GDPMPC1872 14/4/2009 11:55 Preliminary version Do not quote
Transcrição
GDPMPC1872 14/4/2009 11:55 Preliminary version Do not quote
GDPMPC1872 14/4/2009 11:55 Preliminary version Do not quote Income per capita of Brazilian municipalities in the 1870´s∗ Eustáquio J. Reis [email protected] This paper estimates income per capita of Brazilian municipalities in the 1870´s. Estimations use data on wages of municipal civil servants in 1876 published by the Statistical Report of 1878 (Brasil 1878); demographic data from the Census of 1872 (Brasil 1876); and geographic data obtained from various sources. The use of data from different years justifies the loose reference to the period of estimation. The ostensible lack of reliable statistics for sub-national levels of the Brazilian economy in the XIX c. validates the estimation exercise. The few estimates available are based upon scanty data using methodologies which combine simple macroeconomic hypothesis about growth rates for the country and its main regions with microeconomic data for specific localities (municípios or freguesias) and time periods lacking a satisfactory statistical coverage (Contador and Haddad 1980; Neuhaus 1980; Santos 1980; Goldsmith 1986; Maddison 2003; Bértola, Willebald et al. 2006; Monasterio and Zell s.d.). The models used in the estimation assume that wages of municipal civil servants were set autonomously by local authorities and reflected the labor productivity which, according to a production function, was determined by the demographic conditions (distribution of population according to sex, age, occupation and the free or slave juridical status) and the geographic characteristics (distance to sea, altitude, climate and soil attributes) of each municipality. The basic idea is thus to filter the data on average wages of municipal civil servants by the available information on the ∗ Paper to be presentado at the UCLA Von Gremp Wrokshop in Economic History, April 22, 2009. Project undertaken by Research Network on Spatial Models (www.nemesis.gov.br) based at IPEA, Rio de Janeiro, under the financial support of Faperj/CNpq (Proc. E52 168.171/2006 /Pronex). Grateful acknowledgements to both institutions, as well as to Márcia Pimentel for the computer assistance and Bárbara Bravo for the data compilation. With the usual disclaims, the paper benefited from comments of Cláudio Ferraz, Ajax Moreira, Felipe Fernandes and the participants of the seminars at IPEA/RJ, and at the XVII Seminario de Economia Mineira held in Diamantina, November 2008, where a slightly modified version of the paper was presented. 2 productive structure of the municipality in an attempt to clear them from eccentric values. The first section of the paper makes some analytical considerations on the specification used for estimation. The second section describes data and variables constructed for the analysis. The third presents estimation results. The last section makes an assessment of the estimates of municipal and provincial income in 1872. Analytical considerations The basic hypothesis is that municipal civil servants remuneration were set autonomously by local authorities. The main motivation for this assumption is the empirical finding that average wages of municipal civil servants in the Statistical Report of 1878 display significant variance both within and between provinces. The presumption is therefore that there were no strict provincial rules in the determination of municipal civil servant wages. Historical evidence on the decision processes about municipal budgets and civil servants hiring are sparse and ambiguous. Among contemporary opinions it is possible to find non compelling support to the hypothesis that the municipal economic regimen was under the discretionary decision of municipal assemblies.1 A second hypothesis is that average wages of municipal civil servants reflected the average income of the labor force in the municipality. The rationale for this assumption are, on the one hand, that competition in the labor market would preclude local authorities from paying wages to civil servant far below the median income of free workers in the municipality.2 On the other hand, the perception of injustice and political opposition would preclude wages of municipal civil servants far above the median income of free workers in the municipality, in particular if wage payments relied on local taxation. 1 Thus, authors like (Carreira 1980; Bessa 1981; Tavares Bastos 1997) made no reference at all to the ways and means of municipal budgets or to the determination of municipal public servants pay. (Rebouças 1875):136 mentions that the economic and municipal regimens were prerrogatives of municipal assemblies. “Estas corporações tem receitas próprias para acudir ás competentes despezas... e, têm, pelo Acto Addicional, os meios para occorrer (sic) às despezas de seus municípios.” However, he contradicts himself in the next page by describing as atributes of Provincial Assemblies “ [f]ixar as despezas provinciais e municipais, as primeiras sobre o orçamento do presidente da província, as segundas sobre orçamento das respectivas câmaras; decretar os impostos e meios para a receita provincial e municipal ...; crear e supprimir empregos provinciais e municipais, decretar as obras públicas da mesma natureza (p. 135). 2 The Provincial Report of Rio de Janeiro in 195? present some evidence on municipalities which were not able to hire municipal civil servants because wages were too low. 3 The third hypothesis specifies the determinants of production and distribution of an agrarian society. The average income of the labor force was mainly determined by the productivity of labor in agriculture which depended on the demographic and geographic factors describing, respectively, the quantity and quality of the labor force and land availability in each municipality. Slavery played a very important role affecting both the productivity of labor and the rules conditioning the distribution of output.3 The main demographic variables used are the number of (free) households, size of population, as well as its distribution according to age, gender, litteracy, professional occupation, and juridical status (free or slave). Unfortunately, the distribution of slave property which was an important aspect in the determination of both output and distribution of income was not compiled in Census of 1872. The main geographic characteristics of the municipality are area; altitude; soil attributes, erosion susceptibility, and agricultural suitability; temperature and precipitation; among other. The distance to the sea is another important geographic attribute in the determination of the average income of a municipality to the extent that it affects the transport costs to both major domestic and international markets. Additionally, it is legitimate to assume that administrative and political factors played an important role in the determination of the number and wages of municipal civil servants. In this way, it is reasonable to assume that provincial differences in terms of tax base and fiscal institutions would be reflected in the number and wages of municipal civil servants. This is particularly true in the case of the Imperial Capital (Rio de Janeiro) and provincial capitals which, for that reason, were entitled to larger shares of the transferences of imperial and provincial taxation revenues and provided a bigger and more diversified bundle of public goods. Other important political factors were perhaps the concentration of power as reflected in available indicators of political participation of the population which probably affected both the number and the remuneration of civil servants. Finally, it should be kept in mind that wages of municipal civil servants are used to estimate the average income of the (free and slave) labor force of the municipalities. To 3 If y is called output per labor force population, p, f, s, the share of proprietors, free workers, and slaves in the labor force, and αf and αs, the labor productivity of free worker and slaves, respectively; assuming that proprietors do no work, it is possible to show that y = (1/h).