The impact of the China shock on the manufacturing labor market in Brazil

Paz, Lourenço Senne Working Paper The impact of the China shock on the manufacturing labor market in Brazil IDB Working Paper Series, No. IDB-WP-01085 Provided in Cooperation with: Inter-American Development Bank (IDB), Washington, DC Suggested Citation: Paz, Lourenço Senne (2019) : The impact of the China shock on the manufacturing labor market in Brazil, IDB Working Paper Series, No. IDB-WP-01085, Inter-American Development Bank (IDB), Washington, DC, https://doi.org/10.18235/0002047 This Version is available at: https://hdl.handle.net/10419/234674 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/3.0/igo/legalcode The Impact of the China Shock on the Manufacturing Labor Market in Brazil Lourenço S. Paz IDB WORKING PAPER SERIES Nº IDB-WP-01085 Inter-American Development Bank Integration and Trade Sector December 2019 The Impact of the China Shock on the Manufacturing Labor Market in Brazil Lourenço S. Paz Inter-American Development Bank Integration and Trade Sector December 2019 Cataloging-in-Publication data provided by the Inter-American Development Bank Felipe Herrera Library Paz, Lourenço. The impact of the China shock on the manufacturing labor market in Brazil / Lourenço S. Paz. p. cm. — (IDB Working Paper Series ; 1085) Includes bibliographic references. 1. Imports-Brazil. 2. Manufacturing industries-Brazil. 3. Labor market-Brazil. 4. Informal sector (Economics)-Brazil. 5. Wages-Brazil. 6. Industrial productivity-Brazil. 7. Brazil- Commerce-China. 8. China-Commerce-Brazil. I. Inter-American Development Bank. Integration and Trade Sector. II. Title. III. Series. IDB-WP-1085 JEL Codes: F1, O1 Key words: Brazil, China, employment, import penetration, informality, wages http://www.iadb.org Copyright © 2019 Inter-American Development Bank. This work is licensed under a Creative Commons IGO 3.0 Attribution- NonCommercial -NoDerivatives (CC-IGO BY-NC-ND 3.0 IGO) license (http://creativecommons.org/licenses/by-nc- nd/3.0/igo/legalcode ) and may be reproduced with attribution to the IDB and for any non-commercial purpose, as provided below. No derivative work is allowed. Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration purs uant to the UNCITRAL rules. The use of the IDB’s name for any purpose other than for attribution, and the use of IDB’s logo shall be subject to a separate written license agreement between the IDB and the user and is not authorized as part of this CC -IGO license. Following a peer review process, and with previous written consent by the Inter -American Development Bank (IDB), a revised version of this work may also be reproduced in any academic journal, including those indexed by the American Economic Association’s EconLit, provided that the IDB is credited and that the author(s) receive no income from th e publication. Therefore, the restriction to receive income from such publication shall only extend to the publication’s author(s). With regard to such restriction, in case of any inconsistency between the Creative Commons IGO 3.0 Attribution -NonCommercial- NoDerivatives license and these statements, the latter shall prevail. Note that link provided above includes additional terms and conditions of the license. The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors, or the countries they represent. 1 The Impact of the China Shock on the Manufacturing Labor Market in Brazil Lourenço S. Paz* Baylor University Abstract The vigorous growth of the Chinese economy together with its growing role in international trade has raised fears of deindustrialization among developing countries. This study draws on the large increase in the international trade exposure of the Brazilian economy from 2000 to 2012 to assess the impacts of trade on its manufacturing sector. In this period, import penetration increased by 25%, and at the same time, China’s share of import penetration increased from 3% to 20%. Using household survey data that encompasses both formal and informal workers, this paper’s estimates indicate that higher import penetration reduces the employment level, the share of employment in the population, the hourly wage, the interindustry wage premium, and the share of informal employment. The industry-level results indicate that a rise in import penetration from either China or the rest of the world (ROW) reduced the employment level, hourly wage, and share of informal employment while increasing the interindustry wage premium. The worker-level results suggest that industry-level import penetration from China and the ROW raised workers’ wages and reduced the likelihood of their being informally employed. The state-level estimates imply that Chinese and ROW imports per worker initially reduced the employment level, the hourly wage, and share of informal employment, but these effects were reversed after 2008. Chinese imports per firm had a negative impact on the share of informal employment and a positive one on average years of schooling. Before 