Trade impacts households differently depending on their exposure to trade-driven price changes of the goods and services they consume, and the proportion of household expenditure allocated to these goods and services. This chapter examines the impact of a stylised, hypothetical tariff simulation where the seven Latin American countries covered in this Review simultaneously increase applied tariff rates on all imported goods and services from all trade partners to 25%. If the trade-driven price increases modelled in the scenario are concentrated in the goods and services predominantly consumed by lower-income households or those headed by women, for example, then they may exacerbate income or gender-based inequality.
Trade and Gender Review of Latin America
Abstract
This chapter examines the impacts of trade on households, and to the extent possible on women within those households. Following the framework described in (Luu et al., 2020[1])1, this analysis uses the OECD computable general equilibrium (CGE) model METRO (OECD, 2023[2]) to examine the extent to which different households are exposed to trade-driven changes in consumer prices resulting from a trade policy change.2 A stylised hypothetical tariff simulation is conducted where the seven Latin American countries covered in the review simultaneously increase applied tariff rates on all imported goods and services from all trade partners to 25%.3,4 The resulting change in consumer prices for the commodities in the model are subsequently linked to expenditure data in national household surveys5, which provide detailed information across different socio-economic groups. Linking the model results with household survey data allows for a comparison of exposure—measured by changes in purchasing power—across different household characteristics such as household income level and, where available, the gender of the head of household.
The impact across household types may not be uniform since a household’s exposure to trade-driven policy changes will depend on the price changes of the commodities they consume and the proportion of household expenditure allocated to these goods and services. If price increases are concentrated in the goods and services predominantly consumed by lower-income households or those headed by women, for example, then trade-driven changes in prices may exacerbate income or gender-based inequality.
The focus of this simulation is on household exposure; some methodological limitations should be noted. This approach, which links the model results with household surveys through price changes does not fully take into account potential behavioural responses, such as households adjusting their consumption patterns in response to changes in prices and income.6 Additionally, the analysis primarily focuses on household exposure as consumers and does not capture the full welfare effects. Thus, it does not account for the fact that income changes may vary across household types, particularly if sector-specific impacts of the policy disproportionately affect income earned in certain industries.
The simulation increases the tariffs to 25%, therefore the size of the tariff shock is related to the effectively applied tariffs rates7 (herein tariff rates) in the model8, among the Latin American countries covered in the review. On average, tariff rates on goods are low (3.3% across the goods sector), but the range of tariffs rates is broad in most countries except for Peru and Chile (Annex Figure 4.A.1). Brazil and Argentina have the highest tariffs on average 8.7% and 7.7% respectively (Table 4.1). Chile and Peru have the lowest tariffs on goods (0.3% and 0.7% respectively).
In the model database, rates on manufactured products in the seven countries under review were slightly higher on average than the average tariff rate on agrifood products, primarily due to high tariff rates in Brazil, Argentina, and Colombia. Notably, imports of textiles, wearing apparel, and leather products face some of the highest rates among manufactured goods in all countries except for Chile where import tariff rates across goods are uniform. Some agrifood products face very high tariffs in certain countries. For example, rice imported into Colombia faces a 40% tariff rate while dairy, processed rice, and sugar in Costa Rica have tariff rates of 25, 20, 15% respectively.
On average, imported food is taxed at 3.1% while agriculture products have a tariff rate of 1.3% with variation across countries. In Brazil, Argentina, Costa Rica, and Mexico, imported food has a higher effective tax rate than basic agriculture products. Imported manufactured goods are taxed at 3.5% with the highest average tax rate among the regions applied by Brazil, Argentina, and Colombia. Moreover, these countries apply a higher tax rate on advanced manufactured goods, such as computer and electronics, than they do on basic manufactured products.
Services account for a large portion of household consumption, ranging from 44 to 67% among the seven countries in the analysis. Services are not subjected to import tariffs. Services are mostly supplied domestically although there are exceptions: accommodation and food service as well as financial services and communications, for example, which also account for a relatively larger share of household consumption. Other heavily imported services include air transport, construction, and business services but these sectors do not account for a large share of household consumption.
Percentage
Note: The table shows trade weighted (value) average of ad valorem tariffs. Specific tariffs are not shown. They are also not shocked in the analysis. Food covers: Bovine meat products; Meat products nec; Vegetable oils and fats; Dairy products; Processed rice; Sugar; Food products nec; and Beverages and tobacco products. Advanced manufacturing includes: Computer, electronic and optical products; Electrical equipment; Machinery and equipment nec; Motor vehicles and parts; Transport equipment nec; and Manufactures nec. For the full list of sectors covered in the model, see https://www.gtap.agecon.purdue.edu/databases/v11/v11_sectors.aspx.