(f. αf + s. αs) where h + f + s = 1 4 obtain the income per capita of the municipalities we have to multiply it by the size of labor force and to divide the product by the size of total population of the municipalities. Sources and data The Statistical Report of 1878 (Brasil 1878):105-116) reports the number (NFPM76) of and annual expenditures (STFPM76) with municipal civil servants which had fixed wage contracts for 667 Brazilian municipalities in 1876. The total number of municipal civil servants was 4,449, out of which 3,504 had fixed wage contracts and the remaining 945 perceived a percentage of what they collected of municipal taxes. On average, therefore, municipalities had 6.7 civil servants, out of which 5.2 had fixed wage contracts and 1.5 received a percentage of municipal taxes.4 The annual expenditure with payment of wages fixed by contract in 1876 was 1,032,661 mil-réis, thus implying a national average wage of 295 mil-réis. The amount paid as percentage of tax revenues was not published for any municipality with the excuse that there was no way of calculating it. If we assume, as seems legitimate, that there was no intersection between the set of municipal civil servants under fixed wage contracts and those which received a percentage of taxes, then the average fixed wages are not distorted on this account. The other main source of data is the Census of 1872 (Brasil 1876) which was actually undertaken from 1872 to 1874 depending on the province in case (Puntoni ; Puntoni (coord.) 2003). The cross identification of municipalities in the Census and the Report of 1878 resulted in a sample of 615 observations. For that sample, the number of municipal civil servants with fixed contract was 3,664 and the national average wage was 212,8 mil-réis. From the Census of 1872 the following demographic variables were obtained: the number of “fogos” or households headed by free men (FOG72); the size of total population (POP72) categorized by free men (HLTi,e, o), slave men (HETi,e,o), free women (MLTi,e), slave women (METi,e,o) according to classes of age (i = 11 a 15, 16 a 4 In the summary tables of the 1978 Report reference is made to 4,682 civil servants, out of which 3,718 paid by fixed wage contracts and 964 paid by a contract of tax revenues collected but municipal data are not available for all municipalities: for some municipalities of the provinces of Amazônia and Pará fiscal workers from suburban districts were not included. In the administrative division of Brazil of the report (Brasil, 1878:173), the total number of municipalities referred is 738. Probably, some of the municipalities were installed yet or they did not report the number and the expenditure with wages of municipal civil servants. In the 1872 Census, the number of municipalities is 643. 5 40 e 41 a 60 anos), and litteracy (e = literate or illiterate); the distribution of masculine population according to 10 classes of professional occupation (HTPROFo) ; and, finally, the population of non-African foreigners (ESTRL72). The geographic variables available at municipal level were restricted to latitude (LAT_GMS) and longitude (LONG_GMS), altitude (ALT_M) and the distance to the sea (DSHOR) of the seat of municipalities. To include other geographic variables, municipios have to be aggregated in minimum comparable areas between the Census of 1872 and 2000 (AMC 1872-2000) because georeferenced information at municipal level were not available before the Census of 1991.5 Thus, only for the Census of 1991 and 2000 it became possible to superimpose geographical attributes on the maps of municipal networks to obtain measures like the geographic area of minimum comparable area (AMC1872-2000); the proportion of this area in different classes of altitude in meter (PALTx), soil declivity (in degrees) or susceptibility to erosion (PEROx), soil agricultural aptitude (PPTNCx); geo-morphological classes of soil (PSOLOx); the average precipitation (PRE30) and the average temperature (TMP30) of municipalities in the different seasons of the year, namely, summer (December to February), autumn (March to May), winter (June to August), and spring (September to November) Soil attributes were obtained from georeferenced interpretations of satellite images from recent decades (IBGE 2003; Anderson and Reis 2007) . Climate variables refer to seasonal averages for the period 1961-90 obtained from interpolation of observations from Brazilian meteorological stations (Anderson and Reis 2007). Since the figures for climate variables are average values for a thirty year period, it is fair to assume, that they are time invariant, as in the case of other geographic variables. Other sources of data are the electoral lists published in the Statistical Reports of 1873, 1874 and 1875 ((Brasil 1873; Brasil 1874)) providing for each parish the number of citizens qualified as voters (VOTANTE74) and elegibles (ELEGIVEIS74) to the electoral colleges which, in a second turn, nominated representatives to the Provincial and Imperial Assemblies, and to the Senate. The number of voters and elegibles in each parish were aggregated to municipal level thus making possible to construct measures 5 In constructing AMC 1872-2000, it should be noted that the State of Acre was not part of Brazil in 1872. In addition, the states of Mato Grosso and Rondônia are joined with Amazonas and the North Region; and the state of Tocantin is included in Region Center-West (Reis, Pimentel et al. 2007) 6 of political participation or political power concentration using the proportion of adult population which is able to express themselves in the political system or to access its decision echelons. In addition to civic status criteria – like to be married or be older than 25 and autonomous, or else to be a liberal professional, priest, civil servant, or military -- the electoral legislation imposed annual income requirements of 200,000 mil-réis for a voter and 400,000 mil-réis for an elegible. Thus, in addition to political participation, the proportion of voters and elegibles give at least a rough idea of income distribution in each municipality. Finally, variable dummies were used to capture the differential effects of provinces (DUFxx where xx is the IBGE state code), provincial capitals (UFCAPITAL) and the Imperial Capital (or Municipio Neutro located in the city of Rio de Janeiro, RJ) on the wages paid to municipal civil servants. Tables A.1 and A.2. in the Appendix present, respectively, the list of variables and the basic statistics for them in the most comprehensive sample of municipalities with data. Table 1 present the distribution of the number of municipalities and the civil servants with fixed wage contracts according to the classes of value of average wages. Figure 1 maps the geographical distribution of these classes for the minimum comparable in the period 1872-2000 showing that, in 1876, more than 63 percent of the municipalities paid annual average wages below 200 mil-réis, 25 percent between 200 and 400 milréis, and 12 percent higher than 400 mil-réis. In terms of number of municipal civil servants, analogous figures were 58, 26 and 26 percent, respectively. Geographically, municipalities which paid higher salaries were concentrated in the South and CenterSouth regions of the country terms, and in provincial capitals which paid annual average wages above 800 mil-réis. Figures in Table A.2. show that, in 1872, Brazilian population was 9.9 million persons living in 1.33 million free headed households dispersed in 636 municipalities which were located, on average, 204 km from the sea at an altitude of 336 meter. The labor force of the country was composed of 1.1 million male and female slaves (older than 10 7 and younger than 60 years old) and 2.48 million free men (older than 15 and younger than 60 years old).6 The non-African foreigners represented only 2% of the population. Model estimation results Figures of Table A.2. also show that the number of municipal civil servants represented a negligible 0.18% of free labor force in the country. Thus it is legitimate to assume they