2. BUILDING THE BASKET IN LATE COLONIAL SOUTH BRAZIL: COMPOSITION AND CALORIC INTAKE

One of the foremost difficulties involved in measuring living standards in pre-statistical times is to build reasonable consumption baskets. Although it is a fairly simple task to identify the main staple foods, serious problems arise when estimating specific proportions or quantities. This is due to the fact that depending on the purchasing power of the individual, or on their ethnic or social class group (an important point in a strongly hierarchical society, as was the case of Rio Grande), the amount or kind of each food could dramatically change, not to mention cultural or age constraints. Namely, an accurate determination of the type and quantity of goods available is vital to estimate the caloric content and cost of a basket and subsequently to make inferences about the subsistence level of a person or family group.

To this end, sources derived from charitable institutions (asylums, hospitals, religious communities), mendicants (convents) or public or military (hospitals, prisons) have offered quite useful results to the literature (Gelman and Santilli Reference Gelman and Santilli2015a, Reference Gelman and Santilli2016, Reference Gelman and Santilli2018; Moraes and Thul Reference Moraes and Thul2017). However, some precautions are required. For example, hospital diets were determined by the medical practices of the time, and as they had a therapeutic role, they included foods of high relative value, better quality or not commonly consumed by healthy people (poultry, e.g.). Moreover, in highly hierarchical societies, the consumption of religious communities could only represent the daily diet of very small groups of the population, leaving a large part of the population out of the analysis, particularly those ethnically or economically segregated.

As we intend to represent the social and ethnically segmented consumption baskets, we are combining two types of documents: expense bills of the Military Hospital of Rio Grande and rations' records of public and private workers, prisoners and Azorean residents. The first one is quite a valuable source, which includes monthly data on the expenditures made between 1792 and 1803. Although these records only cover 12 years of the period under analysis, they provide a very clear picture of consumption patterns and of the elasticity of the food purchases. Having processed all the monthly entries of each item, we found that fresh meat, wheat bread, manioc flour (farinha, hereafter), rice and sugar monopolised the food purchases and the caloric intakeFootnote ⁶. However, their relative weight varied in relation to price fluctuations. This evidence allows us to know the strategies developed by the hospital staff to change diet components, in order to save costs in case of inflation.

As previously mentioned, meat was very cheap and abundantFootnote ⁷. It provided food security in Rio Grande, since cattle was not only an economic asset but also a strategic resource, carefully protected and moved to safer places when wars broke out. This abundance constituted a reservoir for times of scarcity, especially during agrarian crisis. Even in the course of the worst droughts, while cereal prices soared, the meat supply could increase, as more animals were sent to markets due to their breeder's concerns about their survival in a context of grass shortage. Although ranches in Rio Grande were not as large as those in the neighbouring provinces of the Banda Oriental or Entre Ríos, nor were their herds so overabundant, cattle ranching was always one of the main rural activities (Osório Reference Osório1999). In short, there is a sharp similarity with the cases of Buenos Aires and Montevideo, where the weight of fresh meat in the population's daily diet has been already stressed by the literature (Johnson Reference Johnson1990; Moraes and Thul Reference Moraes and Thul2017; Gelman and Santilli Reference Gelman and Santilli2018).

The hospital expense records show that, when bread prices went up, meat and, to a lesser extent, farinha operated as partial substitutes (see Table A1). It is significant that according to this source, between 1792 and 1803, bread prices were constant (always at 40 reis the unit of 0.575 kilograms), so we assume that price was regulatedFootnote ⁸. Thus, probably this drop in bread purchases is due to an increase in its price, which per unit was not registered at the source; perhaps because the supplier was under contract and obliged not to break the laws in force. To compensate for this fall, in their purchases of food, officials not only bought more meat, but also farinha and rice, looking to roughly balance the total caloric intake.

This argument is reinforced when analysing the weight of each food in relation to the monetary expenditures, supporting the idea that substitution was due to price variations. At the same time, the role of rice in hospital spending was interesting, since its high value in monetary terms did not translate into a proportionally high caloric intake (see Table A1). Evidently, rice operated as a leading carbohydrate substitute for bread, even exceeding in amount of what was spent on meat and farinha. In this sense, it needs to be explained why the purchases included rice, not relying only on farinha, the cheapest alternative. This was probably due to cultural or mercantile aspects, since rice was probably more widely available at the local level than farinha, imported from other Brazilian Atlantic ports (at least during part of this period).