Source: METRO model, reference year 2017.
The imposition of a 25% rate in the stylized simulation leads to a more pronounced increase in tariff rates among sectors in regions with initially lower tariffs. For instance, base level tariffs in Brazil and Argentina on agrifood and manufacturing sectors are higher and not uniform compared to other regions, which translates to a larger variation in size of the tariff rate increase applied in the simulation. In contrast, most sectors in Chile and Peru experience a 25 percentage point increase in tariffs. Some sectors in Colombia, Costa Rica, and Mexico also see significant tariff increases in the simulation, although not as many as in Chile and Peru.
How prices respond to the tariff increase will depend on the extent to which the supply of the sector is imported and how easily the imports can be substituted by domestic production in the region. In general, the tariff increases lead to a rise in commodity prices in the simulation, with sectors subject to larger tariff hikes experiencing larger price increases, particularly in those sectors with large shares of imports.
Across sectors in the different countries, a few common patterns emerge regardless of the country’s tariff profile. On average, manufactured products experience a more substantial increase in prices compared to other sector categories(i.e. Agrifood, Extraction, and Services). This outcome is not surprising, as many manufacturing supply chains are highly internationalized, making them particularly vulnerable to tariffs on inputs that could subsequently affect production costs. In addition, most Latin American consumers rely heavily on international markets for manufactured goods. For instance, in Costa Rica, nearly all advanced manufacturing commodities consumed by households, such as electronics, machinery, and vehicles, are imported (93.6% on average). In Chile, almost three-quarters of these goods are sourced from abroad, while in Mexico and Peru, the proportion is approximately 43%. Brazil stands out as an exception, importing only 8.4% of its advanced manufacturing commodities.
The agrifood sector, in contrast, exhibits some variability in import dependency across the countries under review. Brazil, one of the world's largest agricultural producers, predominantly sources its agrifood products domestically. However, in Costa Rica, Mexico, and Chile, a significant proportion of agrifood products consumed by households is imported, with average import shares ranging from 17% in Mexico to 25% in Chile. As a result, in the simulation, the consumer prices of agrifood products in Costa Rica and Chile increase on average 10% (Annex Table 4.A.2) whereas Mexico experiences a comparatively smaller increase of 6.6%.
The services sector, within the stylized simulation, encounters some of the most substantial tariff increases when the initial base tariffs of zero9 are increased to 25%. However, the resulting price increases in services are smaller than in other sectors facing the same tariff hike. Most services are produced and consumed domestically and are therefore less affected by the tariff increase. Additionally, the way services are produced makes them less sensitive to tariff changes. A significant portion of the inputs used in the services sector is labour and capital (value-added), rather than intermediate inputs which are more directly affected by the tariff increase. Moreover, many of the intermediate inputs that are used in the services sector come from the sector itself, further reducing its exposure to the tariff shock. There are some exceptions. In certain service sectors, a large portion of supply is sourced internationally, leading to a stronger price response. For example, on average, a quarter or more of construction, water transport, air transport, business services and accommodation and food services consumed by households are imported. However, aside from accommodation and food services, most of these services make up only a small share of household consumption (on average less than one percent).
Lastly, the extractive sectors, which include commodities such as coal, natural gas, and other mining products, are generally subject to low levels of taxation. The simulation, therefore, introduces significant tariff increases for these sectors, which lead to a significant rise in prices. While these products constitute only a small fraction of household expenditures, their price increases have indirect effects. In particular, the rise in energy input costs subsequently elevates the production costs of domestically produced goods, which are captured by the model.
In order to assess the impact of the tariff increase across different household types, the consumer price changes from the model simulation for each sector are mapped to available household survey data. Data from household surveys not only record product-level spending, but also other information about the household such as income and in some cases the gender of the head of household. Linking the model results with household survey data enables an assessment of the distributional impacts of trade‑driven price changes, using changes in purchasing power10 to measure household exposure as consumers. Of the seven countries in the analysis, Costa Rica, Mexico, Peru, and Chile have sufficient detail available in their household surveys to distinguish households by income deciles (or quintiles in Chile’s case).11 Costa Rica also identifies the gender of the head of household allowing for an assessment of exposure of women-led households.
