirement pension, but already for up to 35 years. 25% Servico) prefer to stay in the if eLrployed for more than 35 years. labour-force. Lasts until retirement.

7. Lump-sum, pay- Paid to insured who is Sum of contributions paid by em- ment (Abono) over 60 (or to his de- ployee, corrected by inflation pendents, who has joined rate. Monetary correction ex- I.N.P.S. after he was 60 cludes contributions made in years of age and has left previous 12 months- employment or who has left a job at which he was working after returning to employment, even though of pen- sionable age. - 17 -

Name Beneficiary Value of Benefit Contribution Requirements ~~~~~~~~I. 8. Monthly income Paid to those over 70 and 50% of Brazil's highest regional Minimal: suppleisent to invalids, who are un- minimum wage. Maximum value of (i) 12 (Renda months of contributions at men- able to work and who do 60% of local minimum wage. any time, sal vitalicia) even if they were not not have any other means continuous or of support. It cannot (ii) the exercise of any activity be added to any other for five years, now included in existing transfer INPS regime, even if not in- payments. sured during that time or (iii) entry into INPS after the age of 6Q

9. Family allow- Paid to employee for each 5% of local minimum wage for No requirements as to minimum num- ances (Salano- child up to 14 years of each child per month; not ber of contributions. Familia) age or who is invalid. paid when worker becomes un- It is paid to any em- employed ployee or self-employed person insured under I.N.P.S., to any employee receiving sickness benefit, to any person on any form of I.N.P.S. retirement benefit.

10. Sickness Monthly sum paid to em- 70% of earnings during pre- 12 months of contributions. Benefit ployee who is sick for vious 12 month plus 1% for (Auxilio- more than 15 days; each year of employment doenca) during first 15 days, covered by insurance con- employer is supposed tributions up to a maximum to pay. of 20%. Minimum value: 75% of lo- calfminimum wage. Name Beneficiary Value oF Benefit Contribution Requirements

11. Maternity Lump sum paid to insured One local minimum wage; payment made 12 months of contribution. Benefit on birth of child, to after eight month. In case of mul- (Auxillo- insured or to insured's tiple births there is one mininmaum Maternidade) wife or companion. paid for each child.

12. Funeral Benefit Paid to insured's depen- Twice the local minimum wage to de- No minimal requirements in (Auxilio- dent or to other person pendent. To other person, cost of terms of contribution. funeral) who arranges funeral. the funeral or twice the minimum wage whichever is least.

13. Prison Bene- Paid to dependents of 50% of the pension tu which insured 12 months of contribution. fit (Auxilio- insured person who is would have had rights at date of reclusao_ imprisoned, imprisonment plus 10% for each dependent up to a maximum of 5. Minimum: 60% of local minimum wage.

14. Accident Payable to employee Value equal to % reduction in capa- No minimum requirements- Benefit following an acci- city x salary on day of accident. (Auxilio- dent at work which Can be summed to other Income and Acidente) reduces capacity counted towards pension rights etc. by more than 25%; payable even if worker returns to work. - 19 -

Name Beneficiary Value of Benefit Contribution Re uirements 15. Sickness Benefit Payable to insured who is 92% of salary on day of accident None. Following Accident at totally or temporarily in- payable immediately. Worker has Work (Auxilio-doenca capacitated from doing any to undertake some form of pro- acidentario) form of work following an fossional rehabilitation other- accident at work, wise he may lose benefit,

16. Lump sum (following Single lump sum paid to Value equal to 20% or 25% of 72 None- accident at work) (i) dependents in case of x the minimum wage, (Peculio accidente do death; (ii) worker who trabalho) has suffered a reduction in work capacity up to 25% and (iii) to worker when ordinary retirement pension exceeds 90% of retirement pension due to work accidents

17. Retirement Pension Payable to insured who is 100% of salary on day of accident None, due to Work-Acci- forced into retirement by + further 25% if permanent assis- dent (aposentadoria work accident, tant required. por ithialidez acidentaria)

18. Dependent Allowance MIonthly allowance payable 100% of salary on day of accident. None, (Pensao to dependents of insured Maintained until dependcnts are no Acidentaria) who has died following longer sick. 6 accident at work# - 20 -

Table 6: INPS CASH BENEFITS PAID OUT BY INPS

Retirement, Average-Y Widows' and Average Maternity Old-Age and Annual Dependent Annual Invalid Value Old-Age Retirement Invalid Relatives' Value Maternity Sickness Pensions (Cr$ 1975) Pensions Pensions Pensions Pensions (Cr$ 1975) Benefit Benefit

1971 988,709 6,610 137,320 308,260 511,089 716,797 2,708 854,167 448,190

1972 1,096,016 8,047 n.a. n.a. n.a. 778,264 3,271 892,009 462,305

1973 1,132,398 8,472 n.a. n.a. n.a. 815,781 3,586 n.a. n.a.

1974 .-1,305,823 8,752 166,421 357,680 688,059 912,836 4,084 969,696 506,551

1/ Annual values of the highest regional minimum wage (13 pohths) was Cr$6,955 in 1975.

Source: INPS: Mensario Estatistico. - 21 -

30. Currently 1.he average benefit is just above the minimum wage and thus bears more-or-less the same relationship to average q nual earnings (between ½i and 1/3) as retirement pensions in Europe and North America. The pension basically is earnings-related, which means that its redistributive impact amongst people of the same age cohort is quite limited. Ihis statement should be slightly modified since there exists a lower limit (90% of the local minimum wage) to old-age, retirement and invalid pensions, and an upper limit (18 local minimum wages) to the retirement pensions. Thus, the major redistributive effect is an inter-generational one. from wage-earners to the old.

31. The recently-introduced monthly income supplement (renda mensal vitalicia) goes some way to remedying the narrow coverage of existing pension rights and addresses itself to the problem of poverty in old-age. The scheme also goes a long way towards introducing the notion of "need" and in diluting the "contributory," principle, which governs other pensions. The income supplment, equal to 50% of the highest regional minimum wage, can be claimed by anyone over 70 who is either infirm or unable to work and without any means of support; it cannot be accumulated with other pension benefits. The requirements are very min±maL and could be satisfied by many old people: either 12 months of contribution at any time to INPS, even though the contributions may not have been continuous, or the exercise for at least five years of any activity currently included under the INPS regime or participation in INPS after the age of 60. Recipients of the monthly income supplement also have access to the INPS health program. During the first year of operation of the program (1975), 352,979 old people began to receive benefits under the scheme, and it is clearly acting as a very important minimum income net for an extremely needy group.

32. The second group of cash benefits -- maternity, sickness, funeral, prison and accident benefit -- absorb far fewer resources than pensions. In 1975, about one million maternity benefits, each equal to one minimum wage, were granted, implying that a great deal less than 50% of all urban births were accompanied by a maternity benefit. Nevertheless, the flat-rate nature of this benefit does mean that it has a moderate redistribution impact, if the benefit side alone is considered.

33. The next most important benefits, numerically, are sickess benefits; these were taken up by 500,000 workers during 1974. The takeup rate of the funeral benefit seems to be very low. In general, these benefits have a minor redistributive impact, and one suspects that, attractive though al:L of the benefits appear on paper, the takeup rates amongst those eligible to receive benefits could be considerably increased. Unfortunately, there are no data available on this aspect. - 22 -

34 INPS is responsible for paying family allowances directly to "insured" workers. The value of the allowance per child, under the age of 14, is equal to 5% of the minimum wage. Disregarding any relationship between family size and household income, the flat-rate nature of familj allowances means that, considering the benefit side alone, these allowances have a very definite equalizing effect on the household distribution of income, especially given the magnitude of the sums involved: an estimated Cr$4.0 billion in 1976.

35. The overall redistributive impact of all of these benefits, taking account of both benefits and the incidence of the contribution burden, needs to be considered. As yet, no one has ever undertaken such an empirical investigation: to begin with, no existing survey on income distribution has taken explicit account of the magnitude and nature of such transfer payments. Most of the benefits outlined above are financed by ioint employer/employee and self-employed contributions; however, family aiowances and accident benefits are financed by employer- only contributions. W4hile employee contributions mainly take the form of direct taxes, it is unlikely that employer contributions are paid largely, if at all, out of profits. In making a tentative judgment on the likely redistributive impact of cash benefits, it must be remembered that the pensions scheme is strictly earnings-related and only slightly modlfied by floors and ceilings on the receipts side. This issue of the incidence of the present system of financing social welfare programs and the distribution of benefits merits further study as the basis for formulating redistributive measures.

Health Care. 36. All workers insured with fIPS and their dependents have access to the health services financed by the Institute. These, together with the public sector services provided by the State and Municipal Governments and the public health program sponsored by the Ministry of Health, are the major public initiatives in this field.

37. In recent years, there has been a rapid growth in the share of INPS ex$pnditures devoted to medical assistance. The reasons for have been (i) the high elasticity of ITPS receipts; (ii) the stagnation and even reduction in real terms in other public sector expenditure on both curative and preventive medicine; and (iii) the rising demand for health care in urban areas. The quantitative indicators of medical assistance furnished through INPS do point to a rapid incrase in availability, particularly since 1974.

38. However, it is clear that, at least between 1970 and 1974, INPS was not equal to the task of providing health care for the vast members of those who were becoming newly eligible to receive these benefits. Between 1970 and 1974, the number of those eligible for care grew by 70%, while the number of in-patient consultations increased by only 35%, and hospital cases by 45%. As the INPS is only too ready to admit, if there was any imbalance between supply and demand in 1970, it was undoubtedly accentuated by 1974. Thus, in spite of a rise in the volume of resources devoted to health care, those already in INPS - 23 -

probably experienced a decline in the quality of their treatment in recent years. 39. Although the INPS is a semi-autonomous public agency, it must be understood from the outset that the provision of medical assistance is not modelled on a Nittonal Health Service of the British or Scandinavian type. In general the Institute's funds are used to incrase the coverage of the system of private medical care. Nevertheless, there are several ways in which care is available under DINPS. 40. First, the INPS operates its own network of outpatient clinics and hospitals. Many of these facilities (especially the hospitals) were inherited from the former institutes which were founded when the guiding philosophy had been that of the'public provision of health - services. They are staffed by doctors who are typically contracted on a part-time basis and by the INPS' own ancillary staff. The T2PS's own out-patient clinics performed 57% of the total number of outpatient consultations financed by INPS during 1975. However, INPS hospitals only account for between 3 and 5% of the total number of hospital beds in Brazil and so, consequently, during 1975, INPS hospitals were responsible for onlyg 4.3% of all the cases of inpatient treatment financed by INPS. The substantial disparity in the relative importance of INPS's own services in these two branches of treatment can possibly be explained by the greater relative profitability and, therefore, atractiveness to the private sector of inpatient care. In any case, present government policy is to freeze the INPS's own hospital network at its present size: no new hospital construction is contemplated. There is substantial excess demand for the services of outpatient clinics largely stemming from_the_great mass of the population who cannot pay private medical fees (even when INPS pays some proportion of them). Excess demand results in inevitable queuing and low quality of treatment. The system of contracting doctors means that the clinics do not have a regular, full-time staff, and the absence of continuously staffed medical teams tends to aggravate even futher the length of queues. Given that, at least in urban areas, there is an excess supply of doctors, INPS could obviously hire more doctors and, since 1974, the Institute has been following a more aggressive hiring policy. Shortages of medical supplies, and of spare parts to repair broken down equipment are common, as are shortages of specialist ancillary staff, such as rediologists. Such oonditions contrast very strongly with those prevailing in private sector clinics.

41. A second war in which INPS finances medical assistance is through contracts (convenios) signed with a variety of entities: trade unions, university mediical schools, enterprises and municipal and state governments. Under such agreements, the INPS provides a block grant to such institutions whLch bears some relationship to the numbers of INPS me'mbers receiving treatment. Convenios with such bodies were responsible for 25% of all outpatient treatment and 6% of all hospital cases financed by WJ?S in 1975. Outpatient consultancies at clinics - 24 -

located in business enterprises are of major importance since such clinics perform 15% of all outpatient treatment.

42. A third way in which INPS finances medical assistance is through its financial support for private sector medicine. There is clearly a very strong lobby, both within the medical profession and MPAS favoring the continuing "privatization" of INPS. The private medical facilities account for only 18% of outpatient consultations, but 90% of all hospital cases financed by INPS. In its short run plan for increasing the availability of medical assistance, the Geisel Government has emphasized the need to contract more services from the private sector and, given the freeze on INPSts own hospital building program, the tendency in favor of incrasing "privatization" is bound to continue. Medical services for private outpatient treatment are contracted in small towns, where INPS does not have its own clinics; however, INPS is also being encouraged to purchase more outpatient services from private group practices in big urban areas. So far as private hospital care is concerned, the payment by INPS is reckoned on the basis of the "unit of service performed". This can be a source of waste and inefficiency in the use of scarce medical resources in so far as it leads the performance of unnecessary medical procedures there are strong indications that much greater control of costs and of clinical procedures by INPS is required in order to prevent abuse.

43. Two recent measures act in the direction of universalizing the availability of medical assistance. The first, outlined in the Plano para Pronta Acao, means that even members of the urban population not covered by INPS can get emergency treatment at INPS's own facilities; INPS will also reimburse private clinics for the costs incurred in this kind of emergency treatment. To some extent, this change was merely official recognitian of existing practice. In 1975, the number of cases - 62,025 out of 5,227,505 cases of hospital treatment - was not very substantial. The second measure means that INPS will finance the health care of the over-seventy's; this is the health counterpart to the supplementary income program. Both measures work in the direction weakening the principle that "only he who contributes can benefit".

Free Pharmaceuticals Program

44. The free pharmaceuticals program (Projeto de Assistencia Farmaceutica) is undoubtedly one of the most audacious innovations in Brazilian social policy in recent years. The program is financed entirely from the resources of INPS and is being administered by INPS in con- juction with CEME (Central de Medicamentos). It began on July 15, 1975, as an experimntal project, covering the insured members of INPS and their dependents in the exclusively urban area of Greater Rio de Janeiro. - 25 -

Of an estimated total population of 7,059,621 in this area, 4,941,700 (or 705Q) are reckoned to be potential beneficiaries of the program. The aim is to provide, free of charge, pharmaceuticals which have been prescribed by doctors working in the INPS's own outpatient clinics. Fifty-foiurof the most frequently prescribed drugs will be available under the program.

45. The organJ1zation of the program is as follows: all the purchasing of drugs will be undertaken by CEME - largely from private pharmaceutical laboratories. Stocking, packaging and the supply of distribution outlets will be undertaken by CDM (Centrais Distribuidoras de Medicamentes); the CDM are also responsible for ensuring that there are adequate stocks throughout the whole system. Prescriptions will be filled by UAFs (Unidade Assistencial Farmaceutica); initially these will consist largely of the INPS's awn dispensaries, though gradually private pharmacies will be integrated into the system: private pharmacies will receive a certain sum for each prescription filled. In the long run, as and when the program is gradually extended to the rest of Brazil, the private pharmacy will be the crucial element in the dispensing of drugs, since INPS does not feel it has the infrastructure to be able to mount such a distribution network.

46. It is beLieved that a large majority of INPS members who do manage to see a doctor in an outpatient clinic do not have the money to purchase the drugs which constitute an integral part of their treatment. Such . patients tend to return to the clinic, and may even need to be hospitalized, in order to obtain the drugs, free of charge, which they cannot afford to purchase.

47. During the! experimental project, this benefit is not means-tested, except in the case of employees of enterprises operating their own clinics under convenios wilth INPS. In such cases, free pharmaceuticals will only be available to those earning less than two (regional) minimum wages. The cost of the ex)erimental program over a six month period was estimated as follows: drugs - Cr$16.0 million; equipment and wages of personnel - Cr$7.4 million; cost of dispensing through private pharmacies - Cr$3.0 million; total - Cr$26 .5 million. Following an evaluation of the experimental project, the INPS plans to extend the program by stages to make it available to all members of INPS throughout Brazil within two years. Since the experimental project covered approximately 10% of the total INPS population, the annual cost of a nationwide scheme, at 1975 prices, would be Cr$0.5 billion. As the program expands, it could be expected to become more selectlve, thereby concentrating resources on thcse in greatest rLeed. It is estimated that 1.,500,000 prescriptions would be filled during the first six months of operation. In fact, during the first 41/2 month period, 1,030,535 prescriptions were filled containing about three items each. Thus, the program seems on target so far. - 26 -

III. FUNRURAL (Rural Worker Assistance Fund)

48. FUNRURAL (Fundo de Assistencia ao Trabalhador Rural) is the agency of the Ministry of Social Welfare in charge of administering social welfare programs in the rural sector. Its activities began in the mid-1960s, but the program was given a massive injection of financial resources in 1971, with the establishment of PRORURAL. In terms of the volume of resources available - between Cr$5b and Cr$b in 1976 (or about 0.7% of GDP) - FUNRURAL must be adjudged a serious attempt to replicate in rural areas the kinds of benefits available in urban areas through INPS.

49. FUNRURAL has two main sources of receipts: the more important source comes from a levy of 2.i4% on the urlan wage-bill. This levy is paid by employers to INPS, which transfers the funds to FUNRURAL. In 197T, this source provided 66% of FUNRURAL's income. The second main source comes from a 2% levy (recently increased to 2.5%) on the sales Of rural output. The levy is paid either by the wholesaler, cooperative or processing firm, which purchases crops from the rural producer or by the producer, when he is selling directly to the cnnsumer or processing the produce himself. It is almost certain that small producers do not contribute. This source provided 30% of FUNRURAL's receipts in 1974. Total receipts have increased very fast; 1972: Cr$2.1 billion; 1973: Cr$2.8 billion; and 1974: Cr$3.5 billion (at 1974 prices) and they are likely to show similar elasticity in the future..

50. The beneficiaries of the programme are "rural workers" and their dependents, but this category includes rural wage labourers, as well as small proprietors, subsistence farmers, sharecroppers, tenants, etc. Seasonal wage labourers and workers in factories processing rural produce are also entitled to FtNRURAL's benefits. The rural population was estimated at 43.1 million in 1975, and this represents the social group which FUNRURAL is designed to reach.

51. Benefits are of two kinds: transfer payments and medical assistance. Transfer payments are of five types. (i) Old age pensions: these can be claimed as of right by the "rural worker" who has completed 65 years of age. Ofily one old-age pension can be granted per family. In order to be a beneficiary, the minimum qualification is to have spent three years (even if not continously) or 12 months continuously as a "rural worker". The value of this old-age pension is 50% of Brazil's highest minimum wage. At the end of 1974, almoa one million people were receiving old-age pensions. Since the total rural population over 65 (both men and wemen) was estimated to be 1.2 million in 1973, FUNRURAL would appear to have achieved nearly universal coverage within three years of the inception of the programmee. (ii) Invalid pensions: these are paid to rural workers who, as a result of illness, find themselves incapacited for work. Currently the pension is worth 50% of Brazil's highest minimum wage; there were 46,017 of these pensions being paid at the end of 1974; (iii) Widows' and dependents' pensions: these are paid to the dependents of a rural worker on his death, whether he was - 27 -

at the retiring age or not. Originally only worth 30% of the highest regional minimum wage, as from January 1974, it now stands at 50-0 There were 77,884 of these benefits being paid at the end of 1974; (iv) Funeral aid: currently worth one months's minimum wage, it is paid to whoever arranges the funeral of a rural worker or his wife; (v) Sickness benefitq: from July 1975, FUNRURATJ began paying sickness benefits in cases where accidents bad been sustained at work. These are currently equal to 75% of Brazil' s highest regional minimum wage. The cost of this has been met by an additional 0.5% levy on rural sales. Retirement pensions mid widow's and dependent's allowances, following an accident at work, wrill be paid at the same rate.

