3. The economic impact of joining the Information Technology Agreement

Brazil has a relatively low use of ICT goods (and services) in the production of exports, including in sectors of comparative advantage such as mining and agriculture (Chapter 2). This reflects, in part, that Brazil might not be a competitive producer of IT goods. Brazil relies on imports to access IT goods, but these are subject to relatively high tariffs and non-tariff measures (Chapter 2).

Against this backdrop, this chapter looks at the economic effects and transmission mechanisms associated with Brazil hypothetically joining the Information Technology Agreement (ITA), an agreement that sets tariffs to zero on a range of IT goods. It does so by building an analytical framework to estimate the value of ITA trade in Brazil, and simulating the impact of ITA accession using a Computable General Equilibrium model (OECD METRO model).

Information Technology (IT) goods form an integral part of the physical infrastructure that enables the digital transformation and participation in digital trade. As a result, and with a view to “encouraging technological development and fostering the expansion of the world economy” (WTO, 1996[1]), some WTO members have taken steps to liberalise IT goods trade through the Information Technology Agreement (ITA) of 1996 and its expansion agreement of 2015 (also known as ITA II). The ITA sets tariffs to zero on a Most Favoured Nation (MFN) basis on a wide range of IT goods including computers, software, telecommunications equipment, scientific instruments, semiconductors and parts and accessories thereof (Box 3.1).

Despite an overlap, ITA goods differ from ICT goods at the product level (Figure 3.1). In this chapter, ITA goods are considered to be those that were liberalised through the 1996 WTO Information Technology Agreement.1 By contrast, the definition of ICT goods used in previous chapters relates to goods that must primarily be intended to fulfil or enable the function of information processing and communication by electronic means, including transmission and display (OECD (2011, p. 20[2]); see Annex 3.A for greater detail).

Practically, this means that ITA goods2 and ICT goods cover similar items. Subset A (goods contained in the ICT list but not in the ITA list) include consumer electronic equipment such as microphones, headphones or monitors, as well as television and radiobroadcasting-related products.3 Subset B (goods contained in both the ICT and the ITA lists) includes computer and peripheral equipment, communication equipment (e.g. mobile phones) and electronic components (e.g. semiconductor devices). Subset C (goods that are in the ITA list but not in the ICT product list) mostly includes scientific instruments, cables, electrical capacitors and resistors, and electrical machinery.

Globally, the share of ITA goods in total trade peaked around the year 2000, thereafter witnessing a decline reflecting falling prices of ITA goods (WTO, 2021[3]). The value of trade in ITA goods increased again starting from around 2015,4 although the change in Brazil’s import mix appears to be rather moderate compared to the global average (Figure 3.2). Brazil’s imports of ITA goods also appear to be lower as a share of total imports than for the world on average.

On average, MERCOSUR countries maintain relatively high tariffs on ITA goods – in 2018 Brazil had an (unweighted) average MFN tariff of 11.6% (20th highest in the world) and Argentina an average MFN tariff of 10.2% (29th in the world) (Figure 3.3). The issue of the MERCOSUR tariff level is also compounded by tariff perforation issues.5 This, combined with the fact that Brazilian imports of ITA goods are considerable, representing about 11% of total imports, suggests that Brazil may have important trade gains from participating in the ITA.6

In addition, Brazil’s binding overhang – the difference between its bound and applied duties – remains high (Figure 3.4, Panel a). This can create additional uncertainty when trading goods (WTO, 2015[17]), further affecting Brazil’s ability to participate in digital trade. In comparison to other non-ITA members, Brazil also maintains high effectively applied duties on ITA goods. Other major non-ITA members have indeed lowered tariffs on ITA goods either unilaterally, as is the case for Mexico (Anderson and Mohs, 2010[8]) or through preferential agreements, as is the case for Chile (Harrison, Rutherford and Tarr, 2002[18]; 2005[19]), or by eventually joining the ITA as part of broader policy objectives as, for example, Colombia (Figure 3.4, Panel b).

Breaking down the tariff regime at the product level shows that Brazil’s tariffs are lower for electronic integrated circuits (1.6% ad valorem) and semiconductor devices (1.7%), which together account for about 29% of the total value of ITA imports.7 However, tariffs remain high on electrical transformers, converters, and inductors (17.4%), electrical capacitors and electrical resistors (14.3%), insulated cables and conductors (14.2%), and media storage devices (14.1%). These products account for about 8.4% of the value of ITA imports in Brazil. In addition, MFN tariffs are also high on telephone sets (13%), computer products (12.3%) and parts of radio apparatus (12%). These products account for 38% of the total value of ITA imports in Brazil (see Annex Table 3.A.1 for greater detail).

Tariff barriers on these products may be more costly in light of growing digitalisation of economic activities, including the growing role of ITA goods as inputs into the production of many different goods and services. Indeed, more that 90% of ITA goods imported by Brazil were intermediate and capital goods in 2018 (Figure 3.5) – compared to 76% for the World on average. While tariffs on ITA products remain higher for consumer goods than for intermediates, the average tariff on ITA intermediate goods is around 10% and that on capital goods is 12.5% for Brazil (Figure 3.6), significantly higher than for the world average (at 5.7% and 5.3% respectively). These duties increase the cost of importing technology-intensive inputs, affecting the ability of Brazil to leverage digital technologies for trade.

