Profits and inequality during an export boom. Evidence from tax records in Lima, Peru

ABSTRACT

Some studies have shown that inequality increased in pre-1930 Latin America as the economy and foreign trade grew. Others have shown that there was no clear relationship between economic growth and inequality. The information on nineteenth-century Peru is useful to estimate income inequality during a period of rapid economic growth. This article estimates the distribution of profits in Lima between 1838 and 1859 using information from tax reports. Mobility across social classes was possible for some guild members. However, financiers and large merchants experienced higher growth in profits than artisans and small traders. In addition, a comparison of profits and wages shows that the richest entrepreneurs experienced a rapid increase in profits during a period of declining real wages. Thus, the income gap between the richest and poorest residents of Lima widened during this period.

KEYWORDS:

• Profits
• inequality
• Latin America
• Peru

JEL CODES:

• N36
• N86
• N96

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 Economic growth could affect inequality, but inequality could also affect economic growth (Forbes 2000; Galor and Zeira 1993; Neves and Silva 2014; Thewissen 2014).

2 Some studies show that inequality increased during industrialization (Ashton 1955; Lindert 1986; Steckel and Moehling 2001; Williamson 1976; Williamson and Lindert 1980), followed by a decline (Milanovic 2016; Piketty 2014). In some countries, however, inequality has increased in recent decades (Milanovic 2016; Piketty 2014). In others, inequality has maintained a long-term upward trend (Bleynat, Challú, and Segal 2021). Some studies even show that inequality may decrease in the early stages of economic development (Kimhi 2010) or that inequality may remain the same as the economy grows (Li, Squire, and Zou 1998).

3 Wage inequality can also increase as the economy grows. Workers may not be able to move easily across sectors (Bound and Holzer 2000; Kambourov and Manovskii 2009), especially if they need to learn new skills (Barro 2000). In the presence of labour market segmentation, sectoral productivity shocks in the modern sectors of the economy may lead to an increase in wages in the more productive sectors, but not in other sectors. As Jeffrey Williamson pointed out, ‘the more labor markets were segmented in the nineteenth century, the greater would have been the impact of exogenous shocks on the pay structure, and the longer would such disequilibrium attributes have persisted. In that case, influence on the course of inequality would have been more pronounced’ (Williamson 1980, 473).

4 However, if there are large differences in productivity across entrepreneurs, the mobility of resources could increase inequality. If markets are integrated, highly productive entrepreneurs could benefit from multiple opportunities, increasing income concentration (Aghion et al. 2019; Rosen 1981).

5 A positive productive shock in agriculture could also increase the income gap between landlords and workers in the countryside (Bértola, Gelman, and Santilli 2015). If landowners are richer than workers, this productivity shock could increase inequality.

6 These are just some of the studies on income inequality in Latin America. Galli, Theodoridis, and Rönnbäck (2023) provide an overview of recent studies on inequality in Latin America and Africa.

7 Several scholars have argued that institutions contributed to inequality in Latin America. For instance, Kenneth Sokoloff and Stanley Engerman argued that the economies with large native populations as Mexico and Peru became more unequal than the economies with large endowments of land. Over time, the elites in unequal societies were able to establish legal frameworks that maintained an unequal distribution of wealth and income (Sokoloff and Engerman 2000).

8 Another study shows that export growth did not necessarily lead to a rise in inequality (Arroyo-Abad and Astorga-Junquera 2017). On the other hand, within some groups, income inequality did not increase as foreign trade expanded. For instance, Zephyr Frank showed that inequality between skilled and unskilled workers in Brazil remained practically the same in the late nineteenth and early twentieth centuries, a period of export growth (Frank 2001).

9 The Guano Era is a period between 1840 and 1879, in which Peru exported large quantities of guano, a highly valuable natural fertilizer.

10 Agriculture had a large participation in total output. In 1850, agriculture represented 42% of the total gross domestic product of Peru (Seminario 2015).

11 Data on imports in the 1830s and 1840s are only available for 1831, 1832, 1839 and 1845–48.

12 The figure probably includes all categories of merchants, not only the owners of commercial establishments, but also employees. Information on the labour force comes from Fuentes (1858) and Zegarra (2020).