52. Transfer payTlents are by far the most important item of FUMRURAL expenditures, absorbing 75.32% of its total resources. The distributive effect is undoubtedly positive, since the pensions benefit the poorest members of rural society; on the cost side, the benefits are being financed by a rather regressive tax on the mass of urban consumers. Even so, taking the d".stribution of both b-nefits aid costs into account, the program of transfer payments has a positive effect on the distribution of income. In the absence of any significant redistribu'ion of land, the expectation of what are, by rural standards, quite large monetary transfers in old age helps to reduce insecurity and to foster social peace in the countryside. The exiqtence of a minimum income net in old age will become increasingly anpropriate in the future if the mass of rural poor become transformed ini;o landless labourers. A comparison between the regional distribution of FUNPURAL's receipts and of its transfer payments demonstrates that the program brings about a fairly significant interregional transfer of -income. The State of Sgo Paulo contributes 45% of receipts but receives less than 'O% of total benefits, while the contributions benoiit ratio for Guanabara is 100:1. The najor beneficiaries are the pour states.

53. The program of medical assistance is much less well-established and absorbed only r6.1% of FUNRURALD5 budget in 1974 (Cr$631 million). FUNRURAL does not provide any medical assistance directly, but it is gradually making contracts with public (i.#. state or municipal) hospitals, charitable institutions, private clinics and rural unions to provide medical services to rual workers and their dependents. FUNRURAL's financial contribution consists of a monthly subsidy to such institutions, calculated according to the rural population covered by the institutions and independent of the number of consultations which they perform. IENRURAL also provides medical equipment, free of charge, to these institutions. Both of these measures are designed to bring about a long-rurL change in the regional distribution of medical facilities. FUNRURAL also constructs small hospitals which are tlan handed over to local communities to administer, together with a morthly subsidy. In addition, AUNRURAL is equipping a number of mobile clinics to provide a variety c,f preventive and curative assistance in more remote areas. - 28 -

TABLE 7: FUNRURAL: CASH BENEFITS PAID OUT BY FUNRURAL 1972-1974

Old-Age and Average_ An- Average An- AAverage Invalid nual Value Dependents nual Value FUNRURAL Value Year Pensions (Cr$ 1975) Pensions (Cr$ 1975) Benefits (Cr$ 1975)

1972 581.300 2.852 10.085 1.713 10.621 299

1973 454.416 2.956 28.063 1.779 30.874 300

1974 195.448 2.789 45.496 3.034 51.981 464

1/ The annual value (13-months) of the minimum wage in 1975 was Cr$6.955

Source: Mensario Estatistico do FUNRURAL - 2 -9

54. In terms of its coverage, this program encounters the well- known problems of attempting tc intrcduce urban medical systems, with their bias in favor o:' advanced technology and their relative lack of mobility, into rural areas. Althcugh thiere are some aspects of the program which do attempt to deal with these problems, FUNRURAL's lack of penetration so far can be seen from the fact that in 1974 FUNROPAL provided an average of only 0.15 cocnsultations a year for its eligible population, as compared with an average of 1.6 a year made available by INPS. Similarly, only 300,000 out of an estimated number of between 1.5 and 2.0 million rural bit-ths were attended under the FUNPURAL programme in 1974. So far as the type of medical. assistance which is being provided, it is not clear that the emphasis of the present program is entirely correct. The degenerative diseases which are of importance in urban areas (cardio- vascular illness, cancer, etc.) are not soimportant in the rural sector simply because of the much lower average life expectancy. Contagious and debilitating diseases and malnutrition, are the major problems in -ural areas. The coEst-effectiveness of programs addressed to the real causes of rural ill-health is likely to be much greater than conventional medical programs.

55. In this respect, it is worth mentioning the PIASS (Prokrama de Interiorizacgo das Areas de Sau'de e Saneamento) along with its Projeto de Medicina Simplificada (or _"barefoot doctor'' program).__Though still under study, the initial pilot-program would select areas in the North-East and establish small medical posts (unidade simplificador de saude) in villages of between 200 and 500 inhabitants. These posts would be manmed by local paramedics who would have had a few months of basic training; typically the person chosen for the job might be the local midwife. The construction materials for the post would, in most cases, be financed from State funds, although the community would probably undertake the coastruction of the post and establish a small levy to pay the paramedic's salary (say Cr$1 a month); such arrangements could encourage a greater feeling of community independence. The job of the paramedic would be to: (i) make a register of certain target groups, such as pregnant women and young children; (ii) perform vaccinations; (iii) provide treatment for illnesses requiring straightforward treatment and act as a sieving process for cases needing more professional treatment; (iv) spread good sanitary habits amongst the population i.e. teach people how to construct septic tanks; (v) organize small children to act as village healthguardians and to plant useful, fruit-bearing trees. State funds would also be used to provide a public faucet of fresh water. Not only would this be an important sanitation measure, but it has nutritional implications as we]l, since a great deal of women's energy, currently spent in transporting water, would be saved. In the initial programme 1,175 villages will be included, concentrated mainly in Rio Grande do Norte and subsequently spreading out to Para and Ceara. Rio Grande do Norte already has some experience of this approach due to a previous Peace Corps Program. Linked to the "Unidades Simplificadores de Sau'de" will be a "Programa de Unidades de Apoio;" these support units will receive the more serious cases from the "mini-postos." Treatment in these posts could be financed either by FUNRURA.L or INPS. The administration of the program is to be under the Ministry of Health, and the Ministry will operate through the State Health Secretaries, who rill be the main agents in the operation. The initial Federal budgetary contribution amounts to Cr$150 million, demonstrating serious interest in the program. However, in the long-run, the degree of Government commitment to private medicine probably will determine whether more resources will be put into an expansion of public health facilities of this kind.

56. The administration of FUNRURAL presents certain special features which are worth mentioning in relation to a discussion of the program's effectiveness. * ERURAL has a permanent staff which numbers no more than 1500 throughout the whole of Brazil. This means that FUNRURAL's control over expenditure under this very substantial program must be somewhat limited; in fact, there is a legal limit on the share of administrative costs in total expenditure set at 10%. The program is largely being implemented by two types of agents. First, "local representatives," of which there were 845 throughout Brazil in 1974; these representatives are private individuals or firms and are responsible for processing the applications of rural workers who have rights to FUNRURAL's benefits. Local representatives are also supposed to direct people to where theycan get medical treatment. They are paid a fixed sum according to the size of the population they are covering. Second, rural unions (of both employers and employees) also play an important part in informing rural workers of the benefits to which they have access under FUNRURAL.

IV. PRONAN (National Food and Nutrition Program)

57. One consequence of Brazil's unequal distribution of income and of the absolutely low levels of income in both urban and rural areas is poor nutrition. In 1974, the Geisel Government took an important step to alleviate nutritional deficiencies by establishing PRONAN, the Nation- al Food and Nutrition Program. PRONAN is at one and the same time a basic nutrition program and a program for increasing food production.

A. Food Distribution

58. On the side of food distribution, the aims of PRONAN are: (1) to provide 1.9 million pregnant and breast-feeding women by 1979 with 30% of their basic nutritional requirements, as defined by international standards; (2) to provide 4.8 million children in the age group 0-6 with 30% of their basic nutritional requirements by 1979 and (3) to expand the existing school meals program (CNAE - Campanha Nacional de Alimentagco Escolar) to encompass 9.1 million children by 1979. The program originally contained a workers' mid-day meal program; however, this program has been separated from the welfare foods program proper. It now consists of a fiscal incentive whereby enterprises can deduct from taxable profits a sum which is twice their combined expenditure on food and on restaurant facilities. - 31 -

59. The food supplement to be receited by women and young children would consist of a ration of non-perishables such as corn, potato and manioc flour, sugar, soya beans and powdered milk: there would also be some food supplements consisting of certain fortified foods. The budgetary resources to be put into the food distribution program are very' substantial. In the four year period 1976-79, the Federal Government aims to spend Cr$8.7 billion (at 1975 constant prices); this will amount to more than the whole of the budget of the Ministry of Health during this period. The coverage of the program is also quite ambitious: by 1979 about 50% of the women and children comprising the target groups in categories (1) and (2) above will be included, if the plan is fulfilled. So far as its income distribution impact is concerned, the program will be open to every family, regardless of income-level, although better-off families are less likely to take advano tage of it. As regards costs, financing the program from the Federal budget represents a move in the right direction. To conclude, the welfare foods program has ambitious targets, especially given the absence of previous administrative experience of action in this area.

B._ Food Production 60. The aim of the food production part of PRONAN (PAPP - Programa de Apoio ao Pequeno Produtor) is to orient this substantial demand for basic foodstuffs towards a particular class of small, independent rural producers'; specifically, those operating units of land of between 5 and 50 hectares. Top officials in INAN believe this to be the most important aspect of PRONAN. lJsing additional resources (Cr$3.2 billion, at 1975 prices, between 1976 and 1979), the plan aims to provide the credit, transportation, mar]ceting facilities and modern technological inputs with which to equip the nedium-to-small sized producer to be the major beneficiary of the plan. The surpluses supplied by small producers are at present the major source of many basic items of urban food consumption. However, the productivity of such small producers, and hence their ability to increase the markcetable surplus of basic food items, is held back by a pattern of dependent relations in the rural markets for land, credit, labor and matrketing facilities, tying the poor producer to the rich landlord. The aim of the plan is to use state agencies such as Banco do Brasil and Banco Nacional de Credito Cooperativo (credit), EMBRATER (technologr), CFP (minimum prices), COBAL (purchasing and ratailing) and CIBULZEM (warehousing) to break down existing patterns of dependence. Successful implementation of PRONAN would mean a revolution in Brazilian agriculture, achieving a considerable redistribution of production and income without modilying the land henure system. The plan would also mark a major reversal of recent agricultural policy which in ternms of credit and the use made of minimum prices - support policy, has tended to favor the larger commercial farmers. - 32 -

PAPP aims to stregthen the ma.re resourceful and responsive elements among independent small farmers. PAPP is due to start in 1976 in the Agreste tXegion of the North-East, where it is considered that the ideal type of the small, independent, peasant producer envisaged in the plan more nearly corresponds to social reality. Where possible, PAPP wil1 attempt to steer clear of "latifundist" areas, where the- dependence of the small producer on the big landlord is much greater. The pilot program in 1976 will attempt to reach 100,000 producers in 6 municipios in each of 9 North-eastern states, with the aim that these beneficiaries will became community leaders, in terms of best-practice techniques. Following the North-East, the Central-West will be the next region to be included in the program.

61. PAPP is an extremely ambitious program and there are doubts both about the possibilities for effective and full implementation and the conception of the program itself: (l) thie results of previous initiatives designed to reverse the concentration of rural economic power (INCRA, PROTERRA) have been disappointing and it is not clear_ if the Government has either the administrative structure required to strengthen the rural poor; (2) it is not clear how far the plan corresponds to the social and economic realities of the Brazilian country-side in several important respects. First, it is not certain how important a social force the independent small producer really is. Second, the dependency relationships in terms of land-ownership {tenancy arrangements, share- cropping) may be such that too large a proportion of the benefits from improved productivity may accrue to landlords, thus weakening the small proprietor's incentive to modernize and the program's redistributive effects, (3) there is also a transitional problem; the food distribution program has already started: 200,000 recipients by the end of March, 1976 and 1,100,000 by the end of 1976. The food being allocated now is being obtained from existing commercial channels; such a supply network is likely to become structured into longer-term supply arrangements, and large producers and multinational manufacturers of -particular kinds of fortified and processed foods may turn out ot be the major beneficiaries rather than small farmers.

62. One final obstacle to the implementation of both the food distributionand the production program concerns the administrative structure. INAN will be responsible for a program whose ambitions targets and financial resources suggest that it should be the task of a separate Ministry: INAN, however, remains under the Ministry of Health. Both the food distribution and production program require the close cooperation aad coordination of a vast array of government agencies. The State Health Secretaries will be the main agents of the distribution program, utilizing their network of health posts. INPS will, however, be responsible for the program in large metropolitan areas. State Education Secretaries also will be in charge of the extension of the free school meals programs. These agencies will receive funds from INAN to acquire the food from COBAL. The food production side of the equation is: more complicated still: here, the Ministry of Agriciulture - 33 -

(EMBRATER) (TechnoLogy), SUPLAN (planning), CFP. (price support), Banco do Brasil, Banco Nacional de Cr6dito Cooperativo (credit), PROTERRA CIBPAZEM (warehousing) and COBAL (retailing) will all be involved, under the coordination of INAN.

V. CSU (Urban Social Centers)

63. In a recent decree, the Government has introduced a plan to set up 600 social centers in urban areas. The aim of those centers will be to compensate for the virtual absence of cultural and leisure amenities on the peripheries of large and medium-sized cities. The centers will provide five types of amenity, depending on their size: (1) education and culture: libraries, cinema, theatre, music, courses; (2) sports; (3) health and nutrition: sanitary education, immunizatinn, medical and dental consultations, mother and child health care, nutritional education, nutrition supplement; (4) employment and social welfare: profes' sional training, employment agencies, distribution of work permits, social work amongst the old and amongst abandoned children; 'Legal-aid: (5) recreation and leisure: swirming-pools.

64. The estab:lishment of these centers depends on initiatives being taken by state and municipal authorities, and a good part of the finance both for construction and the running cost of the centers has to be met from state and municipal resources. For example, while the plan forsees an expenditure of Cr$2.15 billion (at 1975 constant prices) during the period i1975-1979, only Cr$900 million is to come from the Federal budget; the total of Federal funds plus funds from FNDE (Fundo Nacional de Desenvolvimento da Educagco) cannot exceed 50% of the total construction costs. While each respective ministry (Education, Health, etc.) will contribute some proportion of total running costs, states and municipios will also have to make a contribution. Meanwhile, the construction of centers will only be sanctioned if there is a guarantee of sufficent resources to cover running costs. In the light of the fiscal weakness of state and municipal governments and of the known deficiencies of their administration adid planning personnel, it is possible that these interesting plans may not materialize.

VI. FAS (Social Support Fund)

65. FAS, Fundo do Apoio Social, is a recent innovation in government policy and was set up with the aim of creating a line of credit, at lower cost than was available in the financial system, to finance projects of a social nature, both in the public and private sectors. For administrative purposes, FAS is subordinated to the new Conselho de Desenvolvimento Social. - 3 4-

66. FAS has' several sources of f':nds: out of a total F:nd of Cr$3.0 billion proposed fcr 197T, Cr$1.0 billion Ln tr.sfers from Banco Nacional de Hab4taqgo, Cr$1.0 bIllion from Caixa Lconomica Federal and, at zero cost, Cr$1.0 billion from the Federal Budget (Cr$600 million), the Loteria Esportiva Federal (Cr$200 million) and CEF profits (Cr$200 million). In the first year, however, the Federal contribution was sharply reduced and only Cr$300 million is to come from the Budget in 1976.

67. In terms of its lending policy, FAS is able to practice a degree of interest rate discrimination between borrowers. In general, there is a tendency to charge more to private sector borrowers, to favour the interior as against State capitals, charitable as against profit-making institutions, and moor as against rich regions. There are five different lines of credit: (i) at zero costs; (ii) 61 plus 20% of mcnetary correction; (iii) 6% plus 40% of mmetary correction; (iv) 6% plus -C,< of mcnetar7 co.-,ection and (-v) 6,' plus I02% of monetary correction. Though a detailed distribution is not available. it seems that most loans so far are in categories (iv) and (v). The present level of applications for the Fund's resources indicates that the demand is far in excess of available resources: between June and December, 1975, projects valued at Cr$15.0 bill 4on were presented to FAS. Of the total of Cr$15 billion of projects submitted by 31st Januar-y, 1976, Cr$9.0 billion is from the public sector and Cr$6.0 hillion from the private sector. In the public sector, projects in the field of educatior. (CrS8.0 billion) were more imoortant than in health and, in the private sector, health projects (Cr$4.0 billion) exceeded education projects in importance. To some extent, this distribution reflects Government policy options, especially in health; but one reason why state health secretaries have been slow to submit projects appears to be the lack of qualified personnel in this field. The bigger projects tend to be in education, with the outstanding aase being the Ministry of Education (MC) project for Cr$1.2 billion to complete the University Campus program. Health projects are typically much smaller, with only Cr$800 million out of CrS5.3 billicn for &I health projects worth over Cr$100 million. The number of projects in areas of social policy outside of health and education is very few. Up to the end of February, 1976, a total of Cr$3.9 billion of projects had been approved by the Special 'rorking Group of the FAS. Of these, Cr$3.3 billion were in the field of education, Cr$500 million in health and Cr$100 million for projects submitted by unions. ANNEX III

URBANIZATION IN BRAZIL:

TRENDS, PROBLEMS AND POLICIES

ANNEX III: URBANIZA1'ION IN BRAZIL: TRENDS, PROBLEMS AND POLICIES

Page No.

SUMMARY AND CONCLIJSIONS ...... i-vii

Introduction ...... 1

I. Urban Growth Trends in Historical Perspective, 1870-1970 ...... 2

II. Characteristics and Composition of Recent Trends, 1950-1970 ...... 7

III. Interrelations Between Industrialization and Urbanization, :L950-1970 ...... 25

IV. The Organization, Planning and Finance of Brazilian Cities ...... 42

SJMMHRY ANn 'UNCTLUSIONS

1. Brazil's arban growth in the last zwo decades has beern, on the whole, both rapid and uabalanced. In the 1960s, urban pcpulation increased at an an=atal race of 5.0% (5.5% for cities of 20,000 and over), significantly higher tban the 2.9% rate of overall population increase. An alarming feature of recent urban growth is that the urban population has grown much more rapidly than employment in the industrial sector, whose expansion appears to have triggered it. Since the ability to support added population, directly through jobs as well as in providing the needed irLfrastructure from tax revenues, depends heavily on the industrial sector, this imbalance has become a major concerr to Brazilian planners. The substantial regiona:L imbalances in distribution of urban population and economic activity are a further source of concern. The traditional dominance of the Southeast, especially Sao Paulo and Rio de Janeiro, has increased in this period and has created problems both for these areas and for relatively backrard regions like the Northeast, which have lagged behind in the expansion.

Recent Urban Growth Trends (1950-1970)

2. The overall trend has been toward Lncreasirng growth of the larger cities (250,000 or more) and the newly designrated regional metropolitan areas (Rio de Janeiro, S'o Paulo, Belo Liorizonte, Recife, Salvador, P6rto Alegre, Belem, Curitiba and Fortaleza). In addition, there has been a strong shift in the momentum of growth in the regional metropolitan areas from the central municipalities to the periphery. Except f'o- 'gE' non-metropolitan cities like Brasilia and Campinab, urban growth rates in the small and medium-size cities have been slowing down, on the average.

3. The general pattern of Brazilian urban growth in the last two decades masks some important regional differences. Growth rates have been highest in the Central-Western region whose overall population increase has been inflated by recent migration to new settlement areas and the rapid expansion of Brasilia. The region with the lowest average growth rates is the Nort,neast, with the major exception of--the peripheral areas of the region's three metropolitan areas, Fortaleza, Recife and Salvador. The dominance of the Southeast in urban population has been offset somewhat by the growth of cities in the new settlement areas of the Central-West (and Parana in the South), but not enough to overcome the large imbalance between the Southeast and other regions in the ratio of urban to total population shares. - ii -

4. Brazil's size, regional diversity, and the lack of data make it difficult to sort out the demographic components of recent urbanization trends. In addition to large regional differentials in rates of fertility and mortality for both urban and rural areas there is a rather complex pattern of interregional and rural-urban migration flows. On the basis of information available on interstate migration flows, two sets of net gaining regions emerge: (i) the industrial-metropolitan southeast (Sao Paulo and Rio de Janeiro); and (ii) the new settlement areas (Parana, Mato Grosso, Goia"s, and possibly the Amazon). In the case of Rio de Janeiro and Sao Paulo, migration offset the lower rates of natural increase to give a higher than average rate of total population increase. In the new settlement areas, on the other hand, migration and high rates of natural increase contributed to the higher differential growth. Data on the role of migration in growth of the cities are not readily available. On the basis of scattered estimates, it seems the share of migrants in total population of the larger cities has tended to decline. However, it is important to note the indirect effect of migration on city growth, that is, the effect of migrant natural increase. Migrants have added to the weight of higher fertility groups in large urban populations, and the reproductive force of the migrants is an increasingly important factor in overall urban growth.

Industrialization, Urbanization and Inequality

;. Employment in non-agricultural activities has failed to keep pace with the rapid growth of the urban population. While industrial production expanded at an average annual rate of nearly 8% in 1950-1970, industrial employmett grew at only half that pace. Similarly, industrial employment growth rates are 50% below those of the urban population. Growth in other non-agricultural employment was somewhat higher, but not enough to compensate for the industrial employment differential.