2008, Chinese imports per firm increased the share of workers with both high-school and college educations, and the net impact on both shares became negative after 2008. Estimates using actual imports per worker and per firm did not impact state-level labor-market outcomes. Finally, these effects were modulated by the labor intensity of the industry, the state-level initial share of manufacturing in the gross domestic product, the availability of a sea harbor in the state, and the implementation of the Nova Matriz Econômica policies in 2008. Keywords: Brazil, China, employment, import penetration, informality, wages JEL codes: F1, O1 * E-mail: lou[email protected]. I would like to thank Bruno C. Araújo, Joe McKinney, Danielken Molina, Maurício Mesquita Moreira, Nina Pavcnik, Peri da Silva, Christian Volpe Martincus, and Jim West for their useful suggestions. 2 1. INTRODUCTION Over the last three decades, China has experienced an impressive economic transformation involving rapid economic growth and increasing participation in international trade. In 2000, China accounted for 3.35% of world imports and 4% of world exports. By 2012, these figures stood at 9.77% of world imports and 11.42% of world exports. Given that world trade flows grew by 75% in this period, China achieved an unprecedented expansion in both its trade flows and participation in world trade, which has become known as the “China shock.” This rapid ascension of China as a major manufacturing powerhouse—about 90% of its exports are made up of manufactured goods—raised fears of deindustrialization in developing countries, especially those in Latin America. Such concerns are grounded in the fact that China is a populous country and that a substantial amount of its labor force is still employed in agriculture. This huge labor endowment makes China labor-abundant relative to other countries in the developing world. Moreover, the size of its economy leads to economies of scale that are important in several manufacturing industries. On top of this, Moreira (2006) points out that although in the early 2000s China presented lower productivity levels than some Latin American countries, its wages were more than proportionately lower than those in Latin American countries. All these features give China a strong competitive edge in world markets. This naturally raises the question of whether Latin American economies and, more specifically, their manufacturing sectors have been impacted by the China shock. In this vein, the case of Brazil is emblematic. Besides being the most populous country in Latin America, it has the largest economy and a sizable manufacturing industry. In 2000–2012, as illustrated by figure 1, Brazil experienced a 25% increase in its manufacturing import penetration and, at the same time, a sixfold increase in China’s share of such imports, which went from 3% to 20%. This made China the largest exporter to Brazil with a 20.4% share of the total Brazilian imports. Interestingly, Facchini et al. (2010) point out that Chinese manufacturing goods seem to be close substitutes for those produced in Brazil. Furthermore, China became the second-largest destination for Brazilian exports, with a share of 14.6%. These trade flows are uneven in terms of their contents, though. Manufactured goods constitute less than a third of Brazil’s exports to China, but more than 90% of China’s exports to Brazil. Additionally, China’s trade expansion may also have affected Brazil on the export side of the economy. Figure 2 shows that China gained substantial market share in foreign markets already served by Brazilian exporters. In fact, Brazil’s share in world trade increased by roughly 50%. This poor performance also extended to the manufacturing share in Brazil’s GDP, which declined from 18% to 13%. These observations suggest that the China shock is a good candidate for explaining the weak performance of the manufacturing sector, which suggests a deindustrialization of the Brazilian economy. This is an important question because many observers point out that the manufacturing sector is the driving force of economic growth and development and also typically pays higher wages than jobs in agriculture or services. Given the importance of manufacturing, there is a surprising dearth of research on the impacts of the China shock on developing countries, especially those in Latin America. 