Brazil’s hypothetical accession to the 1996 ITA can be assessed in the framework of Computable General Equilibrium (CGE) models, which are especially well suited for ex-ante analysis of potential policy changes. CGE models have their advantages and limitations. On the one hand, CGE models, such as the OECD METRO model (OECD, 2021[20]), provide a comprehensive picture of the input-output linkages within Brazil’s economy and with its trade partners, which allows gauging the consequences of accession on different sectors of economic activity. CGE modelling also helps identify the channels of transmissions of policy changes, mapping their impact on macroeconomic variables like government expenditure and GDP. On the other hand, CGE models are less suited to capturing product specificities, since analysis tends to be at the sectoral level. They also work under a set of relatively strict assumptions about the structure of the economy and their economic interactions.

This means that the results presented herein need to be viewed in the context of the limitations of the model.8 While the model can provide a useful framework for thinking about some of the consequences of joining the ITA agreement, including mapping some of the transmission channels, it only provides a partial view.

Owing to the way the ITA is set up, the value of Brazil’s imports that can be considered as falling within the Agreement can be qualified using a lower and an upper bound estimate.9 10 Brazil is also a member of MERCOSUR, which limits its ability to unilaterally reduce its tariffs on ITA products without introducing some flexibilities in the common external tariff. This leads to four possible modelling scenarios (Figure 3.7).11 The main scenarios of reference for this exercise are scenarios 3 and 4 – providing an estimate of a hypothetical tariff drop in Brazil and Argentina in those sectors affected by the ITA.

The tariff drop resulting from ITA Membership is calculated at the HS 6-digit product level and then aggregated to the METRO sectors. In practice, for Scenario 4, if Brazil were to join the ITA its trade-weighted average tariff on electronic equipment would fall by 49% relative to the baseline value (from 12% to 6%); its tariff on machinery and equipment would fall by 6% (from 12% to 11%) and tariffs on Mineral products would fall by 1% relative to the baseline values.12 By virtue of the common external tariff, Argentina would witness a similar reduction in tariffs (49% in electronic equipment, 5% in machinery and equipment and 2% in mineral products relative to the baseline value). Tariff rate decreases would be lower with a conservative interpretation of what constitutes trade in ITA goods.13

The impact of ITA accession would differ across industries (Figure 3.8). Sectors such as other manufacturing, transport equipment and business services would witness an important expansion in value added, with a strong positive impact for activities in agricultural and primary sectors (natural resources, oil seeds, other agriculture). This expansion of activity relative to the baseline scenario (i.e. a scenario where Brazil and Argentina do not accede to the ITA) is likely to arise as a result of the greater use of imported inputs that become cheaper with tariff liberalisation.

However, ITA accession would also result in concentrated value added losses in the electronic equipment industry – registering a contraction of 3% relative to the baseline value – owing to import competition. Losses would also be recorded in the other services sector arising from adjustment mechanisms in response to customs revenue losses.14 This sector makes up around 43% of total household expenditure, and is therefore particularly exposed to reductions in spending.15

The impact on value added in the other services sector is sensitive to assumptions on the government account closure, i.e. to how the government would respond to the decrease in customs revenues. Indeed, when government expenditure is allowed to be flexible, the reduction in tariff revenue as a result of ITA accession results in even greater negative value added changes in other services. This is because, while the sector accounts for an important share of household expenditure, it also makes up for an even larger share of government expenditure (around 63% of total expenditure) and is therefore particularly exposed to reductions in public spending.16

Regardless of the model closures, however, the emerging picture is one where electronic equipment is likely to see its production volume shrink to the benefit of other economic activities using electronic equipment as inputs, while the impact on the other services will be driven by how the government responds to ITA accession. Overall, in the main scenario, the net change in domestic value added is slightly negative (USD -340 million), due to the large reductions in electronic equipment (USD -917 million) and in other services (USD -207 million), coupled with reductions in demand as a result of higher household savings, especially in insurance and transportation services, food, and textiles. However, where productivity gains driven by tariff reductions might arise, this negative change can be attenuated (Box 3.3).

As a result of ITA accession, overall import demand is projected to increase by 0.20%-0.48% with imports of electronic equipment growing between 3% and 7.6% relative to their baseline value. Bilateral import prices for intermediates in electronic equipment from all trade partners decline in the range of-1.37% to -4.86% depending on the exporter, with similar decreases for capital goods, goods for government consumption and goods for private consumption.

In the electronic equipment sector, imports increase most from G20 partners except Argentina and Mexico, in light of the preferential arrangements currently in place with those countries (trade reorientation effect) (Figure 3.9). In absolute terms, imports from China would grow most followed by imports from the United States.