In sum, we have developed a fairly comprehensive overview of the Rio Grande's Hospital basket at the dawn of the 18^th century. However, the food quantities have yet to be estimated, both for higher social groups (those theoretically represented by hospital consumption) and the lower ones. Therefore, we will present below (Table 1) three alternatives to real baskets, which would characterise them more accurately.

TABLE 1 DAILY CONSUMPTION BASKETS IN RIO GRANDE, 1772-1823

Source: Computed with data from: AHRS, Fazenda, Maço 63, Despeza diaria feita no Hosp.^al Real Militar da V^a. Do Rio Grande nos 3 mezes de Janeiro a Março, year 1803; See Arquivo Nacional, Códice 104, Vol. 2, fols. 222; AHRS, Fondo Fazenda, Año 1800, Tomo 57, Año 1800, Plano das Prassas empregadas na olaria que resseberam municio de farinha e carne; AHRS, Fundo Fazenda, Tomo 74, Año 1806, Relação das Pessoas que unicamente devem vencer Municio de Carne e Farinha pela Real Fazenda do 1ro de Janeiro do anno fucturo de 1806 em diante; AHRS, Fundo Fazenda, Maço 522, Año 1835, Despeza com o alimento dos presos pobres da cadeia civil desta Villa.

The first basket was calculated from the average quantity of food consumed by hospital patients during the first quarter of 1803Footnote ⁹. Considering the broadness of supply and the relatively high consumption of each good, we assume that this diet would represent upper classes' consumption patterns. It also coincides, in general terms, with the percentages of the hospital for the whole period, calculated in Table A1 of the Appendix. Even though these diets were in fact prescribed at that time, they should not be taken restrictively to build consumption baskets. We must consider that this hospital was a public institution, hence not only did it offer more calories than an ordinary daily diet, but it is also probable that its expenditures did not fully reflect the impact of large price fluctuations on the general population.

Based on the above-mentioned rations' records, we present two other model baskets, applicable to urban workers. The first one was for free white/mestizo, and the second for slaves/Indians. This division arises from the differences in wages and rations that we have found among these groups, a common feature of a deeply segmented society, not only in the legal or ethnic status of the population, but also in food intake and job positions/earnings available in the labour market. With these diets, we aim to present a more accurate view of the local consumption patterns for the lowest income groups, which are essentially the most significant variable in the analysis of living standards. The only real difference between these two «actual» baskets was the amount of farinha. While free urban workers received a tenth of a quarter of alqueire, the Indians or slaves had their share reduced to 2/3 of a tenth, according to our sources. We have also found differences between the wages of slaves and Indians hiredFootnote ¹⁰.

Thus, we have two baskets that are very close to the total daily recommended caloric limit for male adults undergoing moderate/intense work (3,000 calories a day). The first one exceeds it slightly, and the other is a little below. Nevertheless, the caloric content does not imply an accurate calculation since, as stressed in the Appendix methodological note, meat presents large differences in terms of calories according to the animal in question, cutting and discarding.

Having settled the composition and caloric content of Rio Grande's daily consumption patterns, we calculated two baskets, each reaching a total of only 1,941 calories, based upon Allen's methodology (Allen Reference Allen2001; Allen et al. Reference Allen, Murphy and Schneider2012). Both baskets not only include food, but also other basic consumer goods such as cleaning, lighting and fuel (Table 2).

TABLE 2 RESPECTABLE AND BARE BONE BASKETS IN RIO GRANDE, 1772-1823*

¹ Spending percentages were calculated from average market prices of seven benchmark years (1772, 1792, 1802, 1809, 1816, 1819 and 1823). Data sources in Appendix.

² Estimations from Moraes and Thul, Reference Moraes and Thul2017, p. 16.

*Since we did not have prices for firewood for the entire period, we took the percentages from Moraes and Thul (Reference Moraes and Thul2017, p. 16). This is a frequent problem in many of the studies that opted for Allen's method, especially for comparative approaches on Latin America. Although we had an average price for textiles and candles for Rio Grande taken from the hospital sources (1792-1803), we preferred to use Montevideo spending shares (which are in fact similar to those found by Gelman and Santilli for Buenos Aires), as we have not found reliable data for the quantity and cost of firewood. Considering that the purpose of this methodological exercise is to compare with other places, this seems to be the most reasonable methodological option. Of course, this is debatable in terms of historical accuracy.