6. A major characteristic of industrial expansion in Brazil is its unequal spatial distribution. Studies show that the Rio-Sao Paulo axis accounted for 57% of Brazilian industrial employment in 1950, and approximately 62% in 1960 and 1970. These figures suggest a leveling off of the concentration process in 1960-70. By 1970, however, the share of Rio and Sao Paulo in total industrial employment still exceeded their share in total population by nearly 40%. Even greater imbalance is observed in the distribution of salaries and value added in the industrial sector. However, the trend for 1960-70 appears to have been toward a slightly better balance between employment and population, especially in smaller cities. There are of course, important regional variations in these patterns. For instance, the share of industrial employment in the Northeast declined even more than its share in population, resulting in a greater imbalance in relative shares in 1970 than in 1960. 7. The corcentration of industrial employment, salaries and value added and the imbalance between these magnitudes and the distribution of urban population raise important questions about the effects of recent industrial growth on employment and income distribution in urban areas. Recent studies of trends in Brazilian income distribution have called attention to the growing inequality in the urban sector, and suggested that the working population did not share equally in the employment and earnings generated by the economic expansion.

Migration and Urban Poverty

8. Migration flows, which contributed heavily to the growth of large city population, have consisted mainly of less skilled workers. Substantial proportions of these have found lower paying work outside the modern industrial sector, in what has been called the informal sector. This sector has p'layed a key role in recent urban labor absorption in Brazil. It functions as a buffer for the excess between labor supply and the demand for labor generated by the modern industrial sector. Indications are, however, that the labor-absorptive fun-tion of the informal sector is not restricted exclusively to migrants, but also extends to younger adults among the native population as well. The data available show no clear ralation between migration and employment in the informal sector. Lower migration rates in the Northeast have not brought lower informal employment, nor have the higher migration rates which characterize the periphery municipalities of metropolitan areas led to inflated informal employment ratios.

9. There is also no clear relation between migrant proportions in the labor force and either the level of unemployment or migrant/non-migrant differences in unemployment rates. Data on unemployment in each of the nine regional metropolitam areas suggest that the regional factors are more important. Unemployment rates are lowest in the South and generally high in the Northeast. W7ith respect to earnings, on the other hand, migrants show a consistently higher proportion of workers earning less than the monthly minimum wage set for each region. Age, however, is also an important factor in this differential.

10. These findings have some bearing on the issue of possible detrimental effects-of recent urban migration on employment and income distribution. They do not support the view that these problems can be attributed specifically to migration. It is true that the demographic impact of migration flows over and above the natural growth of urban areas has augmented their population increase substantially, and that this has resulted in pressures on urban services, job creation, etc. However, examination of differentials between migrants and non-migrants suggests that socio-economic characteristics (particularly age and education) of migrants rather than migration status per se account for most of the observed differentials. In fact, the urban poor of metropolitan areas are not likely to have a disproportionate representation of migrants - iv - among their ranks. Thus, migration is only a "cause" of urban poverty in the sense of its being one of the main factors in the rapid demographic increase of the urban population. It is certainly not a cause in the sense that the marginalized masses in the cities consist entirely of poor migrants who have swarmed in from the countryside.

Urban Infrastructure

11. One of the reasons for the importance of the relation between urban population growth and the distribution of industrial activity stems from the fact that the main source of local revenue is industrial production. Hence, the imbalance between urban population growth and the growth of industrial employment and output has a double effect. Urban population growth creates increased demand for services, and the imbalance makes it more difficult to provide these services with resources available at the local level. Consid- erable imbalance is evident, as a result, in the supply of social infrastructure.

12. Looking at an indicator of financing capacity (industrial value added), there is some narrowing of the regional differences between 1960 and 1970, although the Northeast continued to be about 25% lower than average in 1970. Surprisingly, the peripheral municipalities of the metropolitan regions appear to be relatively well off in terms of financing capacity. -This is in part due to the considerable imbalance, within the peripheral municipalities of each metropolitan region, between the location of industrial establishments and the residences of their workers.

13. The importance of regional factors again must be stressed. Although there are a number of problems common to all the groups and regions, important aspects of these problems are peculiar to each region and to each metropolitan area. The issue of center-periphery balance, for instance, is more important for Rio de Janeiro and Belo Horizonte, *hereas balance among municipalities on the periphery is more of an issue in Sao Paulo.

Municipal Finance

14. The basic unit of urban government in Brazil is the municipality, above which are the state and then federal governments. The metropolitan areas are a new element in the Brazilian system of urban government and still in a very early state of development. Over time, Brazilian federalism has tended to place increasing power in the central government, especially as regards the power to collect taxes. Municipalities can tax only urban property and those services not taxed by state or federal authorities. State taxes include real property transfer taxes and a tax on value added within the state (Imposto sobre Circula.igo de Mercadorias - ICM). ICM revenues depend on the amount of economic activity in the state. Differences in this tax base are at the root of disparities in the relative financial power of different state governments. All other taxes are collected by the federal government. 15. The changes in government structure which took place in the late 1960s tended to put added stress on the capacities of local governments to finance the activities for which they were made responsible. To bolster municipal government finance, an important system of direct revenue transfers from the federal and state governments to the municipalities was introduced. In practice, this system put local governments into a more direct relationship with the federal government than before, and reduced accordingly the relative influence of the state governments. The principal mechanism for the transfer of federal funds to the municipalities is the Municipal Participation Fund. A percentage of federal revenues from income and industrial production taxes (which will rise to 10% by 1980) goes to this Fund.An equal percentage goes to a state revenue sharing fund, which is distributed among the state governments on the basis of state area, population, and per capita income. Of the Municipal Fund, 10% goes to municipalitLes of the state capitals, and the remainder is divided among the remainLng municipalities according to size (with an upper limit of 120,000 inhabitants in 1974, which penalizes non-capital cities larger than this), A similar size criterion applies in the distribution between state capitals. In addition to the general municipal and state participation Lunds, there is also a Special Fund for the development of the poorer states, and funds for sectoral programs like roads, airports, etc., which are financed with revenue from special excises (fuel and electr:Lcity taxes, etc).

16. Parallelling thke system of transfers of federal revenues to states and municipalities is a system for transferring a share of state revenue from the ICM to the municipalities. Each municipality is entitled to 20% of the ICM collected on economic activity carried on within its boundaries. A considlerable controversy has arisen over the disequilibrium in financing of local government because of the way in which the ICM revenue sharing works. The disequilibria are both inter and intra- regional. Because of the trend toward growing concentration of industrial activity in the Southeast, especially Sao Paulo, municipalities in poorer, less industrialized states collect and share in less revenue. Furthermore, tax exemptions to promDte industrialization in these areas reduce even more their revenue from state tax transfers. Other sources of local revenue - the tax on services and real estate - are also based on wealth and income, and yield less in poor regions.

17. A further dimension of the municipal finance problem is the disequilibrium that has arisen within states, particularly in the metropolitan municipalities. Industrial activities are not spread evenly among the municipalities that comprise the metropolitan areas. Thus the shares in ICM revenues vary significantly between predominantly industrial municipalities and adjacent "dormitory" communities within the same metropolitan area. The size limit on shares in the federal revenue fund likewise works against the larger and more rapidly growing of these municipalities and, when the population consists mainly of low income groups, the capacity to generate revenues from munlicipal taxes is also limited. These municipalities are thus endowed with very poor financial capacity to deal with the problems of "suburban" povertyr. - vi -

Urban Policy

18. During most of the post-war industrialization period, urban problems were given low priority relative to maintaining the overall level of economic expansion. The functional ministries and their subsidiary agencies have generally implemented their particular sectoral strategies with little or no concern for the overall effects on the evaluation of the urban system. The 1970s have, however, brought growing recognition of the detrimental effects on the urban system of viewing urban policy as a relatively exogenous by-product of industrial and sector development strategies. A conscious effort to develop an urban policy began with the creation of a national commission on metropolitan areas and urban policy and the preparation of the II Plano Nacional de Desenvolvimento (II PND).

19. The new commission on urban policy (CNPU) was established in June 1974 to oversee the implantation of Brazil's new system of metropolitan regions, formulate a national urban policy and insure coordinated implementation of urban programs and projects by the various government agencies involved in the execution of this policy. Resources for implementation of the urban strategy derive from various funds at the disposal of ministries that make up the CPNU. In addition, two special funds were established late in 1975 specifically for projects making up the new urban strategy. The first is a national urban development fund (FNDU), designed to provide some financial leverage on the allocation of money in other funds. The second is a fund for urban transport systems (FDTU), linked to a new Brazilian Urban Transport Company (EBTU). Finally, the possibility of achieving a major change in the management of large cities in Brazil was created in 1973-74 by the establishment of nine metropolitan regions. Each region consists of a group of municipalities selected on the basis of economic and demographic affinity to the main metropolitan municipality, and is endowed with metropolitan government institutions responsible for the elaboration of integrated development plans and the coordination of programs relating to common services. The latter include economic and social planning; water, sewerage and sanitation; urban land use; transport and highways; production and distribution of piped natural gas; water resources and pollution.

20. The main framework of Brazilian urban development policy is elaborated in the II Plano Nacional de Desenvolvimento (1975-79), which was published in 1974. Its main policy objective is the establishment of a better structured urban system. The plan recommends: i)-setting up the metropolitan areas and identifying the functions for metropolitan area government; ii) defining secondary nuclei that would help to reverse the tendency toward concentration of urban growth; iii) achieving a reversal of the concentration that has occurred in Greater Rio and Sao Paulo; iv) revitalizing the economic base of the metropolitan areas in the North- east; and v) promoting urbanization in the new settlement areas in the Northern and Central-Western regions.

Conclusion

21. Brazilian urban policy is still at a very formative stage. What has been done represents an important step in the direction of rationalizing the development of Brazil's urban system. At the same time, important - vii - institutional limitations stand in the way of effective implementation. The statutory power of the urban planning organs is limited. In addition, budget decisions and execution of projects affecting urban areas are still very much in the hands of the federal ministries and their subsidiary agencies. The position of coordination, either at the federal or state level, is bureaucratically weak. The CPNU was established to overcome this fragmentation, but it is still not clear how effective this mechanism will be for putting order into the system. Finally, the new urban policy is characterized by a high degree of centralization, and it remains to be seen whether this approach will be successful in reaching solutions at the municipal level, where the urban problems have their roots.

URBANIZArION IN BRAZIL: TRENDS, PROBLEMS AND POLICIES

1. One of the most striking characteristics of post-war industrialization in Brazil is the rapid growth of urban population that accompanied it. The decade 1950-1960 was that of Brazil's most rapid rate of overall population growth, over 3% per annum, and one half of this growth was absorbed by cities of 20,000 or more inhabitants. These cities accounted for only a quarter of the population in 1950. Indeed, on the eve of the recent industrialization period, Brazil's urban population proportion (as measured by an international standard definition of 20,000 or more inhabitants) was low relative to both the world and Latin American average (see Table 1). From a 1940 level of only 16%, it has increased to nearly 40% in 1970. As officialy defined,the urban population passed the 50% mark in the 1960s (56% in 1970) and officials are predicting that it will reach two-thirds by 1980. The recent urban population growth rate (5.5% for cities of 20,000 and over) was 80% above the high rate of overall population increase.

2. The urban population has groun much more rapidly than employment in the industrial sector, whose expansion appears to have triggered it. Since the ability tD support added population, directly through jobs as well as in providing the needed infrastructure from tax revenues, depends heavily on the industrial sector, this imbalance has become a major concern to Brazilian planners. Of further concern is the substantial regional imbalance in distribution of urban population and economic activity. The traditional dominance of the Southeast, especially Sao Paulo and Rio de Janeiro, has increased in this period and has created problems both for these areas and for relatively backward regions like the Northeaat that have lagged behind the expansion.

3. This report is a summary of recent urban growth trends in Brazil and of efforts in urban planning and policy formulation that have evolved in response to problems created by these trends. The report is divided into six parts. Part I La a brief review of Brazilian urban growth in longer term (1870-1970) perspective, drawing contrasts between Brazilian and other Latin American experience. Part II focuses in greater detail on the patterns of urban growth between 1950 and 1970, giving special attention to regional and city size differentials in this pattern. Basic projections of the urban population for 1970-1990 are also discussed. Part III addresses interrelations between urban growth and economic changes in the post-war period, with special attention to the problems of balance between industrial employment and urban growth and to the problem of creating urban infrastructure to match the increase in population. Part IV examines t:he structure of urban government and finance, the urban policy outlined in the recent National Development Plan, and institutional limitations on urban policy implementation deriving from relations between planning, financing, and executing agents in the -2-

government structure.

4. The scope of the report is very broad in the sense that it touches on a wide range of questions relating to urban growth in Brazil, and at the same time it is very limited in that it barely touches upon a number of issues that deserve comsiderable attention. Some of these are the subject of other reports being prepared in conjunction with the recent World Bank Mission to Brazil, and more detailed discussions are found in those reports. This is particularly true of the questions of urban employment and income distribution, which are stressed as being central to the urban problem but not analyzed in great detail in this report.

I. Urban Growth Trends in Historical Perspective, 1870-1970 a) Urban Growth Prior to 1870

5. The historical roots of Brazil's urban system contrast with those of Spanish speaking Latin America. In the early colonial period of the 16th and 17th centuries the population resided in a string of isolated coastal settlements, military outposts and their environs, which had closer links to Portugal than with each other. There was no highly developed pre- existing Indian civilization such as the Aztec and Inca cultures, whose networks of cities and roads were taken over by the Spanish. Not until the gold and diamond boom of thel8th century did Portuguese settlement push into the vast interior regions, and the growth of Brazilian cities of this period was patterned on the rise and decline of the extraction of precious minerals for export.

6. Toward the close of the 18th century the mining boom slackened and in the early 19th century urban growth dropped to insignificant levels, except for Rio de Janeiro and Recife (see Conniff et al.). Rio gained stature through the transfer of the Portuguese Royal Court to Brazil during the Napoleonic wars and opening of world trade to Brazil. Rio then supplanted the northeastern city of Salvador, which had been the major city and capital during most of the colonial period. Recife gained through the revival of sugar exports from 1830 to the mid-1850s, which attracted both internal and European migrants. From 1850 to 1880, however, urban growth declined even in these centers, and was low (generally below 2% per year) in the remaining cities of the country, a trend consistent with the relatively low pace of economic progress in this period. - 3 -

Table 1: COMPA]RATIVE DAkTA ON PERCENTAGES OF POPULATION LIVING IN CITIES

OF 20,000 OR MORE :N LATIN AMERICA AND OrTER REGIONS, 1900-1970

1900 1920 1940 1950 1960 1970

Latin America Argentina 24 37 41 52 58 65 Brazil 10 13 16 21 29 39 Chile 19 28 37 38 50 53 Colombia 8 7 14 21 30 43 Cuba 25 25 30 35 42 48 Mexico 9 11 18 25 32 41. Peru 6 6 13 18 26 33 Venezuela 8 11 19 31 42 56

Regional Averages Latin America n.a. 14 20 25 33 n.a. North America n.a. 41 46 51 58 n.a. Europe n.a. 35 40 41 44 n.a. World 9 14 19 21 25 n.a.

Sources: 1900, Latin American Countries, Sanchez-Albornoz, p. 178; Regional Averages, United Nations, Determinants and Consequences, p. 188; All 1920-1940 figu!res and regional averages for 1950-1960, United Nations, World Urbanization, p. 106; Latin America, 1950-1970, CELADE-PISPAL, "Consideraciones," pp. 32-34. -4-

b) The Export Boom, 1870-1920

7. Only in the last quarter of the 19th centurv. when Brazil was abruptly changed by export specialization (in rubber and coffee) and immigration, did urban growth clearly affect the predominantly rural character of the country. Only 8% of the 1872 population resided in localities having 20,000 or more inhabitants, which consisted of the coastal state capitals and Sao Paulo. Only three cities (Rio de Janeiro, Salvador, and Recife) had populations of over 100,000 and none exceeded 500,000. The city which emerged in this period was Sao Paulo. It grew from 65,000 in 1890 to 240,000 in 1900, a growth rate of over 14% per annum that nearly quadrupled its population in the decade, largely reflecting the growing importance of the interior of that state in an agricultural production especially coffee. In the same period, the rubber port of Beldm in the north doubled its population. Migrants to the rubber area were attracted almost exclusively from the Northeast. Cities in the Northeastern region, in contrast, declined in relative importance in the period because of stagnation of the area's sugar and cotton economy.

8. This period was one in which many Latin American countries experienced a strong increase in the "primacy" of their urban system, i.e., a rapid increase in the growth of the main (usually the capital) city relative to the remaining cities. Comparative analysis of this feature in the Brazilian system is complicated by the extraordinary growth of the city of Sao Paulo. One of the ironies of the immigrant-subsidization policies of the imperial and republican governments of the late 19th and early 20th centuries which sought to built up the agricultural labor force needed for production of coffee for export, is that they contributed as much or more to the growth of the country's burgeoning urban labor force as to the rural labor force. Intra-regional migration streams at the turn of the century in the Center-South were directed to the city of Sao Paulo. Both the foreign born and many natives who had been displaced in other areas by the foreign born made up these flows. Thus the main characteristic of urban growth in this period of Brazilian history is that it originated in the export boom, and was fed both directly and indirectly by international migration.

9. The regional allignment of Brazil's export economy introduced a modest primacy into its urban growth pattern. Throughout most of the 19th century the population living in the 4-5 largest cities (Rio, Salvador, Recife, Belem, and Sao Paulo) stayed very close to 6% of the national population. By 1900 this increased to 8%, with Sao Paulo and Belem contributing to the increase. From 1872 through 1890, Rio de Janeiro was approximately eight times the size of Sao Paulo. In 1872 there were seven cities in Brazil larger than Sao Paulo. By 1900, however, Sac Paulo had risen to second place and closed the gap to the point that Rio was only three times as large. Together Rio and Sao Paulo doubled their relative share in the total population from 1872 to 1900. By the mid-twentieth century, they were roughly equal and since 1960 Sao Paulo has pulled ahead. -5-

10. It would appear reasonable to characterize the pattern of urban growth which predominated in the last 30 years of the 19th century as one with features of growing "primacy," in that the two or three largest cities (which were major export centers) grew much more rapidly than the remaining urban localities in Brazil. Brazil's primacy pattern, never very substantial, probably reached its peak in this period. In contrast to countries like Argentina., Chile and Cuba, Brazil's urban growth pattern since that time has been. considerably less characterized by primacy.

c) The Industrializaticn Phase, 1920-1970

11. Though coffee exports continued to be fundamental to the Brazi- lian economy until well into the post World War II era, the export boom also laid the foundation for Brazilian industrialization, which can be dated from the First World War period. The tendency toward primacy that had characterized the export growth phase did not persist--at least in terms of population size. From 1900 to 1970 Rio and Sao Paulo increased their percentage share of total population by a multiple of 1.8, while the smaller city-size thresholds (which by definition also include Rio and Sao Paulo) increased their shares by multiples of 3.6 (for the measure 100,000+) and 3.9 (for 20,000+). A more satisfactory indication of declining primacy can be seen from the fact that while Rio and Sio Paulo represented approximately 38% of the total urban population in 1872 and 61% in 1900, by 1940 this dropped to 47% and in 1970 it had fallen to 28%. The more recent observation over- states the decline, however, because these percentages omit the rapidly growing suburbs of these two metropolitan regions. Such an omission is really important only after 1950, when suburban growth started to overtake that of the central cities. Prior to that, the central municipalities were dominant. However, the declining primacy trend is evident even if account is taken for suburban growth.