Across the four countries, the share of expenditure on products differs across income groups (Figure 4.1). Lower-income households allocate a larger share of their expenditure on essential items such as food, transportation services, and goods and services for routine household maintenance, while higher-income households direct more of their spending towards durable goods and services. For instance, households in the lowest income decile allocate over 40% of their expenditure to food, compared to just 17% for those in the highest decile in Mexico. In Costa Rica, food accounts for 33.7% of expenditures in the lowest-income households compared to just 12.6% among the highest-income group. In contrast, wealthier households across the four countries spend a larger share of expenditure on vehicle purchases, travel, catering services (which could include restaurants), and telephone equipment compared to lower-income households. The differences in household spending patterns across the income groups means that price changes could have distribution effects.
In the stylised simulation, sectors with a high share of imported supply, such as computer, electronics and optical products, motor vehicles and parts, and business services—categories primarily consumed by upper-income households—experience some of the stronger increases in price. In Mexico, Chile, and Peru nearly all computer, electronic, and optical products consumed by households are imported, leading to a 20% price increase. Similarly, the motor vehicles and parts which are consumed more by higher income households experience a sharp increase in prices particularly in Chile and Costa Rica where these products are mainly sourced from abroad. In contrast, agrifood products, which account for a larger portion of lower-income spending, see only relatively modest increases in price, as most agrifood products are supplied domestically. One exception is bovine meat in Chile, where two-thirds of supply comes from abroad, resulting in an 15% price increase in the simulation.
As a result of the tariff increase, when measuring the loss of purchasing power based on the expenditure approach, purchasing power across all household groups declines. However, the decline is slightly stronger for higher-income households since they consume more imported products (Figure 4.2). For example, in Mexico purchasing power decreases by 8.6%, across all household types. However, the purchasing power of households in the highest-income decile is reduced by 9.6% while those in the lowest-income decile is declines by 8.0%, almost 2 percentage points less. A similar pattern emerges in the other countries, where wealthier households experience greater absolute losses due to their higher expenditures on vehicles, electronics, and business services. However, because higher-income households allocate a smaller proportion of their income to consumption, they are in a better position to absorb these purchasing power losses.
When purchasing power is computed on the basis of total income, losses are much greater for poorer households because they spend a larger portion of their income. In Peru for example the poorest households experience a 20% reduction in purchasing power, compared to just 9% for the wealthiest. In Chile and Costa Rica, the lowest-income households lose nearly two and three times the purchasing power of the highest-income households, respectively. As a result, trade-drive price changes have an uneven distributional effect thereby increasing the inequality across households.12
Note: Includes only COICOP 1999 sectors at the 3-digit level that account for at least 3.5% (4.5% for Peru and Costa Rica) of total household expenditure in any income decile.
Source: Household Surveys of individual countries.
Note: The purchasing power (‘PP’) is computed both based on consumption of a product as a share of total expenditure (Exp) and consumption of a product as a share of household income (Inc). Available income estimates were used to compute changes in purchasing power based on share of income. Specifically for Mexico: current income; Peru: gross income; Chile: disposable income; and Costa Rica: gross monetary income.
Source: METRO model and Household Surveys of individual countries.
According to the International Labour Organisation (ILO), 61% of the working-age population, or roughly 2 billion workers worldwide, are engaged in informal employment – either by working in jobs in unregistered or unincorporated businesses or working in positions without formal contracts, leaving workers without access to social benefits or legal protections (ILO, 2019[3]). These conditions expose informal workers to a greater array of risks, from job instability to poor working conditions, compared to their counterparts in the formal sector (OECD, 2019[4]). This lack of protections and increased risk adds additional layers of economic insecurity to an already vulnerable, low-income household if one or more of the members are employed informally.
In Latin America and the Caribbean, 54% workers are employed informally (ILO, 2019[3]), however over 70% of households in select Latin America countries had at least one member in informal employment (OECD, 2019[4]). Furthermore, households that rely on informal work are not distributed evenly across income deciles.
In Peru, 78% of households have at least one member in informal employment, with this share exceeding 80% in lower income deciles, compared to less than 60% in the highest decile (Figure 4.3). This pattern underscores how informal employment can intensify the vulnerability of low-income households, making them particularly susceptible to economic shocks and price increases.
Source: Authors’ calculation based on the Household Survey of Peru 2022.
Moreover, women are overrepresented in informal employment across Latin America and Caribbean (Berniell, Fernandez and Krutikova, 2024[5]), and in particular among the upper-middle- and low-income LAC countries where the rate of informal employment is higher for women than for men (Table 4.2). Among the seven countries in the review, only Brazil has a higher informality rate among men. Peru and Mexico have the largest gender gap among the seven.