3 FIGURE 1. MANUFACTURING IMPORT PENETRATION IN BRAZIL AND CHINA’S SHARE IN IMPORT PENETRATION PANEL A. MANUFACTURING IMPORT PENETRATION AND CHINA’S SHARE IN IMPORT PENETRATION PANEL B. THE FOUR MEASURES OF IMPORT PENETRATION 0510 15 20 2000 2004 2008 2012 Year Import penetration(%) Chinese share(%) 10 15 20 25 30 2000 2005 2010 2015 year Import penetration Alt. import penetration Adj. import penetration Alt. adj. import penetration Import Penetration (%) 4 FIGURE 2. SIMPLE AVERAGE ACROSS INDUSTRIES OF THE RATIO BETWEEN CHINESE AND BRAZILIAN MARKET SHARE IN FOREIGN MARKETS This study represents a step toward filling this gap by studying how the China shock affected Brazil’s manufacturing labor market in 2000–2012. More precisely, a rigorous empirical analysis is conducted to examine how the level, skill, informality, sector, and regional composition of manufacturing employment in Brazil were affected by the changes in the industry-level import tariffs, import penetration, and the Chinese market share in foreign markets served by Brazilian firms. The empirical exercise utilizes data from the Brazilian Household Survey (Pesquisa Nacional por Amostra de Domicilios, PNAD) and from the Brazilian demographic census. This pooled cross-sectional household-level data contains detailed demographic and employment information. Most importantly, it encompasses both formal and informal workers. This is a major advantage of this data because the share of informal workers is larger than 20% in manufacturing as a whole, and approximately a third of workers are informal in industries like furniture and other products. This paper’s estimates at the industry level indicate that a rise in import penetration from either China or the ROW reduces the employment level. Chinese imports had a stronger effect on labor-intensive industries. Interestingly, Chinese imports in upstream industries increase the employment level in downstream industries. Import penetration from both the ROW and China reduced the industry-level hourly wage and share of informal employment, while they increased the interindustry wage premium. The implementation of the Nova Matriz Econômica (hereafter NME) policies in 2008 dampened these effects. The worker-level results suggest that industry-level import penetration from both China and ROW raised workers’ wages, with a greater effect for ROW import penetration in coastal states and a larger effect for Chinese import penetration on labor-intensive industries. The implementation of the NME magnified the effects of Chinese import penetration and dampened that of ROW import penetration. Industry-level Chinese and ROW import penetration reduced the likelihood of informal employment. These effects are smaller in states with a large manufacturing sector, and Chinese import penetration showed a larger effect on labor-intensive industries. However, the use of state-by- industry trade exposure measures led to some different results. In fact, the ROW import penetration effect switched 5 from being negative to positive after the implementation of the NME, and Chinese import penetration showed the exact opposite behavior. Analogously to the informal employment indicator, the sign for the measure for ROW import penetration also went from negative to positive in response to the implementation of the NME, while Chinese import penetration again showed the exact opposite pattern. The last set of estimates were obtained using state-level identification following Autor et al.’s (2013) methodology. These results imply that before the NME was put into effect, both the Chinese and ROW imports per worker reduced the employment level but that these effects became positive after 2008. In the pre-NME period, the net effect of ROW and Chinese imports per worker decreased the log of the hourly wage and the share of informal employment, but both effects became positive after the NME. Moving to trade exposure measured as imports per firms, this study finds that Chinese imports per firm had a negative impact on the share of informal employment and a positive one on average years of schooling. Before the implementation of the NME, Chinese imports per firm increased the share of workers with both high-school and college educations, but the net impact on both shares became negative after the NME policies went into effect. The effects of ROW imports per firm on these two outcomes the exact opposite pattern. Finally, the estimates using actual imports per worker and per firm did not impact state-level labor-market outcomes. At the end of the day, this paper provides new evidence that the China shock affected manufacturing in Brazil in a nontrivial way, making it difficult to assert that the shrinkage in the participation of manufacturing in Brazil’s GDP is the result of the China shock. Most importantly, there is evidence suggesting that the effect of Chinese imports is different to those of imports from other countries. These impacts also differed depending on the characteristics of both the industry and the state. Finally, the new policies introduced by the implementation of the NME policies substantially altered the effects of the changes in trade exposure measures on labor-market outcomes, substantially, in some cases even reversing these. These findings are a significant contribution to policymakers facing the challenge of addressing these new, potentially harmful effects of the China shock. The remainder of this paper is organized as follows. The next section provides an overview of trade-related policies in Brazil since the 1990s, introduces a theoretical framework to guide the empirical exercise, describes this project’s dataset, and presents some raw data patterns to illustrate the effects of increased trade exposure on labor-market outcomes. The empirical methodology developed to assess causal effects of trade on manufacturing labor markets is laid out in section 3. Section 4 reports the estimates and discusses the results. Finally, conclusions are drawn in section 5. 