Source: Calculated with data from Table 1 and Moraes and Thul (Reference Moraes and Thul2017, p. 16).

The baskets shown in Table 2 present the consumption structure of individual subsistence baskets (Respectable and BBB). In order to calculate the estimates of living standards and presenting international comparisons, the methodology proposed by Allen annualises the total cost and multiplies it by a theoretical family group (3.15 people) (Allen Reference Allen2001, p. 427). In the next section, we will analyse the evolution of its cost over the period under review.

3. MEASURING LIVING STANDARDS: WELFARE RATIOS AMONG URBAN WORKERS

Once the baskets for the period under analysis have been identified, we then calculated their monetary cost (Table 3). As it is very difficult to build comprehensive series for each year, we have chosen those years when prices for all basket products were available. There are seven benchmark years between 1772 and 1823, an adequate amount of evidence, although we were not able to complete data for the 1780s.

TABLE 3 FOOD BASKETS COST EVOLUTION IN RIO GRANDE, AT CURRENT PRICES (IN REIS)

Source: Tables 1 and 2, and sources cited in Appendix.

Table 3 offers the evolution of the three types of real consumption baskets presented in Table 1, and the Respectable baskets and BBB, based on Allen (Reference Allen2001)Footnote ¹¹. It shows the evolution of food costs in different groups of Rio Grande's society between 1772 and 1823. An inflationary trend in local currency is evident, although quite divergent, mainly due to the different composition of baskets. For example, in the 1770s, the hospital basket's value started at a relatively high level, because Rio Grande was under Spanish dominion (after Pedro Cevallos's invasion in 1762, only overthrown by Portuguese troops in 1776). For this reason, the population was dispersed among other freguesias, or established new settlements some kilometres north. This affected food supply from other regions, particularly imported rice.

The following benchmark year, 1792, shows the effect of relative political and economic stability since the end of the 1770s. There is a steady drop in the price of the basic hospital basket. However, at the turn of the century, food prices began to rise for different reasons. By 1802, higher costs were again basically due to rice prices. As early as 1809, farinha and to a lesser extent, meat was responsible for the rise in prices. However, in the following years, a new tax on meat (5 reis per pound), imposed by the Portuguese Crown, had a major impact on the cost of living. We have no evidence of the consequences of this new tax in other Brazilian regions, but in Rio Grande, the rise was sharp due to the importance of meat in the basket. Consequently, the inflation impact was evident, even when it was alleviated by a small fall in rice and farinha prices. The price of meat continued its upward trend in 1819, but the Respectable basket's total cost was practically the same as in the previous year because the price of rice experienced a very steep fall. We could not include the 1823 Respectable basket's cost, as the price of bread was not available. The importance of meat as a strategic consumption good affecting all groups is clear, although with a stronger impact among the lower income population.

This is particularly clear when considering the subsistence baskets, that is, those that we identified as representative for different categories of the lowest group of urban workers. It depicts juridical and ethnical conditions, using available information on consumption for each group, considering thus not only income, but also social constraints. It is difficult to fully reflect all the complexities of a society composed of slaves, wealthy contratadores and a wide array of family arrangements. However, the basic components of subsistence for each level are generally shown.

Two main aspects need to be stressed. The first one is related to the consistent increases in the cost of living from 1809. Besides the impact of the previously mentioned tax on meat (which will only be evident from 1816 onwards), prices experienced an upward trend at least since as early as 1802. Nonetheless, the cost of living remained low, as throughout the period, these baskets were easily attainable even by the lower levels of salaried workers. This was due to the extremely low price of meat, showing its strategic role as a balancer among the basket's components. In turn, fresh meat played a key role in an asymmetric food price structure, where price changes in one of the main components of the basket did not necessarily affect the others. Available accounts, which included data on slaves' food prices, show strong coincidences. For example, probate inventories of Manoel Jesus Da Costa and Felipe J. de Andrade registered daily food expenses for their slaves of around 60 reis between 1820 and 1824. This figure is very similar to the actual baskets' costs in our tablesFootnote ¹². It should also be noted that the BBB follow the same inflationary path as the «real» baskets, meaning that the BBB should accurately represent the evolution over time of actual costs of living. This is relevant, because even if we accept the «artificial» nature of the welfare ratios made upon Allen's (Reference Allen2001) methodology, these indicators must provide a realistic approach to living standards in the past, particularly in the long run, when key features of the food economy structure are shownFootnote ¹³.