According to OECD (2019[4]), women in the informal sector typically face more precarious work conditions, shorter working hours, and lower wages, with the gender pay gap often more severe than in the formal economy. This overrepresentation of women in the informal economy not only exposes women workers and women-headed households to greater economic insecurity but also exacerbates existing inequalities.
A similar analysis was undertaken for New Zealand in 2022 to measure the distributional effects on different types of households of increasing New Zealand’s tariffs to 25% (OECD, 2022[6])). The study found that the most exposed household type is the one-parent household with dependent children, which are in their vast majority headed by women. The New Zealand study could not explicitly measure the impact of trade-driven prices changes on women-led households due to data limitations. In contrast, Costa Rica’s household survey enables the identification of women-led households, allowing for a direct assessment of how trade driven price changes affect women and the extent of any gender-related differences.
4.4.1. Examining differences in exposure based on gender of the head of household
In Costa Rica, 38.4% of households are headed by women.13 This share is above the median percentage globally (28.0%) but in line with the median proportion (36.2%) in Latin American and the Caribbean (LAC) (Saad et al., 2022[7])). Globally, most female-led household types are single-parent households with children, but in the LAC region a considerable number of female-led households have husbands present, although the presence of a husband does not translate into increased wealth. Women-led families, however, are also not necessarily worse off. Liu, Esteve and Treviño (2017[8]) found that the union status of the head of household (i.e. single, divorce, married, cohabitation) more than sex of the head of household is more telling of the living conditions14 of the household in many Latin American countries, including Costa Rica. Once factors such as the union status, education, urban or rural setting, presence of children in the household, and ownership of the dwelling are accounted for, the sex of the head of household becomes statistically insignificant in the case of Costa Rica (Liu, Esteve and Treviño, 2017[8]). Not surprisingly, while there are some small variations in the share of women-led households by income decile in Costa Rica, the differences are not statistically significant (Figure 4.4) confirming that women-led households are not more prevalent among any one income group.
Note: Whiskers on each bar represent the 95% confidence intervals.
Source: Costa Rica Household Survey 2018.
Moreover, there are not strong differences in consumption patterns between female-led and male-led households that could contribute to distributional impacts when consumer prices changes. Among the 108 products in the analysis, most (83 products) are consumed in similar shares of expenditure regardless of whether the household is headed by a female or male. Additionally, in those cases that were found to be statistically different, the differences in the shares were less than 1 percentage point. There were, however, some exceptions. For example, as a share of expenditure, purchase of motor vehicles and items used to maintain these vehicles, such as fuels and lubricants were higher for men-led than for women-led households on average (with differences of 0.7 percentage points for cars and 1.9 percentage points for fuels and lubricants). Women-led households, on the other hand, seem to rely more on transportation services rather than cars, as reflected in a larger share of their total expenditure allocated to this service (almost 0.8 percentage points more than men-led households). Women-led households also spend a larger share of the household expenditure on personal care items, footwear, utilities, and pharmaceutical products relative to male-led households, but the differences are small (less than half a percentage point) (Figure 4.5 Panel A).
Because the consumption patterns between the two types of households are almost identical, the purchasing power loss resulting from the tariff increase are the same between households led by women and those led by men. Both types of households experience a loss in purchasing power of 13.0% using the expenditure-based approach. Moreover, while women-led households on average spend less money in total compared to men-led households, their gross monetary income is also lower. Overall, the share of gross monetary household income that is spent is similar between the two different household types (about 70%) translating to a purchasing power loss of rough 9% for each household type (Figure 4.5 Panel B).
Note: Panel A includes only COICOP 1999 sectors at the 3-digit level that account for at least 2.5% of total household expenditure in any income decile. In Panel B, the purchasing power is computed both on the basis of consumption of a product as a share of total expenditure (Exp) and consumption of a product as a share of household income (Inc). Gross monetary income was used to compute change in purchasing power based on share of income.
Source: METRO model, Household Survey of Costa Rica 2018.
4.4.2. Analysing the combined effects of income and gender on household exposure
As seen in the previous subsection, the level of household income influences consumer consumption patterns. However, the gender of the head of household does not impact consumption patterns across income deciles. Although the exact proportions of spending on certain products may vary slightly between income groups in both types of households, the overall consumption patterns by income are comparable. In both male- and female-headed households, lower-income groups tend to allocate a higher proportion of their income to essential items than higher income groups. For example, food expenses account for a much larger share of household expenditures for lower income households. For households in the lowest income decile, food expenditures are approximately three times greater than those in the highest income decile, with a ratio of 2.7 for women-led households and 3.0 for men-led households. Similarly, transportation services account for nearly double the share of total expenditure in the lowest income decile compared to the highest, with no significant differences based on the gender of the household head.