2. POLICY BACKGROUND, THEORETICAL FRAMEWORK, AND RAW DATA DESCRIPTION AND PATTERNS In this section, I provide a brief overview of the changes in trade-related policies that took place in Brazil from the 1990s onward. Next, I introduce a theoretical framework to facilitate the understanding of the possible labor market ramifications of the changes in Brazilian trade policies. This is followed by an explanation of the source of each component and the assembly procedure used to prepare the dataset employed in the empirical exercises of this project. Finally, I present descriptive statistics about the evolution of the trade exposure of the Brazilian economy and its labormarket outcomes during the 2000 to 2012 period. A. Policy background In the 1970s and 1980s, the Brazilian economy had a very small degree of openness that was a consequence of very high import tariffs coupled with several binding nontariff barriers (NTBs). At the end of the 1980s, this highly protective trade policy started to change. In 1988, the Brazilian government unilaterally decided to change its trade policy by cutting import tariffs to reduce the level of redundant protection. This means that tariffs were reduced to a level close to the domestic–international price differential. Despite these lowered tariffs, NTBs were not eliminated (cf. Kume et al., 2003), and this lowered protection level was still high enough to severely curb imports. 12 other industries. 13 By the same token, increased trade exposure on downstream industries may impact the demand faced by Brazilian manufacturers of intermediate inputs. This type of trade exposure can be calculated according to the equation below 𝐼𝑃𝑗𝑡 𝐷𝑜𝑤𝑛𝑠𝑡𝑟𝑒𝑎𝑚 = ∑ 𝑣𝑟𝑗 ∑𝑣𝑟𝑗𝑟 𝑟𝐼𝑃𝑟𝑡 where 𝐼𝑃𝑗𝑡 𝐷𝑜𝑤𝑛𝑠𝑡𝑟𝑒𝑎𝑚 is the import penetration in year t of the industries that are downstream of industry j, 𝑣𝑟𝑗 is the cost share of inputs from industry j in the output of industry r. Note that again for r=j, 𝑣𝑟𝑗 ≡0. The descriptive statistics for the labor-market outcomes are reported in table 4. The manufacturing industry’s share of employment only fell in the textiles and wood products industries. 14 This reduction is more than offset by the expansion of employment in the remaining industries, especially in food and beverages. Changes in the share of employment in total manufacturing employment indicate labor reallocation among manufacturing industries. Of the 26 industries, 13 natural-resource-intensive and capital-intensive industries—such as biofuels, food and beverages, appliances, automobiles, trucks, and buses—exhibited an increase in their employment share within manufacturing. The remaining industries, which are mostly labor-intensive, displayed a decline in the share in employment. This decline was considerable for apparel, textiles, wood products, footwear, and furniture. The natural logarithm of the real hourly wage increased in seven industries only and diminished substantially in 13 industries. 15 The interindustry wage premium represents the premium attributed to the worker’s industry affiliation as a percentage deviation from the average manufacturing wage. 16 We can see in table 4 that the wage premium variation is positively correlated to that of the average log of the hourly wage. The log wage variation thus cannot be entirely attributed to changes in workers’ skill levels. Table 5 displays the average worker’s characteristics at the industry level for 2000 and 2012. In this study, an informal job is defined as an employment relationship in which the employer does not comply with social security contributions, as in Paz (2014a). This question is included in both the censuses and the PNAD surveys. The share of workers with informal jobs increased in nine industries, and this growth was robust in traditionally labor-intensive industries, such as apparel, wood products, and furniture. There was a sharp increase in the average years of schooling and the share of workers with a high-school diploma in all industries. The share of workers with a college degree experienced more modest growth. Part of this skill upgrade may be due to an expansion in the supply of workers with high-school and college educations in the 1990s and in the 2000s. Nonetheless, the growth in the shares of workers with both high-school and college educations was heterogeneous across industries, which indicate that this supply increase is not the sole driver of this observed skill upgrade. In fact, in some cases, the share of workers with college educations actually decreased in industries like auto parts, automobiles, trucks, and buses. In the next subsection, the analysis shifts to the