2. SOURCES OF INFORMATION

From the late 19^th century, Customs published annual detailed reports of foreign trade. Unfortunately, some of those reports have not survivedFootnote ² . Summaries of these foreign trade reports were made available by the Ministerio de Hacienda y Comercio Footnote ³ . For instance, information about the current value of exports for 1887 and 1890-1930 is available from the statistical report Extracto Estadístico del Perú 1931-1932-1933, published by the Ministerio de Hacienda y Comercio (1935). This publication also contains detailed information about the volume and value of the main exports for 1887, 1891-1892 and 1896-1930Footnote ⁴ . Jorge Hohagen (1927), Chief of Exports at Customs, collected information on exports for a large number of products for 1918-1926. Later, Hunt (1973) also collected information from Customs from 1900.

For most of the 19^th century, Customs did not publish annual reports. Some secondary sources, however, contain occasional official information about exports. The encyclopedic Anales de la Hacienda Pública by Emilio Dancuart and José Manuel Rodríguez contain interesting information. In vol. VI of the Anales, Dancuart (1904) reported the total value of exports and the composition for 1857 and 1859-1860. Dancuart argues that the information comes from official sources, but it is not clear whether those figures come directly from Customs or are simply government estimations. Moreover, the figures for 1857 are clearly incomplete as they do not include guano and nitrate exports. In vol. XXI of the Anales, Rodríguez (1920) reported the value of exports from Peru in 1877. In 1866, Luis Cisneros published information on exports, collected from yearbooks of Ministries of Hacienda to Congress for 1820 and a few years in the 1850s and 1860s. In 1921, José Rodríguez published figures for the total value of exports for several years in the previous 100 years (Rodríguez 1921a). Although he did not indicate the sources of information, one must not discard the information provided by RodríguezFootnote ⁵ . As he was head of Customs for several years in the late 19^th and early 20^th centuries (Contreras 2016), it is reasonable to assume that much of the information he reported came from official recordsFootnote ⁶ .

Newspapers provide another valuable source of information. The official newspaper El Peruano reported the volume and price of several exports for 1837-1839Footnote ⁷ . The report is a reliable source for export prices, but not for the volume of exports: figures refer to Peruvian and Bolivian exports. Fortunately, information for some products is classified by port, which helps us distinguish between Peruvian and Bolivian exports. In addition, the list of exports includes an item called minerals and specie. As specie should not be counted as part of exports, it is not possible to obtain reliable figures for the total value of exports from these sources. The newspaper El Comercio also published Peruvian prices of several exports for some years in the late 1830s, 1840s, 1860s and 1870s.

In the absence of continuous series of volumes of exports for the 19^th century, Shane Hunt made a major effort to fill the gaps of knowledge. In Price and Quantum Estimates of Peruvian Exports, 1830-1962, Hunt (1973) collected information for many products from foreign sources and recurred to data for exports of Peruvian products to her main trade partners, in particular the United Kingdom, France, Germany, the United States and Chile. Annual trade volumes were available for the United Kingdom, the United States and France from 1830, for Chile from 1844 and for Germany from 1880. From these countries, he collected information on the volume of sugar, cotton, coffee, bark, cacao and wool, and also estimated the volume of guano, nitrate, gold and silver exports from a variety of sources.

One potential problem with Hunt´s recollection of data from foreign sources is that Peru did not only export to these five countries. Also, some of those exports may have actually been Bolivian, shipped from a Peruvian port. Fortunately, some Peruvian sources allow us to determine whether Hunt´s figures are close to the actual Peruvian exports. In particular, I compare Hunt’s figures with information from Peruvian sources for 1837-1839, 1847, 1849-1850, 1862-1866, 1887 and 1891-1900. In the case of cotton, Hunt’s figures are much lower than the official figures. For 1862-1866, for example, total exports were 5.3 times Hunt’s figures. In 1837-1839 the ratio was 2.9. In 1887 and 1891-1892, however, the ratio was much lower — around 1.3. In the case of wool, the ratio was 1.8 for 1862-1866 but declined in 1887 and 1891-1892. For coffee, the ratio was 2.9 for 1862-1866, also declining by the late 1880s and early 1890s. In the case of sugar, in the 1860s and the 1890s, Hunt’s figures were not very different from Peruvian sources.