There are some notable exceptions, particularly when examining detailed product categories (Figure 4.6). For example, in terms of personal care, women-led households in the highest income deciles spend nearly twice as much on hair salons and personal grooming compared to women-led households in the lowest income deciles. In contrast, for male-led households, the ratio is only 1.6 between the households in the highest and lowest income deciles. Additionally, the difference in the share of spending on products and services related to cars and personal transportation, such as spare parts and fuel, is larger between high-income and low-income women-led households. This is likely because low-income, women-led households are less likely purchase vehicles15 reducing the need for car maintenance items. One area where low-income women-led households are especially vulnerable is the purchase of water. According to household expenditure data used in the scenario, the share of total expenditure these households spend on water is almost four times more compared to their high-income counterparts. Water supply purchases on average account for 3.9% of total expenditure among women-led households in the lowest income deciles, 3.0% among the lowest income male-led households, but only 1.1% of the total expenditure in the highest income deciles regardless of gender.
Although some differences in consumption patterns by income and gender of the household head are statistically significant, overall spending habits within the same income decile are similar between male- and female-led households. As a result, in a simulation where tariff increases lead to rising prices for manufactured goods compared to food and services, the loss of purchasing power is nearly identical between male- and female-led households. Women-led households lose 12.8% of expenditure-based purchasing power when averaged across the different income deciles, while men-led household the loss is 12.9% with similar losses across income deciles regardless of the gender of the household head (Figure 4.7).
Moreover, the proportion of income spent by households across different deciles does not vary significantly between those led by men and those led by women. For both types of households those in the lower income decile spend a larger portion of their income than households in the upper income deciles.
Consequently, when calculating purchasing power loss using the income-based approach, lower income households experience greater losses in purchasing power compared to their higher income counterparts, whether those households are headed by men or women.
Note: Includes only sectors that account for at least 5.0% of total household expenditure in any decile for each gender respectively.
Source: Household Survey of Costa Rica 2018.
Note: ‘TT’ refers to expenditure shares across all income categories.
Source: METRO model, Household Survey of Costa Rica 2018.
The simulation implemented in this section of the Review increases import tariffs to: i) examine their price impact on household expenditures, and ii) determine if certain types of households are more exposed than others to price changes. How the price responds to an increase in tariffs depends on the extent to which the supply of the sector is imported and how easily the imports can be substituted by domestic production. Household exposure depends on the price changes of the commodities they consume, and the share of household spending allocated to those goods and services. The analysis based on the Costa Rican household survey, shows that household exposure is primarily related to income level rather than the gender of the head of household, as consumption patterns differ across household income rather than by gender of the household head.
Across the four countries with available data on household expenditure —Mexico, Chile, Peru, and Costa Rica—lower income households allocate a large portion of their expenses to essential products such as food, electricity, personal care items, and often transportation services. In contrast, households in the upper income deciles tend to spend a relatively larger share of their expenditure on durable and semi-durable manufactured items and services. When tariffs are raised to 25%, prices of manufactured goods increase more than products from other industries, particularly those of advanced manufacturing goods such as electronics and vehicles, since manufactured products are generally sourced internationally. Agriculture and food products, on the other hand, are generally supplied domestically, and their price increases are not as pronounced. Services consumed by households and industry are generally supplied and consumed domestically, therefore, despite the strong increase in tariffs, their prices do not increase as much as manufactured goods.
Purchasing power losses due to the tariff increases differ by household income levels. When measured using the expenditure-based approach, upper income households experience slightly higher losses in purchasing power (a 1.6 percentage point difference or less) due to price increases in manufactured goods such as motor vehicles and electronics as well as travel services like air transport and accommodation, products and services consumed more by higher income households. However, when examining the loss of purchasing power by looking at expenses as a share of income, lower income households are more negatively exposed to the tariff-induced price increases. Households in the highest income category spend a fraction of their income, while households in the lowest income category spend more than 100%. Consequently, purchasing power losses in the lowest income households are at least double, sometimes greater, than households in the highest income deciles thereby increasing the inequality when purchasing power is measured using the income-based approach.