regional level. ii. State-level data Brazil is a large country that has 26 states and a federal district, where its capital is located. Its manufacturing activity exhibits significant spatial heterogeneity. About 80% of the manufacturing output comes from just seven states, which are in the southeastern and southern regions of the country. Moreover, the state-level descriptive statistics reported in table 6 reveal that approximately 31% of the manufacturing firms are in São Paulo state, followed by Minas Gerais and Rio Grande do Sul with 13% each. At the other end of the spectrum, northern states like Acre and Roraima 13 The cost share of inputs come from the IBGE’s 2000 input-output matrix for Brazil, which provides data at the Level 56 classification level. 14 A better measure would be the employment share of the economically active population. Unfortunately, no annual, nationwide data is available on Brazil’s economically active or working-age population. 15 The hourly wage consists of the monthly wage divided by 4.3 times the number of hours worked in a week, adjusted for inflation as in Corseuil and Foguel (2002). 16 The wage premium is estimated as follows. For every year of the sample, hourly wages are regressed on educational and demographic controls, industry fixed effects, and state fixed effects. The state fixed effects are included to account for differences in labor regulation enforcement and to account for the state-specific minimum wages that were implemented in 2002. Once the Haisken-Denew transformation is applied to the estimated industry effects, they represent the wage premium as a percentage deviation from the wage of the base industry. This makes the estimated wage premia comparable over time. The sum of the wage premium is zero every year. 13 accounted for just 0.14% and 0.09% of firms, respectively. The manufacturing industry composition at the state level also varies considerably. Table 6 reports the Herfindahl index for the industry-level share of firms for each state. Manufacturing in states like Rio Grande do Sul and São Paulo is diversified, as is reflected by the low Herfindahl index of 0.09, whereas manufacturing in Pará and Alagoas is much less diversified, with Herfindahl indices of 0.17 and 0.22, respectively. In terms of employment, the state-level share of employment in the manufacturing industry increased in almost every state. This increase was more prominent (in excess of 20%) in the southern and southeastern states. The last two columns of table 6 illustrate the evolution of the state employment share in overall manufacturing employment, where we can see that the relative share of states like São Paulo, Rio de Janeiro, Rio Grande do Sul, and Amazonas increased, while other states’ shares declined. This means that manufacturing employment became spatially more concentrated. 17 According to Autor, Dorn, and Hanson (2013)—hereafter, ADH—the trade exposure of local labor markets (Brazilian states) is measured by imports per worker, which can be calculated as the weighted average of imports experienced by the manufacturing industries in that state, using employment as weights, as shown below: ∆𝐼𝑃𝑊𝑠𝑡 =∑ 𝐿𝑠𝑗𝑡 𝐿𝑗𝑡 ∆𝑀𝑐𝑗𝑡 𝐿𝑠𝑡 𝑗 where ∆𝐼𝑃𝑊𝑠𝑡 is the change in the imports per worker in state s between year t and t-1, 𝐿𝑠𝑗𝑡 is the number of workers employed in industry j in state s in year t, and ∆𝑀𝑐𝑗𝑡 is the change in industry j imports from country c between year t and t-1. A similar measure of trade exposure can be computed using the number of firms—that is, the local level of imports per firm. ∆𝐼𝑃𝐹𝑠𝑡 =∑ 𝐹𝑠𝑗𝑡 𝐹𝑗𝑡 ∆𝑀𝑐𝑗𝑡 𝐹𝑠𝑡 𝑗 where ∆𝐼𝑃𝐹𝑠𝑡 is the change in the imports per firm in state s between year t and t-1, and 𝐹𝑠𝑗𝑡 is the number of firms employed in industry j in state s in year t. Figure 5 and 6 show the changes in the Chinese and ROW imports per worker for the Brazilian states, respectively. The top map in both figures represents the changes between 2006 and 2000, whereas the bottom map illustrates the change between 2012 and 2006. We can see in both top maps that between 2006 and 2000 imports per worker grew considerably in almost all states. The largest increases took place in the southern or southeastern states, which account for most of Brazil’s manufacturing. Surprisingly, these maps also show a very large increase in imports per worker in states with dismal levels of manufacturing, like Maranhão. An additional pattern that emerges from these two maps in the top is the similarity in the change in imports per worker from China and the ROW. 17 An alternative driving force for the state-level share of employment are state-specific demographic trends, which changes the unobservable economically active population. Unfortunately, no data is available to evaluate this possible explanation. 