12. A further way of characterizing the longer run evolution of the Brazilian urban system is in terms of its stability, i.e., the relative positions of various cities in the urban hierarchy over time. From 1872 to 1920, there was some destabilization of the city system growing out of the impact of expanded foreign trade. The major change was Sao Paulo moving from tenth to second pla.cel/. Looking at the top 20 state capitals in 1920 and 1970, and ignoring for the moment the entrance of new cities into this group, the ranking appears even more stable. The major changes are the movement of Natal from 1.8th to 13th place, and Belo Horizonte from 13th to third place during this 50th year period. At first glance, it appears that the city system has been rather stable through time. If however, the planned capital cities of Goiania and Brasllia are included in the analysis, the results change substantially, these two cities did not exist in 1920 but were ranked eleventh and tenth respectively in 1970. If, in addition, the emergence of the non-capital industrial cities of Santos and Campinas in the top 20th by 1970 is taken into account, the stability appears even lower. Thus, Brazil presents a relatively stable rank order

1/ The Shearman rank correlation coefficient of the top 19th state capitals is .77 for the period 1872-1920. 6

Table 2: LONG RUN TRENDS OF URBAN GROWTH THROUGH SELECTED MEASURES OF

URBAN POPULATION--BRAZIL, 1872-1970

Percent of Total Population % of City Population Official Cities Cities Cities Municipios (20,000+) Definition of of of of Rio and Rio & Sao Paulo Date of Urban 20,000+ 100,000+ 500,000+ Sao Paulo Center Metro Area

(1) (2) (3) (4) (5) (6) (7)

1872 n.a. 7.9 5.1 0.0 3.0 38.4 n.a. 1900 n.a. 10.0 8.1 4.0 6.0 60.6 n.a. 1920 n.a. 13.0 9.7 6.3 6.3 52.8 64.1 1940 31.2 16.0 12.8 7.7 7.5 47.1 57.3 1950 36.2 21.1 15.6 11.1 8.8 41.8 53.0 1960 45.1 28.8 21.6 16.2 10.2 34.9 47.1 1970 56.1 38.8 29.5 23.6 11.0 28.1 41.0

Source: Data from 1872 through 1920 derived from census tabulations in The Urban Development of Latin America, 1750-1920 edited by Richard M. Morse, Center for Latin American Studies, Stanford University, 1971, p.37. Data for 1940 through 1970 available through official census tabulations reproduced in "Urbanization Trends in Brazil, 1940-1970" by George Martine and Cesar Pelaez (mimeo 1971) Rio de Janeiro. Data for Rio and Sao Paulo from Anuario Estatlstico, various years. -7-

of cities in the 20th century only if the analysis, is restricted to the change in the rank order of the 20 state capitals (and Rio de Janeiro). If new cities (Brasilia and Goignia) and smaller cities which grew to substantial size by 1970 (Santos and Campinas) are included,the Brazilian city system appears much less stable over time. This pattern has been conditioned more by the growth of the frontier and of domestic markets for industrial products and commerce than through the impact of foreign trade. The influence of the latter on the pattern of urban growth declined after 1900.

II. Charactieristics and Composition of Recent Trends, 1950-1970

a) The Data and Their Limitations

13. Data are more readily available for analysis of the spatial and size characteristics of Brazil's urban structure for the period after 1940. However, care must be taken in interpreting these data and to define exactly what is being measured, since definitional and compositional problems persist. The official definition of the "urban population" is based on administrative status (a county seat, "sede de municipio", is an urban place) rather than a minimum population size. This results in a higher proportion reported as urban than would be obtained using a more conventional minimum size criterion, such as the 20,000 and over base found in many United Nations publications. Where possible,this report will follow the minimum size (20,000 and over) definition rather than the official administrative criterion. Since rural/urban breakdowns of data in Brazilian census publications follow the official definitions, some tables (as indicated) will necessarily follow the official definition.

14. However defined, urban growth can result both from increases in the population of places defined as urban at the beginning of the measurement interval and from the addition of new cities that passed the urban threshold in the interval. For most analytical purposes it is the first aspect of urban growth that is important, and this can be controlled by specifying the places classified as urban as of a particular census date. A similar problem &rises ia the analysis of urban growth by city size class due to shifts of individua,l cities from one class to the next as their population increases. Againithe analysis can be focused on increases in population for cities of specific size by specifying the size distribution as of one census date. This analysis will be focused on Brazilian cities of 20,000 and over and sizea classes as of the 1970 census.

15. The increase in the size of particular cities through expansion into adjacent localities is an additional aspect of the measurement of urban growth trends requiring careful refinement of published data. In Brazil, this has become the major characteristic of growth trends in the larger -8- Table 3: GROWTH OF BRAZILIAN URBAN POPULATION,

1950-1970 BY SIZE CATEGORY

Population in Thousands Growth Rates Size Group 1950 1960 1970 1950/60 1960/70 # of Cities

Rio and Sa-o Paulo (a) Municlpios of Capitals 4,320 6,472 9,557 4.1 4.0 ( 2) (b) Periphery 1,060 2,052 5,128 6.8 9.6 (c) Total 5,380 8,524 14,685 4.7 5.6

Regional Metropolitan AreasL/ (a) Municfpios of Capitals 2,183 3,740 5,702 5.5 4.3 ( 7) (b) Periphery 391 726 2,056 6.4 11.0 (c) Total 2,574 4,466 7,758 5.7 5.7

Remaining Cities of 250,000 or more (in 1970)/2 925 1,717 3,286 6.4 6.7 ( 9)

Cities of 100,000 to 249,000 (in 1970) 1,467 2,610 4,422 5.9 5.4 ( 30)

Cities of 50,000 to 99,000 (in 1970) 1,020 1,810 2,954 5.9 5.0 ( 44)

Cities of 20,000 to 49,000 (in 1970) 1,623 2,903 4,706 6.0 5.0 ( 154)

Subtotal (a) 12,989 22,030 37,811 5.4 5.6 ( 246)

Remainder of Urban (Official Definition) 5,793 9,796 14,274 5.4 3.8 (3,588)

Subtotal (b) 18,782 31,826 52,085 5.4 5.0 (3,952)

Rural (Official Definition) 33,162 38,293 41,054 1.4 0.7

Total Populatio..L3 51.944 70,021 93.139 3.0 2.9

/1 Belo Horizonte, Recife, Porto Alegre, Salvador, Curitiba, Belem & Fortaleza.

/2 Brasilia, Santos, Vitoria, Joao Pessoa, Natal, Campinas, Manaus, Pelotas (RS), Goiania.

/3 1960 urban population total corrected to reflect populaiao residente, as in 1950 & 1970,

Sources: Data compiled in Fox, Urban Population Growth in Latin America, Tables 27 & 28, and Fundacso IBGE, 1970 Population Census. -9- cities, nine of which are now officially defined as metropolitan areas: Sao Paulo, Rio de Janeiro, Belo Horizonte, Recife, Salvador, Porto Alegra, Fortaleza, Belem and Curitiba. The metropolitan areas consist of the central municipality of each metropolitan area plus a number of surrounding countries that have become absorbed in the metropolitan expansion. Data in census publications are still published by municipality (so that Santo Andre, a suburb of Sao Paulo, appears as the 11th largest Brazilian "city" in 1970), and recomposition of the data for such agglomerations is required to provide an accurate portrayal of the changing urban structure in recent decades. This becomes progressively more difficult (though statistically less significant) as one moves back in time because of frequent changes in boundaries. Fortunately, the major task of sifting through municipality data to reconstruct the metropolitan areas and other important agglomerations from the 1950, 1960, and 1970 censuses has been performed by Robert W. Fox, whose compilation of data on the population of Brazilian cities (20,000 or more) appears in Urban Population Growth in Latin America. Fox's definitions of the major metropolitan areas differ slightly from the official list since he followed the one originally proposed by the Fundacao IBGE rather than that finally adopted in the new metropolitan area legislation. He also defined other lesser sized agglomerations that have not been adopted in official publications. The definition of urban place in hi.s data refers i:o agglomerations of 20,000 or more in 1970. This definition includes :Ln the list of urban places a number of county seats which would not have qualified had they not been part of an agglomera- tion. b) Distribution of Urban Growth by City Size Category

16. The main features of urban growth in the period 1950-70 are as follows:

(i) in the 50s, the rate of growth of population either as officially defined urban areas or in cities of 20,000 or more was 5.4% per annum, compared to 1.4% for the rural population and 3.0% for the total population. Comparing the growth ratesin the 60s to those in the 50s indicates that total population growth and both urban and rural growth declined. However, the growth rate for the urban population defined in terms of cities of 2:0,000 and over increased in the 60s. Small town growth (officially defined urban areas with less than 20,000 inhabitants in 1970) delcined substantially from the 50s to the 60s. This was part of a broad trend wich seemed to indicate declining growth rates in smaller cities (the 20,000-49,000 group and the 50,000-99,000 group) in the same period. However, this was in part a result of the way in which cities were grouped by size categories. A number of smaller localities which would have increased the growth rate of these categories were located in the peripheries of the metropolitan areas, and have been counted in the metropolitan area categories. The cities which remain in the 20,000 to 100,000 groups are those located outside of the major agglomerations. - 10 -

(ii) The most rapidly growing category in both decades was the larger non-metropolitan cities in the 250,000 and over class. Cities belonging to this group were capital cities along the coast (Natal, J6ao Pe^ssoa and Vitoria), Santos and Campinas in Sao Paulo, Pelotas in Rio Grande do Sul, and three rapidly growing interior cities, Brasllia, Goiania, and Manaus. The very high growth rates recorded for this group derive principally from the last three, suggesting that the regional dimension may be as important as size in the overall growth pattern.

(iii) The next most rapidly growing cities were the regional metropolitan areas, which recorded an average rate of 5.7% per annum in both decades. As in the national metropolitan areas, there has been a strong shift in the momentum of growth from the central municipalities to the periphery in the regional metropolitan areas. While the central municipality average declined from 5.5% to 4.3% between the 50s and 60s, the periphery increased from 6.4% to 11%.

(iv) In Rio de Janeiro and Sao Paulo,growth of the central municipalities averaged 4.0%, while growth of the peripheral areas increased from 6.8% to 9.6% and raised the overall growth rates from 4.7%ito 5.6% between the two decades.

(v) The overall trend from the 50s to the 60s has been toward increasing growth of the larger cities (250,000 or more, including the metropolitan areas). This has reversed the tendency toward a decline in the weight of the larger cities in the urban population that occurred between 1950 and 1960. All three categories of cities in the 20,000 to 249,000 groups increased their share from 1950 to 1960, and then fell back in 1970. The bulk of the increase in the urban share in the total population has also shifted to the larger cities. Of the 6.4 percentage point increase in the share of population in cities of more than 20,000 in 1950-1960, 3.9 (61%) went to larger cities. In 1960-70, 72% of the increase went to cities of over 250,000. - 11 - Table 4: PERLCENTAGE DISTRIBUTION OF URBAN POPULATION

BY CITY SIZE CATEGORIES, 1950-1970

1950 1960 1970 Percentage of Percentage of Percentage Group 20,000+ Total 20,000+ Total 20,000+ Total (a) (b) (a) (b) (a) (b)

Rio and Sao Paulo (a) Munici(pios of Capitals 33.2 8.4 29.4 9.2 25.2 10.3 (b) Periphery 8.2 2.0 9.3 2.9 13.6 5.5 (c) Total 41.4 10.4 38.7 12.1 38.8 15.8

Regional Metropolitan Areas (a) Municfpios of Capitals 16.8 4.3 17.0 5.3 15.1 6.1 (b) Periphery 3.0 0.6 3.3 1.0 5.4 2.2 (c) Total 19.8 4.9 20.3 6.3 20.5 8.3

Remaining Cities of 250,000 or more (in 1970) 7.1 1.8 7.8 2.4 8.7 3.5

Cities of 100,000 to 249,000 (in 1970) 11.3 2.8 11.8 3.7 11.7 4.7

Cities of 50,000 to 99,000 (in 1970) 7.9 2.0 8.2 2.6 7.8 3.2

Cities of 20,000 to 49,000 (in 1970) 12.5 3.1 13.2 4.1 12.5 5.1

Subtotal 100.0 25.0 100.0 31.4 100.0 40.6

Remainder of Urban (Official Definition) 11.2 14.0 15.3

Subtotal 36.2 45.4 55.9

Rural (Official Definition) 63.8 54.6 44.1

Total Population 100.0 100.0 100.0

Sources: Data compiled in Fox, Urban Population Growth in Latin America, Tables 27 & 28, and Fundacio IBGE, 1970 Population Census. - 12 -

c) Regional Differentials in Urban Growth

17. The general pattern of Brazilian urban growth in the last two decades masks some important regional differences. Growth rates have been highest in the Central-Western region (and to a somewhat lesser extent the Northern region and the South, because of Parana) whose overall population increase has been inflated by recent migration to new settlement areas. The cities in these areas, though few in number, have grown very rapidly since 1950. The Northern region's single metropolitan area, Belem, follows the pattern of the other regional metropolitan areas, with its increased growth deriving from the peripheral municipalities.

18. The region with the lowest average growth rates is the Northeast. The major exception in this region is the very rapid growth (7.7% in 1950- 1960 and 9.1% in' 1960-70) of the peripheral areas of its three metropolitan areas: Fortaleza, Recife, and Salvador. Growth rates for the remaining categories have been around 5%, with evidence of decline in rates for cities of less than 250,000.

19. Average rates for the Southeastern and Southern regions fall in between. Smaller (20-99,000) cities' growth rates declined from the 50s to the 60s and the medium (110-249,000) also declined, specially in the Southeast. The larger cities in the Southeast (Santos, Campinas and Vito'ria) all grew at about average rates (5.6%), whereas Pelotas, in the South, was much slower.

20. These regional differentials in growth patterns have led to a moderate regional redistribution of the urban population. The Southeast, with around 43% of the national population, has had about 63% of the urban (20,000+) population and nearly 70% of the large city (250,000+) population. The Northeast, in contrast, had about 35% of total population in 1950 (declining to 30% in 1970) and less than 20% of the urban, and an even lower share in the large city population. The dominance of the Southeast in urban population has been offset somewhat by the growth of cities in the new settlement areas of the Central-West (and Parana in the South), but not enough to overcome the large imbalance between the Southeast and other regions in the ratio of urban to total population shares.

21. Differential growth rates have also resulted in redistribution of the population by city size within each region. In the North, growth of smaller cities has started to offset the concentration in Belem. In the other regions with metropolitan areas (Northeast, Southeast, South), the differential growth has resulted in a redistribution of the population within the metropolitan areas toward the periphery, but has left the distribution between the metropolitan areas and remaining categories relatively unchanged. In the Central-West, Brasilia and Goignia have gained substantially. - 13 -

Table 5: REGIONAL DIFFERENTIALS IN URBAN GROWTH RATES, 1950-1970

Region (N) 1950/60 1960/70

North (a) Metropolitan-Center C 1) C 4.8 4.8 Periphery C 1.0 8.4 (b) 250,00+ ( 1) 5.3 6.4 (c) 100-249,000 ( 0)) (d) 20-99,000 ( 5) 7.8 7.4 (e) Total C 7) 5.1 5.8

Northeast (a) Metropolitan-Center ( 3) ( 4.8 4.0 Periphery ( 7.7 9.1 (b) 250,000+ ( 2) 4.6 5.1 (c) 100-249,000 ( 9) 5.2 4.8 (d) 20-99,000 ( 40) 5.4 5.1 (e) Total ( 54) 5.2 5.0

Southeast (a) Metropolitan*-Center ( 1) ( 6.6 5.8 Periphery ( 6.9 7.9 (b) 250,000+ ( 3) 5.6 5.6 (c) 100-249,000 ( 14) 6.1 5.3 (d) 20-99,000 (105) 5.9 4.6 (e) Total** (135) 5.2 5.5

South (a) Metropolitan-Center C 2) C 6.5 3.7 Periphery ( 4.5 14.9 (b) 250,000+ C 1) 3.7 2.2 (c) 100-249,000 C 6) 7.3 6.8 (d) 20-99,000 C 39) 6.2 5.0 (e) Total C 48) 6.1 5.6

Central-West (a) Metropolitan C 0) - - (b) 250,000+ C 2) 21.1 12.8 (c) 100-249,000 C 1) 7.2 7.4 (dN 20-99,000 C 9) 8.2 7.3 (e) Total ( 12) 12.7 10.4

* txcluding Rio and Sao Paulo

** Including Rio and Sao Paulo

Sources: Data compiled in Fox, Urban Population Growth in Latin America, Tables 27 & 28, and Fundacao IIIGE, 1970 Population Census. - 14 -

Table 6: DISTRIBUTION OF URBAN POPULATION BETWEEN REGIONS,

BY SIZE CATEGORY, 1950-1970

1950 1960 1970

Large Cities (250,0004 and Metropolitan)

North 3.9 3.7 3.6 Northeast 17.0 17.1 16.0 Southeast 70.0 68.2 67.8 South 8.6 9.2 9.0 Central-West 0.5 1.8 3.5

Small and Medium-Sized Cities (20,000-249,000)

North 1.1 1.3 1.7 Northeast 25.9 24.3 23.8 Southeast 52.5 52.5 51.0 South 17.4 18.4 19.2 Central-West 2.9 3.5 4.3

Total City (20,000+)

North 3.0 2.9 3.0 Northeast 19.8 19.5 18.5 Southeast 64.5 63.0 62.5 South 11.5 12.2 12.2 Central-West 1.2 2.4 3.8

Total Ebpulation

North 3.6 3.7 3.9 Northeast 34.6 31.5 30.2 Southeast 43.4 43.8 42.8 South 15.1 16.8 17.7 Central-West 3.3 4.2 5.4

Sources: Data compiled in Fox, Urban Population Growth in Latin America, Tables 27 & 28, and Fundapao IBGE, 1970 Population Census. - 15 -

Table 7: DISTRIBUTION OF POPULATION BY CITY SIZE,

WITHIN REGIONS, 1950-1970

1950 1960 1970 N

North (a) Metropolitan-Center 57.4 55.7 50.3 ( 1) Periphery 6.4 4.4 5.5 (b) 250,000+ 24.4 24.9 26.3 ( 1) (c) 100-249,000 - - ( 0) (d) 20-99,000 11.8 15.0 17.9 5) (e) Total 100.0 100.0 100.0 ( 7)

Northeast (a) Metropolitan-Center 42.9 41.3 37.4 ( 3) Periphery 7.3 9.2 13.5 (b) 250,000+ 8.4 7.9 8.0 ( 2) (c) 100-249,000 21.9 21.7 21.3 ( 0) (d) 20-99,000 19.5 19.8 19.9 ( 40) (e) Total 100.0 100.0 100.0 ( 54)

Southeast (a) Metropolitan-Center 55.6 51.3 45.2 C 31) Periphery 13.5 15.8 23.3 (b) 250,000+ 5.1 5.3 5.4 ( 3) (c) 100-249,000 8.1 8.8 8.7 ( 14) (d) 20-99,000 17.7 18.8 17.4 (105) (e) Total 100.0 100.0 100.0 (135)

South (a) Metropolitan-Center 34.6 35.9 29.9 ( 2) Periphery 6.9 6.3 14.5 (b) 250,000+ 9.9 7.8 5.6 ( 1) (c) 100-249,000 13.0 14.4 16.2 ( 6) (d) 20-99,000 35.6 35.7 33.8 ( 48) (e) Total 100.0 100.0 100.0

Central-West (b) 250,000+ 25.0 51.3 63.3 ( 2) (c) 100-249,000 20.0 12.1 9.2 C 1) (d) 20-99,000 55.0 36.6 27.5 9) (e) Total 100.0 100.0 100.0 ( 12)

Sources: Data compiled in Fcox, Urban Population Growth in Latin America, Tables 27 & 28, and Fundacao IIiGE, 1970 Population Census. - 16 -

22. It is clear from this examination of regional differences in city growth by size categories that an important part of the overall pattern of differential growth by size derives from regional differences. Because of their very rapid overall growth, the city size growth patterns in the North and Central-West are more varied. While Belem has declined in relative weight in its region, Brasilia and Goiania have gained substantially Smaller cities in these regions have grown much more than the national average, and thus while showing a declining share of the regional total, show a rising share of the national total. Parana, which is located in the South, has followed the Central-West pattern, and accounts for the somewhat different behavior of the South. In the remaining regions,the principal trend has been the rapid growth of the peripheries of the metropolitan areas, which has maintained their share in the overall and regional population distributions. Smaller and medium-sized cities in these regions have increased less rapidly, and have thus declined in weight relative to the metropolitan areas and smaller cities in the Central-West. d) Demographic Components of Urbanization: Migration and Natural Increase

23. Brazil's size, regional diversity, and lack of comprehensive vital registration data make it difficult to sort out the demographic components of recent urbanization trends. In addition to large regional differentials in rates of fertility and mortality for both urban and rural areas,there is a rather complex pattern of interregional and rural-urban migration flows. However, indirect measurement of vital rates from census data permit "order of magnitude" estimates of the relative weight of migration vs. natural increase, at least at the regional level. Table 8 shows estimated birth, death and natural increase rates for ten regional groupings of states from the 1970 census tabulations. Combining these with data on overall population increase gives one estimate of the net interregional migration flow in 1960-70. Another estimate can be derived using the survival ratio technique. Both estimation techniques show a number of the main features of interregional Brazilian migration flows. Using either method two sets of net gaining regions emerge: (1) the industrial-metropolitan Southeast (Sao Paulo and Rio de Janeiro) and (2) the new settlement areas (Parana', Mato Grosso, Goias) and possibly the Amazon, if the survival ratio data are accepted). The main losing region is Minas Gerais-Espirito Santo, followed by a number of Southern and Northeastern states. In percentage terms, the new settlement regions lead the destinations, although because of their larger population base this masks the higher volume inflow into Sao Paulo and Rio de Janeiro. Migration and high rates of natural increase contributed to the higher differential growth of the new settlement areas, whereas migration actually offset the lower rates of natural increase to give a higher than average rate of total.increase for Sao Paulo and Rio de Janeiro. Despite this, migration accounts for only a third of the overall growth of these two destination regions (compared to about 40% in the Central- West). Survival-ratio data suggest that the leadership in net miRration losses shifted from the Northeastern states to Minas Gerais between the 50s and 60s. The frontier destination leadership shifted from Parand to Mato Grosso and Goia's. - 17 -

Table 8:. NEr INTERREGIONAL MIGRATION ESTIMATES, 1970/1 (number per 1,000)

Crude Crude Birth Death Rate Rate Natural Total 1960-1970 1950-1960 (196)-1970) Increase Increase A B B

Amazonian States 46 12 34 34 0 3 0

Maranhao-Piaui 44 15 29 24 - 5 - 6 2 Northeastern States 50 23 27 26 - 1 - 5 -13

Bahia-Sergipe 49 18 31 23 - 8 - 7 -11

Mines Gerais 35 11 24 15 _ 9 -13 - 6 Espirito Santo

Rio de Janeiro (Incl. Guanabara) 31 10 21 31 10 9 12

Sio Paulo 33 9 22 33 11 8 8

Parana 45 10 35 50 15 19 43

Santa Catarina Rio Grande 36 8 28 25 - 3 - 5 - 4

Goias-Mato Grosso Distrito Federal 44 10 34 56 22 30 22

/1 Aggregative estimates of both rural and urban movements.