Was this evolution of the basic consumption basket followed by changes in workers' wages? Qualitative sources show that access to basic food for a young family group (i.e. two adults in their 20s) was relatively easy among workers in the lower level of the social structure. In the words of an English merchant about Rio Grande slave workers:

Leaving aside prejudices or subjectivity of the observer, this testimony shows once again that fresh meat was an inexpensive and abundant subsistence good in Rio Grande. The excellent ratio between meat's price and its nutritional value granted high subsistence levels even for low-income workers. Likewise, quantitative data also point in this direction, as can be seen in Figure 1 which confirms the high standards of living of Rio Grande. Following Allen's methodology, the bars show the annualised family cost of the baskets, between 1772 and 1823. The lines account for the nominal annual income of construction workers, with different degrees of qualifications. Collected sources from bills of the Real Fazenda and probate inventories offer daily wages of urban workers, mostly carpenter's assistants and masons, generally in three categories: Master, Official and Servente Footnote ¹⁴. Unfortunately, there is not enough data to build Masters series, and consequently, we have only included examples from the middle (officials) and the lowest group (serventes/unskilled labourers). Although there are information gaps, we consider that these two categories represent both workers with some qualifications and unskilled ones. The annual income was calculated from daily wages multiplied by 250 working days per year (Allen Reference Allen2001, p. 427).

Source: See data sources in Appendix.

FIGURE 1 RESPECTABLE AND BBB TOTAL COST AT CURRENT PRICES AND NOMINAL INCOME OF MASONS AND CARPENTERS IN RIO GRANDE, 1772-1823. ANNUAL VALUES, IN LOCAL CURRENCY (REIS).

The evidence confirms again the inflationary cycle, particularly after 1816 in Respectable baskets (which included a wider and more varied range of goods and was therefore subject to price variations). However, it can also be seen that wages generally matched inflation, even in the least qualified groups. The welfare ratios covered subsistence levels in almost all benchmark years, as can be seen in Table 4.

TABLE 4 CARPENTERS AND MASONS WELFARE RATIOS EVOLUTION USING ANNUAL RESPECTABLE AND BARE BONES BASKETS (1772-1823)

Source: Calculated with data from Table 2.

Results can be summarised in three main points. First, it is evident that all workers could reach subsistence levels throughout this period, even considering a family group of 3.15 people, as proposed by Allen. However, it must be stated that this methodology has important issues, particularly in cases such as the serventes category. As discussed in previous research (Djenderedjian and Martirén Reference Djenderedjan and Martirén2016), salaries, regional diets and housing had different structures in 18^th-century Latin America vis-à-vis other regions of the world. Among other problems, it should be considered that in particular, monthly salary, as a category, had diverse meanings and scopes throughout the regions, as it also included some extra benefits (both in food and other supplementary goods (called ‘vicios’ in Spanish)). Working time was not the only thing to be paid for; a wide range of other aspects, such as loyalty, productivity or even submission, was also rewarded. Therefore, it is important to highlight that the monthly salary category is problematic to estimate real wages or living standards at that time. The relationship between masters and workers was not necessarily governed only by economic matters; the duties and rights of both sides should be understood under a strong moral framework, as society itself was still defined within religious beliefs and the weight of traditional corporate bodies, where only those at the head of a «great house» had the power to take decisions. Patrons, for example, should take care of a wide range of spiritual issues when dealing with their workers. They were responsible for their conduct, matching the hierarchical relationship usual in those times between persons of different «quality». Patrons' responsibilities included even completely private matters, such as their workers' ability to marry. This was so even in major cities, more suitable places for the spread of purely economic interests. Besides, as most young people did not live on their own but as dependents in great houses, salaries in the lower part of the workers' scale were not necessarily intended to maintain a whole familyFootnote ¹⁵. Daily wages, even considered in the context of this traditional framework, were more suitable to represent the monetary price of hours worked. On the one hand, because of the short-term contract they involved; on the other hand, since daily wages were undoubtedly less contaminated by «moral» constraints such as loyalty or others that clearly affected the long-term relationship between patrons and peons. This is why we use daily wages and not monthly salaries.