14 FIGURE 5. STATE-LEVEL IMPORTS FROM CHINA PER WORKER Turning to the imports-per-firm trade measures, figures 7 and 8 show the changes in imports per firm for China and for the ROW, respectively. In contrast to the imports-per-worker maps, we can see that the imports from China did not significantly impact firms located in the southeastern states. As before, both figures also show large changes in imports per firm in states with limited manufacturing activity, like Rondônia and Maranhão, for instance. 15 FIGURE 6. STATE-LEVEL IMPORTS FROM ROW PER WORKER 16 FIGURE 7. STATE-LEVEL IMPORTS FROM CHINA PER FIRM 17 FIGURE 8. STATE-LEVEL IMPORTS FROM ROW PER FIRM The international trade data extracted from the Aliceweb system contains import volumes according to the destination state of imports. I used this data to calculate imports per worker (hereafter, actual imports per worker) and imports per firm (hereafter, actual imports per firm) at the state level. Figures 9 and 10 show the changes in actual imports per worker and in actual imports per firm, respectively, between 2012 and 2000. The top map is for imports 18 from China while the bottom map for imports from the ROW. The pattern that emerges from the maps in both figures is that the states with little manufacturing activity are the ones exhibiting the largest increase in actual imports per worker and actual imports per firm. This unexpected result is in stark contrast with the maps depicting the ADH-style trade exposure measures. This casts some doubts on the representativeness of the ADH-type measures of trade exposure at the state level. I now turn to describe the empirical methods used to evaluate the testable hypothesis outlined earlier and to distinguish the effects of each trade exposure measure on labor-market outcomes. 19 FIGURE 9. STATE-LEVEL ACTUAL IMPORTS PER WORKER PANEL A. IMPORTS FROM CHINA PANEL B. IMPORTS FROM ROW Note: No data means that the level of imports in 2000 is missing. 20 FIGURE 10. STATE-LEVEL ACTUAL IMPORTS PER FIRM PANEL A. IMPORTS FROM CHINA PANEL B. IMPORTS FROM ROW Note: No data means that the initial level of imports per firm (year 2000) is missing 3. EMPIRICAL METHODOLOGY The empirical methodology developed in this section exploits the industry-level, the state level, and the stateindustry-level variation present in the data in order to estimate the impacts of the trade exposure measures on the 21 following industry-level outcomes (𝑦𝑗𝑡): the natural logarithm of the employment level, the share of the population in manufacturing industry employment, the share of manufacturing employment, the average log of the real hourly wage, the interindustry wage premium, the share of informal employment, workers’ average years of schooling, the share of workers with a high-school diploma, and the share of workers with a college degree. A. Industry- and worker-level specifications Starting with the industry-level variation, the first econometric specification is used to investigate whether Chinese import penetration (𝐼𝑃𝑗,𝑡−1 𝐶ℎ𝑖𝑛𝑎) has an effect that differs from that of ROW import penetration (𝐼𝑃𝑗,𝑡−1 𝑅𝑂𝑊), as depicted by equation (2). 𝑦𝑗𝑡 =𝛼+𝛽1𝐼𝑃𝑗,𝑡−1 𝐶ℎ𝑖𝑛𝑎 +𝛽2𝐼𝑃𝑗,𝑡−1 𝑅𝑂𝑊 +𝛾𝑗+𝛿𝑡+𝑢𝑗𝑡 (2) where 𝛾𝑗 and 𝛿𝑡 are industry and year effects respectively, and 𝑢𝑗𝑡 is the error term. Note that equation (2) can also incorporate additional control variables such as China’s market share in the export destinations served by Brazilian firms, the effective applied import tariffs, and upstream Chinese and ROW import penetrations. 18 Hypotheses 1 and 2 indicate that labor-intensive industries are more likely to be affected by imports from China, and to assess that I estimate equation (3), where 𝐿𝐼𝑗 is an indicator variable that is “1” for labor-intensive industries (food and beverages, textiles, apparel, footwear and leather products, wood products, nonmetallic minerals and products, and furniture and other products), and “0” otherwise. 𝑦𝑗𝑡 =𝛼+𝛽1𝐼𝑃𝑗,𝑡−1 𝐶ℎ𝑖𝑛𝑎 +𝛽2𝐼𝑃𝑗,𝑡−1 𝑅𝑂𝑊 +𝛽3𝐼𝑃𝑗,𝑡−1 𝐶ℎ𝑖𝑛𝑎 ×𝐿𝐼𝑗+𝛽4𝐼𝑃𝑗,𝑡−1 𝑅𝑂𝑊 ×𝐿𝐼𝑗+𝛾𝑗+𝛿𝑡+𝑢𝑗𝑡 (3) Hypothesis 1 implies 𝛽3< 0 and 𝛽4= 0 whenever the dependent variables are the share of the population in manufacturing industry employment and the share of manufacturing employment. Hypothesis 2 implies 𝛽3<0 and 𝛽4= 0 if the dependent variables are average log of the real hourly wage and the share of workers with a college degree. The next step is to evaluate how the implementation of the NME policy altered the response of the labor-market outcomes to changes in trade environment variables. This assessment is implemented by augmenting equation (2) to incorporate interactions between the import penetration measures and the 𝑁𝑀𝐸𝑡 indicator variable, which is “1” when the NME is active in 2009 and onwards, and “0” otherwise, as depicted by equation (4). 19 𝑦𝑗𝑡 =𝛼+𝛽1𝐼𝑃𝑗,𝑡−1 𝐶ℎ𝑖𝑛𝑎 +𝛽2𝐼𝑃𝑗,𝑡−1 𝑅𝑂𝑊 +𝛽5𝐼𝑃𝑗,𝑡−1 𝐶ℎ𝑖𝑛𝑎 ×𝑁𝑀𝐸𝑡+𝛽6𝐼𝑃𝑗,𝑡−1 𝑅𝑂𝑊 ×𝑁𝑀𝐸𝑡+𝛾𝑗+𝛿𝑡+𝑢𝑗𝑡 (4) There are some aspects of the above econometric specifications that merit further discussion. The first issue is that it can take some time for Brazilian producers to react to changes in market conditions, hence I utilize lagged trade exposure variables to address this. The second issue is the simultaneity between the outcomes and the import penetration measures, since the value-added is part of the industry output that is used to calculate import penetration. 