Some attempts have been made to estimate the value of exports in constant prices, although not in current prices. Hunt did not estimate the F.O.B. (free on board) value of exports in current dollars because he did not have access to F.O.B. prices for most of the period. He estimated an index of export quantum for 1830-1900 and for 1900-1962 using information on export prices for 1900 and 1929, respectively. One problem with this method is that relative prices changed substantially over the 19^th centuryFootnote ⁸ . A recent attempt to estimate the quantum of exports was conducted by Seminario (2015), who employed the Laspeyres method and relied on Hunt’s volume estimates for the 19^th century. He estimated the export quantum in constant prices from 1700. Like Hunt, Seminario only relied on one year’s prices to estimate the index of export quantum for most of the 19^th century, even though relative prices changed over time. On the other hand, Seminario estimated an index of export prices. However, he only relied on the prices in the countries of destination. The evolution of F.O.B. prices in Peru could have been different from those in international markets. Transportation costs, for example, could have changed over time. Moreover, the prices of one commodity in New York or London may reflect the average price of different types of such commodity. Differences in quality make those international prices an imperfect proxy for Peruvian prices.

In the late 1970s, Heraclio Bonilla reported information on the volume and value of Peruvian exports to Great Britain. In Los Mecanismos de un Control Económico, Bonilla (1978) reported information on the volume of the main Peruvian exports to Great Britain in 1854-1919, including guano, nitrates, copper, silver, sugar, cotton, wool, rubber, bark, coffee, leather, among others. Bonilla’s estimates are interesting as they reflect a similar phenomenon to the estimates of Hunt and Seminario. Bonilla’s figures, for example, also show a significant decline in the quantum of exports from 1880, confirming the collapse of the export sector following the War of the PacificFootnote ⁹ .

3. DATA SET

I estimated the value of exports of the main export products for 1830-1930 in current dollars and then corrected the estimates to obtain the total value of exports. To estimate the index of export quantum, I first estimated the index of prices using the Fisher method. Table 1 shows the main export products of Peru taken into account for the estimation of the value and quantum of exports.

TABLE 1 GOODS INCLUDED IN THE CALCULATION OF THE VALUE OF EXPORTS AND THE INDEX OF EXPORT PRICES

Notes: The table reports the main export products for each period. I first took into account the information of those products for the estimation of the value of the main exports and the index of export prices. I then adjusted the estimates of the value of the main exports to obtain the total value of exports.

It is important to mention that export prices need to be Peruvian prices (prices obtained by Peruvian exporters), not prices in the countries of destination. Peruvian prices and those in the countries of destination are different because of transportation costs, taxes and other fees, as well as differences in quality. For the 20^th century, for some products I collected official information on Peruvian export prices. For the 19^th century, I collected information on Peruvian prices from newspapers and secondary sources. When Peruvian prices are not available, I estimated those prices using information about foreign prices and making some adjustments. In particular, I estimated the ratio Peruvian price/foreign price and then multiplied foreign prices by the ratio to obtain domestic pricesFootnote ¹⁰ . All prices were first expressed in gold dollars, and then converted to current dollars.

4. ESTIMATIONS

The value of the main exports in current dollars was estimated using the information on the volume of exports and Peruvian prices. As all prices were first expressed in gold dollars, I transformed these prices to current dollars. The Appendix reports the value of exports in dollars and sterling.

The value of the main exports was adjusted to estimate the value of all exports, taking into account the value of minor exports. I estimated k — defined as the ratio value of minor exports/value of the main exports — and $\tilde{k}$ — defined as the ratio value of minor exports/value of the main exports, excluding guano — to adjust our estimatesFootnote ³² . For 1857 and 1859-1860, the Anales, vol. VI reports information on the value of the main exports and other exports (Dancuart 1904). For 1857 and 1859-1860 the ratios k and $\tilde{k}$ are 2.4 per cent and 6.2 per cent respectivelyFootnote ³³ . For 1865, the value of minor exports was obtained from Cisneros (1866). The value of k was estimated as 1.6 per cent. For several years in 1887-1930, I calculated the value of minor exports as the official value of exports minus the value of the main exports and minus the value of exports of specie. The ratio k is 16.1 per cent for 1887-1899, 12.3 per cent for 1900-1910, 8.1 per cent for 1913-1917 and 3.2 per cent for 1921-1925. To estimate the value of minor exports, I multiplied the value of the main exports by k for 1857-1930 and the value of main exports, excluding guano, by $\tilde{k}$ for 1830-1856. I assumed that $\tilde{k}$ in 1830-1856 was the same as in 1857-1860, that is 6.2 per cent; and that k in 1926-1930 was the same as in 1921-1925, that is 3.2 per cent. The value of k was interpolated to fill the missing years. Figure 1 shows that there are minor differences between the value of the main exports and the total value of exports, except in the late 19^th century (when k is above 10 per cent).