do not introduce problems of endogeneity in the estimation since their wages were unlikely to affect the wages or any other attribute of the occupational distribution of the labor force of the municipality. However, since the number of municipal civil servants show wide variance, estimations were made using a weighted least square method using the number of municipal civil servants with wages fixed by contracts as weights. The first estimation (Model I) uses data at the municipal level and, therefore, can not include geographic factors, except for latitude, longitude, distance to the sea and the dummies for capitals. Table A.3. presents the estimation results showing that the model is able to explain, approximately, 79% of the total variance of average wages fixed by contracts. Since the main purpose is to estimate labor productivity based upon demographic and geographic factors the goodness of fit is encouraging. Though there is no particular interest in effects of specific variables, it is interesting to observe that the following parameters were significant at the 5% level of significance (the signal of the parameter is reported in the parenthesis): dummy for imperial capital (-), dummies for provincial (+) capital, the distance to the sea (-), the altitude (-); the size of population (+); the longitude (-); the percent of non-African foreigners in population (+); percent of population in industrial and commercial professions (+); percent of population in salaried positions (-). In addition, given the likelihood of multicolinearity among explanatory variables, tests of significance were made for the joint effects of some subsets of variables. Results were significant for the joint distribution of the classes of occupational professions; the age classes for male slaves; and provincial dummies, but not significant for the age classes of free men and women as well as for slave women. A suggested interpretation of the results is that major explanatory variables are proxies of "urban" activities and locations near sea ports. The exceptions are given by 6 Since women were mainly dedicated to household activities, they were excluded from the labor force to make the estimation of output consistent with figures from later Census years. 8 negative sign of the Imperial Capital which is probably related to the large number and low wages paid to municipal civil servants, and the percent of population in salaried positions which lacks an obvious explanation. On the other hand, the age distribution of male slaves was more likely to be a proxy of rural activities, in particular related to coffee production in the South-Center regions. Figure A.1 presents the scatter diagram of the estimated and observed values of the logarithm of average wages of municipal servants in 1876. The diagram shows strong heteroskedasticity with underestimation in the case of high wages municipalities and overestimation in the case of low wage municipalities. The second estimation (Model II) uses data at the level of minimum comparable areas in the period 1872 to 2000 (AMC 1872-2000) thus making possible to include the all the available geographic variables in the specification. The results presented in Table A.4. show significant improvements compared to the previous estimation. In terms of goodness of fit, corrected R2 is now close to 0.83. Keeping the same urban flavor of the previous model, the significant variables are the dummies for imperial (-) and provincial capitals (+) the size of population (+); the number of households (+); the number of municipal civil servants (-); the percent of non-African foreigners in the population (+) probably related to ports; the percent of slaves older than 10 and younger than 16 years old (+); free women older that 40 and younger than 60 (-) which are out of the labor force; percent of population in industrial and commercial occupations (+); percent of population in agricultural occupations (-); and 7 out of 20 provincial dummies; latitude (+); precipitation in spring (September to November) (+), highest class of altitude (-). Tests of the significance for joint distribution of some subset of variables show that the distribution of slave men and free women according to age classes; professional occupations; seasonal temperature and precipitation; agricultural aptitude of soil; and provincial dummies are significant factors. On the other hand, the distribution of free men (probably captures by professions) and slave women according to age classes, classes of soil suitability to agriculture and susceptibility to erosion did not come out as significant factors in the determination of wages. Figure A.2 shows that the introduction of geographic variables brings a significant reduction in the heteroskedasticity problem observed in the previous estimation. The map in Figure 2 show estimated values of the average wage of municipal civil servants 9 which, by assumption, is made equal the income per worker of Brazilian municipalities in 1872. Figure 2.A presents the spatial distributions of the residuals of the regression. Finally, estimation of a third model (Model III) including the degree of political participation and concentration of political power in 1874 as explanatory variables did not show significant improvements when compared to Model II. In addition, the introduction of these variables implied a significant reduction in the sample of municipalities and brought a possible endogeneity problem. Thus, results are not reported and the decision is to keep the results of Model II in the remaining analysis. Municipal income estimates Income estimates are assumed to refer to 1872, despite the fact that wages of municipal civil servants are from 1876. The use of the demographic structure of the Census of 1872 justifies the assumption. A supplementary assumption required is that changes in wages and prices between the two reference years were negligible. Giving support to this assumption, according to (Catão 1992) wholesale price index, inflation from 1872 to 1876 was merely 0.65%.7 Another main assumption for GDP estimation is the definition of labor force as the sum of free male population between 16 and 60 years old with male and female slave population between 11 and 60 years old. For the whole country, this hypothesis implies a labor force equivalent of 34% of total population. The age classes selected for the definition of the labor force as well as the exclusion of free women are quite arbitrary assumptions. A possible alternative candidate would be to define the labor force based upon the population with declared professions in the Census of 1872. That implies excluding from total population all the persons under the heading “Without profession” which includes both persons in the labor force who had not declared a profession and the persons in the age classes outside the labor force, that is, children and old age persons. In this case, the labor force would be equivalent to 60% of total population. The differences between the two definitions are thus quite significant and the first definition is preferred because it sounds more reasonable when compared to other Censuses years. 7 However, (Lobo, Canavarros et al. 1971; Lobo, Canavarros et al. 1973) report disparate CPI inflation rates in Ro de Janeiro ranging from 24% to 2.5% depending on the weights used in the calculations (the larger values refer to the slave food budget); (Graça Filho 2002) report inflation rates in São joão Del Reis of 10%, approximately, based upon a simple mean of 14 food items; analogously, (Queiros Mattoso 1986) reports 14% for Salvador, Bahia, based upon a sample of 16 food items. 