20 This is also alleviated by employing the first lag of the import penetration measures. The third aspect is the omitted variable bias—more precisely, omitted factors that may affect both the outcome and the trade exposure measures. For instance, a government averse to unemployment may protect more labor-intensive industries. As a result, this industry characteristic affects import penetration and the industry employment share (outcome). This renders the estimates inconsistent. The industry-specific and time-invariant omitted variables are controlled for by industry effects. The year fixed effects account for time-varying factors that affect industries equally, such as business cycles. For example, if firms employing formal workers are more likely to reduce employment during 18 Unfortunately, the downstream import penetration variables are highly correlated with the import penetration variables. This leads to a severe multicollinearity problem, and estimates could not be obtained for the downstream import penetrations. 19 A shortcoming of the empirical specification used to assess the effects of the NME is that besides the response to the NME policies, the interaction coefficients may also capture the effects of the 2007–2008 world crisis. 20 I consider the Chinese share in the markets served by Brazilian exporters (𝐶𝐸𝑋𝑃𝑗,𝑡−1) to be exogenous because the Brazilian share is small relative to the market size and also much smaller than the Chinese market share in several industries. As a consequence, a supply shock in the Brazilian industry is highly unlikely to displace Chinese producers. 28 estimated coefficients in table 21 do not show much of a change for the different options of fixed effects, and only the statistical significance of Chinese import penetration seems to be affected. 29 The next estimates are based upon augmented versions of equation (5) to control for industry-level effective applied tariffs, upstream import penetrations, and interactions between import penetrations and the labor-intensive industry indicator, the manufacturing state indicator, and the coastal state indicator. The OLS estimates for the log of the hourly wage using industry, state, and year effects are displayed in table 22. The coefficient for Chinese import penetration is positive and is only statistically significant in column (6). The estimated coefficients for ROW import penetration are positive and have a similar magnitude to the coefficients in table 21, which are statistically significant in most cases. Of the additional regressors, the only estimated coefficients that are statistically significant are the positive coefficient for the interaction between Chinese import penetration and the labor-intensity indicator, and the positive coefficient for the interaction between ROW import penetration and the coastal state indicator. Table 23 reports similar estimates for informal employment status. In almost all cases, the import penetration coefficients are negative. The coefficients for Chinese import penetration are significant solely in columns (6) and (7), while the coefficients for ROW import penetration are significant in all specifications except for column (5). The effect of the Chinese market share abroad on informality is negative and significant, as can be seen in column (2), and so is the effect of the effective applied tariff. An increase in upstream ROW import penetration is found to reduce the likelihood of the worker having an informal job. The interaction between ROW import penetration and the NME indicator is positive and significant, even though the net effect of ROW import penetration after the implementation of the NME is still negative. The interaction between Chinese import penetration and the labor-intensity indicator has a negative coefficient, which indicates that workers that remain employed in a labor-intensive industry are less likely to be informal in the event of a rise in Chinese import penetration. The IV estimates based on equation (5) with industry, state, and year effects are reported in table 24 for the log of the hourly wage, and in table 25 for the informal job indicator. In table 24, not a single coefficient is statistically significant even at the 10% level. Looking at the KP statistics for these estimated specifications, column (1) has a statistic of 27, column (3) of 12, and column (4) of 19. The remaining columns exhibited KP statistics of below 4 and thus are plagued by the weak instruments problem. The null hypothesis of exogeneity for the import penetration measures in columns (1), (3), and (4) could not be rejected. In fact, the coefficients for column (1) are very close to those obtained by means of OLS reported in table 22. The estimates in table 25 for the informal job indicator show a similar pattern; however, Chinese import penetration is negative and significant at the 10% level in column (1) and significant at the 5% level in column (7). Again, the null of exogeneity could not be rejected in columns (1), (3), and (4), which are the specifications with first-stage KP statistics above 12. In view of the results of the statistically significant coefficients for Chinese import penetration in columns (3) and (6) of table 21, I decided to conduct further estimates using state-industry and year effects. 