Brazil’s accession to the ITA is projected to lead to export gains across all sectors except electronic equipment (Figure 3.10). In absolute terms, other manufacturing, natural resources and meats exports would see the largest gains, whereas transport equipment, machinery and equipment and nonferrous metals would see the highest increase relative to their baseline value.17

While the value of exports would increase in traditional sectors of comparative advantage such as natural resources, export gains are also observed in business services and other services, highlighting the complementarities between access to technological hardware inputs and export competitiveness in services sectors. Overall, ITA accession can have a positive impact on export diversification.18

The overall effects of ITA accession are sensitive to different model closures. For instance, the medium term impact on GDP can range between -0.03% when government expenditure is fixed to +0.01% when government expenditure is allowed to adjust, pointing to an overall impact on GDP that can be considered as neutral. However, these results are based on estimates that do not include possible productivity gains from ITA accession, or other secondary effects such as on investment.

To capture the productivity effects of ITA accession, a targeted shock, based on the 2020 OECD Brazil Economic Survey, is implemented (Figure 3.11 and Box 3.3). The results show that medium-term real GDP may increase by +0.17% in Brazil and +0.04% in Argentina as a result of joining the ITA in the presence of positive productivity dynamics. These figures are, however, based on a particular estimate of the elasticity of labour productivity with respect to tariffs (OECD, 2020[27]). Indeed overall GDP impacts are relatively sensitive to the elasticity of labour productivity with respect to tariffs, that is, the extent to which tariff liberalisation affects productivity. Nevertheless, the modelling shows that GDP gains also arise in the context of lower productivity shocks (Figure 3.11).19

The analysis also points to potential differences in the economic impact of accession between Brazil and Argentina. Owing to a smaller electronic equipment sector, Argentina’s accession to the ITA is unlikely to have a negative impact on real GDP, even in the absence of positive productivity dynamics. However, it is also unlikely to lead to very high GDP effects, even when higher productivity effects are modelled. By contrast, Brazil displays a steeper curve in Figure 3.11, highlighting a higher potential for benefits from ITA accession in the presence of higher labour productivity dynamics, but, at the same time, greater potential to witness a GDP contraction if these productivity gains do not materialise due to pressures from import competition.

Beyond affecting macroeconomic results such as real GDP, the productivity shock also affects sectoral value added, imports and exports leading to somewhat different results to those reported in the body of the report. In general, strong productivity effects in the electronic equipment and machinery and equipment sectors increase factor demand for those activities. With mobile factors across sectors but fixed overall endowments, this can negatively affect output in other sectors of the economy, especially in agriculture, services and primary activities (see Annex 3.A for a discussion of the impacts).

In terms of other macroeconomic variables of interest, ITA accession is likely to lead to decreases in prices and slight increases in household income. However, the degree to which households will be able to translate higher income into expenditure and welfare gains is also sensitive to closures in the government account. When government expenditure is flexible, households increase expenditure (+0.05%) and decrease their savings rate (-0.52%). However, when government expenditure is fixed, households adjust savings (+1.81%) and decrease their expenditure (-0.09%) to compensate for changes in the government balance. This means that the impact on household welfare can vary from a net aggregate gain of USD 807 million (flexible government expenditure) to a net loss of USD 1.39 billion (fixed government expenditure).20

In terms of government revenue, ITA accession is expected to lead to a 4% to 10% reduction in revenues from tariffs (equal to about USD 780 million to USD 1.9 billion) depending on whether the lower or the upper bound estimate is used.21 This estimate is not sensitive to different model closures or productivity dynamics. While sizeable, this shock should be put in the perspective of the revenue generated form tariffs as a share of total government revenues. In the baseline scenario, import tariff revenue accounts for 2.6% of Brazil’s government revenue (USD 19 billion),22 with a much larger role played by factor use taxes (29.7% of total revenue – USD 220 billion), sales taxes (23.47% ‒ USD 174 billion), income taxes (20.3% ‒ USD 150 billion USD) and value added taxes (19.6% ‒ USD 146 billion).

Nevertheless, the government account can be a key transmission mechanism for the impact of ITA accession on Brazil’s economy. In this regard, a scenario in which Brazil maintains its current government expenditure would lead to the costs of accession being spread across economic activities and households, whereas a decrease in government expenditure is likely to result in significant reductions in the value added of the other services sector, in which government expenditure plays an important role.

While more comprehensive and complex modelling approaches may be required to determine the optimal tax policy response to ITA accession for Brazil, tax policy considerations should play a role if the prospects of ITA accession are discussed. This includes in terms of how government expenditure would respond as well as whether alternative revenue sources should be considered to compensate for the losses from tariff revenues.

Computable General Equilibrium models provide a useful framework for ex-ante analysis of policy changes. They are especially useful in highlighting the channels of transmission arising from different policy choices and for alerting policy-makers to areas they might have not anticipated. However, as has been shown, the results are relatively sensitive to the assumptions made in the closure of the model, this is why it is important to point to some of the limitations of this modelling approach.

  • While productivity-enhancing effects are likely to result from increased access to ITA goods, they are difficult to model. This chapter has shown the sensitivity of macroeconomic results to different sizes of the productivity effect. More need to be done to better capture productivity enhancing effects of access to IT goods across other sectors.