Source: Crude birth rates and rates of increase from Merrick, "Interregional Differences in Fertility in Brazil," Demography (August 1974):425, crude death death rates from Santos, unpublished m.s., "B" migration estimate from Graham and Merrick, "Papulation and Economic Growth in Brazil," unpublished m.s., March 1975. - 18 -

24. The lack of reliable vital statistics makes it even more difficult to account for the relative contribution of migration and natural growth'in the case of the urban population and particular groups of cities. However, some estimates based on indirect measurement techniques applied to census data are available. Brazilian statisticians (at SERFRAU and the FUNDAqAO IBGE) have prepared a set of estimates for the nine metropolitan regions covering the 1960-70 period. These estimates range from 34Z for the migration share in Belem (which had low migration combined with higher natural increase) to 81% in Peorto Alegre (a contrasting case of higher migration and low natural increase). Additional estimates suggest that the migration share has shown a tendency to decline in other larger cities. In Sao Paulo the migration share declined from over 70% in the 1940s to less than 60% in the 1960s, and a similar pattern has been observed in Rio de Janeiro.

25. In assessing these trends,it is important to note the indirect effect of migration on city growth, that is, the effect of migrant naturai increase. Migrants have added to the weight of higher fertility groups in large urban populations, and the reproductive force of the migrants is an increasingly important factor in overall urban growth. When the indirect effect of migration is taken into ac-count in the Belo Horizonte data, for example, the share of growth attributable to migration increases to 72Z, and is expected to rise to 75% in 1970-19a0.

26. A number of direct questions on migration appear in the census. Because the 1960 census tabulations were never completed, these data are available only for 1970 and 1950. The Brazilian census defines as a migrant anyone not residing in the municipio of his/her birth at the time of the census. Data have been tabulated both for individuals residing in other than the native municipality and for those whose present residence is also in a different state from the one in which the individual was born. Questions on length of residence at the present residence and (officially defined) rural or urban place of previous residence help to determine the nature and volume of more recent flows. However, there remains a fundamental difficulty with migration measures derived from these data in that return migration and intervening steps in individual migration experience are over- looked. Over time, those who appear as "migrants" in the data are, in fact, those who succeeded in assimilating to the destination environment, with those who failed having moved on or returned to their origin.

27. Migrants can be defined either as those residing in a different state from the one of birth or in a different municipality. Of the 93.1 million total population in 1970, 14.4 million (15.5%) were residing in a different state, and 30.3 million (32.5%) in a different municipality. Migrants can also be distributed by duration of present residence, i.e., less than one year, 1-5 years, 6-10 years, and 11 or more years. A separate - 19 -

Table 9: DISTRIBUTION OF MIGRANTS BY ORIGIN - DESTINATION AND DURATION OF RESIDENCE, 1970 BRAZLLAN CENSUS DATA

Distribution by A. Interstate B. Intermunicipality Previous/Preser.t Duration of Residence Duration of Residence Residence Total 0-1 1-5 6-10 11+ Total 0-1 1-5 6-10 11+

Urban/Urban 50.5 53.6 50.6 48.2 50.7 50.4 48.6 48.1 49.1 53.0 Rural/Urban 17.4 16.5 15.4 16.3 18.8 17.9 15.6 16.3 18.0 19.7 Urban/Rural 5.6 8.3 7.0 5.4 4.8 6.0 9.3 7.4 5.5 4.4 Rural/Rural 26.5 21.6 27.0 30.1 25.7 25.7 26.5 28.2 27.4 22.9

'rotal 100.0 100.0 100.0 100.0 100.0 100.0 200.0 :O0.0 100.0 100.0

Distribution by A. Interstate B. Intermunicipality Durat-lon of Durat'on of Residence Duration of Residence Residence Total 0-1 1-5 6-10 11+ Total 0-1 1-5 6-10 11+

Urban/Urban 100.0 8.2 21.9 17.3 52.6 100.0 11.2 27.3 18.1 43.4 Rural/Urban 100.0 7.3 19.4 16.9 56.4 100.0 10.1 26.0 18.6 45.3 Urban/Rural 100.0 11.4 27.1 17.4 44.1 100.0 18.1 34.9 16.7 30.3 Rural/Rural 100.0 6.3 22.2 20.6 50.9 100.0 12.0 31.4 19.8 36.7

Total 100.0 7.7 21.9 18.1 52.3 100.0 11.7 28.5 18.5 41.3

Ratio of Intermunicipality/Interstate

Total 0-1 1-5 6-10 11+

1. Urban/Urban 2.09 2.87 3.23 2.19 1.73

2. Rural/Urban 2.16 2.99 2.90 2.38 1.73

3. Urban/Rural 2.24 3.55 2.88 2.15 1.54

4. Rural/Rural 2.03 3.88 2.87 1.96 1.47

5. Total 2.10 3.16 2.74 2.15 1.65 - 20 - category for migrants of less than one year residence is given because it is a relatively large group, and probably represents a higher proportion of temporary and/or returning migrants. Finally, migrants can be classified by their origin and destination, i.e., rural-rural, rural-urban, urban- urban and urban-rural.

28. The distribution of migrants by origin and destination shows that about one-half of both inters:ate and intermunicipality migrants are classed as "urban-urban". Po3sible changes in composition over time can be determined from this. Setting aside the 0-1 years of residence category because it includes a large nu-ber of temporary migrants, there is no clear trend evident comparing the I-5 and 6-10 year classes to the 11 and more years of residence categories for the interstate migrants. However, there is a change in the case of intermunicipality migrants. For the "urban-urban" category the 11 and over group is larger (53%) than the shorter duration groups (about 48%), suggesting that this type of migration was relatively less important in the 1960s thin it was in the 1950s.

29. This appears to be related to the increase over time in the share of the "urban-rural" and "rural-rural" groups. The increase in the "urban- rural" flow over time also shows up at the state level, but not the "rural- rural". This raises a question as to the nature of these increased flows to rural destinations. Are they movements to new settlement regions, temporary migration of agricultural workers, resettlement of older areas, or part of the shift to the peripheral (but not yet urbanized) parts of the metropolitan regions?

30. State-by-state analysis o' the destinations of urban-rural and rural- rural migrants suggests that a substantial part is either temporary or permanent migration to or within the new settlement areas (Goias, Mato Grosso, Parana, and the Northern region). For the inter-municipality data, 41% of rural 'estinations for Brazil as a whole are in these areas. Sao Paulo and Minas Gerais account for another 22%. Resettlement and/or temporary migration may also bc~involved in these states. Analysis of duration of residence at the state (and county) level would shed further light on this, but goes beyond the scope of this presentation.

31. Less than 18% of migrants at either the state or municipal level are reported as "rural-urban',': and there is a declining trend for both types. The most important explanatory factor here is that "origin" refers to the place of last residence, which may well have been an urban step in an individuals's rural-urban move. Furthermore, Brazil's broad administrative definition of an urban place, would give added weight to smaller places in the step migration process. Examination of the distribution of migrants by duration of residence suggests caution in interpreting the time trend. The less-chan-one-year group is quite large relative to the 1-5 category, suggesting that temporary migrants weigh heavily in it. The 6-10 group is - 21 - relatively low in comparison with others. It is true that 1960-64 was a period of recession in EBrazil (especially in such sectors as construction, which are important sources of migrant employment), but it is also possible that response errors in the duration question represent a preference for either recent (less tharL 5) or long time (11+) residence categories at the expense of the 6-10 group.

32, Comparison of the time patterns for interstate and inter-municipal data suggests that migration flows since 1960 have been of shorter distance than those before, sincet the 11+ category is lower in the inter-municipal data. A similar conclusion is drawn from the ratios of inter-municipal to interstate flows. The t:endency is more accentuated with respect to rural destinations. However, caution is warranted in this interpretation because of the nature of the dat:a. Both stage migration and temporary (and return) migration could be of shorter distance, and the differential time pattern in the state vs. municipal results could be the result of the municipal data picking up a larger relative amount of these types of migrants in relation to the total flows recorded for them.

33. A further source of information on recent migration flows are the special tabulations of t:he census data on migration for the nine metropolitan areas prepared as part of the Brazilian Government's national study of internal migration. Again, the definition of migration in the census complicates interpretat:ion of the data, since inter-municipal moves within the metropolitan regioni; are included among their migration totals. In any event, these data indicate that non-natives account for about one-half of the population of most metropolitan areas, with the proportion falling as low as 22% in Belem and reaching to 53Z in Sao Paulo. The proportions of non- natives in the adult (anid therefore, working,) population are higher, since migrants generally arrive as adults and the children of migrants (born after arrival) count as natives. The distribution of migrants by duration of residence shows Rio de Janeiro with the highest proportion (53%) of longer term residents, followed by Sa"o Paulo. Porto Alegre and Curitiba have the highest proportions of imore recent migrants, followed by Belo Horizonte and Recife. Very recent migrants (less than one year residents) are lower in the three Southeastern metropolitan regions, however the significance of this is difficult to determiiie. It is possible that the Southeast represents a more "final" destination, and that the temporary and/or stage migration indicated by this duration category is lower in the Southeast.

34. There is considerable variation in the percentage of each metropolitan area's non-natives who reside in their peripheral municipalities. In part, this reflects the number of municipalities included in the metropolitan area definition, which in turn reflects the degree to which each region has undergone the "metropolitanization" process, i.e., shifts of population, economic, and other urban activities from the core municipality - 22 -

Table 10: DISTRIBUTION OF MIGRANTS TO METROPOLITAN AREAS

BY DURATION OF RESIDENCE, 1970

Percentage Distribution Periphery Native Nonnative by Duration of Residence (Yrs) Share Total 0-1 1-5 6-10 11+ % Index

Belem 78.2 21.8 100.0 11.2 28.8 17.3 42.7 5.9 1.7

Fortaleza 48.8 51.2 100.0 13.5 26.8 16.5 43.2 8.0 0.6

Recife 51.5 48.5 100.0 12.5 29.6 15.4 42.5 47.5 1.2

Salvador 70.5 29.5 100.0 12.1 26.9 16.2 44.8 11.4 0.9

Belo Horizonte 49.6 50.4 100.0 9.6 30.1 19.1 41.2 22.9 1.0

Rio de Janeiro 54.7 45.3 100.0 8.8 22.3 15.7 53.2 43.8 1.1

Sao Paulo 47.1 52.9 100.0 9.7 25.7 18.5 46.1 34.0 1.2

Curitiba 59.0 41.0 100.0 13.5 33.2 16.9 37.4 15.4 0.6,

Po^rto Alegre 51.0 49.0 100.0 10.1 39.7 17.3 32.9 47.0 1.1

Note: See text for explanation of index.

Source: Ministry of Interior,SERFHAU, 'Mudan9as na Composic,o do Emprego e Distribuipa"o da Renda", Relatdrio No.2.1.-, Tables Ic and A4. - 23 -

to outlying areas. Be.Lm shows the lowest share of migrants in the periphery, and Recife, Rio., and Porto Alegre the highest. However, a better guide. to the extent to which a region's migration streams are becoming periphery-oriented is the ratio of the periphery migrant share to the periphery share in tota population. Belem has a relatively small share of migrants in the periphery, but this well exceeds its peripheral population share and' suggests thai: migration is contributing substantially to the peripheral growth. ThiLs is also true of Recife and Rio de Janeiro, and much less so in Fortaleza anid Curitiba. Thus,while there are a number of comon growth patterns in the recent expansion of the metropolitan area populations, there are also a number of features that are specific to each metropolitan area case.

e) Basic Urban Populat:ion Projections, 1970-1990

35. To conclude review of recent urban population growth trends, a set of basic population projections for 1980 and 1990 for cities with 20,000 inhabitants or more in 1970 will be examined. The methodology for the projections is the ratio-trend method, which employs a weighted average of the trend in the logarithm of each city's share of population in previous decades to determine the share in the next decade. The method starts with an assumed trend in nat:ional population growth and works by steps down to each city's share. These projections employ regional shares as the intermediate step, since recent Brazilian migration flows (especially for the metropolitan areas) are more regional than state oriented. For example, Minas Gerais is a large state which hasi lost migrants to Sao Paulo, Rio, and the Central- West. However, Belo Horizonte, its metropolitan region, has shared in the dynamics of the Southeastern region's metropolitan area growth. Similarly, Recife, in Pernambuco, has drawn heavily from neighboring Alagoas and Paraiba because the Northeastern states run in narrow strips from the Atlantic coast. The Northeastern region is the more appropriate "context" for establishing Recife's growth trend.

36 . The national population growth trend is assumed to decline from 2.9% in 1960-70 to 2.7, and 2.6% in 1970-80 and 1980-90. The regional and city projections show a stabilized or dampened trend. All the projected growth rates decline from their 1960-70 levels, following the national trend. Particular cities and city size categories follow their regional trend, but the overall distribution remains quite similar to that found in 1970. Within the metropolitarn regions, the peripheral areas continue to grow in relative importance, and there is a very modest decline in the weight of Rio-Sao Paulo relative to the other metropolitan areas and medium and larger cities. The overall city population share increases from 40% in 1970 to 48% and 54% in 1980 and 1990. The non-city residual grows at 1.4% per annum in the interval. The projections reflect the dampening of past trends inherent in the logistic method. They are thus conservative estimates. - 24 - Table 11: BASIC URBAN POPULATION PROJECTIONS, 1970-1990

Implied Growth Rates Distribution Group 1970 1980 1990 70/80 80/90 1980 1990

Rio and Sao Paulo (a) Municipios of Capitals 9,557 13,194 17,652 3.3 3.0 22.7 20.8 (b) Periphery 5,128 8,719 13,437 5.5 5.1 15.0 15.8 (c) Total 14,685 21,913 31,089 4.1 3.6 37.7 36.6

Regional Metropolitan Areas (a) Municfpios of Capitals 5,702 8,376 11,953 3.9 3.6 14.4 14.0 (b) Periphery 2,056 3,690 6,034 6.0 5.0 6.4 7.1 (c) Total 7,758 12,066 17,987 4.5 4.1 20.8 21.1

Remaining Cities in 250,000 and Over Class as of 1970 3,286 5,548 8,784 5.4 4.7 9.6 10.3

Cities of 100,000 to 249,000 (in 1970) 4,422 6,823 10,082 4.4 4.0 11.8 11.9

Cities of 50,000 to 99,000 (in 1970) 2,954 4,505 6,601 4.3 3.9 7.8 7.8

Cities of 20,000 to 49,000 (in 1970) 4,706 7,162 10,481 4.3 3.9 12.3 12.3

Total 20,000+ 37,811 58,017 85,024 4.4 3.9 100.0 100.0

Total Population Base 93,139 121,508 157,644 2.7 2.6

Source: Base data from Table 3; projection method described in text. - 25 -

III. Interrelations Between Industrialization and Urbanization, 1950-TO a) Growth of the Industrial Labor Force and its Distribution in the Brazilian UTrban System

37. Growth-in induLstrial employment in Brazil since the Second World War has not kept up with urban population increase. While industrial production expanded at an average eanual rate of nearly 8% during the period 1950-TO, industrial employment grew at only half that rate. Using the broader population census definition which includes construction and may also pick-up more small- scale and traditional activities, industrial employment increased at a rate of 4.15% per annum between 1950 and 1970. From the industrial census data, the rate was 3.51%. Both measures show a slower growth in employment in 1950-1960, when implantation of imported technology appears to have had a larger labor- saving bias. The increase in employment growth in 1960-70 is greater for the population census data, suggesting that construction and smaller scale, more traditional activities contributed substantially to the increase. Comparing industrial employment growth rates to those of urban population increase indicates that the latter was about 50% greater (depending on the definition of "urban") over the interval. Growth in other non-agricultural employment was closer (4.5%) to the increase in urban population, but not enough to compensate for the industrial employment differential. Thus the overall trend was one in which employment in non-agricultural activities failed to keep pace with the rapid growth of the urban population.

38. A second major characteristic of industrial expansion in Brazil is its unequal regional distribution. The tendency toward spatial concentration dates back to the earliest phases of Brazilian industrialization at the beginning of this century. By 1950, 39% of all Brazilian industrial employment (industrial census data) was located in the state of Sao Paulo. The Rio-Sao Paulo axis accounted for 57%, and an additional 24% was located in three other states: Minas Gerais, Rio Grande do Sul, and Pernambuco, giving a five state total of 81%. A recent study of trends in industrial concentration reveals that industrial concentration increased between 1950 and 1960. Saco Paulo alone accounted for 66.4% of new industrial employment in that decade, and the other four states listed above for an additional 20%, thus leaving only 15% for the remaining 17 states. However, the data also show a leveling off of the coacentration process in 1960-70. The 1970 share of the Rio-Sao Paulo axis was 61.6%, down just a little from its peak level of 62.2% in 1960.

39. The bulk of the industrial activity reported for the State of Sa.o Paulo is located in the mnetropolitan area of the city of Sao Paulo. Unlike Rio deJaneiro which was until recently a separate state, data for the Sao Paulo metropolitan area mnust be reconstructed from municipality data. Since this is a progressively mnore difficult task with each earlier census because of municipality boundary changes, industrial census data for Sao Paulo, as well as other metropolitan areas and cities of 20,000 or more could be reassembled only for 1961) and 1970 (Table 13). - 26 -

Table 12: GROWTH OF OUTPUT, EMPIOYMENT AND URBAN POPULATION, 1950-1970 (per cent)

1950-1960 1960-1970 1950-1970

Industrial Product 7.79 8.18 7.99 Industrial Employment-Y a) 2.35 5.97 4.15 b) 2.92 4.10 3.51 Other Non-Agr. Product 6.75 7.90 7.35 Other Non-Agr. Employment 5.24 3.83 4.52

Urban Population (i) Offical Definition 5.42 3.05 5.23 (ii) Cities of 20,000 + in 1970 5.43 5.55 5.49

Source: Population & Industrial Censuses FGV National Accounts Data.