In Costa Rica, the only country with gender-specific data on household expenditure and income, the spending patterns of women-led households closely resemble those of men-led households across income deciles. Both male- and female-led households in lower income deciles allocate a larger share of their income on essential goods such as food and transportation, while higher-income households spend more on discretionary items. Although some differences exist in the share of expenditure allocated to specific categories, such as personal care, car related items, and water supply, between women-led and male-led households, these variations are minor compared to the overall influence of income. As a result, purchasing power loss due to the tariff increase is nearly the same for male- and female-led households.
Even when using the income-based approach, purchasing power losses are less pronounced but still negative. In Costa Rica, lower-income households, regardless of the household head's gender, spend a larger share of their income, leading to a decline in purchasing power three times greater than that of higher-income households.
However, in cases where a trade policy change disproportionately affects sectors that are more heavily consumed by women, such as personal care or transportation services, the outcome could differ. Relatively stronger price changes in these sectors where spending patterns diverge between the gender of the head of household could lead to different impacts on purchasing power depending on the gender of the head of household. Additionally, in countries where household composition or household consumption patterns differ based on the gender of the head of household, gender-based disparities in exposure to trade-induced price changes could emerge.
References
[10] Aguiar, A. et al. (2022), “The Global Trade Analysis Project (GTAP) Data Base: Version 11”, Journal of Global Economic Analysis, Vol. 7/2, pp. 1-37, https://doi.org/10.21642/jgea.070201af.
[13] Benz, S. and F. Gonzales (2019), “Intra-EEA STRI Database: Methodology and Results”, OECD Trade Policy Papers, No. 223, OECD Publishing, Paris, https://doi.org/10.1787/2aac6d21-en.
[15] Benz, S. and A. Jaax (2022), “The costs of regulatory barriers to trade in services: New estimates of ad valorem tariff equivalents”, Economics Letters, Vol. 212, p. 110057, https://doi.org/10.1016/j.econlet.2021.110057.
[14] Benz, S. and A. Jaax (2020), “The costs of regulatory barriers to trade in services: New estimates of ad valorem tariff equivalents”, OECD Trade Policy Papers, No. 238, OECD Publishing, Paris, https://doi.org/10.1787/bae97f98-en.
[5] Berniell, I., R. Fernandez and S. Krutikova (2024), ““Gender inequality in Latin America and the Caribbean”, NBER Working Paper w32104.”, https://www.nber.org/system/files/working_papers/w32104/w32104.pdf.
[11] Cadot, O., J. Gourdon and F. van Tongeren (2018), “Estimating Ad Valorem Equivalents of Non-Tariff Measures: Combining Price-Based and Quantity-Based Approaches”, OECD Trade Policy Papers, No. 215, OECD Publishing, Paris, https://doi.org/10.1787/f3cd5bdc-en.
[12] Gourdon, J., S. Stone and F. van Tongeren (2020), “Non-tariff measures in agriculture”, OECD Food, Agriculture and Fisheries Papers, No. 147, OECD Publishing, Paris, https://doi.org/10.1787/81933f03-en.
[3] ILO (2019), “Women and Men in the Informal Economy: A Statistical Brief.”, https://www.ilo.org/sites/default/files/wcmsp5/groups/public/%40ed_protect/%40protrav/%40travail/documents/publication/wcms_711798.pdf.
[18] Instituto Nacional de Estadística y Censos (2018), Instructivo para el personal entrevistador: Encuesta Nacional de Ingresos y Gastos de los Hogares (ENIGH), 2018-2019, https://sistemas.inec.cr/pad5/index.php/catalog/244/download/2794.
[8] Liu, C., A. Esteve and R. Treviño (2017), “Female-Headed Households and Living Conditions in Latin America”, World Development, Vol. 90, pp. 311-328, https://doi.org/10.1016/j.worlddev.2016.10.008.
[1] Luu, N. et al. (2020), “Mapping trade to household budget survey: A conversion framework for assessing the distributional impact of trade policies”, OECD Trade Policy Papers, No. 244, OECD Publishing, Paris, https://doi.org/10.1787/5fc6181b-en.
[9] McDonald, S. and K. Thierfelder (2013), Globe v2: A SAM Based Global CGE Model using GTAP Data, Model Documentation, http://cgemod.org.uk/Global%20CGE%20Model%20v2.pdf.
[2] OECD (2023), “METRO v4 Model Documentation”, TAD/TC/WP/RD(2023)1/FINAL., https://one.oecd.org/document/TAD/TC/WP/RD(2023)1/FINAL/en/.