  • The current model set-up does not take into account possible benefits arising from commitment on tariff rates (e.g. ‘commitment effects’ in WTO (2015[10]). Increased certainty arising from binding tariff commitments are likely to result in positive impacts on investment and could also promote GVC integration.

  • The model set-up does not capture the benefits of reducing tariffs to zero on a number of products, as the tariff rate shock is applied as a weighted average in broader sectors in METRO. Zero tariffs on some electronic equipment would lead to gains in trade facilitation including streamlined customs procedures and reduced delays (WTO, 2015[10]). These effects cannot not be captured by reducing average tariff rates in broader industries.

  • Beyond limitation related to the general approach, particular model closures also have an important impact on results. These include, for instance, assumptions on what the government would do to respond to the drop in tariff revenues resulting from accession, including for instance whether it would raise additional revenue through other tax instruments or rather reduce government expenditure. Results are therefore reported noting the sensitivity to different model closures in the government account and in the current account.

  • ITA accession is likely to result in benefits across many industries. This is because lower prices for electronic equipment and related items lower input costs across a range of sectors. However, import competition leads to concentrated value added losses in domestic IT good producing sectors.

  • Lower import prices contribute towards greater ICT input use. Consequently, imports of electronic equipment increase between 3% to 7.6% following accession.

  • Accession can also lead to greater exports, especially in primary activities but also in services and manufacturing activities. In this respect ITA accession has the potential to contribute to export diversification.

  • When productivity effects are incorporated, ITA accession could increase Brazil’s real GDP in the medium-term by 0.17% and Argentina’s by 0.04%.

  • However, ITA accession could lead to a 4% to 10% decrease of tariff revenue, and while this accounts for a small share of overall government revenue, fiscal policy considerations should accompany discussions on the region’s accession to the agreement.

The Ministerial declaration on trade in information technology products of December 1996 (WT/MIN(96)/16) is used to identify the original list of ITA goods.23 The declaration contains two attachments: Attachment A – which is itself subdivided in Section 1 and Section 2 ‒ is a list of products identified by specific HS 6-digits codes in the 1996 nomenclature. Attachment B contains product descriptions for items included in the agreement but not identified by specific HS codes.

Ninety-five of the 190 items contained in the ITA are accompanied by a notation (ex) which indicates that the items set to attract zero tariffs are found at a more specific level than the HS 6-digits level (i.e. an 8 or 10 digits sub-heading in national schedules) (WTO, 2012[9]). Since the HS classification system is harmonised internationally only up to the 6-digit nomenclature and comparative trade statistics can only be obtained at this level of aggregation, the value of trade in ITA goods can be identified by using one of three different approaches (WTO, 2012[9]).

A first approach would involve taking into account only the value of trade for the HS 6-digits codes under which all goods are set to attract zero tariffs, ignoring those products liberalised with ex-outs. This would result in an underestimation of the value of trade in ITA goods.

A second, “mixed” approach involves taking into account all the HS lines that are fully liberalised and some of the codes reported with ex-outs but that contain a high proportion of ITA products (WTO, 2012[9]; WTO, 2017[12]). This approach leads to a smaller underestimation of the value of trade in ITA products.

The third approach, which is the one adopted in this report for the descriptive section, consists in taking into account the full value of trade under HS codes that are affected by the ITA, regardless of partial coverage in some of the sub-headings (Bora, 2004[28]). In the absence of detailed information, this approach allows to avoid making judgements about which HS 6-digits lines contain a high proportion of ITA products in HS1996, as well has having to make such judgement for the different nomenclatures of the HS.

While this approach may lead to an overestimation of the value of trade in ITA goods (WTO, 2012[9]) it is adopted mutatis mutandis for all countries in the main body of the report.24 It is also not adopted in the economic modelling section, where findings are reported depending on whether a “short” or “long” list of ITA products is used, producing a lower bound estimate and an upper bound estimate.

For the trade data, Annex Figure 3.A.1 below reports the value of imports and exports of ITA products using the first approach (considering only fully covered lines) and the third approach (considering all concerned lines at the 6-digit level) respectively. Since comparable tariff data is also only available at the HS 6 digits level, similar considerations apply for the analysis of tariff data.

The list of products covered by the ITA was affected by each revision to the HS system, which undergoes periodic updating in light of developments in technology.25 Amendments affected the ITA product list to different extents, with the revision to the 2007 Harmonised System of classification bringing the most changes (WTO, 2017[12]; Anderson and Mohs, 2010[8]). While difficult, tracking the classification of ITA goods in more recent versions of the HS system is necessary to ensure that tariff data reflect the regime faced by the goods liberalised with the 1996 Declaration, regardless of where they are re-classified in successive HS nomenclatures.

In order to transpose the original model list in successive nomenclatures, the correspondence between HS1996 and HS2002 is obtained using the WTO document G/IT/W/22: ‘List of changes due to the introduction of HS2002 Nomenclature in relations to the ITA products’. Lists of ITA products in HS2007, HS2012 and HS2017 revisions are obtained using UN correlation tables and World Customs Organisation (WCO) correlation tables. Hence, the number of unique HS 6 digits codes covered in the analysis changes according to different HS revisions, producing five different model lists for the different time intervals: 154 codes (HS 1996), 155 codes (HS 2002), 112 codes (HS 2007), 113 codes (HS 2012) and 115 codes (HS 2017) for the upper bound estimate.