13 The figure probably includes different categories of artisans, not only masters.

14 Contrary to dependency theorists, the Peruvian economy did not experience any kind of foreign domination; on the contrary, domestic firms had great power to limit foreign competition.

15 Decree, 10 August 1826 (Oviedo 1870, 390–91).

16 Decree, 11 August 1826 (Oviedo 1870, 390–91). The reports of guild officers also indicate that the tax rate was 4%.

17 Each guild had two ‘clases’. The word ‘clase’ in Spanish was translated as ‘category’ in English to avoid confusion with the terms ‘class’ and ‘social class’, which are also used later in this article.

18 Decree, 29 December 1826 (Oviedo 1870, 390–91). On 4 October 1826, a decree lowered the tax rate to 3%. A decree of 9 July 1829 repealed the decree of October 1826 and raised the tax rate to 4% (Oviedo 1870, 390–91). In 1827 and 1831, the guild tax was abolished, but it was reinstated shortly after (Oviedo 1870, 390–91).

19 Decree, 11 August 1826 (Oviedo 1870, 301–03).

20 It seems it was easier for guild officers to determine sales than profits. In 1846, the government accused some leading consignees of being unwilling to report their actual profits and issued a decree establishing a profit rate of 5% on sales for the guild of consignees. Taxes would be calculated based on these profit estimates (Oviedo 1870, 409–10). Information from the tax reports indicates that profits were first estimated for the upper and lower categories; then, in cases where the guild had more than two categories, profits of middle categories were estimated proportionally to the difference in profits between the upper and lower categories. In 1851, a law included a formula to estimate profits. In the case that the guild had four categories, first the highest profits in the first category and the lowest profits in the fourth category were calculated. Denote those amounts as $M$ and $m$. For the first category, taxes were 4% of $M$. For the second category, taxes were 4% of $m+\frac{2}{3}\left(M-m\right)$. For the third category, taxes were 4% of $+\frac{1}{3}\left(M-m\right)$. For the fourth category, taxes were 4% of m. Prior to 1851, however, such a method was also commonly employed.

21 Decree, 28 June 1837. https://www.leyes.congreso.gob.pe/Documentos/LeyesXIX/1837038.pdf.

22 Decree, 3 October 1844 (Oviedo 1870, 407).

23 Law, 3 July 1851 (Oviedo 1870, 412).

24 Decree, 1 April 1852 (Oviedo 1870, 104–05). The 53 guilds included the ones exempted in 1844.

25 In fact, by looking at the data set, it is clear that those guilds were not excluded from the tax reports. Apparently, the decree did not refer to all categories of those guilds, but only to the categories that would pay less than 8 pesos per year.

26 Errors, however, could affect any year, not only 1838 and 1843.

27 For 1838 and 1843, missing observations represent only 6% of the total number of guild establishments. To estimate the proportion of missing observations, I assumed that the number of establishments was the same in 1838 and 1843 if information was missing for one of these years. The number of establishments was assumed to be the same as in 1834 if information was missing for both years.

28 In addition, actual profits may differ from estimated profits, for two main reasons. First, the profits of all members in the first and fourth categories of a guild were assumed to be the highest and lowest profits in this guild, respectively. This practice tends to over-estimate the differences in profits between the first and fourth categories. Second, taxes were the same within categories of the same guild, so profits were assumed to be the same too, thereby under-estimating profit gaps within each category. Despite these problems with the tax methodology, estimated profits provide valuable insights into profit inequality in Lima. If actual profits were very unequal, estimated profits (like actual profits) may show large differences between the richest and the poorest guild members. In addition, if there was a significant increase in profit inequality, estimated profits (like actual profits) may show increasing differences between the richest and poorest guild members.

29 In fact, the population increased by around 2% per year in 1836–50 (Gootenberg 1991).

30 The percentage of guild members was calculated with respect to the total population of Lima in 1838, 1843, 1852, and 1859. The figure of 2.2% (corresponding to 1852) was the lowest percentage. The population was taken from Gootenberg (1991) and then interpolated for missing years.

31 Some guilds changed names, were divided into two or more guilds, or were included in other guilds. These guilds were not excluded from sub-population I.