FIGURE 1 VALUE OF EXPORTS (MILLIONS OF CURRENT DOLLARS). Notes: The figure depicts three series of value of exports. These series are (a) the total value of exports (the value of the main exports plus the value of minor exports), (b) the value of the main exports and (c) the total value of exports using the alternative method for estimating export prices. For a discussion of the sources of information for the calculation of the value of exports, see the text.

As explained in the previous section, information about Peruvian export prices comes from different sources. It might be argued that using different sources for prices affects the reliability of the series. To address this issue, then, I followed an alternative method for estimating prices for the 19^th centuryFootnote ³⁴ . For most products, I relied on the available figures to estimate the average ratio Peruvian price/foreign price and then used the average (together with foreign prices) to estimate the entire series of Peruvian pricesFootnote ³⁵ . In the case of guano and wool, I noticed that the ratio experienced important changes over time; I then allowed the ratio for these two products to have a steady increasing trend (in the case of guano) or decreasing trend (in the case of wool) for part of the period. This method reduces the variability of the ratio domestic price/foreign price and decreases the possible biases as a result of the use of different sources; but at the same time it does not take into account the fact that in some years the ratio domestic price/foreign price may have departed from the average due to temporary changes in transportation costs or in other factors that affected the ratio. The evidence shows that the alternative method of estimating export prices yields very similar estimates to the original method (Figure 1). Both our original and the adjusted estimates show growth of exports during the Guano Era, a collapse of exports during the War of the Pacific and a recovery in the late 19^th century. As the differences are usually not large, I opt for using the original method of estimating prices.

Figure 2 depicts the evolution of the value of exports in current dollars and reports three series. The first series refers to our estimates of the value of exports. A second series is the official series of exports from 1887. A third series comes from Rodríguez (1921a), Cisneros (1866) and the Anales de la Hacienda Pública.

FIGURE 2 VALUE OF EXPORTS (MILLIONS OF CURRENT DOLLARS). Notes: The figure depicts three series of value of exports. These series are (a) our estimations of the value of exports, (b) official value of exports, reported by the Ministerio de Hacienda y Comercio and Customs, (c) value of exports according to Rodríguez (1921a), Cisneros (1866) and Anales de la Hacienda Pública del Perú. All series are expressed in current dollars. Rodríguez, Cisneros and Anales usually do not report figures for the same year, with the exception of 1859-1860, 1863 and 1877. For those years, I report the simple average of the available figures. Cisneros reports figures for 1820, 1852, 1857, 1859-1860, 1862-1865; Rodríguez reports exports for 18311832, 1839, 1845-1848, 1851, 1853-1854, 1858-1861, 1863, 1866-1867; and Anales reports exports for 1859-1860 and 1877. For a discussion of the sources of information for the calculation of the value of exports, see the text.

A first limitation of official statistics is that they are usually not available for 1830-1880. Another limitation is that they include exportation of specie. For the 19^th century, exports of specie were not negligible. For 1865, for example, exports of specie were worth 3.4 million current dollars, whereas the total value of exports was 60.4 millionFootnote ³⁶ . So specie exports accounted for 5.6 per cent of total exports. For the 20^th century, specie accounted for a smaller portion of exports. In 1925, for example, exports of specie were worth around 1.2 million dollars, whereas total exports were more than 93 million dollars; specie exports then accounted for less than 1.5 per cent of total exports.