[6] OECD (2022), Trade and Gender Review of New Zealand, OECD Publishing, Paris, https://doi.org/10.1787/923576ea-en.
[4] OECD (2019), “Tackling Vulnerability in the Informal Economy”, https://www.oecd.org/en/publications/tackling-vulnerability-in-the-informal-economy_939b7bcd-en.html.
[16] OECD (2018), Trade Facilitation and the Global Economy, OECD Publishing, Paris, https://doi.org/10.1787/9789264277571-en.
[17] OECD/KIPF (2014), The Distributional Effects of Consumption Taxes in OECD Countries, OECD Tax Policy Studies, No. 22, OECD Publishing, Paris, https://doi.org/10.1787/9789264224520-en.
[7] Saad, G. et al. (2022), “Paving the way to understanding female-headed households: Variation in household composition across 103 low- and middle-income countries”, Journal of Global Health, http://10.7189%2Fjogh.12.04038.
Notes: *Gender of the head of house is available for Costa Rica. **Even when surveys use the COCIOP 1999 classification there was not always a perfect match between the household survey product code and the correspondence table used to match COICOP products and model sectors. In those cases, judgement was used to realign the COICOP code of the household survey. For the full list of COICOP 1999 classification codes see https://unstats.un.org/unsd/classifications/Econ/Download/In%20Text/COFOG_english_structure.txt.
Source: Authors’ compilation.
Note: Countries ordered by average overall tariffs. Trade weighted average. The box represents the interquartile range. Circles above the box are considered outliers. Orange dash represents the trade weighted average of the Ad Valorem Tariff. Specific tariffs, which are in the model database are not shown. They are also not shocked in the analysis.
Source: METRO model database, reference year 2017.
The source of the applied tariff protection in the GTAP (and therefore METRO model) database with a reference year of 2017 is the Market Access Map database produced by the International Trade Centre. The tariff rate is aggregated from the HS 6-digit level to the GTAP sector level using trade weighted average based on three-year average imports and concordance between the HS6 and GTAP sectors.16
Comparing the latest (2024) WTO tariff profiles17 and latest available tariff rates (2022) from WITS18 with the tariff rates in the model database, the differences in weighted averages for each region are small (within 3 percentage points). The one exception is Chile where the model database average tariff rate is close to zero while the average tariff rates for Chile in the WTO tariff profile is 6%. In all regions in the study, apart from Brazil and to some extent Costa Rica, the tariff rates in the model database are, on average, lower than the most recent tariffs (2022) in the WITS data. However, since the average tariff rates in the WTO profiles are still relatively low, the tariff shock to 25% be similar using the latest tariff levels, albeit slightly lower given that the most recent tariff rates are higher than what is in the model database.
At the sector level, the tariff rate has increased more since the database reference year in the agrifood sector compared to the manufacturing sectors in Brazil, where average tariff on manufactured goods decreased, as well as in Argentina, Colombia, and Mexico. Tariffs on imported dairy and meat products in particular saw strong increases in these regions. In Mexico, tariffs on imported grains and fruits and vegetables were 1% or less in 2017 but increased between to between 6.7 (Wheat) to 16.5% (fruits and vegetables) by 2022. For the remaining regions, changes in the applied tariff rates since 2017 by Costa Rica, Peru, and Chile across sectors were either generally small or homogenous.
Note: This table shows the weighted average of the percent change of absolute consumer prices. Average consumer price change weighted by volume of consumption. Average percentage point increase in tariff rates weighted by import volumes.
Source: METRO model.
The METRO model (OECD, 2023[2]) is a multi-country, multi-sector computable general equilibrium (CGE) model, based on the GLOBE model (McDonald and Thierfelder, 2013[9]). The model traces international interdependencies in a theoretically and empirically consistent framework incorporating key features of GVC participation such as trade of intermediate and final products and trade in value added (TiVA) concepts.
The model consists of an individual economies interlinked through trade, with a price system focusing on relative price changes, which is common in CGE models. Each region has its own numeraire, typically the consumer price index, and a nominal exchange rate (an exchange rate index of reference regions serves as model numeraire). Prices between regions change relative to the reference region.