In this respect, and in order to further refine the classification of ITA goods, an exercise was undertaken to verify the list of ITA goods in the HS2012 and HS2017 nomenclatures against tariff information for ITA Members in the TRAINS database. This ensures that the ITA model lists are consistent with observable bound and applied tariff rates and contributes to distinguishing ‘ex-out’ products from those entirely covered by the agreement in the more recent HS nomenclatures.

It is worth noting that while the approach based on model lists allows for comparative analysis, it may lead to apparently inconsistent results in specific instances when comparing the information in the model list to the applied tariffs for ITA members. This depends on, inter alia, different choices on transposition of ITA HS codes at the national level (WTO, 2012, p. 97[9]).26

For trade data extracted from the UN COMTRADE database, the automatic conversion function available in WITS is used for the transposition of HS codes, as the value of trade extracted therefrom closely tracks the value of trade extracted through the different model lists. Hence, the different models lists are mainly used to obtain tariff and trade data from the TRAINS database, as well as to undertake the CGE exercise using the METRO model.

A further obstacle to the analysis of trade and tariffs on ITA products involves goods classified “in” and “for” Attachment B of the 1996 Ministerial Declaration, since these goods are either not identified by specific HS codes or subject to classification divergences (WTO, 2012[9]; WTO, 2017[12]; Anderson and Mohs, 2010[8]; Ezell and Wu, 2017[16]). Nevertheless, agreement was found on the classification of some of these products into specific HS categories in WTO decisions G/IT/27 of 2013 and decision G/IT/2016 of 2016.

For products “in” attachment B ‒ not identified by specific HS codes in the original Ministerial declaration ‒ Decision G/IT/27 of 2013 classifies Monitors and Optical disc storage units in HS1996 codes 847160 and 847170 respectively, and decision G/IT/2016 of 2016 classifies Computers in HS2007 codes 847130, 847141, 847149, and 847150. In order to compare product data across time, the transposition of Monitors and Optical disc storage units from HS1996 to other HS revisions is obtained using UN and WCO correlation tables. For computers, the transposition from HS2007 to HS1996 and HS2002 is obtained using the WTO document G/IT/W/40, while the transposition for HS2012 and HS2017 is obtained using UN and WCO correlation tables.

This approach, however, does not lead to the identification of additional HS 6 digits codes to those already covered in the existing model lists. For the category of products identified by both specific HS1996 codes and the notation “for” attachment B in the original Ministerial declaration, the original HS codes are used for identification27.

This chapter also differentiates goods according to their different end-uses. This classification is obtained through the correspondence key of the OECD BTDIXE database, which allows for the differentiation between capital goods, intermediate goods and goods destined for household consumption, among others.

ITA goods are defined in this report as the unique HS 6-digits goods covered in the WTO Information Technology Agreement of 1996. In contrast, ICT goods are defined as the HS 6-digits list of goods that must be intended to fulfil the function of information processing and communication by electronic means, including transmission and display (OECD, 2011, p. 30[2]).

In broad terms, the Information Technology agreement covers computers, software, telecommunications equipment, scientific instruments and semiconductors as well as most of the parts and accessories of these products.28 ICT goods include computers and peripheral equipment, communication equipment, consumer electronic equipment, electronic components and miscellaneous ICT goods (OECD, 2011, pp. 40-46[2]).

The total number of goods covered and the difference between the two product lists vary depending on the HS nomenclature. Taking the HS2007 nomenclature for illustrative purposes, ITA goods include a list of 112 unique HS codes whereas ICT goods include 95 unique HS codes. The main category of goods that is in the ICT list but not in the ITA list is consumer electronic equipment, which was carved-out of the ITA negotiation process (Fliess and Sauvé, 1997, p. 27[29]). Goods that are in the ITA list but not in the ICT list include electrical capacitors and resistors, insulated wire, cable and conductors, calculating machines, scientific equipment and semiconductor-related equipment. At the HS 6 digits level, the two lists mainly overlap in Computers and Peripheral Equipment (17 instances) and Electronic Components (15 instances).

Translating the hypothetical tariff drop following ITA accession from the product-level to the sector level in METRO requires making use of a correspondence. The correspondence used in this exercise related goods in the HS2012 nomenclature to METRO sectors. On this basis, the trade-weighted average tariff drop is calculated based on the sectorial aggregation. The use of the ITA short list and long list leads to the differentiation between the lower bound estimate from the upper bound estimate in the exercise.

Including the labour productivity enhancing effects of ITA accession in the METRO simulation would lead to more reduced gains in value added for most economic activities. Gains would mostly remain positive in manufacturing activities as well as some other sectors (Annex Figure 3.A.4), as manufacturing processes can more effectively benefit from greater labour productivity in related production activities and greater access to imported intermediate inputs. Greater household income and expenditure stemming from productivity gains in turn supports demand for some of the unrelated production activities (e.g. financial services and communication). Including the productivity shocks in the simulation also leads to a smaller contraction of sectorial value added in the electronic equipment sector, and a smaller net value added decrease in total.