Second, it is evident that a high skill premium was in force throughout this period, implying wide margins between workers' salaries. This situation reflected a strong social and juridical segmentation, which, however, allowed income levels to be high enough for all to cover basic needs. Third, our results show a slight downward trend in Respectable baskets WR through time, particularly among qualified carpenters between 1792 and 1816Footnote ¹⁶. Besides their enduring ability to purchase a consistent amount of food subsistence baskets, it is remarkable that workers experienced a clear decline in their purchasing power. This was due not only to increases in meat prices (thanks, in part, to the previously mentioned tax), but also to the fact that farinha broke, to some extent, the traditionally asymmetrical food supply structure. In this regard, it is important to mention that serventes did not experience significant losses in their purchasing power, as they were able to access practically the same number of baskets throughout the period, despite the inflationary process. This shows that, even under the double impact of the slight devaluation that affected the Brazilian currency between 1820 and 1823, and the rise in food prices, earnings (or at least daily wages) of low-income workers supported these changes well, therefore allowing people not to suffer significant changes in their living standardsFootnote ¹⁷.

In short, the food basket analysis provided us with some preliminary results on living standards in southern Brazil at the end of the colonial regime. Beyond subtleties on limits and the potential of hospital baskets or the different methodological issues referred to in previous pages, it is important to stress that all data and calculations show a relatively low cost of living, easily within the reach of workers at even the lowest levels of income (except slaves, for obvious reasons). Despite not including the worst part of the high inflationary cycle of the 1820s, basket coverage was widespread. These results allow us to understand better the lack of subsistence crises or food shortages even in a period when the traditionally asymmetrical food supply structure was challenged by the commoditisation of some of its components, as meat began to be widely exported as charque (Bell Reference Bell1993; Berute Reference Berute2011; Vargas Reference Vargas2016). Our results also shed light on how an extremely divided, unequal and hierarchical society was able to deal with these constraints, as low-income and young people without the support of a family were, apparently, able to survive without too many problems.

Why food prices remained at lower levels than those in the Rio de la Plata neighbours? Was the relative abundance of low-cost labour (slaves) in Rio Grande the main reason for this? In the next section, we will offer some possible explanations.

4. A REGIONAL COMPARISON: RIO GRANDE, MONTEVIDEO AND BUENOS AIRES

Were these characteristics unique of Rio Grande or were low food prices due to the abundance of meat a common feature in the Rio de la Plata region during this period? To answer this question, we compare Rio Grande with the other two main harbour cities studied by Gelman and Santilli (Reference Gelman and Santilli2018) and Moraes and Thul (Reference Moraes and Thul2017): Buenos Aires and Montevideo, respectively. Both studies used the 1,941 calories (Bare Bones) basket and welfare ratios according to Allen (Reference Allen2001). It is important to mention that despite the utility of Allen's method, it has many problems when implemented without considering regional particularities, which can even make international comparisons failFootnote ¹⁸.

Many of this method's starting points are deeply subjective, so results can be different with minimal changes in only one variable. The River Plate region raises several considerations, particularly concerning the importance of meat in the diet since it was not only the main food, but also a strategic staple; the cheapest in relative terms. Therefore, including more or less meat in the diet can substantially change the results. This is clear in the cases of Montevideo and Buenos Aires: Moraes and Thul's (Reference Moraes and Thul2017, p. 16) basket includes 23 per cent of calories from meat, but the corresponding figure in Gelman and Santilli's (Reference Gelman and Santilli2018, pp. 9-11) is 62 per cent. These variations resulted in significant differences affecting the total cost. This does not imply that the figures are wrong, but when dealing with comparisons, quantities of each good must be similar, or even equal, if the entire region shares a comparable consumption structure with meat being always the cheapest staple food. Using fixed baskets for a long period of time also underestimates the cross-price elasticity of some products. As can be seen in Table A1, the hospital continuously used replacement goods.