The database relies on the GTAP v11b data (Aguiar et al., 2022[10]) in combination with the OECD Inter-Country Input-Output Tables, UNCOMTRADE data and OECD’s Bilateral Trade in Goods by Industry and End-Use to distinguish trade by end-use. The database contains 160 countries and regions and 65 sectors Available policy information include tariff and tax information from GTAP and OECD estimates of non-tariff measures on goods (Cadot, Gourdon and van Tongeren, 2018[11]); (Gourdon, Stone and van Tongeren, 2020[12]), services (Benz and Gonzales, 2019[13]); (Benz and Jaax, 2020[14]); (Benz and Jaax, 2022[15]), trade facilitation (OECD, 2018[16]) and export restricting measures.
For this analysis, the database was aggregated to 15 regions and 65 sectors.19 Additionally, capital and labour stocks are assumed fixed, and factors are mobile between industries, but not between economies. All factors are fully employed and returns to land and capital and wage rates are flexible. Tax rates are fixed. Governments adjust spending to maintain their pre-simulation fiscal position. The trade balance is fixed, while exchange rates are flexible. Investments are savings driven.
← 1. This framework was also used the Trade and Gender Review of New Zealand (OECD, 2022[6]).
← 2. See Annex 4.B for details about the METRO model, database, and analysis setup.
← 3. Tariff rates above 25% are not changed. Tariff rates between Argentina and Brazil, countries that are part of MERCOSUR, also remain unchanged.
← 4. The model database is aggregated to 15 regions (Argentina, Brazil, Chile, Colombia, Peru, Costa Rica, Mexico, Canada, United States, China, European Union, rest of Europe, rest of Asia, rest of Latin America, rest of World) but maintains all 65 sectors.
← 5. Household survey information is available for four countries: Costa Rica, Mexico, Peru, and Chile. See Annex Table 4.A.1 for more information of the household surveys used. Gender of head of household is available only for Costa Rica.
← 6. One representative household in the METRO model decides which commodities to consume using a Stone-Geary utility function, meaning that households have a minimal level of consumption of essential goods and services, and then decide how much to consume based on (constant) preferences and remaining income. The price changes in the model take the income changes into account, but the household survey data on expenditure allocation by household demographic is static.
← 7. The applied tariffs rates in the GTAP database come from the Market Access Map (MAcMap) which compiles bilateral measurements of applied tariff duties at the Harmonized System (HS) 6-digit level product classification and account for regional agreements and trade preferences. The tariff rates are aggregated to the GTAP sector using a trade-weighted average. See the GTAP documentation for more information: https://www.gtap.agecon.purdue.edu/uploads/resources/download/12097.pdf.
← 8. Reference for the model database is 2017. See Annex 4.B for a description of the difference in tariff rates between the reference year and the current period.
← 9. Under the current international trade framework, customs duties typically apply to tangible goods that cross international borders. Accordingly in the model database, there are no tariff duties applied to services trade, which in the model covers cross-border supply of services and consumption of services abroad.
← 10. The analysis follows the approach in Luu et al (2020[1]), which computed the change in purchasing power based on the compensating variation approach. The purchasing power is computed both on the basis of consumption of a product as a share of total expenditure (referred to as the “expenditure base approach”) and consumption of a product as a share of household income (referred to as the “income base approach”.
← 11. See Annex Table 4.A.1 for the list of household surveys for each country.
← 12. These findings are similar to other studies looking at the distribution effects of taxes using the same approach. Luu et al. (2020[1]), OECD (2022[6]), and OECD/KIPF (2014[17]) found small and somewhat neutral effects on an expenditure basis and regressive effects on an income basis in OECD Member countries.
← 13. The household survey of Costa Rica defines the head of household as the person who has the greatest responsibility in decision-making and generally contributes the majority of the household's economic resources, although not necessarily. In some households (such as non-family groups) the head is the person who has the maximum authority, who carries out the administration, who has resided there the longest or, finally, the oldest. If a person lives alone, he or she is the head of the household (Instituto Nacional de Estadística y Censos, 2018[18]).
← 14. Based on the number of physical assets a household has, such as various building materials of dwelling and household amenities, relative to total possible assets in the country.
← 15. Women-led households in the lowest decile did not purchase a motor car during the survey year.
← 18. World Integrated Trade Solution (WITS) is a database that includes trade and tariff data produced by the World Bank in collaboration with the United Nations Trade and Development, and in consultation with other international trade organisations. Tariff year 2022 is the most recent data available.
← 19. The model data specification is available from the authors upon request. For detailed descriptions of the sectors and geographical coverage of GTAP version 11 see https://www.gtap.agecon.purdue.edu/databases/v11/v11_sectors.aspx and https://www.gtap.agecon.purdue.edu/databases/regions.aspx?Version=11.211.