5. Enabling Brazilian firms to export in the digital era
This chapter uses detailed firm-level data to investigate different aspects of the digital transformation for Brazilian firms. The analysis shows that imported digital inputs contribute to the export competitiveness of Brazilian goods exporters. These are especially important for micro-sized firms. It also highlights that the digital services trade restrictions faced by Brazilian exporters significantly affect their capacity of export digitally deliverable and non-digitally deliverable services, especially in the case of smaller exporters.
Although only 0.46% of firms in Brazil were engaged in international trade, those that do command a disproportionate share of economic activity (8.9% of total employment).
Out of those that engage in trade, most, 67% do so through a single trade channel, whether import or export of goods and/or services. The remaining 33% of firms, employing 59% of workers engaged in trade are simultaneous traders.
Large firms represent only 7% of all trading firms but they employ most workers (72%) and represent the highest share of exports (59%), they also tend to be simultaneous traders. By contrast, micro firms represent the largest share of firms (52%) but the smallest share of workers (2%) and an intermediate share of exports (17%). They also tend to be single traders.
Enabling Brazilian exports in the digital era requires facilitating access to ICT inputs and reducing barriers to digitally enabled trade in partner countries, this is especially important to enable smaller firms to benefit from trade in the digital era.
Export growth can be a driver of economic growth (Brenton and Newfarmer, 2007[1]), which is why understanding export competitiveness is key to informing policy action to enable benefits from digitalisation. Ultimately, it is firms that export, which is why it is important that analysis on export competitiveness be undertaken at the firm level.
This chapter uses micro-data to shed light on different aspects of the digital transformation of Brazilian trading firms. The next section provides an overview of the characteristics of Brazilian traders. Section two follows with an analysis of the role of imported ICT inputs in enabling export competitiveness. Section three reviews how digital services restrictions affect the ability of Brazilian firms to export digitally deliverable services. The final section concludes with some observations and policy recommendations.
5.1.1. Profile of Brazilian trading firms
As is the case across most countries, trading is relatively rare among firms in Brazil. In 2019, only 0.46% of firms were engaged in international trade (whether importing or exporting goods or services). However, these firms accounted for 9% of total employment, underscoring that trading firms command an important share of economic activity (Figure 5.1).
Source: Own calculations based on data from SECEX, SISCOSERV and RAIS.
The largest share of trading firms is in wholesale and retail representing 34% of Brazilian trading firms and 20% of employment in trade. This sector is also one of the most engaged in international trade representing 29% of goods imports, 19% of goods exports and 11% and 8% of services imports and exports.1 The other business services sector also represents a considerable share of both trading firms (14%) and employment in trade (12%) and is widely engaged in services trade as importer (11% of services imports) and exporter (16% of services exports).
Firms in the coke and refined petroleum sector were the largest services importers (32% of total services imports)
Firms in financial and insurance activities were the largest services exporters (30% of total services exports)
Firms in the wholesale and retail trade; repair of motor vehicles sector were the largest goods importers (28% and 29% of total goods imports respectively)
Firms in extractive industries and wholesale and retail trade were the largest goods exporters (19% of total goods exports each).
Most Brazilian firms that engage in trade do so only as importers of goods (33.3%) while 17.2% of trading firms engage only as goods exporters, respectively employing 21% and 9% of the workforce engaged in trade (Figure 5.2). Most trading firms, 67%, only engage in one, single, type of trade channel, whether import or export of goods or services.2 However, simultaneous traders, which are firms engaging in more than one channel of trade (whether as exporters of goods and exporters of services or importers of goods and importers or services or combinations thereof), although representing a lower share of traders (33%), they occupy a high share of employment (59%) and an even higher share of exports (75%).3
Note: The left panel shows the share of firms according to different categories of traders. The right panel does the same but using employment data. Each category is mutually exclusive so that, those in the “Export Goods” category are firms that only export goods. Firms that export goods but also import goods appear in the “Imports and Exports of Goods” category. The “Other” category represents permutations of goods and/or services imports and/or exports, not represented in the figure.
Source: Own calculations based on data from SECEX, SISCOSERV and RAIS.
Focusing on simultaneous traders (Table 5.1), the most common activity of firms is the simultaneous import and export of goods, although this is closely followed by imports of goods and services. Over 31% of the value of Brazilian exports involves firms that import goods and services and export goods, category which also occupies the largest share of employment (21.3%).