In our research, we built both baskets (Respectable and Bare Bones of 1,941 calories) following real diets found in local sources. However, when comparing with Buenos Aires and Montevideo datasets, we see that our BBB is extremely simple, but our Respectable Basket is similar to the Buenos Aires and Montevideo BBB. Is this a valid method? On the one hand, it is, as it deals with the particular baskets of each city. On the other hand, however, it should be acknowledged that this is not a precise comparison, which is impossible at the moment.

It is difficult to make welfare ratio comparisons when the benchmark years and occupations used are not the same. For Montevideo, Moraes and Thul (Reference Moraes and Thul2017) offer averaged information by decades, of various kinds of construction workers and for rural and maritime occupations. Regarding Buenos Aires, the available evidence provides data on skilled construction workers («oficiales») and only one of the benchmark years, 1825, is comparable with our evidenceFootnote ¹⁹. This is important when we are not sure about the comparability of local occupations. Was a servente of Rio Grande the same kind of worker as a labourer («peón») in Montevideo? Both represent the unskilled sector (and therefore the comparison is interesting), but it is difficult to appreciate the differences between them clearly, if such differences existed.

Having considered these methodological issues, we used data from baskets of both cities for comparisons with Rio Grande. Welfare ratios were calculated on the base of similar professions: for Montevideo, we considered data for unskilled labourers («peón») (Moraes and Thul Reference Moraes and Thul2017, p. 19), to make comparisons with the «serventes» of Rio Grande. They do not necessarily belong to the same category, as the Rio Grande serventes were mostly youngsters, probably slaves, and their earnings had a wider gap with respect to those of building officers than those existing in Montevideo between unskilled labourers and higher categories. We have also compared building officials since we have evidence for the three cities. For carpenters, information is offered for Rio Grande and Montevideo. Despite all these difficulties, we believe this is the most accurate way to make useful comparisons (Table 5).

TABLE 5 WELFARE RATIOS OF URBAN WORKERS IN THE RIVER PLATE REGION AT THE END OF THE COLONIAL PERIOD

Sources: Montevideo: Moraes and Thul (Reference Moraes and Thul2017, p. 19); Buenos Aires: Gelman and Santilli (Reference Gelman and Santilli2018, pp. 9-11). For Rio Grande, Figure 1 and Table 4.

Table 5 shows several differences, but also some similarities. It is clear that the standards of living in Rio Grande resembled more closely those of Buenos Aires than those of Montevideo, where welfare ratios were lower in all cases. The most remarkable point is, no doubt, the situation of skilled workers (building and carpenters' officials). Even though in all cases they were relatively well positioned, it seems clear that Rio Grande's welfare ratios were better than those of its neighbouring harbour cities in some years between 1770 and 1810. It seems to be a counterintuitive result, as these cities were relatively similar areas in geography, production, society and demography. As Moraes and Thul (Reference Moraes and Thul2017, pp. 24-25) pointed out, low Montevideo standards are probably displaying some of the problems of Allen's methodology. It does not reflect the actual conditions of food access properly, at least for not qualified workers. At that time, Montevideo and its hinterland were experiencing a strong economic expansion. Therefore, what would happen if the simpler BBB of Rio Grande was applied to Montevideo? The results would probably take a positive trend and resemble the Brazilian case. This shows how a small shift in the basket structure can radically change the welfare indicators.

Rio Grande, as previously observed, seems to be more similar to Buenos Aires, at least during the 1820s. According to Gelman and Santilli's (Reference Gelman and Santilli2018) estimations for 1825, annual earnings of a building official could purchase 3.28 BBB, while for 1823, according to our data for Rio Grande, a similar employee could reach 3.68 Respectable baskets and 5.02 BBB. If we consider that Gelman and Santilli's BBB was close (in composition) to the Respectable basket for Rio Grande, the WR levels would be very similar.

It is important to settle some preliminary hypotheses on this, apparently, higher standard of living in Rio Grande, compared to the neighbouring cities. It should be noted that the larger differences are found in the composition of the BB basket, which is much simpler in Rio Grande. As a possible explanation, we could mention that Rio Grande had a lower relative labour cost as a large share of its workers were slaves (about 30 per cent of its population were slaves vs. less than 10 per cent in Buenos Aires)Footnote ²⁰. This lower cost of the main productive factor at that time probably resulted in cheaper food, both for locally produced and for imported from other Brazilian regions (as was the case of farinha). Therefore, cheaper food allowed better standards of living for the free population, even though placed at the lowest social and productive levels, and consequently with the lowest incomes.