32 The discussion in this section mainly focuses on the full data set, in which all guild members are included. In some cases, I discuss the results when some guild members are excluded.

33 These figures are obtained when all guild members are included.

34 Classes A and B were equivalent to the original classes 1 and 2, respectively. Original classes 3, 4 and 5 were grouped into class C, original classes 7 and 9 formed class D, and original classes 10 and 11 formed class E. No person belonged to classes 6 or 8. Some guild members had occupations that belonged to different classes. I included them in the class for which they received the largest earnings.

35 I consider them managers.

36 Excluding the companies whose owners could not be identified, the concentration ratios are slightly lower.

37 Concentration ratios are useful to determine the share of the guilds’ total profits that goes to the wealthiest guild members. Those ratios, however, have a limitation. They only measure the differences in profits between the richest guild members and the rest of the guild members. Thus, the concentration ratios do not consider the differences within the richest guild members, nor do they account for the differences within those who are not at the top. In order to account for all differences in profits, it is possible to use the Gini coefficient. This coefficient takes values between 0 and 1, where a higher value means greater inequality. The Gini coefficient is calculated as follows: $Gini=2\frac{{\sum }_{i=1}^N{\rho }_i{y}_i}{N{\sum }_{i=1}^N{y}_i}-\frac{N+1}{N}$, where N is the number of guild members, $y_i$ is the profit of guild member i, and ${\rho }_i$ is the position of guild member i in the ranking of guild members, from the bottom to the top. In particular, ${\rho }_i$ is equal to 1 for the poorest guild member and N for the richest guild member.

38 Including individuals and companies with identified owners, the Gini coefficient decreased slightly to 0.52 in 1838 and 0.51 in 1843. Using only individuals, the Gini coefficient was also 0.52 in 1838 and 0.51 in 1843.

39 The conclusions are similar when analysing sub-population I or sub-population II.

40 The price index was estimated using the cost of a fixed subsistence basket from Zegarra (2021).

41 These figures were obtained including all guild members in sub-population I.

42 If the growth process were pro-poor, the growth incidence curve would have a negative slope (Ravallion and Chen 2003).

43 Excluding the companies whose owners were not identified, the 10% concentration ratio increased from 0.38 in 1838 to 0.42 in 1852 and 0.43 in 1859. For sub-population II, concentration ratios also increased over time. For example, using information about all guild members, the 10% concentration ratio increased from 0.38 in 1838 to 0.42 in 1843, 0.54 in 1852 and 0.58 in 1859.

44 Excluding the companies whose owners were not identified, the Gini coefficient increased from 0.5 in 1838 to 0.63 in 1859. The results for sub-population II also indicate that the Gini coefficient increased over time. For example, using all guild members from this sub-population, the Gini coefficient increased from 0.46 in 1838 to 0.65 in 1852 and 0.67 in 1859.

45 The evidence for other Latin American economies suggests that income inequality could increase or decline as the economy grew, depending on a number of factors. For example, in land-intensive economies, income inequality between landowners and workers could increase if the terms of trade increased and decrease if the frontier expanded.

46 Mobility was estimated in short time periods (1838–43, 1843–52 and 1852–59). In calculating social mobility for 1838–52 or 1838–59, the proportion of guild members who moved up or down increased. However, for these periods, the proportion of members who do not appear in either report also increases substantially. For this reason, I opted to only analyse mobility in short periods.

47 In 2001, Robert Allen assumed that labourers worked 250 days per year. Other scholars made the same assumption. Recently, however, some have questioned this assumption. In Lima, labourers were paid for a day of work. However, other unskilled workers received wages for months of work. Information on the annual wages of servants is available for Lima. On average, the annual wages of servants were around 248 times the daily salary of labourers in 1825–70 (Zegarra 2020). In equilibrium, there may not have been systematic differences in total payments to different types of unskilled workers, such as labourers and servants. It is reasonable to assume that labourers worked around 250 days per year.

48 I consider all guild members.

49 It could be argued that the Peruvian economy was in an early stage of the Kuznets curve: as the economy grew, inequality grew. Unfortunately, since the period of analysis ends in 1859, it is not possible to know whether inequality continued to grow.