Note: 1/ Industrial production data for 1949, 1959 and 1970. 2/ "A": Population Census "B": Industrial Census. - 27 -

40. The entire Rio-Sao Paulo axis (consisting of the states of Rio de Janeiro, Guanabara, and Sao Paulo) accounted for 62% of total Brazilian industrial employment in 1960, while the 1960 share of the two metropolitan areas alone in the national total was 46%. Moreover, they accounted for 56% of industrial employment in cities of 20,000 or more. (Industrial. employment in cities of 20,000 or more was 82% of overall industrial employment in 1960 and 84% in 1970). There is a slight decline (3 percentage points) in the share of industrial employment in the metropolitan areas in Rio and Sao Paulo from 1960 to 1970. But their share in total salaries paid declined less (about 1 percentage point), and value added increased slightly. Their population shares remained virtually unchanged. The distribution of industrial activity within these metropolitan regions shifted from the central municipality to the peripheries, parallelling the population trend, but at a slower pace. Even with declining shares, Rio and Sao Paulo account for nearly 40% more industriil employment than their share in population -- indicating a higher level of primacy in industrial activity than was evident in population. Greater asymmetry is observable in salaries and value added.

41. The balance between industrial activity and population shows a corresponding deficit in the other city size categories. In 1960, employment was 35% lower than the population share in other metropolitan areas, and nearly 50% lower in other large cities. It should be noted that Brasflia, with relatively little industrial employment, weighs heavily in the latter class. The deficit was iess in smaller cities. Salaries and value added were even more unequally distributed, reflecting the added productivity advantage of Rio and Sgo Paulo. The overall trend for 1960-1970 appears to have been toward a slightly better balance between employment and population. The improvement is especially marked in smaller cities, and the 20-49,000 class achieved a one-to-one ratio in 1970.

42. There was important regional variation in these patterns. There is a clear break between the South and Southeast, with larger relative shares of employment, and the North, Northeast, and Central-West, with lower relative shares. The regional concentration appears to have remained stable or increased slightly from 1.960 to 1970, with relative shares for the Southeast dropping from 124 to 121 and for the South rising from 97 to 121. This has occurred despite a decline in the very high concentration in the peripheral municipalities of the metropolitan areas of these regions, and without substantial increases in the metropolitan area central municipalities. Smaller cities (below 100,000) accounted for the gains in the South and Southeast, Middle sized cities also gained in the South, but not in the Southeast. The share of industrial employment in the Northeast declined even more than its share in population, resulting in a greater imbalance in relative shares in 1970 than in 1960. Smaller cities and the peripheral municipalities of the metropolitan areas experienced the greatest declines in relative shares. In contrast, cities in the new settlement regions, the North and Central-West, show improved relative positions. Except for these cities on the frontier, which are relatively few in number, most of the deconcentration of industrial employment away from Rio 1e Janeiro and Sao Paulo proper thus appears to have been in the direction of -the smaller cities in the South and Southeast. These - 28 -

Table 13: DISTRIBUTION OF INDUSTRIAL SECTOR EMPLOYMENT,.SALARIES, AND VALUE ADDED BY CITY SIZE, 1960-1970

1960 1970 Percentage shares in Industrial Percentage Shares in Industrial Sector Sector Pop. Emplt. Salaries Value Pop. Emplt. Salaries Value added added

Rio & Sao Paulo a) Center 29.4 43.6 49.3 45.5 25.2 38.2 43.2 40.9 b) Pheriphery 9.3 12.3 14.3 14.8 13.6 14.7 19.3 20.0 c) Total 38.7 55.9 63.6 60.3 38.8 52.9 62.5 60.9

Regional Metropolitan Areas a) Center 17.0 8.8 7.0 6.6 15.1 8.8 6.5 6.3 b) Periphery 3,3 4.4 3.5 3.6 5.4 5.5 4.6 5.0 c) Total 20.3 13.2 10.5 10.2 20.5 14.3 11.1 11.3

Other Large Cities (250,000 & over, 1970) 7.8 3.9 3.5 6.4 8.7 4.5 4.4 5.0

Medium Cities (100,000 to 249,000 in 1970) 11.8 10.4 8.8 9.9 11.7 9.4 8.7 9.0

Small Cities (a) (50,000 to 99,000 in 1970) 8.2 6.2 4.6 4.7 7.8 6.3 4.5 4.8

Small Cities (b) - (20,000 to 49,000 in 1970) 13.2 10.4 9.0 8.5 12.5 12.6 8.8 9.0

Total City Population 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0

Source: Industrial Censuses of 1960 and 1970, state volumes. - 29 -

cities may indeed be. offering increasingly attractive cost advantages in comparison with Rio and Sao Paulo because of their proximity and lower land values and labor costs.

43. Indeed there is very little consistency between the main changes in the distribution of industrial employment by region and city size and population shifts. In the peripheral municipalities of the metropolitan areas, the population share increases more substantially than industrial employment, and in the smaller cities of the Southeast there was a declining population share corresponding to their small increase in the share of industrial employment. These trends suggest that migratory redistribution of population, insofar as it was responsive to differential employment opportunities, was more clearly linked to employment outside of the established industrial establishments that are represented in the industrial census data. The next section examines the relation between these migration flows and urban employment in greater detail.

b) Urban Growth and Urban Poverty: Indicators of Inequality in the Brazilian Urban System

44. The concentration of industrial employment, salaries and value added and the imbalance between these magnitudes and the distribution of urban population raises important questions about the effects of recent industrial growth on employment and income distribution in urban areas. Recent studies of trends in the Brazilian income distribution have called attention to the growing inequality in the urban sector. Langoni (1973) has shown that inequalities in earnings (expressed by Gini coefficients) for urban activities increased much more (15%) between 1960 and 1970 than for the agricultural sector (3%). Inequality for the state of Sao Paulo increased much more (24%) than the national average (13.7%), and other Southern regions show a similar trend. In fact, inequality increased most in those regions and sectors in which economic expansion was most rapid. The growing inequality indices suggest that benefits for the working population in terms of employment and earnings were not shared equally. Langoni has attributed this to the unequal distribution of human capital resources and the remuneration paid to scarce skilled labor that has been in great demand during the boom. It should also be noted thaLt migration flows, which contributed heavily to the growth of large city population, consisted mainly of less skilled workers. Substantial proportions of these found jobs in lover paying construction and service activities, which were also in demand because of the boom but which have not benefitted froma the salary increases of scarcer more skilled types of employment.

45. An important characteristic of the Brazilian labor absorption process is employment of large proportions in jobs outside of the modern industrial sector. This feature of urban employment growth in Brazil, as in a number of other developing countries, derives from the existence of a labor market which is parallel to that in the organized, modern sector. Because institutional - 30 -

Table 14 DISTRIBUTION OF EMPLOYTENT AND POPULATION BY CITY SIZE AND REGION, 1960 AND 1970

1960 1970 Ratio of Ind. Employment Region/City Group PercentaRe of Percentage of to Population Population Employment Population Industrial 1960 1970 Employment

NORTH Metro Center 1.63 0.47 1.54 0.59 29 38 Periphery 0.13 0.03 0.17 0.03 23 18 250,000 0.73 0.27 0.79 0.39 37 49 100,00 - - - - - _ 20,000 0.44 0.22 0.54 0.19 50 35 Total 2.93 0.99 3.02 1.20 34 40

NORTHEAST Metro Center 8.06 3.29 6.92 3.24 41 47 Periphery 1.79 1.11 2.49 1.49 62 52 250,000 1.55 0.44 1.48 0.51 28 34 100,000 4.23 1.57 3.93 1.28 37 33 20,000 3.86 1.89 3.68 1.28 50 35 Total 19.40 8.30 18.52 7.59 43 41

SOUTHEAST Metro Center 2.92 1.33 2.98 1.42 46 48 Periphery 0.62 1.33 1.00 1.27 215 127 250,000 3.34 2.28 3.35 2.59 69 77 100,000 5.57 7.22 5.43 6.04 130 111 20,000 11.87 10.33 10.87 11.32 87 104 Total 62.99 78.40 62.46. 75.53, 124 121

SOUTH Metro Center 4.37 3.74 3.66 3.51 86 96 Periphery 0.76 1.90 1.77 2.90 250 164 250,000 0.95 0.86 0.69 0.57 91 83 100,000 1,76 1,54 1,98 1,99 88 100 20,000 4,35 3.82 4.14 5.79 88 140 Total 12.19 11.86 12.24 14.76 97 121

CENTRAL WEST 250,000 1.23 0.08 2.39 0.45 6 19 100,000 0.29 0.09 0.34 0.10 31 29 20,000 0.88 0.28 1.03 0.36 32 35 Total 2.40 0.44 3.76 0.91 18 24

Source: Industrial Censuses of 1960 and 1970, state volumes. - 31 -

factors determine the functioning (and therefore the definition) of this market, the usual census data on labor force characteristics are of limited use in its definition and measurement. The informal sector generally consists of smaller scale, artisan-type establishments with relatively easy entrance and high turnover. In Brazil, regulation by social security legislation is an important factor distinguishing the formal and informal sectors. Empirical study of the informal sector in Belo Horizonte indicates that it accounts for about a third of all employment in that city; and a recent ILO study of Sao Paulo indicates about 25%. Construction and personal services represent large proportions of informal employment in both cases.

46. InformAl employment has played a key role in recent urban labor absorption. This sector functions as a buffer for the excess between labor supply and the demand for labor generated by the modern industrial sector. Reported open urban unemployment in Brazil is quite low (partially because there is no reason to report oneself as unemployed, since there is no unemployment compensation) and informal sector employment has probably taken up a good part of the slack.. It has also allowed lower income urban households to increase overall income through the employment of additional family members. The urban poverty case study for Belo Horizonte shows that lower income families who "survived" economically are those with additional sources of earnings other than the head of household. A further finding of the Belo Horizonte study is that younger adults among the native population are as likely to be found among informAl employees as recent migrants, indicating that this sector performs a broader labor absorbing function than one restricted exclusively to migrants.

47. Since sample dtata on informAl activities similar to those used in the Belo Horizonte stutr are not available for other Brazilian cities, some proxy measurements will be attempted. The difference between industrial employment as measured in the population census and industrial census can be used as an indicative masure of the informAl group, on the assumption that this would consist mainLy of smaller scale activities and construction. In addition, those employedl in commerce and personal services are included in the informal group. The measures are admittedly very rough, and clearly misclassify certain groups -- like professionals in the personal service category. The categories "informal-traditional"I and "formal-modern" are simply labels for these rough measures, and the data are presented here to give order-of-magnituide indications of regional and city size class differentials

48. The ratio of population to industrial sector employment was calculated for both 1960 and 1970. It is a useful supplement to the other data in that it gives azn indication of the relative number of people that the industrial sector worker supports, either directly through earnings distributed through households or irndirectly through taxes levied on industrial earnings and production. The latter are particularly important in Brazil since municipal finances depend heavily on the volume of industrial output carried on in them. - 32 -

Table 15: CENSUS DATA ON MIGRATION AND LABOR ABSORPTION, BY Cm SIZE AND REGION, 1970

Recent Migrants Informal (traditional) Population Region/City Group 1/ Industry Workers per Commerce Industry Worker Industry & Services 1960 1070 1/ 1/

NORTH Metro Center 14.3 1.81 4.71 51.6 42.6 Periphery 7.9 1.35 1.80 67.6 108.7 250,000 10.8 1.71 3.41 40.3 34.2 100,000 - - - _ _ 20,000 32.3 3.65 5.79 63.6 61.1 Total 24.3 2.89 4.86 56.7 61.4

NORTHEAST Metro Center 14.3 1.97 5.23 40.7 37.3 Periphery 23.3 1.11 1.95 120.0 58.7 250,000 17.2 1.74 5.50 59.5 48.9 100,000 14.2 3.01 7.90 75.0 74.5 Total 17.6 3.66 7.23 71.3 68.1

SOUTHEAST Metro Center 14.9 1.05 2.94 19.5 20.8 Periphery 23.8 1.15 1v59 11.5 19.1 250,000 25.6 1.48 3.17 32.5 27.6 100,000 15.9 1.02 2.99 21.3 22.5 20,000 18.4 1.56 3.44 27.3 27.2 Total 18.4 1.48 3.33 26.2 26.4

SOUTH Metro Center 18.8 0.71 2.52 17.4 17.4 Periphery 24.2 0.64 0.87 8.2 13.8 250,000 8.6 0.87 2.30 16.5 20.1 100,000 21.4 1.27 3.90 29.1 32.7 20,000 23.8 1.59 4.46 38.6 36.4 Total 23.1 1.47 4.13 34.9 34.0

CENTRAL WEST 250,000 26.2 3.12 7.61 59.0 91.6 100,000 20.7 2.75 6.61 48.5 56.2 20,000 35.9 3.40 6.79 79.4 58.0 Total 33.0 3.31. 6.91 73.4 63.4

Sources: Industrial Censuses of 1960 and 1970, state volumes.

Note: 1/ X a proportion of total population. - 33 -

The migration rates by region have several important features. The rates are highest in the Central-West and North, indicating that the rural settlement in these regions is being accompanied by substantial urban growth. The lowest rates are for the Northeast and Southeast. By city size, two categories stand out with high migration rates: the smaller (20-99 thousand) cities and the peripheral municipalities of the-metropolitan areas (except for Belem in the North). In addition, the 250,000+group in the Southeast (Camoinas, gantos and Vit&ria) has a high proportion of migrants.

49. Region and eity size appear to be more important influences on the informal-traditional employment ratio than migration. Lower migration rates in the Northeast have not brought lower informal employment, nor have the higher migration rates of the periphery municipalities of metropolitan areas led to inflated infonral employment ratios. One explanation for the latter is that recent industrial expansion in many of the metropolitan areas has taken place in the peripheral municipalities. The only group for which there does appear to be some consistent relation between the informal employment ratio and migration is the smaller city group. They follow their general regional trend, but have higher informal employment as a group than other size classes. c) Metropolitan Area Growth: Migration, Labor Absorption, and Urban Poverty 50. The distribultions of the indicators of migration and labor absorption by region and city size in the previous section mask some important differences in these processes between the metropolitan areas in each region, which is useful to examine in greater detail. Because special tabulations of census data on migration and labor absorption are available for the metropolitan areas but not other cities, it is also possible to examine linkages between urban growth and urban poverty in somewhat greater detail for these regions.

51. Migration rates are higher for the peripheries than the center in the six largest metro:politan areas, and less in Fortaleza, Belem and Curitiba. The center-periphery difference is largest in Rio de Janeiro, Recife and Salvador, whereas Sao Paulo and Belo Horizonte are more balanced. Several intra-regional contrasts in labor absorption patterns are evident. For example, differences emerge when the informal employment ratio for the center and periphery are compared in the respective metropolitan areas, indicating corresponding differelaces in the spatial distribution of formal and informal employment.

52. Special tabulations of metropolitan area census data prepared for the Brazilian Interior Ministry to use in its national migration survey permit a more detailed view of other aspects of labor absorption in the metropolitan areas. Table 17 summarizes several measures for the metropolitan areas found in the Ministry of Interior -- SERFHAU studies. The proportion of migrants in the labor force is shown with "migrant" defined as a non-native with 10 or fewer years of residence in the municipality. In this instance, non- natives with 11 or more years of residence are classed with non-migrants. It should be noted that the placement of the longer term residents vis-a-vis other - 34 -

Table 16: MIGRATION AND LABOR ABSORPTION IN THE METROPOLITAN AREAS, CENTER AND PERIPHERY, 1970

Recent Informal-Traditional Population per Migrants* Industry Workers** Industrial Worker (%) (A) Industry (B) Commerce 1960 1970 and Services (% of Tbtal Population)

(SE) SAO PAULO a) Center 15.9 0.42 1.19 6.82 8.14 b) Periphery 20.5 0.45 0.63 6.47 10.03

RIO DE JANEIRO a) Center 9.9 0.61 2.44 18.72 19.22 b) Periphery 22.6 2.58 3.44 21.21 34.20

BELO HORIZONTE a) Center 19.1 2.13 5.21 32.90 34.89 b) Periphery 28.3 0.42 0.69 6.98 13.13 (S) PORTO ALEGRE a) Center 13.2 0.45 2.44 17.84 17.48 b) Periphery 29.9 0.63 0.89 5.30 9.01

CURITIBA a) Center 24.4 0.97 2.60 16.98 17.23 b) Periphery 18.5 0.65 0.85 34.16 18.62 (NE) RECIFE a) Center 9.4 0.93 3.89 25.58 31.84 b) Periphery 26.7 1.73 2.49 18.10 30.36

SALVADOR a) Center 10.5 3.21 7.06 50.83 55.34 b) Periphery 40.3 0.65 0.71 7.69 5.78

FORTALEZA a) Center 22.9 1.77 4.73 42.54 24.65 b) Periphery 2.9 0.95 2.63 334.16 139.98 (N) BELEM a) Center 14.3 1.81 4.71 51.63 42.63 b) Periphery 7.9 1.35 1.80 67.63 108.73

* Five or fewer years of residence. ** See text for definition

Source: Derived from Population and Industrial Census Data. - 35 -

Table 17: UNEMPLOYMENT AND INCOME DISTRIBUTION DATA FOR THE METROPOLITAN AREAS, 1970

Migrants Unemployment Labor Force Earning Gini Coefficient in L.F. Rate (%) Less Than Regional () 'Total Migrant Non- Minimum Wage Total Migrant Non- Migrant Total Migrant Non-

(SE) SAO PAULO 34.7 3.1 3.6 2.9 32 41 28 0.54 0.51 0.5 RIO DE JANEIRO 25.8 3.7 3.8 3.7 37 48 33 0.56 0.58 0.5 BELO HORIZONTE 37.3 4.0 4.3 3.9 46 53 42 0.61 0.58 0.6 (S) PORTO ALEGRE 31.5 2.4 2.6 2.3 36 40 34 0.55 0.52 0.5 CURITIBA 33.4 1.8 2.3 1.5 36 38 35 0.56 0.57 0.5 (NE) RECIFE 25.3 5.5 5.7 5.4 42 46 41 0.61 0.67 0.5 SALVADOR 23.6 3.9 3.5 4.0 37 47 34 0.58 0.58 0.5 FORTALEZA 24.5 4.8 5.7 4.4 46 48 46 0.55 0.63 0.5 (N) BELEM 16.6 3.6 3.6 3.6 32 39 30 0.56 0.57 0.5

Source: SERFHAU, "Mudanias..,'' Relato'rio 2.3 - 36 - groups in migrant-native comparisons in recent Brazilian census data has an important bearing on the outcome of the comparisons. Longer term residents in the metropolitan area on the average are an older aged population than either recent migrants or natives, and are therefore likely to be more represented in higher income segments of the age-earnings profile than the other two groups. The latter, being younger, are also more likely to be subject to job entry and unemployment problems that are more frequent for the 10-25 year age cohorts.

53. The proportion of migrants in the labor force ranges from 17% in Belem to 37% in Belo Horizonte. Sao Paulo, Porto Alegre and Curitiba are above 30%, and the remaining cities except Belem are about 25%. Unemploy- ment rates are low, which is understandable in the absence of unemployment compensation and the availability of informal jobs for individuals who want to work fewer hours at lower pay rather than have no income at all. There is no clear relation between migrant proportions in the labor force and either the overall level of unemployment or migrant/non-migrant differences in unemployment rates. Migrants do have higher unemployment in Sao Paulo, Belo Horizonte, Porto Alegre, and Curitiba, which have the highest proportion of migrants in the labor force. However, unemployment of migrants is even higher in Recife and Fortaleza, where the migrant proportion is a quarter. Age factors, more than anything, explain the migrant/non-migrant differential pattern, which is relatively consistent for almost all levels of the unemployment rate. Migrants are younger, more likely to be seeking work for the first time, and more likely to have difficulty in doing so. Regional factors appear to be more important in the level of unemployment for both migrants and non-migrants. The rates are lowest in the South and generally high in the Northeast although Salvador, with a lower than average level, is an exception.