This phenomenon also contributed to lower the maintenance costs of slaves, boosting the expansion (or at least the preference) for them to fulfil labour needs (Aladren Reference Aladren2012). It is also remarkable, however, that the gap between skilled and unskilled workers was apparently higher in Rio Grande than in its neighbours, which points to a wider income inequality. This is not surprising if we take into account the extent to which the labour supply in Rio Grande (and in Brazil in general) relied on slaves. It was a deeply hierarchical society, in which slavery was crucial for its operation. This meant that higher categories were clearly differentiated (in several edges) from lower ones, at least more than in Buenos Aires. Slavery pushed down unskilled incomes or boosted those of positions requiring some skills.

This is a good example of a particular problem, which usually affects international living standard comparisons: the structural conditions of each economy have to be considered, especially when they involve labour costs. If attention is not paid to these conditions, comparative efforts will not provide purposive results. Moreover, it is worth stating that the long-run evolution of workers' purchasing power seems to be quite comparable. Workers in Montevideo and Rio Grande had gains between the 1770s and the first decade of the 1800s, which points to some degree of integration between the regional markets. Demographic research has shown that internal migration existed, although it was not so intense as to determine labour demand. Probably, more general processes played a major role; as the whole region benefited from cattle exports in the period, this resulted in similar pressures on labour demand in each city, which in turn contributed to similar evolutions of workers' purchasing power.

To summarise, we must stand out that useful comparisons must be developed using comparable data. To make baskets and salaries comparable across regions, they must belong to analogous categories and be composed of similar amounts of each food. A slight variation could lead to disparate results. In our opinion, questionable assumptions to fit the evidence should be avoided. In this article, we have sought to make regional comparisons, in order to offer a stronger evidence than the usual benchmark literature. In contrast with previous research, this article has attempted to find a better way to understand the differences, or at least to identify which changes in the structural features of a given economy could move living standards in one direction or another.

5. CONCLUDING REMARKS

This study presents, for the first time, a regional standards of living comparison, taking into account a region that shared, as a whole, one common characteristic: abundance of cattle and cheap relative value of meat, which was a strategic staple food and allowed the population to remain practically free of famine throughout its history. For this comparison, we have used data on southern Brazil, from 1772 to 1823, taken from several sources, including the very rich hospital ledgers. Probate inventories complemented price series and salaries have been taken from construction works.

We discussed hospital records as a source for research on living standards and we have been able to identify the main components of real diets, segmented by ethnic, social or legal constraints. Our research presents three types of real baskets to show the consumption patterns of different social groups. We have calculated the caloric intake of each one of these baskets and also their respective monetary cost. We adapted these baskets to the methodology proposed by Allen (Reference Allen2001), seeking to offer new inputs for the literature, as well as to make a regional comparison.

From these estimates, we calculated welfare ratios using salaries from representative occupational categories. Our results showed positive ratios of the BBB even for the least qualified workers' categories. For the most complete baskets, coverage was positive only for the top skilled workers. Results presented here show that even in wartime situations (precisely at the dawn of independence in Latin America), living standards in southern Brazil do not seem to have undergone major changes.

We compare Rio Grande with the two main cities of the region: Buenos Aires and Montevideo. Results show a relative price stability at the end of the 18^th century, followed by a steady increase in food prices from the first years of the 19^th century. Tendencies were similar in the three cities but living standards in Rio Grande seem to have been higher than in the other two locations. We do not yet have enough data to build a definite explanation; however, our evidence stands that the structural labour conditions for the three cases have to be taken into account. Finally, an aspect to consider is that welfare ratio estimates under Allen's (Reference Allen2001) methodology have some problems that must be solved. In our comparison, the different composition of the baskets (much simpler in Rio Grande) leads to some distortion of the results. Thus, data should be collected, controlled and prepared very carefully to achieve this target. Finally, an aspect to consider is that Allen's (Reference Allen2001) methodology has some problems when data are not properly gathered or controlled to be applied to a given case study.