As regards the other economic variables, imports of electronic equipment would still grow although in much smaller quantities (+1.2% instead of +7.6% in the absence of productivity dynamics). Bilateral imports of electronic equipment would grow from Argentina instead of declining (as the country would also experience analogous productivity dynamics) while trade reorientation effects would be much stronger for electronic equipment imports from Mexico, Latin America, South East Asia, and Korea.

Similarly to value added, export gains would mainly materialise in manufacturing activities, with most agricultural, primary and services activities bearing the consequences of greater competition for production factors (especially capital). Electronic equipment would also gain an important export share, which contrasts with its loss of export values in the absence of productivity effects, and as a reflection of a more productive domestic industry and subdued domestic prices.

It should also be noted that these effects are compounded by the METRO model set-up and closures in the factor market: the negative effects on activities outside manufacturing are reduced when allowing for unemployment in the model, which enables greater adjustment of production factors. Capital endowments are also not allowed to increase in the static model, while this would relax the production constraints due to fixed factor endowments.


[22] Aguiar, A. et al. (2019), “The GTAP Data Base: Version 10”, Journal of Global Economic Analysis, Vol. 4/1, pp. 1-27, https://www.gtap.

[8] Anderson, M. and J. Mohs (2010), The Information Technology Agreement: An Assessment of World Trade in Information Technology Products.

[24] Benz, S. and F. Gonzales (2019), “Intra-EEA STRI Database: Methodology and Results”, OECD Trade Policy Papers, No. 223, OECD Publishing, Paris, https://dx.doi.org/10.1787/2aac6d21-en.

[25] 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://dx.doi.org/10.1787/bae97f98-en.

[28] Bora, B. (2004), The Information Technology Agreement and World Trade.

[6] Bora, B. and X. Liu (2008), “Evaluating the Impact of the WTO Information Technology Agreement”, https://doi.org/10.1142/9789814299404_0002.

[23] 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://dx.doi.org/10.1787/f3cd5bdc-en.

[15] European Commission (2016), “The Expansion of the Information Technology Agreement: An Economic Assessment”, https://doi.org/10.2781/11765.

[16] Ezell, S. and J. Wu (2017), How Joining the Information Technology Agreement Spurs Growth in Developing Nations.

[5] Feenstra, R. (2008), “Offshoring in the Global Economy”, The Ohlin Lectures, Stockholm School of Economics, Stockolm, https://www.researchgate.net/publication/242382768.

[29] Fliess, B. and P. Sauvé (1997), Of Chips, Floppy Disks and Great timing: Assessing the Information Technology Agreement, https://www.nomurafoundation.or.jp/en/wordpress/wp-content/uploads/2014/09/19971011_Barbara_Fliess_-_Pierre_Sauve.pdf (accessed on 4 October 2021).

[11] Gnutzmann-Mkrtchyan, A. and C. Henn (2018), “Peeling away the layers: Impacts of durable tariff elimination”, Journal of International Economics, Vol. 115, pp. 259-276, https://doi.org/10.1016/j.jinteco.2018.09.003.

[19] Harrison, G., T. Rutherford and D. Tarr (2005), “Chile’s Regional Arrangements: The Importance of Market Access and Lowering the Tariff to Six Percent”, in Chumancero, R. and K. Schmidt-Hebbel (eds.), General Equilibrium Models for the Chilean Economy, Central Bank of Chile, Santiago, Chile, https://si2.bcentral.cl/public/pdf/banca-central/pdf/v9/303_344harrison_rutherford_tarr.pdf (accessed on 5 October 2021).

[18] Harrison, G., T. Rutherford and D. Tarr (2002), “Trade Policy Options for Chile: The Importance of Market Access”, The World Bank Economic Review, Vol. 16/1, pp. 49-79.

[14] Joseph, K. (2013), Information Technology Agreement of the WTO: Call for a Revisit.

[13] Kallummal, M. (2012), “Process of Trade Liberalisation under the Information Technology Agreement (ITA): The Indian Experience”, Working Paper, No. CSW/WP/200/3, Centre for WTO Studies, New Delhi.

[30] Laens, S. and M. Terra (n.d.), “Deepening Integration of MERCOSUR: Dealing with Disparities”, Integration and Regional Programs Department, Inter-American Development Bank, https://publications.iadb.org/publications/english/document/MERCOSUR-Asymmetries-and-Strengthening-of-the-Customs-Union-Options-for-the-Common-External-Tariff.pdf (accessed on 5 October 2021).

[7] Mann, C. and X. Liu (2007), The Information Technology Agreement: Sui Generis or Model Stepping Stone?, https://www.wto.org/english/tratop_e/region_e/con_sep07_e/mann_liu_e.pdf (accessed on 5 October 2021).

[21] Mcdonald, S. and K. Thierfelder (2013), Globe v2: A SAM Based Global CGE Model using GTAP Data.