The distribution of traders according to firm size follows similar patterns to those of OECD countries.4 That is, large firms employ most workers (72%) and represent the highest share of exports (59%) but they represent the smallest share of trading firms (7%) (Figure 5.3). By contrast, micro firms represent the largest share of firms (52%) but the smallest share of workers (2%) and an intermediate share of exports (17%). Overall, large firms tend to engage more readily in simultaneous trade while smaller, firms, particularly micro-sized, tend to be single traders whether of goods or of services (Figure 5.4).
Note: Large firms defined as those above 250 employees, Medium firms between 100 and 250, Small between 10 and 100 and micro fewer than 10.
Source: Own calculations from SECEX, SISCOSERV and RAIS.
Note: Large firms defined as those above 250 employees, Medium firms between 100 and 250, Small between 10 and 100 and micro fewer than 10.
Source: Own calculations based on data from SECEX, SISCOSERV and RAIS.
5.1.2. Existing evidence of trade in bundled products (simultaneous export of goods and services)
Firms that export goods and services, or bundled products, irrespective of whether they import or not, represent around 13% of exports, 6.7% of employment and 2.3% of firms (Table 5.1). Bundles vary widely but some patterns emerge. In terms of number of bundles these include combination of manufactured goods, from transmission shafts to electrical transformers and rubber and plastics with maintenance services. In terms of highest value bundles, these concentrate in the light aircraft sector with a combination of engineering, consultancy and IT support (Table 5.2).
Existing analysis based on micro-data has shown that firms engaged in trade are not only larger and more productive but also create more jobs and pay higher wages (Melitz and Redding, 2014[2]). However, despite considerable advancements in the empirical findings of this heterogeneous firm literature, the specific role that ICT goods and services play in enabling firms to trade is not well understood. This, despite a wide acknowledgement that ICT inputs have the potential to play a critical role in productivity and in reducing trade costs (see World Bank (2016[3]), Baldwin (2016[4]), WTO (2018[5]) and Box 5.1).
The literature on heterogeneous firms offers important insights into how firms engage in trade and what benefits they draw from this engagement. More productive firms are able to meet the costs of engaging in international markets and, as a result, draw benefits associated with exporting (Melitz, 2003[6]). At the same time, access to more varieties of competitively priced inputs through imports is also associated with greater export competitiveness (Bas and Strauss-Kahn, 2014[7]). On aggregate, these trade benefits can translate into higher productivity, more employment and higher wages, all of which are associated with economic growth and higher living standards (see (Alcalá and Ciccone, 2004[8])and (Sing, 2010[9])).
The role of ICT goods and services in this context is likely to be two-fold. First, a wider use of ICT goods and services should enable firms to increase productivity (Cardona, Kretschmer and Strobel, 2013[10]), by, for example, helping streamline processes or enabling access to, and use of, new digital technologies (e.g. productivity enhancing software and hardware). This will improve firms’ ability to meet the fixed cost of engaging in export markets. At the same time, use of ICT goods and services can also affect trade costs (WTO, 2018[5]),, reducing the cost of gathering information about markets and enabling faster communication with suppliers and consumers, in turn, affecting both fixed and variable trade costs. Therefore, much like access to foreign intermediate inputs is associated with a positive effect on export decisions (Bas, 2012[11]), access to imported ICT goods and services, which embody a range of new digital technologies, could be expected to positively affect exports, including exports of goods, services and bundles of goods and service.
However, despite a wide acknowledgement of the potential contribution that ICT goods and services may play on firm ability to trade, there remains, to date, very little empirical evidence mapping the links between access to ICT goods and services from abroad and the trading behaviour of firms.
This section sets out to identify whether access to imported ICT goods and services, hereinafter ICT inputs, can enable greater export sales for Brazilian firms (noting a number of caveats discussed in Box 5.2).
Before delving into the analysis, there are several caveats to note. While the micro-data used for this analysis is very rich in terms of the trading activities of Brazilian firms, it is not possible to link this information to the characteristics of the firms themselves. This prevents the calculation of firm-level productivity measures (e.g. total factor productivity). It also complicates the use of firm specific controls in the econometric analysis, reducing the ability to identify what specific characteristics might be driving change. Moreover, while the data identifies when a firm imports ICT goods and services, it does not capture the use of domestically produced or procured ICT goods and services. It therefore misses an important channel for ICT adoption related to use of domestic inputs.