54. The proportion of workers earning less than the monthly minimum wage set for each region is a better index of the employment situation than the unemployment rate. This index captures a good part of the informal employment and also is suggestive of the level of urban poverty in each metropolitan area since the minimlm wage, when divided by average family size, was indeed a "bare" minimum if it was the only source of family income at the time of the 1970 census. These proportions range from 32% in Sao Paulo and Belem to 46% in Belo Horizonte and Fortaleza, with migrants having a consistently higher proportion than non-migrants. Age is again an important factor in this differential and the more detailed SERFHAU data from which Table 17 was derived show a steady decline in this proportion as length of residence increases. Another factor in the migrant/non-migrant differentials are the types of jobs sought and found by recent arrivals. For males, civil construction is the most popular, and for females it is domestic services. Both are low paying, but both (particularly domestic service) provide food and lodging for the worker, especially if the individual is single. For this reason, the non-migrant category, which includes natives and older non-natives, is more representative of the individuals who would be supporting households with less than a minimum salary. The proportion of households in this position ranges from 28% in Sao Paulo to 46% in Fortaleza, with Belo Horizonte and Recife also showing higher than average levels. As with the unemployment rates, there is no evident pattern,in migrant/non-migrant differentials that could be associated with the - 37 - proportion of migrants in the labor force. Belo Horizonte has proportions of both, but Sao Paulo is lower than Rio on the minimum wage but higher on migration.

55. Gini coefficients suggest that the migrant population is generally more homogeneous with respect to earnings. The non-migrant coefficients are equal to or higher tharn coefficients for migrants for all regions, except for Rio, Recife and Fortaleza. Rio and Recife had higher proportions of migrants in their peripheral municipalities suggesting that these two metropolitan areas may have especially severe "suburban" poverty problems. Unfortunately the available data do not permit us to explore center-periphery differentials in this respect.

56. Concern for the detrimental effects of recent urban migration on employment and income distribution was one of the major motives that led Brazilian authorities to undertake the national migration study. Neither the data we have reviewed here nor the more detailed results reported in the Brazilian government's studies have been able to attribute these problems specifically to migration. In part, the conclusion reached on this issue depends on how the question is formulated and how a migrant is defined. There is little dispute that the demographic impact of migration flows over and above the natural growth of urban areas has augmented their population increase substantially, and that this has resulted in pressures on urban services, job creation, etc. On the other hand, examination of differentials between migrants and non-migrants suggests that socio-economic characteristics (particularly age and education) of migrants rather than migration status per se accounts for most of the observed migrant/non-migrant differentials in census data. Defining migrants as those with 10 or fewer years of residence brings out migrant/non-migrant differentials because of the age factor. However, the changing employment and earnings profiles of migrants with length of residence, which with time approach and even surpass those of natives, along with age- standardized comparisons found in the Ministry of Interior studies, suggest that migrants as a grou? do quite well in their assimilation to social and economic conditions of B3razil's metropolitan areas. As a result, the urban poor of metropolitan areas are not likely to have an over-representation of migrants among their ranks. Thus migration is a "cause" of urban poverty problems in the one sense of its being one of the main factors in the rapid demographic increase of the urban population, but certainly not a cause if one were to adopt the once popular view that the marginalized masses in Latin American cities consist entirely of poor migrants who have swarmed in from the countryside. d) Implications for Demand and Supply of Urban Services

57. The disequilibrium between changing shares of industrial employment and population has some implications in relation to the problem of providing adequate urban services for the increased population. In the Brazilian fiscal system, the main source of tax revenues for local governments is the levy on value added in industrial production. Hence the growing imbalance between - 38 - population growth and the growth of industrial employment and output has a double effect. Urban population growth creates increased demand for services, and the imbalance makes it more difficult to provide these services with resources available at the local level.

58. Population and housing census data provide some rough indicators of the supply of urban infrastructure, and industrial census information on value added per capita gives further insight into differentials in the capacities of municipalities in various city size classes and regions to meet the further demands generated by population increase as well as to make up for existing deficits. Table 18 is a compilation of average percentages of households by region and city size class reported in the 1970 census to have access to public water supply and electric power, to reside in what were classed as "durable" dwellings, and to have a television set. The latter is included as a measure of overall modernization, in that it is closely related to the degree in which a community is linked to the national market and its products (and is used by Brazilian marketing researchers as an index for this). An index (Brazil = 100) of industrial value added per capita for each class of localities, which is a measure of differentials in their capacities to generate tax revenues,also is shown.

59. The highest proportions of households served by water and electricity are found in the larger cities in the South and Southeast. The noticeable exceptions are the peripheral municipalities of the metropolitan areas. The North and Northeast are lowest on these scales, with the small cities and peripheries of their metropolitan areas being the least served. The pattern is similar in housing durability and ownership of televisions, except that the small city vs. large city differential is somewhat more marked and the center city/periphery gap in the metropolitan areas is less so. The data on value added per capita indicate some narrowing of the regional-differences between 1960 and 1970. The Northeast continued to be about 25% lower than average in 1970, whereas the Southeast lost all of its earlier advantage. In the Southeast, the metropolitan regions lost a substantial part of their 1960 edge over other groups. Surprisingly (for their poor showing on the infrastructure indices) the peripheral municipalities of the metropolitan regions appear to be relatively well off in terms of the financing capacity indicated by this measure. One explanation is that aggregation of the metropolitan areas in each region masks important intraregional variation. Within the peripheral municipalities of each metropolitan region, there is considerable imbalance between the location of industrial establishments and the residences of those who work in them. Thus within the metropolitan region of Sao Paulo, industrial activity is concentrated in a half dozen municipalities that are very well off in terms of tax revenues and have relatively adequate urban infrastructure, and the working population is spread over the remaining thirty, which are much worse off. - 39 -

Table 18: PERCENTAGES OF HOUSEHOLDS WITH WATER AND ELECTRICITY CONNECTION, DURABLE DWELLINGS AND TV, AND VALUE ADDED PER CAPITA, BY REGION AND CITYSIZE, 1970

REGION/CITY Water Eletricity Durability TV Val.p Added GROUP Connection Connnection of Owners Per Caplta Dwellings 1960 1970

NORTH Metro Center 55.8 77.6 69.0 36.3 A1-I 93 Periphery 5.9 29.6 88.6 8.6 72 45 250,000 48.9 62.1 60.7 21.5 92 216 100,000 - - _ _ _ _ 20,000 14.4 25.6 50.5 1.7 72 112 Total 22.8 37.2 58.9 9.4 61 114

NORTHEAST Metro Center 40.4 74.6 74.5 33.9 38 83 Periphery 11.6 37.5 58.1 10.4 225 130 250,000 49.3 38.6 74.8 15.3 23 71 100,000 34.3 46.9 70.1 13.8 27 73 20,000 18.9 33.0 67.5 5.2 39 68 Total 13.2 37.8 68.0 8.7 46 73

SOUTHEAST Metro Center 65.0 91.8 91.9 64.4 164 114 Periphery 43.5 74.8 88.5 41.8 273 127 250,000 79.8 85.1 83.9 54.9 177 207 100,000 64.5 78.9 89.8 44.0 137 115 20,000 57.2 68.3 87.6 31.1 104 94 Total 58.4 70.5 88.0 34.1 118 100

SOUTH Metro Center 65.5 90.5 90.4 61.0 91 92 Periphery 21.5 61.3 87.5 27.4 215 84 250,000 67.3 65.2 85.2 30.9 167 159 100,000 56.2 75.4 84.3 36.1 113 116 20,000 33.5 54.5 79.9 19.4 139 135 Total 37.7 58.9 81.3 23.6 137 129

CENTRAL WEST 250,000 54.3 75.6 78.1 45.1 87 59 100,000 39.8 50.9 82.8 27.2 30 179 20,000 31.8 36.4 64.2 10.3 46 104 Total 36.2 44.2 68.1 17.5 35 94

44.3 58.7 80.4 24.8 100 100

Source: Population and Industrial Censuses 1960 and 1970.

Note: 1/ Brazil = 100 -40 -

60. The problem is somewhat different in Rio, where industrial activity is more concentrated in the central municipalities, and almost all of the peripheral ones are worse off in terms of taxing power and urban infrastructure. On average, urban services in the peripheries of Rio and Sao Paulo are not much different. However, the value added data suggest more clearly that for Sao Paulo the problemis one of balance between municipalities of the periphery, whereas for Rio it is a question of center-periphery balance. Belo Horizonte and Porto Alegre have almost the opposite situation to Rio, with peripheral municipalities having more value added per capita. Like Sao Paulo, the level of urban services in the peripheries suggests imbalance in the location of industrial activity and worker residences in their peripheries. Salvador is an extreme case of this, and explains the unexpectedly high value added ratio for the Northeast as a whole. The Salvador metropolitan area includes the Landulfo Alves refining complex in the municipality of Sao Francisco do Conde. This highly capital-intensive activity employs relatively few workers and has not benefitted the environs of Salvador in any proportion to its impact on the value added ratio for that area.

61. Two points stand out in this review of demographic and economic indices for the metropolitan areas and other city groups. First, although there are a number of problems common to all the groups and regions, important aspects of these are peculiar to each region and each metropolitan area. Thus the issue of center-periphery balance is more important for Rio de Janeiro and Belo Horizonte, whereas balance among municipalities in the periphery is more an issue in Sao Paulo. A second point is that institutional factors impinge heavily on capacities of municipalities and metropolitan areas to cope with the problems generated by their rapid population increase. It is not enough that Sao Paulo has the highest level of value added per capita in urban Brazil. The way in which local government is organized and financed conditions the response to population pressures even when resources are relatively abundant. Obviously the problem is even more acute when resources are relatively scarce, as in the case of Northeastern cities like Recife and Fortaleza. It is these institutional dimensions of Brazilian local government and the recent urban policy response that the next section of this report addresses. - 41 -

Table 19: WATER, ELECTRICITY, DURABILITY OF HOUSING, TV OWNERSHIP AND INDUSTRIAL INDICIES FOR METROPOLITAN AREAS, 1970

Water Electricity Durability TV Value Added Per Capita Connecti,Dn Connection of Dwellings Owners 1960 1970

(SE) Sio Paulo (a) Center 64.2 95.7 94.6 72.0 219 219 (b) Perl.phery 44.3 86.5 88.9 57.1 303 242

Rio de Janeiro (a) Center 82.8 95.1 90.1 70.7 72 92 (b) Periphery 46.5 78.7 88.0 43.3 52 0

Belo Horizonte (a) Center 47.9 84.5 91.0 50.4 31 33 (b) Periphery 39.7 59.0 90.5 25.1 181 147

(S) PSrto Alegre (a) Center 83.4 90.7 89.0 63.0 69 76 (b) Periphery 34.1 77.8 90.1 34.9 206 145

Curitiba (a) Center 47.6 90.2 91.8 59.1 50 70 (b) Periphery 8.6 44.8 84.8 19.9 75 41

(NE) Recife (a) Center 53.1 85.6 74.0 37.1 35 40 (b) Periphery 22.5 57.2 65.8 15.6 48 39

Salvador (a) Center 54.7 79.2 76.3 37.6 23 21 (b) Periphery 6.1 35.9 62.4 10.9 392 506

Fortaleza (a) Center 13.4 58.9 73.3 27.1 16 36 (b) Periphery 6.4 19.5 45.9 4.7 2 4

(N) Belem (a) Center 55.8 77.6 69.0 36.3 20 21 (b) Periphery 5.9 29.6 88.6 8.6 8 8

Sources: Same as Table 18. - 42 -

IV. The Organization, Planning and Finance of Brazilian Cities a) The Government of Urban Localities and its Finance

62. The basic unit of urban government in Brazil is the municiPio, roughly equivalent to a county in the United States, although there is no distinction between the "city" and "county" government as in many localities in the U.S. In Brazil there are nearly 4,000 municipalities, representing the lowest level of the three-tier federal republican form of government. Above the municipal governments are the state and the federal governments. The metropolitan areas are a new element of the Brazilian system of urban government and still in a very early state of development in terms of the disposition of their governmental functions and sources of finance vis-a-vis the municipal, state and federal governments. The metropolitan area legislation will be discussed separately below.

63. Over time, and especially with the changes introduced in the Constitu- tion adopted in 1967, Brazilian federalism has tended to place more and more power in the central government, especially as regards the power to collect taxes. Particularly important to the government of urban areas are the related questions of how responsibilities of governments at each level and the availability to each level of means to finance these activities are distributed and balanced. The changes in governmental structure which took place in the late 19 6 0's tended to put added stress on the capacities of local govere=ents to finance the activities for which they were made responsible. The main feature of the fiscal reforms that followed on this was the introduction of a system of direct revenue transfers from the federal and state government to municipalities. This put local governments in a more direct relationship to the federal government than before, and reduced accordingly the relative influence of the state governments. The degree to which this actually occurred has depended on the fiscal strength of the state government in question. Thus a relatively prosperous state government like Sao Paulo has played a much more important role in the affairs of its localities than poorer Northeastern states. Parallelling, and to a considerable degree extending, the fiscal influence of the federal government in local affairs are the federal agencies and companies that have been set up to promote infrastructural and/or general development in particular regions or sectors. These include both the public corporations like Petrobras, Eletrobras, Telebras and the Federal Railway System, as well as the regional and sectoral agencies like SUDENE in the North- east and the National Housing Bank (BNH) in the urban sector. The latter plays an extremely important role in the urban sector, a topic to which we will return below.

64. The fiscal transfer system is the mechanism by which Brazil has attempted to achieve a more even balance between the finances and the requirements of local government activities. The 1967 Constitution distributed taxing power between the three levels of government in a way that left most instruments in federal hands. Municipalities can tax urban property and services not taxed by state or federal authorities. State taxes include real property transfer taxes and a tax on value added within the state (Imposto sobre Circulacao de Mercadorias-- ICM). The volume of the latter depends on the amount of economic activity in the state and is at the root of differences in the degree of relative financial power of different state governments. All other taxes are - 43 -

collected by the federal government: import and export duties, income taxes, rural property taxes, and taxes on industrial products, financial transfers, transport wad communication, fuels, energy and minerals.

65. The system of transfers to municipalities functions with both federal and state tax revenues. The principal mechanism for the transfer of general federal funis is the Municipal Participation Fund (Fundo de Parti- cipac,o dos Municipios). There is also a State Participation Fund for transfer of federal revenue to the states. Six percent of federal revenues from income and industrial production taxes go to the fund for states, and another six percent to the municip&l fund. These precentages will rise progressively to ten percent by 1980. S3tate fund revenues are distributed according to size, population, and the inverse of per capita income. Ten percent of municipal fund revenues go to muaicipalities of the state capitals, and the remainder is divided among the remaining municipalities according to size, except that there is an upper limit (120,000 in 1974) above which the share does not increase, thus penalizing non-capital city municipalities larger than this. A similar size criterion applies in the distribution between state capitals. In addition to the general municipal and state funds, there is a special fund of the poorer states and funds for sectoral programs like roads, airports, education, etc., which are financed with revenues from specific taxes (fuel and electricity taxes, etc.).

56. Parallelling the system of transfers of federal revenues to states and municipalities is a system for transferring a share of state revenues to municipalities. There are some important differences in the way the state system works. It is based upon the tax on value added (ICM), which is the principal source of state revenues. Each munici'io is entitled to a share of 20% of the ICM collected on economic activity that is carried on within its boundaries. The tax rate varies by region and exemptions from the tax are given as incentives to regional investment. The state revenue sharing plan thus differs from the federal funds in that a municipality's share depends primarily on the amount of tax revenue collected in it and not on distribution formulas.

67. A considerable controversy has arisen over the disequilibrium in financing of local government because of the way in which the ICM revenue sharing works. The disequilibria are both inter and intra-regional. As seen in the previous section, there has been a trend toward growing concentration of industrial activity in the Southeast, especially Sao Paulo. Municipalities in poorer, less industrialized states collect and share in less revenue. Further, tax exemptions to promote industrialization in these areas reduce their current revenues even more. Other sources of local revenue --the tax on services and real estate-- are also linked to wealth and income and yield less in poor regions. For this reason, it has been argued that "?consuming" regions benefit less than "producing" ones in the ICM system. It is beyond the scope of this report to present a detailed study of this problem of ICM revenue sharing and its obvious implications for the execution of urban development policies. - 44 -

68. A further dimension that should be noted is the disequilibrium that has arisen within states, particularly in the metropolitan municipalities. Industrial activities are not spread evenly among the municipalities that comprise the metropolitan areas. Thus the shares in ICM revenues vary significantly between predominantly industrial municipalities and adjacent "dormitory" communities within the same metropolitan area. The size limit on shares in the federal revenue fund likewise works against the larger and more rapidly growing of these municipalities, and when the population consists mainly of low income groups, the capacity to generate revenues from municipal (service and real estate) taxes is also limited. These municipalities are thus endowed with very poor financial capacity to deal with the problems of "suburban" poverty outlined in the previous section. With rapid expansion of population and accompanying demands for urban infrastructure services, governments of the peripheral municipalities of the metropolitan areas have had to turn increasingly to sources other than the established revenue system to meet these needs. Given the growing concentration of technical capacity and financial resources at the state and federal government levels, not to mention their statutory power over infrastructure development, it is hardly surprising that municipalities are looking to higher levels of government for policies and programs to deal with both inter- and intra- regional disequilibria. b) Evolution of Urban Policy and the Structure of Brazilian Government

69. Brazil is currently in the process of articulating an explicit national urban policy. During most of the post-war industrialization period, urban problems were given low priority relative to maintaining the overall level of economic (especially industrial) expansion. The conscious effort to develop an urban policy can be dated from the preparation of the II Plano Na- cional de Desenvolvimento (II PND). The planning and execution of Brazilian development strategy must be seen first in terms of the structure of the federal government system. The growing influence of the federal government in the economy as a whole has been accomplished through the establishment of development banks, planning agencies and public corporations that undertake a wide range of infrastructural activities (transport, power, sanitation, education, and housing), as well as in basic industries like steel and petro- chemicals. As a result, the functional ministries to which these institutions/ enterprises report have become powerful agents in the economy. Planning and execution of specific agency projects have obvious effects on the municipalities in which they are located, and the priorities assigned to allocation of agency budgets have been important determinants of the evolution-of the urban system, whether or not an explicit urban policy significance was attached to them. In general, the functional ministries and subsidiary agencies have implemented their particular sectoral strategies with little or no concern for the overall effects on the evolution of the urban system.

70. The task of coordinating overall development strategy is assigned to the Ministry of Planning and Coordination (now Secretariat of Planning). Various plans have been elaborated by the Brazilian government over the course of the post-war industrialization period, with the II PND being the latest in the series. Brazilian plans are not the exhaustive catalogs of targets and budget allocations for a specific planning period that are typical of centrally planned economies. Documents like the II PND represent more of a synopsis of the dialogue between the technocrats charged with running the public sector - 45 -

and the presidential adninistration with respect to general objectives and priorities to be followed by the functional ministries and their subsidiary agencies. The typical policy response to an issue like urban growth is to set up a special commission consisting of representatives of the functional ministries whose activities bear on the problem, along with the Secretariat of Planning, and to create a special fund (or funds) to be applied in programs in the problem area or sector. This may eventually lead to the establishment of a special bank or agency for the area, which might then be assigned permanently to one of the functional ministries. The Secretariat of Planning is supposed to coordinate the activities of functional ministries in the execution of both their regular and "problem area" programs. However, the effectiveness of this coordinating role depends to a large extent on the relative political influence of the Secretariat with the presidential administration vis-a-vis the functional ministry whose activities are in question. There is a tendency for inertia on the part of established functional ministries to strongly dilute the coordinating role of the planners.