[20] OECD (2021), METRO: the OECD’s Trade Model, OECD Publishing, Paris, https://www.oecd.org/trade/topics/metro-trade-model/.

[27] OECD (2020), OECD Economic Surveys: Brazil 2020, OECD Publishing, Paris, https://dx.doi.org/10.1787/250240ad-en.

[26] OECD (2017), “Economy-wide Impacts of Trade Facilitation: a METRO model simulation”, OECD, Paris, http://www.oecd.org/officialdocuments/publicdisplaydocumentpdf/?cote=TAD/TC/WP(2016)15/FINAL&docLanguage=En (accessed on 18 December 2020).

[2] OECD (2011), OECD Guide to Measuring the Information Society, OECD, https://www.oecd-ilibrary.org/science-and-technology/oecd-guide-to-measuring-the-information-society-2011_9789264113541-en (accessed on 4 October 2021).

[31] The World Bank (2019), Effects in Argentina of a liberalization in the Mercosur common external tariff, https://documents1.worldbank.org/curated/pt/132981561656833211/pdf/Effects-in-Argentina-of-a-Liberalization-in-the-Mercosur-Common-External-Tariff.pdf (accessed on 5 October 2021).

[4] WTO (2021), Information Technology Agreement, https://www.wto.org/english/tratop_e/inftec_e/inftec_e.htm.

[3] WTO (2021), ITA Symposium: 25 Years of the Information Technology Agreement, https://www.wto.org/english/tratop_e/inftec_e/ita_symp_sep21_e.htm?utm_source=twitter&utm_campaign=social&utm_content=organic2021q4.

[12] WTO (2017), 20 Years of the Information Technology Agreement, World Trade Organisation, Gevena, https://www.wto.org/english/res_e/booksp_e/ita20years_2017_full_e.pdf.

[32] WTO (2017), “Trade policy review of Brazil”, WTO Trade Policy Reviews, No. WT/TPR/S/358, World Trade Organization, Geneva, https://www.wto.org/english/tratop_e/tpr_e/s358_e.pdf (accessed on 5 October 2021).

[10] WTO (2015), “The Layers of the IT Agreement’s Trade Impact”, WTO Working Papers, No. 2015/01, World Trade Organization, Geneva, https://dx.doi.org/10.30875/92cd71f9-en.

[17] WTO (2015), “Trade Policy Uncertainty as Barrier to Trade”, WTO Working Papers, No. 2015/05, World Trade Organization, Geneva, https://dx.doi.org/10.30875/6c9ef04c-en.

[9] WTO (2012), 15 Years of the Information Technology Agreement : trade, innovation and global production networks., World Trade Organization.

[1] WTO (1996), Ministerial Declaration on Trade in Information Technology Products (WT/MIN(96)/16), https://docs.wto.org/dol2fe/Pages/SS/directdoc.aspx?filename=Q:/WT/MIN96/16.pdf&Open=True (accessed on 4 October 2021).


← 1. This chapter looks at the liberalisation of 1996 ITA goods as this covers most computers, telecommunication equipment and electronic components, while the ITA II agreement adds to the pre-existing agreement by deepening and broadening its scope, including for instance for some medical equipment and scientific instruments.

← 2. Unless otherwise specified, IT goods refer to goods covered in the 1996 Information Technology Agreement.

← 3. This could be explained by the carve-outs in the ITA on consumer electronic equipment (Fliess and Sauvé, 1997, p. 27[29]).

← 4. Still against the background of falling prices (WTO, 2021[3]).

← 5. Within MERCOSUR, the average MFN tariff on ITA goods varies significantly. MERCOSUR members appear to rely on a range of exceptions to the common external tariffs in their tariff schedules for ITA goods. Data from the tariff schedules of active MERCOSUR countries (Argentina, Brazil, Paraguay, Uruguay) in the NCM nomenclature reveal that MERCOSUR countries maintain a common MFN external tariff in only 155 of the 484 8-digits tariff lines containing ITA goods (upper bound estimate of ITA coverage), whereas at least one of the four countries makes use of exceptions in the remaining tariff lines. The issue of the tariff level on ITA goods is hence compounded by tariff perforation issues, as elsewhere reported by the World Bank (2019[31]) the Inter-American Development Bank (Laens and Terra, n.d.[30]), and the WTO (2017[32]). See the WTO trade policy review of Brazil (2017[32]) for a discussion of IT-related exceptions.

← 6. Note that this is an upper bound estimate of the coverage of the ITA (Annex 3.A).

← 7. This suggests that there may be gains in lowering tariffs to zero on some products that represent a high share of ITA imports while attracting relatively low tariff levels.

← 8. Other caveats can be found in the ‘Limitations’ sub-section below.

← 9. This is a consequence of the structure of the ITA (i.e. product ex-outs). The upper bound estimate includes all goods at the HS 6 digit level that can be considered as falling within the ITA, including those with product ex-outs (covered at the 8 or 10 digit level, for instance). The lower bound estimate only includes those goods that are fully covered by the ITA at the HS 6 digit level of aggregation. Note that the upper bound estimate does not include Appendix B goods, meaning that it may underestimate the maximum possible value of ITA imports. See the methodological annex for more information.