10 The maps in Figures 2, 3, 4, and 5 show, respectively, the geographic distribution of the levels of municipal income per worker (RPL), income per capita (RPC), aggregate municipal income (RM), and the geographic density (mil-réis/km2) of municipal income (RMD). Table 2 presents data on population, labor force and aggregate income for provinces (UF) and major regions of the country. Figure 6 shows the provincial distribution of income per capita and income per worker. For the Brazilian economy estimates of national income in 1872 are 1.075 billions of mil-réis thus implying levels of income per capita (RPC) of 108 mil-réis and income per worker (RPL) of 299 mil-réis. Compared with the available estimate of GDP in 1900 (www.ipeadata.gov.br), the average nominal growth rate of national income from 1872 to 1900 was 4.7% p.a., approximately. Using the wholesale price index (Catão 1992) as deflator, in real terms, the rate of growth was 2.6% p.a.; and, in real per capita terms, 0.5% p.a. The small growth rates estimated seem hard to reconcile with the impressionistic evidence on Brazilian development in the last quarter of the 19th c. Indeed, this period was one of steady growth of exports, particularly of rubber, large government investments in railroads, massive international migration flows, and the rise of industrialization. It is true, however, that the last decade of the 19th c. was characterized by recession and inflation. In addition, the upsurge of international immigration during the decade would take some time to show up in the GDP figures. Finally, a substantial part of the problem might come from the underestimation of Brazilian GDP in the beginning of the 20th c. (Reis 1980; Reis, Blanco et al. 2004) Among the alternative attempts to estimate the Brazilian income per capita circa 1872, Bertola et al. (2007) is perhaps the most systematic and reliable effort. Their methodology is based upon the cross product of the distribution of population according to occupational categories (disaggregated by sex, age, juridical conditions and productive sectors) in the Census of 1872 with the distribution of income according to occupations (disaggregated by the same attributes whenever possible) obtained from diverse sources and for different years of the 1870’s. Their income per capita figure is 113 mil-réis which is surprisingly close of the estimate presented in this paper, especially taking account of the differences in methodology and sources of data. Another often quoted reference is Goldsmith (1986:23) which presents figures of 122 mil-réis for Brazilian GDP per capita in 1872. Though his methodology is not clearly 11 exposed, it seems to be based upon "reasonable" conjectures about GDP growth rates for the Brazilian economy in specific historical periods.8 Looking at the estimates of provincial income, attention should be paid to the concentration of income as well as the relatively high levels of income per capita in the Province of Rio de Janeiro. The provinces of Rio Grande do South and of the North Region follow next in terms of income per capita.9 The provinces of the Center-South, together with Bahia, Pernambuco e Paraná take an intermediate position in the ranking, while the remaining provinces of the Northeast and the Center-West Region display relatively low levels of income per capita. The position of Rio de Janeiro is explained by her importance as the leading coffee producer at the time and also by the commercial, industrial and service privileges derived from the location of the Imperial Capital as well as from other geographic advantages. The prominence of Rio Grande do South is mainly related to the good quality of her soils which gave rise to both the cattle raising activities and the salted meat industry. The North Region provinces during this period profited from the rubber boom. The differences between income per capita and income per worker are obviously related to the diversity of demographic structure among the different provinces, in particular concerning the number of slaves and free women in population. It should be observed, however, that in almost all of the provinces the levels of income per capita are approximately 1/3 of the levels of income per worker. The noteworthy exceptions are perhaps Rio de Janeiro, and to a lesser degree, Minas Gerais. Comparison with the estimates of provincial income of Goldsmith (1986), based upon Buescu,1979) show some significant discrepancies. In particular, his figures for GDP per capita are 198 mil-réis in the Center-South, 62 mil-réis in the Northeast, 49 milréis in South and North regions (Goldsmith, 1986:13) show much wider regional 8 Two other reference are Buescu (1978) and Santos (1980). Assuming that Brazilian exports were equivalent to ¼ of GDP, Buescu (1979:28) estimates in 77 mil-réis the value of GDP per capita in 1872. His figures seem exceedingly low and his model too naive to be defensible. Even less defensible are the hypothesis made to disaggregate the GDP by main regions: y = X.(x. + (p*m/k)) where y, x, m, denotates regional shares in income, exports, and imports respectively, X, value of national exports, and k, the national export coefficient. Santos (1980) estimates Amazonia regional income as 37 million mil-réis in 1875 which is also far below the estimates presented in this paper. 9 It should be observed, however, that the sample of municipalities (AMC1872-2000) in the North region is not statistically representative despite the large size of the geographic area. 12 differences than the results presented in Table 2. At provincial level, he estimates a share of 68% in Rio de Janeiro and Minas Gerais together, thus much higher than the 49% presented in Table 2. Indirect evidence on the regional distribution of GDP are provided by the share of provinces in foreign trade in the period 1872-77 presented in Table 3 and Figure 7. The table suggests that the regional figures of Goldsmith/Buescu (1986) tend to overestimate the share of the Center-South and to underestimate the remaining regions. The graph on provincial foreign trade in Figure 7, on its turn, corroborates the estimates of provincial income presented in Table 2. Another interesting aspect of the data on trade flows are the import (M/GDP) and export coefficient (X/GDP) of Brazilian provinces as displayed in Figure 8. At national level, these coefficients were 15% and 19%, respectively.10 The degree of openness of the economy – (X+M)/GDP – may seem low for a primary export economy, but it is still larger than the values of observed during the whole 20th c. except for the first five years. At provincial level, São Paulo was an outlier with the largest export coefficient (42% of GDP) and one of the smallest import coefficients (9% of GDP). The extreme values make it hard to appraise whether the provincial income level was over or underestimated. A plausible speculation would be that the province was in a phase of booming coffee export and very high domestic absorption associated with the expansion of the agricultural frontier (e.g. imports of slaves from other provinces). In addition, the fact that several other provinces also present disparate values of imports and exports (PB, RN, AL, SE, ES, MT+GO+TO) suggests that imports were to some extent overvalued or were made by large and/or traditional ports (Recife, Salvador, Rio de Janeiro, Belém, etc.). In the case of São Paulo, imports flows destined to the Paraíba Valey and the North Region of the province were likely to come from the ports of Rio de Janeiro (Corte, Angra, etc.). Taxation of exports could be the explanation for the absence of the same effect in the case of exports. 10 Buescu (1979) uses 25% reason why his estimates of income per capita are lower. 