Rodríguez (1921a) reports information on the value of exports but the series is not continuous. For instance, Rodríguez does not report the value of exports for most of 1833-1844 and several other years in the following decades. In addition, as indicated previously, it is not clear whether these figures include exports of specie. Cisneros and the Anales de la Hacienda Pública also report figures for the value of exports for the 19^th century, but only for a few years. Moreover, it is also not clear whether the figures include exportations of specie. For some years, there are similarities between the figures reported by Rodríguez, Cisneros and the Anales. In particular, according to Rodríguez the value of exports is 16.8 million dollars in 1859 and 33.5 million in 1860; whereas the Anales report figures of 16.6 million for 1859 and 34.3 million for 1860 (Dancuart 1904). For 1863, however, the value of exports is 70.6 million dollars according to Rodríguez and 47.6 million according to Cisneros.

Let us now compare our series with official estimates and the series from Rodríguez/Cisneros/Anales (Figure 2, Table 2). Prior to 1890, the series have some similarities. Our figures and the figures from Rodríguez/Cisneros/Anales show clear growth from the 1850s to the 1860s and then a large drop during the War of the Pacific. Our estimates, for example, increased from 9 million dollars in 1850 to 30 million in 1860 and 50 million in 1870, then declining to 44 million in 1878 and 8 million in 1881. Meanwhile, according to Rodríguez/Cisneros/Anales, the value of exports increased from 9.8 million dollars in 1852 to more than 73 million dollars in 1864, but then declined to 31 million in 1877. For 1890-1930, our estimates are similar to the official statistics. For 1900, for example, the current value of exports is 21.8 million dollars using our estimates and 22.4 million dollars using figures from the Ministerio de Hacienda y Comercio. For 1920, the current value of exports is 130.6 million dollars using our estimates and 131.5 million dollars using the official series. As our estimates do not include exports of specie, it should not be surprising that our estimates — although close to the official figures — are usually below them.

TABLE 2 INDICATORS OF THE EXPORT SECTOR, 1830-1930

Notes: RCA: Rodríguez, Cisneros and Anales.

The table reports growth annual rates and annual average levels per decade of the current value of exports and the index of export quantum. Columns 1, 4, 7 and 10 are our estimates. See the text for the discussion of sources. Columns 2 and 8 report official statistics from Ministerio de Hacienda y Comercio and Customs. Columns 3 and 9 come from Rodríguez (1921a), Cisneros (1866) and Anales de la Hacienda Pública del Perú. Columns 5 and 11 come from Hunt (1973), pp. 64-66. Columns 6 and 12 were obtained from Seminario (2015), pp. 888-92, 1061-62. As Rodríguez, Cisneros and Anales do not report export figures for all years, I found the missing observations by interpolation.

To estimate the index of export prices, I differentiated between four periods (Table 1) and relied on the Fisher method to estimate the index of export prices. Recently, Seminario (2015) estimated an index of export prices, using foreign prices and the Laspeyres method. In theory, the selection of the method of estimation may have an important impact on the results. Assume that absolute prices of all exports increase at a different rate from year t to year t+π. Also assume that the volume of all exports increases in this period, but the growth rate of the volume is greater for products with the highest increase in prices. Then the Laspeyres method will underestimate inflation if the base year is a year of low absolute prices (such as year t) and overestimate inflation if the base year is a year of year of high prices (such as t+π), whereas the Paasche method would do the opposite. The Fisher method minimises possible biases in the estimation of price indexes. Figure 3 depicts the indexes of export prices. The indexes show some differences. Considering the method of estimation, it is not surprising that the Fisher index shows lower variability than the other two indexes.

FIGURE 3 INDEX OF EXPORT PRICES (1913=100). Notes: The figure depicts three indexes of export prices, using the methodologies of Fisher, Laspeyres and Paasche. I took into account the main export products for four sub-periods. Table 1 lists the goods taken into account for the construction of the index.

The Fisher index shows that export prices were relatively stagnant in the 1830s and 1840s, before increasing in the 1850s. Although prices declined in the early 1860s, they rose again in the late 1860s. The increase in prices is largely explained by the increase in guano prices. The price of guano increased from 18 dollars per ton in 1849 to 38 in 1857 and remained above 30 dollars in the 1860s and 1870s. The index of export prices then experienced a decline in the 1880s, 1890s and 1900s. During World War I (WWI), prices grew rapidly. The boom in export prices affected practically all exports. Sugar prices, for example, increased from 47 dollars per ton in 1913 to 220 dollars in 1920. Cotton and copper prices also increased substantially. After WWI, however, prices experienced a sudden drop. Sugar prices, for example, fell from 220 dollars per ton in 1920 to 76 dollars in 1921.