Finally, the analysis is unable to control for issues at the intersection of adoption and use. That is, existing literature suggests that data-driven innovation requires both access to new technologies but also the managerial capacity and know-how to drive innovation (Brynjolfsson and McElheran, 2016[12]). Since there is little visibility on the characteristics of firms, it will be difficult to control for such issues.
However, the data does provide a rich source of information, including on access to foreign ICT goods and services helping provide important insights for policy analysis.
5.2.1. Firms that import ICT goods and services tend to export more
Preliminary evidence suggests that Brazilian firms that import ICT goods and digitally deliverable services tend to export more than those that do not, especially in the case of the latter (Figure 5.5). However, it is possible that there are factors which make firms more prone to exporting that also make them more prone to importing these such as technical capacity. A more formal analysis is needed to better identify the links between imports of ICT inputs and export competitiveness.
5.2.2. Imported ICT inputs are especially important for goods exports and for micro-sized firms
A more formal, econometric, analysis is undertaken to better identify the links between imports of ICT inputs and export competitiveness. However, this is not without complications (Box 5.2). In order to enable identification, the analysis exploits variance across firms exporting the same product to the same market, comparing whether those that import ICT inputs have witnessed higher sales. This is done through a gravity-type estimation using firm level data (Box 5.3).
The results show that imported ICT inputs are especially important for firms that export goods, a category of trade that is one of the most important in Brazil (Table 5.3). Indeed, goods exporters that import tend to export more, but an additional boost is given to their competitiveness when they import ICT inputs.
The impact of importing ICT goods and services to increase export performance is identified using a gravity-type setup. Exports of firm i at time t in product p to partner c are a function of four types of variables.
1 Firm characteristics: comprising sector or activity, size and trading status (whether a firm trades goods, services).
2 Import use, including ICT: comprising whether or not firms import any product and if they import ICT goods, services or both.
3 Policy variables: including measures of absolute services restrictions as well as measures of regulatory heterogeneity derived from the OECD’s Services Trade Restrictiveness Index database as well as participation in FTAs.
4 Other trade costs: which include variables such as distance, common language, the presence of common borders and other such.
So that the following baseline specification is estimated using a PPML approach with high dimensional fixed effects:
Four broad categories of Fixed effects (FE) are used in the identification strategy.
The first are year-product-market. The idea is to compare firms that sell the same product to the same market but which are different in their use of ICT.
The second are year-sector-market. This is a similar specification capturing between effects but at the sector rather than the product level. One issue is that differences may be drive by differences in the products traded.
The third looks at within variation through firm-product-partner and year FE. This aims to identify whether firms that have started importing ICT goods/services have witnessed bigger changes in exports than those that have not. To a certain extent, this is a somewhat restrictive model in that it requires for there to be a change in the ICT USE status to identify an impact.
The last uses firm, sector, partner and year fixed effects separately.
The main variable of interest is the coefficient on the ICT variables across different categories of products. A positive impact is expected to show that firms that import ICT goods and/or services, can increase their export competitiveness more than those that do not.
Moreover, the analysis also reveals that imported ICT inputs are especially important for micro-sized firms across services (Table 5.4). That is, micro-sized firms that import ICT inputs witness larger increases in exports than those that do not.
Overall, the analysis highlights that access to ICT goods and services via imports can be an important determinant of export competitiveness of Brazilian goods exporters, especially in the case of micro-sized firms.
While services represent nearly three-quarters of Brazilian GDP, they account for only 12% of overall exports in 2019. This is a recurring pattern across many countries and is due to several factors. The first is that many services have to be provided in person, making these difficult to trade across borders. The second is that there is wide regulatory heterogeneity for services across countries (OECD, 2017[13]) which can make it difficult for firms to operate across markets. The third, are structural factors such as language, culture or differences in time-zones. While the digital transformation has enabled more trade in services, including those that were previously non-tradeable (López González and Jouanjean, 2017[14]), it has also resulted in growing digital trade restrictiveness (Ferencz, 2019[15]).
This section relies on micro-data to identify how services exports, including those that are digitally deliverable, are affected by regulatory measures in destination countries. The aim is to help policy makers get a clear view of the issues that affect Brazilian exporters’ capacity to thrive in the digital environment. The first section sets the scene, providing context to the empirical strategy and describing some overarching trends. The second section provides the main results of the empirical analysis and the third section concludes with some policy observations.