13 Secular convergence of labor productivity and income per capita in Brazil, 18722000 To analyze patterns of spatial convergence of income per capita and labor productivity in Brazil from 1872 to 2000, the income estimates for 1872 are complemented with data on GDP, income, population, and labor force obtained from the Demographic and Economic Censuses of 1920, 1950, 1980 e 2000 (www.ipeadata.gov.br). For the Census of 1920 and 1950 there was no income estimates but only GDP which, to be precise, refer to 1919 and 1949. The model specified is extremely simple. Specifications were restricted to nonconditional models of spatial convergence. The specification of a conditional model with the period of estimation starting in 1872 would pose problems of endogeneity since the income per capita in 1872 was estimated from a model using the set of conditional variables. Only from 1920 on it is possible to specify a conditional model including variables like infrastructure, geographical attributes, human capital, among other. The specification and estimation of these models are postponed for a future version of the paper. In the non-conditional growth equation specified, for each variable -- income per capita or labor productivity -- the growth rate is a simple function of the level of the variable in the initial period. The basic specification of the convergence model is thus: (1) log (yi, t/yi, t-n) 1/n = a + b. log(yi ,t-n) where yi,t= (Yi,t / Popi) is GDP per capita in municipality i, Census year t Yi,t is GDP per capita in municipality i, Census year t Popi,t is population in municipality i, Census year t Estimation was made for the sample of minimum comparable area of Brazilian municipalities in the period 1872 to 2000 (AMC 1872-2000) and for the sub-periods 1872-1919, 1919-1949, 1949-1980 e 1980-2000. Though primarily determined by the availability of Census data, the choice of sub-periods allows a fairly broad characterization of the main phases of Brazilian development. Up to the 1920, growth was driven by the export of primary commodities including coffee and rubber, and to a lesser extent, cocoa and sugar cane. From 1920 to 1950, the 14 country experienced the first phase of the import substitution industrialization which was mainly based upon light consumer goods industry. From 1950 to 1980, the import substitution industrialization deepened going into basic raw material and capital goods industries. By the end of this phase the economy was pretty much closed to trade with an import coefficient of 5%, approximately. After 1980, the economy experiences the debt crisis, hyperinflation, stagnation followed by fiscal adjustment and trade liberalization policies. It is reasonable to assume that patterns of spatial convergence of income per capital and labor productivity were significantly different between these various development phases (Reis et al. 2003). For the whole period 1872-2000, estimations were disaggregated by main regions. – North (NO), Northeast (NE), Center-South (CS), South (SU) e Center-West (CO) – to get a more detailed picture of patterns of spatial convergence of income per capita and labor productivity. The results of estimates by Ordinary Least Square (OLS) are presented in Tables 4 to 7. Table 4 show a quite good adjustment – corrected R2 equal to 0,22 – for the growth of income per capita for the period from 1872 to 2000, The estimate of β is -0,0037, negative and highly significant, thus implying convergence in the distribution of income per capita of Brazilian municipalities from 1872 to 2000. Thus, each 1% more in the level of income per capital in 1872, implies a reduction of 0.37 percentage points in the annual average growth rates in the period 1872-2000. Table 5 presents analogous results for the productivity of labor (income per worker) of Brazilian municipalities from 1872 to 2000. Estimations of the model for the selected sub-periods show that in all of them there was convergence of income per capita among Brazilian municipalities. The values of β were significantly negative in all sub-periods. The absolute magnitude of the parameters show that the speed of convergence was significantly bigger in the periods 1872-1920 and 1980-2000 when [β] > 0,01. In the other two sub-periods, the speed of convergence was smaller, particularly in the period 1919-49 when [β] was ~ 0,0057. The suggested interpretation is that import substitution phases were characterized by the exploitation of economies of scale, urban concentration, and relatively slow decrease in the spatial inequality in both labor productivity and income per capita. On the other hand, export led growth phases were characterized by intense use of natural 15 resources and the spatial dispersion of economic activities thus implying a much faster convergence of the spatial distribution of productivity and income per capita. A complementary observation is that the value of β in all sub-periods is significantly larger than the one estimated for the 1872-2000 period as a whole. Thus, the processes of spatial convergence of income per capita of municipalities in the different subperiods are not reinforcing but reversing themselves. Finally, it should be observed, that average growth rates were much higher in the import substitution phases. Compared to other countries, the historical process of spatial convergence of income per capita in the Brazilian economy seems quite slow. Indeed, estimates β are close to – 0,02, both in the case of personal income in the US states in the period 1950-80 and of income per capita of Japanese in the period 1955-87 (Barro and Sala-i-Martin 1995). Equivalent estimates for income per capita of municipalities in Japan are –0,025 for the period 1951-70, and –0,003 for the period 1970--2000 (Arbbia et al.). Despite all the differences in variables, units of observation, and methods of estimation, the estimates (except for Italy in recent decades) are twice the magnitude of those estimated for Brazil in the periods 1950-80 and 1980-2000. Table 6 and 7 present the regional decomposition of the analysis of spatial convergence of income per capita for the whole period 1872-2000. Though the samples in the case of the North and North regions are relatively small, including 14 and 18 observations, respectively, estimates of β are negative and significant for all regions (at 8% for the Center-West, however). Comparing the magnitudes β in Table 6, the speed of convergences was significantly faster in the South region where β is equal to -0,0086, compared to -0,0069 in the North Region and even lower in the remaining three where β is very similar ranging from -0,0053 and -0,0055. In all the regions, however, the speed of convergence was faster (β were larger) than in Brazil as a whole. That implies a process of regional divergence which reduces counteracts the regional convergence observe inside each region. The concentration of import substitution industrialization in the Center-South region of the country was, undoubtedly, a major factor in the process of regional divergence. Figures 9 and 10 show the results of the process of spatial convergence both in terms of income per capita and productivity of labor in the various sub-periods. 16 Conclusions and extensions The paper uses data of wages of municipal civil servants in 1876 to estimate the productivity of labor and income per capita in Brazilian municipalities in 1872. The estimation presents a satisfactory adjustment, in particular when geographic factors are incorporated into the model. Aggregation at provincial and national levels make the results more reliable. For the Brazilian economy, estimates of income per capita in 1872 are 108 mil-réis, approximately. This figure implies very small growth rates of per capita income (0.5% p.a.) in the last quarter of the 19th c. despite the strong expansion of primary exports in the period. Part of the problem is the massive international immigration and its impact on the demographic growth of the period. Comparing the income estimates with foreign trade data, Brazil appears as a relatively closed economy in the face of the stylized facts associated with primary export economies. From a spatial perspective, estimation results show wide disparities of labor productivity which ranges from 28 to 1,280 mil-réis. The rank of the province of São Paulo with income per capita of 64 mil réis is worth mentioning. Estimations of a non-conditional model show the slow convergence of municipal income per capita in the period 1872-2000. The decompositon of the analysis for the main phases of Brazilian development show that spatial patterns of convergence were relatively more accelerated in the export led growth periods, though the average rates of growth were higher in the import substitution periods. Suggested extensions of research would be a meta-analysis comparing the municipal estimates with those obtained from alternative methodologies applied to specific municipalities or localities ((Buescu 1979; Buescu 1981; Klein 1995; Silva Jr. 2002; Nunes 2003; Leite 2006; Monasterio and Zell s.d.). 