I estimated the export quantum by dividing the value of exports in gold dollars by the Fisher index of prices. Figure 4 depicts our estimates of the index of export quantum (1913=100). It also depicts indexes of export quantum from Hunt (1973) and Seminario (2015). Hunt estimated the index of export quantum for 1830-1900 using the Laspeyres method at constant prices of 1900. For 1900-1930, Hunt estimated the index using the same method, but at constant prices of 1929 (Hunt 1973, pp. 28, 64-66). Meanwhile, Seminario also estimated the index of export quantum for 1830-1900 using the Laspeyres method and Hunt’s estimates of the volume of the main exports. For 1830-1896, Seminario estimated the index at constant prices of 1876. For 1896-1930, Seminario relied on prices of 1954 (Seminario 2015, pp. 888-895, 1061-1062).

FIGURE 4 Index of export quantum (1913=100). Notes: The figure depicts three series of the index of export quantum. These series are (a) Our estimations of the index of export quantum, (b) the index of export quantum estimated by Hunt (1973), (c) the index of export quantum estimated by Seminario (2015). All series are expressed in indexes (1913=100). For a discussion of the sources of information, see the text.

Hunt and Seminario’s estimates may have some shortcomings. First, Hunt’s estimates (and so Seminario´s estimates) are constructed largely based on foreign sources. However, as indicated previously, foreign sources may not depict the volume of exports adequately. Our estimates correct foreign figures by using Peruvian sources. Second, Hunt and Seminario do not deduct specie from exports of gold and silver. Including specie in the calculations introduces biases into the estimates. Third, for most of the 19^th century, Hunt and Seminario rely on a single set of prices to calculate the quantum of exports. The evidence, however, shows that relative prices showed important changes over the 19^th and early 20^th centuries.

In spite of using different methodologies, the three series show some similarities. In the three cases, the quantum of export experienced significant growth in the 1850s, 1860s and early 1870s. The three series also show a significant drop in export quantum due to the War of the Pacific and a slow recovery in the 1880s and 1890s. Also, the three series show a significant expansion in the late 1910s and 1920s. The main differences between our series refer to the boom period 1863-1878 (especially with respect to Hunt’s figures), and in 1921-1930 (with respect to Seminario’s estimates). In general, however, the series are very similar even though I opt for a different method of estimation.

5. DISCUSSION OF THE RESULTS

Several economists and historians have examined the evolution of the Peruvian export sector from independence in the 1820s to the Great Depression in the 1930s. In general, there seems to be consensus around the importance of guano exports until the 1870s, the War of the Pacific, and the posterior recovery interrupted by the Great Depression.

Thorp and Bertram (1985), for example, indicate that the export sector of Peru went through two phases in the 19^th and early 20^th centuriesFootnote ³⁷ . The first was the Guano Era, which ended with the War of the Pacific and the consequent collapse of the export sector; the second era began with the reconstruction in the post-war years, continuing with a phase of stable growth, ending with the Great Depression.

Other studies have also highlighted the importance of guano and the War of the Pacific for the Peruvian export sector in the 19^th century. Based on their estimates of export quantum, Hunt (1973) divides the evolution of Peruvian exports in 1830-1896 into the following periods: the post-independence recovery, the early Guano Era, the late Guano Era, the collapse during the War of the Pacific and the slow recovery after the war. Similarly, Seminario (2015) distinguishes between the early Guano Era, the late Guano Era, the War of the Pacific and the recovery after war. In the same direction, Bonilla (2005) distinguishes between the slow post-independence recovery, the Guano Era and the depression of exports from the late 1870sFootnote ³⁸ . According to these studies, the exploitation of guano and the War of the Pacific shaped the evolution of exports in the 19^th century.