5.3.1. Do barriers to digitally enabled services affect Brazilian firms ability to engage in services exports?
Services trade costs can be large, affecting the ability of countries to sell in foreign markets and reducing access to important services inputs via imports (OECD, 2017[16])). Indeed, according to Benz (2017[17]) the tariff equivalent of services trade restrictions are, on average, between 20% and 300% and can be as high as 2000% for specific sectors. These are in-line with much of the existing empirical literature which also highlights a wide degree of heterogeneity across different sectors.
To make the most out of the evolving digital trade environment, it is important to understand how different regulatory obstacles affect firms’ ability to access markets. This section looks at different facets of this question, looking first at aggregate impacts of different barriers on digitally deliverable and non-digitally deliverable services and then at how these play out across firms of different size (Box 5.4).
The impact of services trade restrictions on Brazil’s exports are identified using a gravity-type setup. Exports of services of firm i at time t in sector s to partner c are a function of three types of variables.
1 Firm characteristics: comprising sector or activity, size and trading status (whether a firm trades goods, services).
2 Policy variables: including measures of absolute services restrictions as well as measures of regulatory heterogeneity derived from the OECD’s Services Trade Restrictiveness Index database as well as participation in FTAs.
3 Other trade costs: which include variables such as distance, common language, the presence of common borders and other such.
So that the following baseline specification is estimated using a PPML approach with high dimensional fixed effects:
Four broad categories of fixed effects (FE) are used to control for unobserved but likely important factors:
The first are those that control for firm-specific characteristics (such as productivity) but which allow variation across markets (year-firm-product and sector fixed effects). This will capture how a particular firm selling across different markets is affected by different DSTRI measures. A downside to this identification strategy is that, while it enables controlling for firm specific effects, it requires that a firm sells across different markets. This means that the sample of firms is restricted to those trading the same products in more than one market.
The second are those that control for year-firm-sector and product fixed effects. The model is like the one above but less restrictive in that it compares the same firm selling different products in the same sector in different countries. Same issue applies in that to identify an impact firms have to sell products in at least two markets.
The third relates to within variation using firm-product-partner and year FE. This will identify how changes in trade of a particular firm in a particular product and marker are affected by changes in the STRI scores of that market. Since it captures within changes, this does not require firms to export in various markets but issues might arise due to low variance in STRI scores in time.
The last also exploits the within variation but this time in the context of firm-sector-partner and year FE. This means it is less restrictive in that it can compare firms that sell different products in the same sector.
The key variable of interest in these specifications will be the STRI variable which captures three main elements. The first is the overall impact of services trade restrictions, the average STRI. The second, the impact of the Digital STRI. The third the disaggregated measures of the DSTRI capturing specific aspects of the digital trade environment (such as infrastructure, payments and other). Interest is also in whether some sectors stand out in terms of negative impacts and if it is possible to identify how digital restrictions affect digitally deliverable services exports, with breakdown according to whether larger firms are more affected than smaller ones.
The analysis is undertaken using detailed firm level data under a gravity-type setup. It exploits variation across countries controlling for year-firm-sector and product specific characteristics. The analysis shows that regulatory obstacles related to digital trade have a significant negative impact on Brazilian firms ability to export services (Table 5.5). The impact is particularly high on digitally deliverables services exports, an area where Brazil has been developing strong comparative advantage in the region.5
5.3.2. What type of barriers matter for Brazilian exporters?
The type of restrictions faced by Brazilian firms matter (Table 5.6). Although difficult to compare across estimations, barriers related to payment systems, intellectual property rights, other barriers and infrastructure and connectivity all constitute important impediments for Brazilian exporters of digitally deliverable services.6
5.3.3. Do regulatory restrictions impact Brazilian services exporters of different sizes in the same way?
Not all Brazilian firms will face the same restrictions in the same manner. Where digitally deliverable services are concerned, restrictions are important across firms of all sizes but they are most trade reducing for smaller firms, especially those between 10 and 50 employees (Table 5.7). In terms of non-digitally deliverable services, restrictions, as identified by the Digital Services Trade Restrictiveness Index, appear to only affect small and micro firm exports (Table 5.8).