17 Referencias bibliográficas Anderson, K. and E. Reis (2007). The Effects of Climate Change on Profitability and Land Uses in Brazilian Agriculture: A Municipal Cross-Section Analysis. Rio de Janeiro, Instituto de Pesquisa Econômica Aplicada - IPEA. Barro, R. and X. Sala-i-Martin (1995). Economic Growth. New York, Mc Graw Hill. Bértola, L., H. Willebald, et al. (2006). An exploration of the distribution of income in Brazil, 1839-1939. XIV International Economic History Congress, Helsinki, Finland. Bessa, A. L. d. (1981). História financeira de Minas Gerais em 70 anos de República. Belo Horizonte, Secretaria do Estado da Fazenda. Brasil, D. G. d. E. (1873). Relatório e trabalhos estatísticos apresentados ao Illm. e Exm. Sr. Conselheiro Dr. João Alfredo Corrêa de Oliveira pelo Director Geral Interino, Dr. José Maria do Couto. Typographia de Hyppolito José Pinto, Rio de Janeiro. Brasil, D. G. d. E. (1874). Relatório e trabalhos estatísticos apresentados ao Illm. e Exm. Sr. Conselheiro Dr. João Alfredo Corrêa de Oliveira pelo Director Geral, Conselheiro Manoel Francisco Correia. Typographia Franco-Americana, Rio de Janeiro. Brasil, D. G. d. E. (1876). Recenseamento Geral do Império de 1872. Rio de Janeiro, Typ. Leuzinger / Tip. Commercial. Brasil, D. G. d. E. (1878). Relatório e trabalhos estatísticos apresentados ao Illm. e Exm. Sr. Conselheiro Dr. Carlos Leoncio de Carvalho Ministro e Secretario de Estados dos Negócios do Império pelo Director Geral Conselheiro Manoel Francisco Correia em 20 de Novembro de 1878. Typographia Nacional, Rio de Janeiro. Buescu, M. (1979). Brasil: disparidades de renda no passado: subsídios para o estudo dos problemas brasileiros. Rio de Janeiro, APEC. Buescu, M. (1981). "No Centenário da Lei Saraiva." Revista do Instituto Histórico Geográfico Brasileiro(330): 176 Carreira, L. d. C. (1980). História Financeira e Orçamentária do Império no Brasil. Brasília, Senado Federal/Fundação Casa de Rui Barbosa. Catão, L. A. V. (1992). "A new wholesale price index for Brazil during the period 18701913." Revista Brasileira de Economia 46(4): 519. Contador, C. and C. Haddad (1980). Economia Brasileira: Uma Visão Histórica. P. Neuhaus. Rio de Janeiro, Editora Campus. Goldsmith, R. (1986). História Financeira do Brasil. Rio de Janeiro. IBGE (2003). Base Cartográfica Integrada Digital do Brasil ao Milionésimo Versão 1.0 para ArcGis Desktop/ArcView. Rio de Janeiro, Instituto Brasileiro de Geografia e Estatística - IBGE, Diretoria de Geociências Coordenação de Cartografia. Klein, H. (1995). "A participação política no Brasil do Século XIX: os votantes em São Paulo em 1880." Dados 38(3): 527. Leite, R. M. (2006). Desigualdades regionais brasileiras: comércio marítimo e posse de cativos na década de 1870. Faculdade de Economia e Administração de Ribeirão Preto. Ribeirão Preto, SP, USP. Maddison, A. (2003). The World Economy: Historical Statistics. Paris - France, OECD. Monasterio, L. M. and D. C. Zell (s.d.). "Uma estimativa de renda per capita municipal na Província de São Pedro do Rio Grande do Southem 1872." Neuhaus, P., Ed. (1980). Economia Brasileira: Uma Visão Histórica. Rio de Janeiro, Editora Campus. 18 Nunes, N. F. M. (2003). ". A experiência em Campos dos Goytacazes (1870-1889): Freqüência eleitoral e perfil da população votante." Dados 46(2): 329. Puntoni (coord.), P. (2003). Os recenseamentos brasileiros no Século XIX: 1872 e 1890, Cebrap,. São Paulo, CEBRAP. Puntoni, P. "Os recenseamentos do Século XIX: um estudo crítico." Rebouças, A. (1875). Provincia do Paraná dados estatisticos e esclarecimentos para os emigrantes, publicado por ordem do Ministerio dos Negocios da Agricultura do Commercio e das Obras Publicas. Rio de Janeiro, G. Leuzinger & Filhos. Reis, E. (1980). "Resenha bibliográfica de Paulo Neuhaus (ed.). Economia brasileira: uma visáo histórica. Editora Campus. Rio de Janeiro. 1980." Pesquisa e Planejamento Econômico 10(3). Reis, E., F. Blanco, et al. (2004). O Século XX nas Contas Nacionais. Estatísticas do Século XX. Rio de Janeiro, IBGE. Reis, E., M. Pimentel, et al. (2007). Áreas mínimas comparáveis para os períodos intercensitários de 1872 a 2000. Rio de Janeiro, IPEA. Santos, R. (1980). História Econômica da Amazônia. São Paulo, TAQ. Silva Jr., A. L. d. M., Paulo Roberto Staudt (2002). Nota preliminar sobre os votantes da paróquia do Rosário (Porto Alegre, 1880). Porto Alegre. Tavares Bastos, A. C. (1997). A Província: estudo sobre a descentralização no Brasil. Rio de Janeiro, Academia Brasileira de Letras. 19 Tablel Distribution of the number of municipalities (AMC), civil servants, wage bill and average wages according to average wages in 1876 Average AMC Civil servants Total wage bill wage Classes of average wage in mil-réis Número % Número % Mil-réis % Mil-réis/ano Sem valor 20 4,6 9 0,3 0 |--- 50 17 3,9 92 2,6 3.484,2 0,3 37 50 |-- 100 90 20,8 538 15,4 39.567,0 3,8 74 100 |-- 200 147 34,0 1033 29,6 146.589,0 14,3 139 200 |-- 400 104 24,1 915 26,2 265.511,2 25,8 273 400 |-- 800 49 11,3 718 20,6 388.098,2 37,7 496 Above 800 5 1,2 183 5,2 185.312,5 18,0 1018 Total 432 100,0 3488 100,0 1.028.562,1 100,0 208 20 Figure 1 Average Wages of Municipal Civil Servants in 1876 (in mil-réis) 21 Figure 2 Municipal income per worker in 1872 (in mil-réis) Figure 2.A Residual of the estimation 22 Table 2 – Brazil: Estimates of income, population, labor force in main regions and provinces (UF) in 1872 (values in mil-réis) Wages of Average Income % Mun. Municipal Municipal wages Income % Labor % Population Labor Civil Income of civil per civil Civil per capita % Pop. force servants Income Region/ in force Servants. Servants in 1872 servants worker in 1872 1872 1872 1872 1872 1872 1876 Porvince Code 1876 1876 1876 1872 479 North 1 393264 133,244 239 117720 63,774,528 493 151 4,0% 3,7% 9,6% 5,9% 219 Northeast 2 4631335 1,476,810 1648 403836 322,852,553 245 70 46,7% 41,1% 33,0% 30,0% 346 Centro-Sul 3 4016922 1,695,315 1299 538580 586,297,344 415 146 40,5% 47,2% 44,0% 54,6% 398 South 4 721337 247,679 396 158688 98,577,636 401 137 7,3% 6,9% 13,0% 9,2% 82 Center-West 5 163456 38,831 85 6716 3,189,936 79 24 1,6% 1,1% 0,5% 0,3% 1,074,691,997 299 108 Brasil 9923253 3,591,879 3664 1225159 334 100,0% 100,0% 100,0% 100,0% 20,744,465 185 1.13% 1.06% 2.37% 1.93% AM+MT 513 112202 79848 87 46815 538 543 109,863 19 0.06% 0.05% 0.11% 0.01% RR 14 5825 2101 4 235 59 59 42,090,405 155 2.74% 2.31% 3.90% 3.92% PA 15 272089 134503 143 69590 487 507 302,079 19 0.03% 0.03% 0.14% 0.03% AP 16 3148 2000 5 1080 216 64 21,508,636 60 3.62% 3.41% 4.97% 2.00% MA 21 359040 106986 182 35523 195 176 9,525,440 53 1.82% 1.95% 2.59% 0.89% PI 22 181089 64968 95 12548 132 136 32,071,989 43 7.49% 5.84% 9.83% 2.98% CE 23 742819 256744,4 360 69366 193 153 12,668,546 54 2.36% 2.14% 2.73% 1.18% RN 24 233979 83599 100 15798 158 165 18,996,492 50 3.79% 2.84% 2.73% 1.77% PB 25 376226 110198 100 22435 224 186 86,331,049 103 8.41% 7.41% 7.53% 8.03% PE 26 834314 331248,4 276 127073 460 324 19,677,033 57 3.51% 2.91% 2.59% 1.83% AL 27 348009 95214 95 16599 175 188 7,811,698 44 1.78% 1.62% 2.81% 0.73% SE 28 176243 60674 103 18535 180 134 114,261,671 83 13.90% 13.00% 9.20% 10.63% BA 29 1379616 434905 337 85958 255 245 191,804,172 94 20.56% 24.06% 9.36% 17.85% MG 31 2039735 650293 343 79759 233 222 8,071,207 98 0.83% 0.91% 1.72% 0.75% ES 32 82137 30862 63 14335 228 246 327,972,342 310 10.66% 13.87% 13.54% 30.52% RJ 33 1057696 588054,3 496 343995 694 658 58,449,623 70 8.44% 8.35% 10.84% 5.44% SP 35 837354 288084,2 397 100490 253 195 9,663,682 76 1.28% 1.18% 2.16% 0.90% PR 41 126722 54165 79 19284 244 227 15,536,356 97 1.61% 1.27% 1.23% 1.45% SC 42 159802 52742 45 17114 380 340 73,377,598 169 4.38% 4.44% 7.42% 6.83% RS 43 434813 232846,7 272 122290 450 460 639,563 19 0.33% 0.28% 0.41% 0.06% TO 17 33111 9551 15 909 61 64 3,078,089 24 1.28% 1.06% 1.83% 0.29% GO 52 127284 37149 67 5427 81 81 Source: Author estimates Obs.: Acre was not part of Brazil. Mato Grosso and Rondônia included in Amazonas and in North Region. Tocantins included in Region Center-West. . 