On the other hand, the Great Depression is usually considered as the main negative shock of the 20^th century. Thorp and Bertram (1978) are not alone in arguing that the post-War of the Pacific recovery of exports ended with the Great Depression. In the book of economic crises of Peru edited by Bonilla (1986), the Great Depression is mentioned as the main case of economic downturn in 1900-1950. Several studies tend to agree that the Great Depression had a significant effect on raw material prices and on the country’s export sector, as well as on exports from other countries (Thorp and Bertram 1978; Thorp and Londoño 2000; Bulmer-Thomas 2003; Zegarra 2014a).

Our results are consistent with the general view that Peru experienced significant growth during the Guano Era, collapse due to the War of the Pacific, and recovery in the late 1890s and early 20^th century until the Great Depression. Our estimates, however, also show that there were important differences within each of those periods.

I have not estimated the value of exports or the index of export quantum in the 1820s, so it is not possible to determine whether the post-independence recovery was slow as other studies suggest. The evidence shows that exports experienced significant growth in value and quantum from 1830 to 1840, and that growth was lower in the 1840s. The evolution of Peruvian exports largely depended on exports of silver. According to our estimates, silver exports accounted for more than 60 per cent of the total value of exports in the early 1830s, between 40 per cent and 60 per cent in 1835-1845, and around 40 per cent in the late 1840s.

Our estimates indicate that the growth of silver exports stagnated in the following decades until the mid-1890s. If the Peruvian export sector had continued its heavy dependence on silver, the total value of exports would not have experienced much growth in the 19^th century. However, other products appeared in the export basket. In particular, guano became the most important export in the 1850s, 1860s and 1870s.

Consistent with previous studies, the figures show that the discovery of guano had an important effect on the evolution of exports. On average, the annual value of exports increased from 7 million dollars in 1840-1850 to 21 million in 1850-1860 and 43 million in 1860-1870. From the late 1860s, however, the value of exports stagnated. The index of export quantum continued increasing in the 1870s. However, as prices declined in the 1870s, the value of exports did not grow as in the 1850s or 1860s. The subsequent decline in the growth of exports occurred prior to the War of the Pacific.

The composition of exports shows important changes in the export sector during the Guano Era. The participation of guano in the total value of exports increased from 2.4 per cent in 1845 to 35 per cent in 1850 and 63 per cent in 1860. In the following years, however, the participation of guano declined, remaining below 50 per cent in 1871-1876. Nitrates increased their participation in the total value of exports from 9 per cent in 1860 to 14 per cent in 1870 and 25 per cent in 1875. Rather than being part of the Guano Era (of the Late Guano Era), the 1870s might more accurately be called the Era of Guano and Nitrates. Our evidence also shows that agricultural products had a significant participation during this period. Overall, sugar, cotton and wool exports accounted for around 20 per cent of the total value of exports in 1875.

Consistent with other studies, our findings show that the War of the Pacific led to a drastic drop in the value and quantum of exports. In particular, annual exports declined from 42 million dollars in 1870-1880 to only 10 million in 1880-1890. Exports in the 1880s were then below the level reached in the 1850s. All exports dropped in the early 1880s. Peru lost territories rich in nitrates, the second largest export, and guano exports fell to almost zero. Sugar and cotton exports also dropped. As a result, the index of export quantum (1913=100) fell from 70 in 1878 to 15 in 1881. On average, the index of export quantum in 1880-1890 was 65 per cent lower than the index in 1870-1880.

Much of the decline in exports in our estimates is explained by the decline in guano and nitrate exports. In total, guano and nitrate exports declined from 33 million dollars in 1878 to 2.9 million in 1883. As they accounted for around 73 per cent of the value of exports in 1878, the drop in these exports had a significant effect on the total value and quantum of exports. It is important, however, to notice that other exports did not drop to the same degree. Sugar exports, for example, declined from 4.9 million dollars in 1878 to 2.4 million in 1883; whereas silver exports declined from 2.8 million dollars in 1878 to 1.9 million in 1883. Most of the drop in exports during the War of the Pacific occurred as a result of the fall in guano and nitrate exports. The destruction of infrastructure and machinery and the disorganisation of the economy due to the war had a much lower impact on mining and agricultural exports.