← 10. See European Commission (2016[15]) for a similar modelling approach.

← 11. Estimates for MERCOSUR accession are only based on Brazil and Argentina simultaneously lowering tariffs, as Uruguay and Paraguay cannot be separately identified in the OECD METRO model.

← 12. The mineral product is HS 702000, ‘Other articles of glass’ included in the ITA model list with a product ex-out. In the original ITA agreement, these are ‘Quartz reactor tubes and holders designed for insertion into diffusion and oxidation furnaces for production of semiconductor wafers’ (HS1996: 701710).

← 13. With the lower bound estimate interpretation of trade in ITA goods, the weighted average tariff in Brazil would fall by 20% in Electronic equipment and 4% in Machinery and Equipment; it would fall by 28% in Argentina for Electronic equipment and 3% in Machinery and Equipment.

← 14. As the government keeps its current level of expenditure fixed, households will consume less and save more in order to compensate for increases in debt. This means they will spend less and other services is a key item of consumption for households.

← 15. Information obtained from Brazil’s Social Accounting Matrix (SAM) in METRO. Communication services and financial services also make up a relatively important share of household expenditure (4.86% and 4.03% of total expenditure respectively) while also showing a low degree of electronic equipment input use, which explains the observed value added contraction in those sectors.

← 16. At the same time, flexible government spending in response to the tariff change would lead to a greater positive impact on other economic activities as adjustments to household savings rates would no longer be required to compensate for higher government debt, meaning that household consumption would be less affected (see Annex Figure 3.A.3 for a comparison of the two scenarios).

← 17. In the current setting, the trade balance remains fixed following tariff liberalisation, while the exchange rate is left to adjust, which is more likely to reflect Brazil’s response to ITA accession (Brazil maintains a floating exchange rate). Changing this assumption and fixing the exchange rate instead of the trade balance would lead to greater losses in electronic equipment exports, together with small net decreases in agricultural and primary sectors’ exports (see Annex Figure 3.A.1. for comparison). Export changes would remain positive only in few activities (Annex 3.A).

← 18. As regards the geographical distribution of exports, exports in dollar terms would grow mostly to China (USD +124 to USD 300 million), the United States (USD +116 to USD 285 million), the European Union (+USD 63 to USD 153 million) and the Rest of the World category (+USD 202 to USD 496 million). In relative terms, they would mostly grow to Mexico (+0.5% to 1.1% relative to the baseline value), South Africa (+0.4% to 1%) and Canada (+0.4% to 0.9%).

← 19. For the estimated values, Brazil’s medium-term GDP estimate relative to the elasticity of labour with respect to tariffs can approximately be described by the function y=0.91x-0.03, where y represents real GDP growth in percentage terms and x stands for the elasticity parameter (n.b., the curve is actually slightly downward sloping when looking at more specific decimals, with the coefficient of the x varying from 0.9164 at elasticity equal to 0.05 to 0.91297 for elasticity equal to 0.3). This means that starting from an elasticity of labour productivity with respect to tariffs equal to 0.0325 (which implies in the model a 0.8% increase in labour productivity in electronic equipment and a +0.1% increase of labour productivity in other manufacturing equipment), Brazil would start observing positive GDP gains from ITA accession over the medium term (in the scenario with the fixed government expenditure closure).

← 20. Upper bound ITA estimate. Households experience greater gains in expenditure and income when the simulation incorporates the productivity-enhancing effects of ITA accession, as they both benefit from greater factor remuneration and lower domestic prices. Using the benchmark estimate of -0.22 for the elasticity of labour with respect to tariffs, household welfare gains equal USD 675 million with fixed government expenditure (instead of USD -1.39 billion in the absence of productivity dynamics).

← 21. Estimates are in a similar range for different government account closures and for the simulation incorporating the productivity enhancing effects of ITA accession.

← 22. A more recent estimate from the OECD Global Revenue Statistics Database also shows that tariff revenue accounted for a minor share of total revenue, at about 1.8% of total taxation in both 2018 and 2019.

← 23. HS 1996 codes 8424.99 and 8456.93 are corrected to 8524.99 and 8466.93 respectively. Although the ITA agreement did not yet enter into force at the Singapore Ministerial Conference, the final list of ITA goods has not been subject to substantial changes after the Ministerial Conference (Fliess and Sauvé, 1997, p. 30[29]).

← 24. Attachment B products are also absent from this upper bound estimate.

← 25. See http://www.wcoomd.org/en/topics/nomenclature/overview/what-is-the-harmonized-system.aspx.

← 26. This may also depend on, for instance, countervailing duties applied on goods that would otherwise attract zero tariffs, or on duties being different from zero as new members to the agreement are in the process of transitioning towards zero bound duties.

← 27. ITA members can chose to classify ITA products “for” Attachment B in different product categories than those reported in the Ministerial declaration (G/IT/W/6/Rev.3).

← 28. See https://www.wto.org/english/tratop_e/inftec_e/itaintro_e.htm.

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

© OECD 2022

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at https://www.oecd.org/termsandconditions.