23 Figura 3 Gráfico II.2: Renda per capita e por trabalhador por UF circa 1876 (mil-réis) Renda por PEA 1876 Renda per capita 1876 700 640 600 509 499 400 369 297 301 282 300 267 235 199 200 141 265 220 226 194 156 155 148 142 158 118 45 45 43 PB 49 RN 50 57 CE 90 66 183 179 160 138 107 100 90 71 93 90 89 105 79 64 39 24 TO RS SC PR SP RJ ES MG BA SE AL PE PI MA AP PA RR AM+MT 0 Brasil Mil-réis 500 24 Figure 4 Income per capita of Brazilian municipalities in 1872 (mil-réis) 25 Figure 5 Aggregate income of Brazilian municipalities in 1872 (mil-réis) 26 Figure 6 Geographic density of income in Brazilian municipalities (mil-réis/km2) 27 Tabela 3: Regional shares of exports, imports , and the volume of trade in the period 1872/73-76/77 (%) Imp Exp Imp+Exp 4,6% 6,4% 5,6% North 30,1% 25,8% 27,8% Northeast 58,7% 61,3% 60,1% Centro-Sul 5,8% 6,4% 6,1% South 0,9% 0,1% 0,4% Center-West Source: Brasil. Directoria Geral de Estatistica. Boletim Comemorativo da Exposição Nacional de 1908. Rio de Janeiro. Typographia da Estatistica 1908, p. 108. 28 Figure 7 Share of income in 1872 (%Y) versus the share of the volume of trade (% X+M) in Brazilian provinces (log scale) Percentual da renda 1876 (% Y) versus percentual do volume de comércio (% X+M) nas provincias do Império do Brasil (escala logarítimica) MT+GO+TO RS SC PR ES SP RJ+MG BA %Y SE AL PE RN PB CE PI MA PA AM % X+M 100,0% 52%48% 10,0% 12% 10% 11% 7% 6% 2% 1,0% 4% 3% 2% 2% 8% 5% 5% 2% 2% 1% 8% 2% 2% 1% 1% 1% 1% 1% 1% 1% 0% 0,1% 0% 0% 0,0% 0% 1% 0% 0% 1% 29 Figure 8 Export (X/Y) and import (M/Y) coefficients for the Brazilian provinces, circa 1872 Coeficientes de importação (M/Y) e de exportação (X/Y) para as províncias do Império do Brasil circa 1876 (%) M/Y X/Y 45% 42% 40% 33% 35% 31% 30% 30% 30% 24% 25% 20% 20% 17% 15% 14% 15% 3% 2% 1%1% 15% 14% 12% 11% 9% 19% 18% 16% 13% 10% 5% 19% 18% 17% 13% 11% 9% 0% 1% 1% 1% 0% 1% 4% 2% 3% BR MT+GO+TO RS SC PR ES SP RJ+MG BA SE AL PE RN PB CE PI MA PA AM 0% 30 Table 4 Brazil: Convergence of the municipal distribution (AMC 1872-2000) of labor productivity (GDP/Labor force) for selected sub-periods from 1872 to 2000 Dependente Log (PIB/ PEA) Log (PIB/ PEA) Log (PIB/ PEA) Log (PIB/ PEA) Log (PIB/ PEA) Região Brasil Brasil Brasil Brasil Brasil Período 18722000 18721919 19191980 19191949 19802000 N,obs, 380 380 427 n.d. 431 R2 corr, 0,15 0,20 0,09 n.d. 0,18 Beta -0,0034 -0,011 -0,0042 n.d. -0,014 Erro padrão -0,0005 0,001 -0,0006 n.d. -0,0014 31 Table 5 Brazil and regions : Convergence of the municipal distribution (AMC 18722000) of labor productivity (GDP/Labor force) In the period 1872-2000 Dependente Log (PIB/ PEA) Log (PIB/ PEA) Log (PIB/ PEA) Log (PIB/ PEA) Log (PIB/ PEA) Log (PIB/ PEA) Região Brasil North Northeast Centro-Sul Sul Centro Oeste 18722000 18722000 18722000 18722000 18722000 18722000 Período N.obs. 380 14 190 134 20 18 R2 corr. 0,15 0,59 0,23 0,27 0,74 0,13 Beta -0,0034 -0,007 -0,004 -0,005 -0,008 -0,005 Erro padrão -0,0005 -0,0017 -0,0006 -0,0007 -0,0001 -0,0027 32 Table 6 Brazil: Convergence of the municipal distribution (AMC 1872-2000) income per capita (GDP/POP) for selected sub-periods from 1872 to 2000 Dependente Log (RPC) Log (RPC) Log (RPC) Log (RPC) Log (RPC) Região Brasil Brasil Brasil Brasil Brasil Período 18722000 18721919 19191949 19491980 19802000 N,obs, 380 380 427 430 431 R2 corr, 0,43 0,36 0,038 0,45 0,01 Beta -0,0046 -0,012 -0,0058 -0,013 -0,002 Erro padrão -0,0003 -0,0008 -0,0014 -0,0007 -0,0008 33 Table 7 Brazil and regions : Convergence of the municipal distribution (AMC 1872-2000) of income per capita (GDP/POP) in the period 1872-2000 Dependente Log (RPC) Log (RPC) Log (RPC) Log (RPC) Log (RPC) Log (RPC) Região Brasil NO NE CS SU CO Período 18722000 18722000 18722000 18722000 18722000 18722000 N,obs, 380 14 190 134 20 18 R2 corr, 0,43 0,59 0,73 0,77 0,83 0,25 Beta -0,0046 -0,007 -0,006 -0,007 -0,006 -0,006 Erro padrão -0,0003 -0,0015 -0,0003 -0,0003 -0,0006 -0,0021 34 Figure 9 Brazil: Geographic distribution (AMC 1872-2000) of labor productivity (PIB/PEA) in 1872, 1919, 1980 e 2000 (units and scales variable) 35 Figure 10 Brasil: Geographic distribution (AMC 1872-2000) of income per capita (PIB/POP) in 1872, 1919, 1949, 1980 and 2000 (units and escales variables). 36 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • Apêndice Tabela A.1 Lista de variáveis POP72 – População total em 1872 FOG72 -- Fogos - Homens livres (domicílios de homens livres) em 1872 HTT72 -- População masculina (livre e escrava) em 1872 HET111572 idade - 11 a 15 anos - Homens escravos (pop. presente) em 1872 MET111572 idade - 11 a 15 anos - Mulheres escravas (pop. presente) em 1872 HLT164072 – População masculina livre entre 16 e 40 anos em 1872 HET164072– População masculina escrava entre 16 e 40 anos em 1872 MET164072– População feminina escrava entre 16 e 40 anos em 1872 HLT416072 idade - 41 a 60 anos - Homens livres (pop. presente) em 1872 MLT416072 idade - 41 a 60 anos - Mulheres livres (pop. presente) em 1872 HET416072 idade - 41 a 60 anos - Homens escravos (pop. presente) em 1872 MET416072 idade - 41 a 60 anos - Mulheres escravas (pop. presente) 3m 1872 ESTRL72 -- Estrangeiros livres, não-africanos em 1872 HTALF72 – Homens livres alfabetizados em 1872 MTALF72 – Mulheres livre alfabetizadas em 1872 HTPROFCAP72 – Capitalistas e proprietários em 1872 HTPROFSAL72 -- Assalariados - criados e jornaleiros - em 1872 HTPROFDOM72 -- Serviço doméstico em 1872 HTPROFLIB72 – Profissionais liberais em 1872 HTPROFMIL72 – Militares em 1872 HTPROFINC72 – Profissões industriais e comerciais em 1872 HTPROFMM72 – Profissões mecânicas e manuais em 1872 HTPROFAGR72 – Profissões agropecuárias em 1872 PROF0105_som_ger_1872,PROF0106_som_ger_1872,PROF0107_1872, PROF0109_som_ger_1872,PROF0110_som_ger_1872,PROF0111 SUM(PROF0101_som_ger_1872,PROF0102_som_ger_1872,PROF0103_1872, ELEGÍVEL75 – Número de cidadãos qualificados como elegíveis na lista eleitoral de 1875 VOTANTE75 – Número de cidadãos qualificados como votantes na lista eleitoral de 1875 LAT_GMS – Latitude da localidade sede a AMC1872-2000 LONG_GMS – Latitude da localidade sede a AMC1872-2000 PALT1 -- Proporção da área do município com altitude de 0 a 99 metros PALT2 -- Proporção da área do município com altitude de 100 a 199 metros PALT3 -- Proporção da área do município com altitude de 200 a 499 metros PALT4 -- Proporção da área do município com altitude de 500 a 799 metros PALT5 Proporção da área do município com altitude de 800 a 1199 metros PALT6 Proporção da área do município com altitude de 1200 a 1799 metros PALT7 Proporção da área do município com altitude de 1800 a 3000 metros PERO1 Proporção da area do município com limitação moderada de erosão = proporção nas classes 2 e 3 (7.5 e 15% inclinação) PERO2 proporção da area do município com limitação acentuado de erosão = proporção nas classes 4 e 5 (30 e 45% inclinação) PPTNC1 Proporção da área de solo com potencialidade agrícola na classe 1 - alta 37 • • • • • • • • • • • • • • • • • • • • • • • • • PPTNC2 Proporção da área de solo com potencialidade agrícola na classe 2 média a alta PPTNC3 Proporção da área de solo com potencialidade agrícola na classe 3 média a baixa PPTNC4 Proporção da área de solo com potencialidade agrícola na classe 4 - baixa PPTNC5 Proporção da área de solo com potencialidade agrícola na classe 5 desaconselhável ao uso agrícola PSOLO1 Proporção da área de solo na classe 1 PSOLO10 Proporção da área de solo na classe 10 PSOLO11 Proporção da área de solo na classe 11 PSOLO12 Proporção da área de solo na classe 12 PSOLO13 Proporção da área de solo na classe 13 PSOLO2 Proporção da área de solo na classe 2 PSOLO3 Proporção da área de solo na classe 3 PSOLO4 Proporção da área de solo na classe 4 PSOLO5 Proporção da área de solo na classe 5 PSOLO6 Proporção da área de solo na classe 6 PSOLO7 Proporção da área de solo na classe 7 PSOLO8 Proporção da área de solo na classe 8 PSOLO9 Proporção da área de solo na classe 9 PRE30DJF – Precipitação média nos meses de verão (dezembro a fevereiro), 196190 PRE30JJA Precipitação média nos meses de inverno (julho a agosto), 1961-90 PRE30MAM Precipitação média nos meses de outono (março a maio), 1961-90 PRE30SON Precipitação média nos meses de primavera (setembro a novembro), 1961-90 TMP30DJF Temperatura média nos meses de verão (dezembro a fevereiro), 1961-90 TMP30JJA Temperatura média nos meses de inverno (julho a agosto), 196190 TMP30MAM Temperatura média nos meses de outono (março a maio), 1961-90 TMP30SON Temperatura média nos meses de primavera (setembro a novembro), 1961-90 38 Tabela A.2.: Basic statistics of variables in the sample of municipalities of the Statistical Report of 1978 identified in the Census of 1872 Variável Soma Média Desvio Mediana Máximo Mínimo N. obs. 213.162,0 336,2 329,1 210,0 1.179,0 1,4 634,0 ALT_M 129.176,9 203,7 259,6 107,9 1.620,3 0,1 634,0 DSHOR 205.412,0 323,0 2.704,7 24,0 66.213,0 0,0 636,0 ESTRL72 1.335.467,0 2.099,8 2.588,9 1.489,5 43.911,0 0,0 636,0 FOG72 74.338,0 116,9 167,6 63,0 1.935,0 0,0 636,0 HET111572 378.822,0 595,6 963,7 262,0 11.394,0 0,0 636,0 HET164072 140.516,0 220,9 457,8 79,0 6.623,0 0,0 636,0 HET416072 1.888.331,0 2.969,1 3.784,8 2.023,5 69.393,0 11,0 636,0 HLT164072 593.246,0 932,8 1.205,3 593,5 20.690,0 0,0 636,0 HLT416072 1.013.587,0 1.593,7 3.151,1 999,0 65.384,0 48,0 636,0 HTALF72 3.239,0 36.421,0 0,0 636,0 HTPROFAGR72 2.948.370,0 4.635,8 4.180,1 37.491,0 58,9 168,8 14,0 2.007,0 0,0 636,0 HTPROFCAP72 1.262,0 55.012,0 0,0 636,0 HTPROFDOM72 1.347.761,0 2.119,1 3.114,0 160.121,0 251,8 1.057,7 104,5 24.303,0 0,0 636,0 HTPROFINC72 82.619,0 129,9 626,2 55,5 14.425,0 0,0 636,0 HTPROFLIB72 50.898,0 80,0 584,9 6,0 13.513,0 0,0 636,0 HTPROFMIL72 861.343,0 1.354,3 1.985,7 760,0 30.909,0 0,0 636,0 HTPROFMM72 576.905,0 907,1 1.761,8 301,5 25.686,0 0,0 636,0 HTPROFSAL72 5.120.221,0 8.050,7 9.328,2 5.658,5 158.766,0 279,0 636,0 HTT72 -9.062,6 -14,3 8,4 -13,0 0,0 -32,6 634,0 LAT_GRAUS -27.567,8 -43,5 5,5 -43,4 -32,4 -64,7 634,0 LONG_GRAUS 65.526,0 103,0 152,2 58,0 2.209,0 0,0 636,0 MET111572 336.309,0 528,8 836,4 248,0 12.282,0 0,0 636,0 MET164072 114.791,0 180,5 328,7 69,5 4.704,0 0,0 636,0 MET416072 1.837.713,0 2.889,5 3.031,9 2.038,0 42.681,0 6,0 636,0 MLT164072 541.237,0 851,0 1.000,9 562,5 14.103,0 1,0 636,0 MLT416072 551.207,0 866,7 1.823,0 484,5 34.101,0 8,0 636,0 MTALF72 4.605,0 7,2 9,3 5,0 187,0 0,0 642,0 NUMFPM78 3.664,0 6,0 8,8 4,0 177,0 1,0 615,0 N_EMPR_VENC FIXO_1878 POP72 SALMFPM78 UF 9.923.253,0 15.602,6 17.103,3 130.653,5 212,8 202,2 19.105,0 29,7 8,2 Fonte: Brasil ( 1876 and 1878). Elaboração do autor 11.352,5 274.972,0 153,3 2.606,4 29,0 52,0 535,0 0,0 13,0 636,0 614,0 643,0 39 Tabela A.3. Estimates of average wages of municipal civil servants in 1867 Weighted least squared by the number of civil servants with fixed wage contract Municipalities in 1872 # Observations Degrees of freedom R2 corrected F-value Root MSE 552 498 0,79 38.5 1.07 Variable Parameter estimate Standard error t value PR > |t| Intercept 1.35 1.37 0.99 0.32 LOG_NUMFPM78 -0.09 0.05 -1.92 0.06 LOG_POP72 0.35 0.06 5.44 <.0001 LOG_FOG72 0.09 0.06 1.52 0.13 P_HTT72 -3.40 1.69 -2.02 0.04 P_ESTRL72 1.49 0.69 2.16 0.03 P_HET111572 16.85 8.69 1.94 0.05 P_HLT164072 0.96 1.29 0.74 0.46 P_HET164072 -0.68 2.66 -0.26 0.80 P_HLT416072 0.75 1.97 0.38 0.71 P_HET416072 7.38 4.01 1.84 0.07 P_HTALF72 0.75 0.89 0.84 0.40 P_MET111572 -13.11 9.27 -1.41 0.16 P_MLT164072 -0.89 1.27 -0.70 0.48 P_MET164072 1.84 3.01 0.61 0.54 P_MLT416072 -2.71 1.96 -1.38 0.17 P_MET416072 0.20 4.49 0.04 0.96 P_MTALF72 -1.06 1.08 -0.98 0.33 P_HTPROFLIB72 4.32 3.81 1.13 0.26 P_HTPROFMIL72 4.22 2.56 1.65 0.10 P_HTPROFCAP72 0.13 3.23 0.04 0.97 P_HTPROFINC72 5.59 2.51 2.23 0.03 P_HTPROFMM72 0.84 0.55 1.54 0.13 P_HTPROFAGR72 -0.41 0.21 -1.98 0.05 P_HTPROFSAL72 -1.24 0.50 -2.49 0.01 P_HTPROFDOM72 0.32 0.34 0.96 0.34 DSHOR 0.00 0.00 -3.33 0.00 ALT_M 0.00 0.00 -3.27 0.00 LAT_GRAUS -0.01 0.02 -0.53 0.59 LONG_GRAUS -0.06 0.02 -2.76 0.01 LOG_AREAMUN -0.05 0.03 -1.63 0.10 DUMMY_MUN_NEUTRO -1.70 0.21 -8.13 <.0001 DUMMY_CAPITAL 0.64 0.11 5.66 <.0001 DUF13 0.47 0.50 0.93 0.35 DUF15 -0.08 0.43 -0.17 0.86 DUF16 0.10 0.69 0.15 0.88 DUF17 -0.53 0.36 -1.45 0.15 DUF21 -0.80 0.36 -2.20 0.03 DUF22 -0.47 0.34 -1.36 0.17 DUF23 -0.36 0.33 -1.08 0.28 DUF24 0.01 0.34 0.02 0.98 DUF25 0.14 0.33 0.41 0.68 DUF26 -0.03 0.30 -0.09 0.93 DUF27 -0.26 0.29 -0.89 0.37 DUF28 -0.29 0.28 -1.02 0.31 DUF29 -0.66 0.22 -3.04 0.00 DUF31 -0.41 0.13 -3.10 0.00 DUF32 -0.28 0.19 -1.50 0.13 40 DUF35 DUF41 DUF42 DUF43 DUF50 DUF51 DUF52 -0.30 -0.37 -0.74 -0.48 0.00 0.12 -0.57 0.15 0.22 0.27 0.25 0.53 0.38 0.25 Obs.: P_ refer to the percent of population for the variable in case. -2.00 -1.67 -2.70 -1.90 0.01 0.31 -2.31 0.05 0.10 0.01 0.06 0.99 0.75 0.02 41 Figure A.1. Observed and predicted log values of wages of municipal civil servants in 1876 (Reg04W3) 8 7 Predicted 6 5 4 3 2 2 3 4 Obse5rved 6 7 8 42 Tabela A.4. Estimates of average wages of municipal civil servants in 1867 Weighted least squared by the number of civil servants with fixed wage contract AMC 1872-2000 Number of observations Degrees of freedom R2 corrected Root MSE 454 379 0,83 1.19 Variable Paremeter Standard t Value Pr > |t| Estimate Error Intercept -4.56 5.63 -0.81 0.42 LOG_NUMFPM78 -0.23 0.06 -3.91 0.00 LOG_POP72 0.38 0.08 4.79 <.0001 LOG_FOG72 0.14 0.07 2.14 0.03 P_HTT72 -4.27 2.34 -1.82 0.07 P_ESTRL72 2.06 0.90 2.28 0.02 P_HET111572 24.36 12.09 2.01 0.04 P_HLT164072 3.16 1.78 1.78 0.08 P_HET164072 2.74 3.41 0.80 0.42 P_HLT416072 2.11 2.74 0.77 0.44 P_HET416072 5.88 5.08 1.16 0.25 P_HTALF72 1.92 1.25 1.54 0.12 P_MET111572 0.07 12.56 0.01 1.00 P_MLT164072 -2.27 1.72 -1.32 0.19 P_MET164072 -6.00 4.34 -1.38 0.17 P_MLT416072 -7.43 2.87 -2.59 0.01 P_MET416072 7.25 6.16 1.18 0.24 P_MTALF72 -2.49 1.60 -1.55 0.12 P_HTPROFLIB72 2.63 5.51 0.48 0.63 P_HTPROFMIL72 9.77 5.10 1.92 0.06 P_HTPROFCAP72 -1.34 5.94 -0.23 0.82 P_HTPROFINC72 12.66 3.61 3.51 0.00 P_HTPROFMM72 0.56 0.69 0.81 0.42 P_HTPROFAGR72 -0.55 0.28 -1.95 0.05 P_HTPROFSAL72 -0.90 0.67 -1.34 0.18 P_HTPROFDOM72 -0.36 0.46 -0.79 0.43 DUMMY_MUN_NEUTRO -2.30 0.31 -7.34 <.0001 DUMMY_CAPITAL 0.52 0.10 5.10 <.0001 DUF13 -1.37 0.72 -1.89 0.06 DUF14 -2.83 1.03 -2.75 0.01 DUF15 -0.31 0.60 -0.53 0.60 DUF16 0.49 0.84 0.58 0.56 DUF17 -0.79 0.47 -1.67 0.10 DUF21 -0.97 0.46 -2.11 0.04 DUF22 -1.07 0.46 -2.35 0.02 DUF23 -0.65 0.43 -1.49 0.14 DUF24 -0.25 0.42 -0.60 0.55 DUF25 -0.18 0.43 -0.41 0.68 DUF26 -0.37 0.40 -0.91 0.36 DUF27 -0.79 0.38 -2.05 0.04 DUF28 -0.53 0.37 -1.42 0.16 DUF29 -0.94 0.29 -3.21 0.00 DUF31 DUF32 DUF35 DUF41 DUF42 -0.28 -0.31 -0.12 -0.66 -0.57 0.19 0.23 0.20 0.30 0.37 -1.51 -1.33 -0.59 -2.18 -1.56 0.13 0.18 0.55 0.03 0.12 43 Tabela A.4. Estimates of average wages of municipal civil servants in 1867 Weighted least squared by the number of civil servants with fixed wage contract AMC 1872-2000 Number of observations Degrees of freedom R2 corrected Root MSE 454 379 0,83 1.19 Variable Paremeter Standard t Value Pr > |t| Estimate Error DUF43 -0.70 0.53 -1.33 0.19 DUF51 -0.04 0.83 -0.04 0.96 DUF52 -0.75 0.33 -2.26 0.02 ALT_M 0.00 0.00 0.38 0.70 LAT_GRAUS 0.08 0.03 2.54 0.01 LONG_GRAUS -0.06 0.03 -2.32 0.02 LOG_AREAMUN -0.15 0.04 -4.22 <.0001 TMP30DJF 0.14 0.17 0.83 0.41 PRE30DJF 0.00 0.00 -1.56 0.12 TMP30MAM 0.01 0.17 0.06 0.95 PRE30MAM 0.00 0.00 0.24 0.81 TMP30JJA -0.20 0.13 -1.61 0.11 PRE30JJA 0.00 0.00 -1.06 0.29 TMP30SON 0.13 0.14 0.88 0.38 PRE30SON 0.01 0.00 4.24 <.0001 PERO1 -0.18 0.36 -0.50 0.62 PERO2 -0.16 0.36 -0.45 0.65 PALT1 0.04 0.19 0.19 0.85 PALT3 -0.23 0.20 -1.11 0.27 PALT4 -0.07 0.25 -0.28 0.78 PALT5 -0.18 0.39 -0.46 0.65 PALT6 0.86 1.10 0.78 0.43 PALT7 -21.36 10.84 -1.97 0.05 PSOLO1 5.63 5.29 1.06 0.29 PSOLO2 5.83 5.28 1.10 0.27 PSOLO3 5.99 5.36 1.12 0.26 PSOLO4 5.31 5.28 1.01 0.32 PSOLO5 5.49 5.22 1.05 0.29 PSOLO6 5.41 5.28 1.02 0.31 PSOLO7 5.94 5.31 1.12 0.26 PSOLO8 5.89 5.31 1.11 0.27 PSOLO9 7.50 5.35 1.40 0.16 PSOLO10 5.68 5.30 1.07 0.28 PSOLO11 5.94 5.21 1.14 0.26 PSOLO12 4.39 21.09 0.21 0.84 DSHOR 0.00 0.00 -1.86 0.06 PPTNC1 -0.76 0.40 -1.92 0.06 PPTNC2 0.17 0.17 0.98 0.33 PPTNC3 -0.83 0.91 -0.91 0.36 PPTNC4 0.34 0.14 2.46 0.01 Intercept -4.55579 5.63219 -0.81 0.4192 44 Figura A.2. Observed and predicted log values of wages of municipal civil servants in 1876 (Reg04AMCW_3L) Fig.A.2. 8 7 Predicted 6 5 4 3 3 4 5 6 Observed 7 8