Our estimates show a slow recovery of the value and quantum of exports in the late 19^th century. The average index of export quantum in 1890-1900 was below the level in 1850-1860. The recovery of the current value of exports in the post-war period was even slower due to the slight decline in prices. Certainly, the fact that the recovery of export quantum was slow does not imply that the volume of all exports was lower in the late 19^th century than in the middle of the century. Sugar exports, for example, were less than 900 tons in 1855 and more than 65,000 tons in 1895. Also, cotton exports increased from 850 tons in 1855 to more than 6,000 tons in 1895; whereas wool exports increased from around 2,100 tons in 1855 to almost 3,000 in 1895. The disappearance of guano and nitrate exports rather than the collapse of other exports explain the changes in export quantum after the War of the Pacific.

In value and quantum, exports grew faster during WWI. Exports reached a peak in 1920, when they were 130 million dollars. According to our estimates, the average index of export quantum in 1910-1920 was around twice as large as in 1900-1910. Several exports experienced an expansion in volume. Sugar exports, for example, increased from 123,000 tons in 1910 to 272,000 in 1919; cotton exports expanded from 14,000 tons in 1910 to 37,000 in 1919; and copper exports grew from 20,000 tons in 1910 to 41,000 in 1919. Export quantum expanded during WWI, but the current value of exports expanded more rapidly due to the growth of prices. On average, the current value of exports in 1910-1920 was around three times the level in 1900-1910Footnote ³⁹ .

With the end of WWI, export prices fell by 36 per cent per year on average between 1920 and 1922. The cumulative drop was almost 60 per cent. Prices then recovered slowly in the following years. This decline in prices in 1920-1922 was even more pronounced than the drop of prices during the Great Depression. Zegarra (2014a), for example, argues that on average export prices fell by 21 per cent per year in 1929-1932, so the cumulative drop of prices was 51 per cent in the 3 years. This may have affected the incentives to produce in the exporting-related activities: export quantum and so the value of exports declined in 1919-1920. However, export quantum expanded after 1920, reaching a far greater level as early as 1922 than prior to or during WWIFootnote ⁴⁰ .

Much has been mentioned about the negative effect of the Great Depression on Latin American economies (See, e.g., Bulmer-Thomas 2003; Thorp and Londoño 2000). The drop in export prices of 1920-1922, however, has not received much attention from the literature. The export price channel in 1929-33 is mentioned as one of the main channels of transmission of the Great Depression into the Peruvian economy. However, if the drop in export prices caused economic recessions in Latin America, including Peru, one would expect the fall in prices in 1920-1922 to have caused a large recession. However, GDP did not decline in any year in 1920-1929Footnote ⁴¹ ; and the volume of the main exports also expanded from 1921 in spite of the far lower prices.

It may be surprising that the drop in prices in 1920-1922 was not accompanied by a decline in GDP, whereas the smaller drop in prices in 1929-1932 occurred at the same time as a decline in GDP. One possible explanation for the differences in the impact is that the structure of production was different in the early 1920s than in the early 1930s. In particular, the sectors most hurt by the fall in prices had a lower weight in the composition of GDP in 1920-1922 than in 1929-1932. The sectors directly affected by the fall in commodity prices were exporting agriculture and mining. Recent estimates by Seminario (2015) show that exporting agriculture and mining jointly accounted for 10.9 per cent of GDP in 1920 and 18.5 per cent in 1929. The weight of the sectors directly affected by the drop in export prices almost doubled between 1920 and 1929. So, the direct and indirect effects of the drop in exporting activity on total GDP were greater in the 1930s. As GDP growth is the weighted average of the growth rates of all economic sectors, the drop in the economic activity of exporting sectors had a greater direct impact on the average growth of GDP. On the other hand, more Peruvians could have been affected by the fall in their incomes in 1929-1932, with the consequent greater decline in the demand for goods and services in other sectors.

Another plausible explanation for the differences in the growth of GDP is that factors other than export prices were more favorable for the growth of the economy in the early 1920s than in 1929-1932. Macroeconomic policies in the 1920s, for example, may have been more effective, or may have counteracted the contraction of private spending better than in the early 1930s. In fact, nominal fiscal revenues fell more sharply in the early 1930s than 10 years earlier, in spite of similar deflation rates; in nominal terms, fiscal revenues fell by 8 per cent per year in 1920-1922 and by 17 per cent per year in 1929-1932Footnote ⁴² . Having very scarce fiscal revenues in the early 1930s limited the ability of the government to follow countercyclical policies.