References
[8] Alcalá, F. and A. Ciccone (2004), “Trade and Productivity”, The Quarterly Journal of Economics, Vol. 119/2, pp. 613–646, https://doi.org/10.1162/0033553041382139.
[18] Ariu, A. et al. (2019), “The interconnections between services and goods trade at the firm-level”, Journal of International Economics, Vol. 116(C), pp. 173-188.
[4] Baldwin, R. (2016), The Great Convergence: Information technology and the New Globalization, Harvard University Press, Cambridge.
[11] Bas, M. (2012), “Input-trade Liberalization and Firm Export Decisions: Evidence from Argentina”, Journal of Development Economics, Vol. 97(2)/March, pp. 481-493.
[7] Bas, M. and V. Strauss-Kahn (2014), “Does importing more inputs raise exports? Firm-level evidence from France”, Review of World Economics (Weltwirtschaftliches Archiv),, Vol. 150(2)/May, pp. 241-275.
[17] Benz, S. (2017), “Services trade costs: Tariff equivalents of services trade restrictions using gravity estimation”, OECD Trade Policy Papers 200, https://doi.org/10.1787/dc607ce6-en.
[1] Brenton, P. and R. Newfarmer (2007), “Watching More than the Discovery Channel: Export Cycles and Diversification in Development”, World Bank Policy Reseach Working Paper, Washington D.C.: The World Bank.
[12] Brynjolfsson, E. and K. McElheran (2016), “The Rapid Adoption of Data-Driven Decision-Making”, American Economic Review, Vol. 106/6, pp. 133-139.
[10] Cardona, M., T. Kretschmer and T. Strobel (2013), “ICT and productivity:conclusions from the empirical literature”, Inform.Econ. Pol., Vol. 5, pp. 109–125, https://doi.org/10.1016/j.infoecopol.2012.12.002.
[15] Ferencz, J. (2019), The OECD Digital Services Trade Restrictiveness Index, https://doi.org/10.1787/16ed2d78-en.
[14] López González, J. and M. Jouanjean (2017), “Digital Trade: Developing a Framework for Analysis”, OECD Trade Policy Papers, No. 205, OECD Publishing, Paris, https://dx.doi.org/10.1787/524c8c83-en.
[6] Melitz, M. (2003), “The impact of trade on intra-industry reallocations and aggregate industry productivity”, Econometrica, Vol. 71, pp. 1695-1725.
[2] Melitz, M. and S. Redding (2014), “Heterogeneous Firms and Trade”, Handbook of International Economics, Vol. 4, pp. 1-54, https://www.sciencedirect.com/handbook/handbook-of-international-economics/vol/4/suppl/C.
[16] OECD (2017), Services Trade Policies and the Global Economy, OECD Publishing, Paris, https://doi.org/10.1787/9789264275232-en.
[13] OECD (2017), Services Trade Policies and the Global Economy, OECD Publishing, Paris,, https://doi.org/10.1787/9789264275232-en.
[9] Sing, T. (2010), “Does International trade cause economic growth? A survey”. The World Economy, Volume 33, Issue 11, November 2010 pp:1517-1564”, Vol. 33/11, pp. 1517-1564.
[3] World Bank (2016), World Development Report 2016: Digital Dividends, The World Bank, Washington DC.
[5] WTO (2018), The future of world trade: How digital technologies are transforming global commerce, WTO publishing.
Notes
← 1. This is likely to include what might be considered as ‘indirect trade’ where goods and services produced domestically or abroad are sold by retail companies.
← 2. Single traders refers to firms that only engage in either exports of goods or exports of services or imports of goods or imports of services. Simultaneous exporters are those that engage in more than one channel, including exporters of goods AND exporters of services or other combinations of goods and services imports or exports.
← 3. These number are comparable to the emerging literature that looks at simultaneous imports and exports of goods and services (Ariu et al., 2019[18]).
← 4. See, for example, https://data.oecd.org/trade/exports-by-business-size.htm.
← 5. These results are robust to different specification including different sets of fixed effects, from year-firm-product and sector-fixed effects to individual year, product, firm, sector-fixed effects.
← 6. Unlike in other estimations, it is difficult to compare the DSTRI scores across different categories because each has to be evaluated against a different mean. For instance, while the coefficient is smaller on the infrastructure and connectivity estimations, this variable has a higher mean than the others.