71. For the execution of their programs, the functional ministries work either directly or through subsidiary agencies with counterpart functional institutions at the state level. For example, Telebras, the new telecommuni- cations company, which reports to the Ministry of Communications, is currently consolidating existing telephone systems into a network of state-wide companies. When this process is completed, as has occurred to varying degrees in power, sanitation, and other areas of urban infrastructure, municipalities must contract with these state subsidiaries for execution of projects no matter whether financing comas from its own regular sources of revenue, an outside source like one of the special area funds, or an international loan. In most types of urban infrastructure these federal-state networks represent a very powerful force for any "coordinating" agency to deal with, whether it be from the federal, state or municipal government level.

72. An institution in this structure that is of particular importance for the urban sector is the National Housing Bank (BNH). This bank has grown so large in terms of program and budget that it overshadows the Ministry (Interior) to which it is attached. BNH was created in 1964 in response to the housing crisis that resulted from a combination of rent controls and recession in the construction industry in the period just prior to the revolution. BNH's financial strength derives from the Brazilian Workers' Guarantee Fund (Fundo de Garantia do Tempo de 3ervipo --FGTS) which draws from employer contributions equalling 8% of the monthly wage. FGTS deposits can be withdrawn by workers for various purposes: illness, temporary job loss, or purchase of a residence. Deposits left in the fund grow with interest accumulation and monetary correction to protect the principal against inflation, and thus represent a relatively secure mode of working cLass saving. Over time, the Fund has grown to a substantial level (roughly US$3 billion at the end of 1975).

73. The FGTS deposits have been placed at the disposal of the BNH for its activities in housing and urban development. BNH channels resources to the housing sector through the National Housing Finance System (SFH) and related specialized BNH housing finance programs. About half of all residen- tial mortgages in Brazil are BNH funds. The distribution of BNH housing loans - 46 - between different income groups has come. under criticism because working class savings were essentially being used to finance middle and upper class housing. A new program for lower income housing (PLANHAP) has been introduced with the objective of obtaining a more equitable distribution of these resources.

746 Specifically in relation to the question of urban development, it is important to recognize that BNH activities reach far beyond housing finance. They include a number of complementary programs in the housing field as well as a broad range of urban development activities. Some notion of the relative importance of the various components can be obtained by examining a breakdown of BNH's 1976 budget proposal:

PLANNED BNH BUDGET ALLOCATION, 1976

1. Basic Housing Finance 38.0%

2. Complementary Programs in Housing 10.1% (land purchase, common areas in multi-unit projects, etc.)

3. Urban Development 35.4Z.

a) Urbanization (4.6) b) Water and Sewerage (19.8) c) Transport (2.4) d) Community Affairs Equipment (2.9) e) Support of Pre-investment Studies (1.1) f) Support of New Growth Centers (4.6)

4. Redevelopment Programs 16.5% (both housing and complementary urban redevelopment)

By far the most important current effort of BNH in the urban development field is the water and sewerage program (PLAKASA). The objective of this plan is to provide water for 80% of the population living in four-fifths of Brazil's urban communities by 1980 and to furnish sewerage services to residents of the main urban centers. The execution of the plan follows-:almost exactly on the model described above: A system of revolving funds (FAES: Fundo de Xguas e Esgostos) has been established in the states which finance approximately 50% of project costs with the remainder coming from BNH loans. The PLANASA program is now moving forward with impressive momentum. c) The Comissao Nacional de Regioes Metropolitanas e Politica Urbana -CNPU

75, The 1970's have brought growing government recognition of the detrimental effects on the urban system of viewing urban policy as a relatively exogenous by-product of industrial and sector development strategies. Discussions of new objectives for national policy during preparations of the II PND heighten- ed this perception. The specific response was tha establishment in June, 1974, of a national commission on metropolitan areas and urban policy (CNPU). The evolution of the CNPU has followed the general pattern described in the - 47 - previous section. It :Is an interministerial commission with members representing the Secretariat of Planning (SEPLAN) and Ministries of Interior, Transport, Finance and of Industry and Commerce. The President of BNH is also a member. The CN?U's mandate, as stateO in the decree law that established it, is: (1) to accompany the implantation of Brazil's system of metropolitan regions;

(2) to propose directives for the national urban development policy and formulate strategy to implement these objectives;

(3) to propose the norms and instrumentalities needed for urban development in Brazil; and

(4) to relate to the Ministries, Regional Superintendencies, and government agencies involved in the execution of national urban policy so as to insure a coordinated implementation of programs and projects.

Besides the actual ministerial level commission, the CNPU functions through an administrative and professional staff (currently 15 people) provided by the Secretariat of Planning and the Ministry of the Interior.

76. Resources for, implementation of the urban strategy derive from various funds at the disposal of ministries (including SEPLAN) that make up CNPU. In addition, two special funds were established late in 1975 specifically for projects making up the new urban strategy. The first is a national urban development fund (Fundo Nacional de Apoio ao Desenvolvimento Urbano -- FNDU), which is described in the document establishing it as being "a catalytic agent and coordinating element with respect to all the other sources of finance for urban areas." This suggests that the FNDU is intended to provide some financial leverage on allocation of money in other funds. The CNDU documents describe a System of Funds for Urban Development comprised of the FNDU, of regional funds (administered by the Bank of Brazil, the Bank of the Northeast, and the Bank of the Amazon) , resources of the BNE, and the National Development Bank, and special funds for social projects and pre-investment studies. The description of this system appeared in the exposido de motivos for the FNDU, but not in the actusl decree law establishing it.

77. At about the same time that the FNDU was being set up, an even larger fund for urban transport systems was being developed. This led to the Fundo de Desenvolvimento de Transportes Urbanos -FDTU and the establishment of the Brazilian Urban Transport Company (EmpreEsa Brasileira de Transportes Urbanos). The application of these funds to urban priorities is supposed to be coordinated by-the CXPU. However, the system is subject to all of the structural limitations described in the previous section. Even with the establishment of the FNDU, the CNPU has no clear statutory power over the constituent funds. Its coordinating leverage depends largely on the administrative and political support given to its executive staff and on the cooperation of member ministries to support communal vs. sector-specific priorities. One - 4 8- indication of the dilemms is that while the voluminous urban transport fund is to be coordinated by CNPU, the Brazilian Urban Transport Company that will spend the money is responsible to the Ministry of Transport.

78. At the time the CNPU was actually established, major steps had already been taken on its two main objectives. The directives on a national urban policy for the II PND had been drafted, and the decree law establishing eight of the nine metropolitan regions had been adopted (the special law for Rio de Janeiro appeared soon after). The initial tasks of the CNPU were to assist in implementation of the metropolitan area laws and to find ways of implementing the urban development strategy of the II PND. d) The Metropolitan Areas

79. The possibility of achieving a major change in the management of large cities in Brazil was created in 1973 by the establishment of eight metropolitan regions. A special law for Rio de Janeiro followed in 1974 after the former state of Guanabara rejoined the state of Rio de Janeiro. The metropolitan regions consist of groups of municipalities selected on the basis of economic and demographic affinity to the main metropolitan municipality. The metropolitan area law named the municipalities for each region and directed that two councils be set up in each one. The first was a Deliberative Council, with 5 members (with a representative from the core municipality, one from the periphery and 3 to be selected by the governor) that was to elaborate an integrated development plan for each region and to coordinate execution of programs and projects of common interest to the municipalities, with the special goal of unifying those relating to common services. The second, a Consultative Council, has representatives of all municipalities, was to keep abreast of and make suggestions regarding the actions of the Deliberative Council. Defined as common services were:

1. Economic and social planning;

2. Water, sewerage and sanitation;

3. Urban land use;

4. Transport and highways;

5. Production and distribution of piped natural gas;

6. Water resources and pollution;

7. Others, as the Deliberative Council decides. Preferential treatment in federal loan programs is promised for municipalities that participate in integrated metropolitan area projects.

80. Work on plans for some of the metropolitan regions (Sao Paulo, Belo Horizonte, Porto Alegre, Curitiba, and Fortaleza) was in various stages of completion when the metropolitan area law was decreed. Though much effort was given to achieving a balance between the socio-economic and physical planning dimensions of metropolitan area development in the plans, the physical - 4 9 -

dimension clearly predomLnated once the metropolitan area agencies turned from planning to implementation of projects.

81. One of the first major acts of the CNPU, when it met as a formal commission, was to issue further directives for the metropolitan area institutions. It set as priorities: (1) restressing the need for socio-economic planning; (2) a common urban land use policy; (3) improvement of mass transport facilities; and (4) priority for water and sewerage among common urban services. Later the Deliberative Councils of the metropolitan areas were strengthened by making the respective State Governor the president of each council.

e) Urban Development Strategy of II PND

82. The II Plano Nacional de Desenvolvimento (1975-79), published late in 1974, presents specific objectives on the evolution of the Brazilian urban system and proposes mechenisms to design and execute an urban development strategy. The explicitness of the II PND contrasts with the loosely worded acknowledgements of urban problems and the need to face them that typified previous plans. As with the Brazilian planning process in general, the II PND's urban component represents the broad concensus of views current within the government structure by mid-1974 on the urban question. It consists of a prognosis of Brazil's urban problem, a regionally oriented urban development strategy, and a set of recommendations for instrumentalities to implement the strategy. The mnin themes of the prognosis of the urban problem are the rapidity of recent urban growth and the disequilibria in the urban system that have accompanied this growth. There is very little detail in the published plan document, and there appears to be no published "diagnostic" analyses (such as those that accompanied some earlier plans), of these disequilibria. There is an unpublished IPEA study of the urban situation from which much of the II PND was drawn.

83. In the II PND, the growth of the metropolitan areas is characterized as "premature", and small city growth as "excessive". In view of the purely demographic growth described in section II of this report, it is difficult to justify these adjectives. Presumably reference is being made to the growth of these areas relative to development of economic (especially employment) and urban service delivery capacities. It is also argued that there is an insuffi- cient number of medium-sized cities in the Brazilian urban system. There is specific reference to the unequal distribution of urban infrastructure, concentration of industrial activity in Greater Rio and Sao Paulo, and corresponding deficits in Recife, Fortaleza, and Salvador. These are clearly related problems given the relation between the local revenue generating capacity and the level of industrial activity. Lls6 cited as a disequilibrating factor is the historical tendency toward concentration of the urban population along the Atlantic coast. The main policy objective is the establishment of a better structured urban system, one that will succeed in "raising the standards of urbanization and the quality of life". For this the plan-recommends: (1) setting up metropolitan areas and identifying the functions for metropolitan area government; (2) defining secondary nuclei that would help to reverse the tendency toward - 50.- concentration of urban growth; (3) achieving a reversal of the concentration that has occurred in Greater Rio and Sao Paulo; (4) revitalizing the economic base of the metropolitan areas in the Northeast; and (5) promoting urbanization in the new settlement areas in the Northern and Central-Western regions.

84. The II PND presents several key concepts which underlie the new urban development strategy. These are as follows:

1. "Decentralization" or, more literally, deconcentration, (desconcentrac,o) refers specifically to containment of the growth of Rio de Janeiro and Sao Paulo, because of what is viewed to be excessive growth of both population and industry there. Decentralization means that further expansion of existing industry and implantation of new industries will not be actively encouraged.

2. "Putting order into the process" (ordenagco) refers to the metropolitan areas of Belo Horizonte, Curitiba, and P6rto Alegre (the remaining ones in the Southeast and South). The objective here is to control their expansion in order to avoid the pattern that has led to the problems evident in Rio de Janeiro and Sao Paulo. The notion extends to medium sized cities in these regions and implies that urban development strategy should seek a more optimal hierarchy of city size in these regions.

3. Revitalizing (dinamizarco) refers to the metropolitan areas (Salvador, Fortaleza, Recife) and cities of the Northeast, with policy measures to restore balance bet-ween their economic base and population increase. It is also applied to middle sized cities in other regions.

4. Promotion (promo2go) of urban development and urban settlement refers to the cities of the Northern and Central-West regions, especially Belem and Brasilia.

85. These policy objectives emphasize the strong regional orientation to the strategy. Cities in each region are classed by size and function in accordance with these aims and planning concepts. For the Southern and South- eastern regions the strategies are similar. The containment policy for Rio and Sao Paulo envisions a slowing of industrial expansion in an area extending about 100 kilometers from both cities and in the coastal strip that runs between them. Special emphasis is given to chnnneling further industrial expansion to the medium sized (50,000 and up) cities of these regions. The regional metropolitan areas (Belo Horizonte, Curitiba, Porto Alegre) are put in an intermediate status between the "containment" of Rio and Sao Paulo and "revitalization" of the medium sized cities. Special treatment is accorded the tourist areas and resort cities. In the Northeast the major emphasis is on expansion of economic activity and infrastructure to catch up with population growth in both the metropolitan areas and medium sized cities. The latter are also mentioned for their potential role in containing migration and supporting agricultural and agro-industrial activities. In the North and Central-West, development of an urban system to accompany the settlement process occurring in those regions is the central theme. Included in the list of priorities are adequate administrative apparatus for the new cities and guidelines for the growth of the Brasilia region. - 51 -

Table 20. SUMMARY OF URBAN GROWTH STRATEGY IN II PND

Region/Function Strategy Cities

SOUTHERN REGION

1. Regional Metropolitan Control Porto Alegre and Curitiba. Areas

2. Sub-regional Centers Revitalization Florianopolis, Londrina, Pelotas, Caxias do Sul, Santa Maria, Blumenau, Joinville and remaining urban centers of 50,000 or more inhabitants.

3. Tourism Areas Disciplined Iguaqu Falls, Paranaguia, Laguna, Expansion and Granado, coastal regions. Revitalization

4. Special Programs Promotion Itaipu.

SOUTHEASTERN REGION

1. National Metropolitan Containment Sao Paulo and Rio de Janeiro. Areas

2. Regional Metropolitan Control Belo Horizonte, Vitoria, Santos, Areas and Conurbaticns Campinas.

3. Sub-regional Centers Disciplined Juiz de Fora, Riberao Preto, Growth and Sorocaba, Campos, Jundiai, Revitalization Volta Redonda, Bauru, Sao Jose' dos Campos, Piracicaba, Uberlandia, Uberaba, Governador Valadares, Sio Jose do Rio Preto and remaining centers of 50,000 or more.

4. Tourism Areas Disciplined Historical cities, hotsprings Growth resorts, coastal regions.

CENTRAL WESTERN REGION

1. Geoeconomic Region Disciplined Federal District and Environs. of Brasilia Growth and Revitalization

2. Sub-regional Centers Revitalization Goiania, An'apolis, Cuiaba', Corumba, Campo Grande, Dourados.

3. New Settlement Areas Promotion Araguina, Aragarcas, Caceres, Rondonopolis. - 52 -

Table 20 (Continued)

Region/Function Strategy Cities

4. Tourism Areas Promotion Goias, Caldas Novas.

NORTHERN REGION

1. Regional Metropolitan Disciplined Belem Area Growth and Revitalization

2. Sub-regional Centers Revitaliztion Manaus, Macapa, Boa Vista, Rio Branco, Santarem.

3. New Settlement Areas Promotion Capanema, Monte Alegre, Maraba, Alenquer, Parintins, Itacoatiara, Cruzeiro do Sul, Imperatriz, Icana, Humaita, Labrea, Con- ceiqiao do Araguaia, Caracarai, Tucurui, Benjamin Constant, Itaituba, Altamira, B^oca do Acre, Tefe, Paragominas, Coari, Vilhena.

NORTHEASTERN REGION

1. Regional Metropolitan Disciplined Recife, Salvador and Fortaleza Areas Growth and Revitalization

2. Sub-regional Centers Revitalization Joao Pessoa, Sao Luis, Teresina, Natal, Maceio, Aracaj6, Campina Grande, Feira de Santana, Caruar'u and other centers with regional growth pole function.

3. Tourism Areas Promotion Historical cities and coastal regions.- _ 53 -

86. The final section of the II PND is devoted to the mechanisms for implementing this strategy. The CNPU is described as an auxiliary organ "whose particular task is to guide and accompany the implantation of the controlling organs of the metropolitan regions, that is, the deliberative and consultative councils." The CNPU has since issued directives for the councils and participated in the move to strengthen them by nominating the state governors to their membership. CNPU (directives have also been issued with regard to the planning and administrative organs of each of the metropolitan regions.

87. Not mentioned in the II PNV or CNPU documents are the state level planning units that have been set up to coordinate urban policy at the state level for the non-metropolitan cities. Several states, (including the two most populous, Minas Gerais and Sao Paulo) have established such units within the State Planning Agencies, In both of these states, documents detailing the overall state urban strategy have been prepared. The work done in Sio Paulo is of particular interest because the director of the planning group had participated in preparatory work on the II PND. Its main objective is identification of medium sized nuclei in the State of Sao Paulo where the deconcentration strategy canl be directed. A broadly similar study has been completed by SUDENE for the Northeastern region. The techniques employed were mainly descriptive (factor analysis), and the list in Sao Paulo is comprised of cities of common social, economic, and locational characteristics. In the Sao Paulo State urban strategy, the same planning concepts in the II PND are applied to different parts of the state that had been selected according to the factor analytic procedure.

88. Another initiative in the urban sector in which the CNPU has had a role and which is important because of its social equity content is the project relating to Urban Social Centers (Centros Sociais Urbanos --CSU). These are to be community centers offering informal education, health care and nutritional help, and cuLltural and leisure time activities in neighborhoods. Approximately 660 such centers are to be constructed over the period 1975-1979, representing an investment cf roughly US$250 million at 1975 prices.

89. Parallelling the urban policy of the II PND and its further articula- tion by the CNPU is the recent statement of national policy on internal migration. This document is the joint effort of a commission of technicians from the Secretary of Planning and the Ministry of the Interior, which has been concerned mainly with research on migration. Members of the commission have all participated in the national research program on inter-nal migration, which has produced a series of studies on migrants based on data from the 1970 census (the focus is on migrant-native differences and on characteristics of migrants by length of residence at destination). The findings suggest that migration per se may not be the root problem in urban poverty. A fundamental qualification on the policy significanpe of these findings is that the census data permit analysis of migrants oriy at their destination, whereas the most relevant questions for policy may be those relating to conditions at the point of origin. The migration policy statement echoes the II PND's directives on the spatial aspects of Brazilian development, but is less specific on directives. _ 54 -

It proposes further research on the rationale for the spatial distribution of the population and migration flows, coordination with other government agencies whose activities could affect these flows, and programs of services for migrants. Coordination of the activities of the migration commission with the CNPU is mentioned briefly in the document, but there appears to have been little coatact up to the present.

f) The New Urban Strategy --An Assessment

90. It is clear from the foregoing that the urban policy described in the II ?ND is still at a very formative stage. What has been done represents an important step in the direction of rationalizing the development of Brazil's urban system. At the same time, important institutional limitations stand in the way of effective implementation of the new urban policy. First, the statutory power of the urban planning organs is limited, a problem that has plagued efforts at general economic planning in Brazil for some time. Secondly, budget decisions and execution of projects affecting urban areas are still very much in the hands of the federal sectoral ministries (or state secretariats) and their subsidiary superintendencies and'autarchies (public corporations for planning and execution of sector projects). The position of coordenacao, either at the federal or state level, is thus bureaucratically weak even when it has the political support of the executive. Recognition of the adverse effects of sectoral fragmentation of activities affecting the urban sector led to the formation of CNPU, but it is still not clear how effective this mechanism will be for putting order into the system. The situation of the CNPU is delicate in that major executors of urban programs, for example the BNI and Ministry of Transport,which are responsible to it, nevertheless have major responsibilities-and implementing agencies of their own in related fields. it remains to be seen whether the CNPU with its fund for urban development can really exercise any leverage over these institutions and implement its urban strategy.

91. A second set of observations relates to that strategy. Many of the recent urban policy proposals are based on quite limited evidence and policy research. While it is recognized that the concentration of industrial activity in Sao Paulo and the problems of urban services in the Northeast are severe, the rationale for seeking the solution in the shift of industrial production to medium cities has not been subjected to thorough empirical analysis which establishes the economic costs and benefits of decentralization. Nor is it clear that "dinamization" really can contain and re-orient migration as is envisioned in the plan.

92. A final point relates to the centralization that characterizes Brazilian urban development policy. Growing centralization (coexistant and not inconsistent with sectoral fragmentation) of government activities affecting the urban sector makes it logical that the institution designed to seek solutions work from the top. Still, the problems have their roots in the municipalities, and it is not clear that the new approach will be successful in reachirng solutions at this level.