30 Table 26 shows the OLS estimates for the log of the hourly wage. These new estimates show more statistically significant coefficients and are comparable in magnitude to those in table 24, except for columns (6) and (7). Now, the Chinese import penetration coefficient is significant at the 5% level in columns (2) and (7). Chinese market share abroad, the effective applied tariffs, and the upstream ROW import penetration all have a positive statistically significant effect on the log of the hourly wage. The coefficients for the interactions of import penetration with the NME indicator show a positive effect for Chinese import penetration, and a small negative effect for ROW import penetration. In column (7), the interaction of Chinese import penetration with the coastal state indicator is negative and significant at the 10% level, while the interaction with ROW import penetration is negative and no longer statistically significant. This means that coastal states experienced a smaller net increase in wages due to an increase in Chinese import penetration. The OLS estimates for informal employment status are reported in table 27. In comparison to table 25, the only salient change 29 Results obtained using the other import penetration measures were comparable, thus I omitted them from this report. 30 Estimates using state-year and industry effects were similar to those using industry, state, and year effects. 29 is in the estimated coefficient for ROW import penetration interaction with the coastal indicator that is now positive and statistically significant at the 5% level of confidence. Turning to the IV estimates for the log of the hourly wage with state-industry and year fixed effects (table 28), we can see that several estimated coefficients are statistically significant at the 5% level, such as ROW import penetration in columns (1), (3), and (4). In column (4), the upstream Chinese and ROW import penetrations had a negative effect on the log of the hourly wage. Note that this specification in column (4) is the only one in this table with a KP statistic of 17, while the other specifications have KP statistics that are below 6, which means that the weak instruments problem affects the latter estimated specifications. In column (7), the interaction between ROW import penetration and the manufacturing state indicator is positive and significant at the 5% level. The null of exogeneity of the regressors is only rejected in the specifications in columns (1), (2), and (5). The IV estimates for informal employment status are reported in table 29, in which the coefficients for neither Chinese nor ROW import penetration are statistically significant. In fact, in the only specification that does not suffer from the weak instruments problem, column (4), none of the coefficients are statistically significant, although the null of exogeneity of the import penetrations could not be rejected. In column (5), we can see a negative effect of the interaction between ROW import penetration and the NME indicator. In column (7), it is the estimated coefficient of the interaction between ROW import penetration and the manufacturing indicator that is negative and statistically significant. Finally, in column (8), the interaction between Chinese import penetration and the coastal state indicator has a negative coefficient that is also statistically significant. The final set of estimates employing the worker-level data uses state-by-industry trade exposure measures and industry, state, and year fixed effects. Table 30 reports the OLS estimates for the log of the hourly wage. In most specifications, the coefficients for ROW import penetration are positive and statistically significant. The effects of the upstream Chinese and ROW import penetration are negative and significant in column (4). Interestingly, the interaction between ROW import penetration and the NME indicator is positive and significant, while the coefficient for ROW import penetration is negative and not significant. This suggests that the positive effect of ROW import penetration on the regressions without controls for the NME comes from the period after the NME was implemented. Finally, in column (6), the coefficient for the interaction between ROW import penetration and the labor-intensive indicator is positive and statistically significant, while the interaction with Chinese import penetration is negative, as predicted by hypothesis 2. Table 31 shows the OLS estimates for informal employment status. Chinese import penetration shows a negative impact on informal employment, except in column (5), in which its interaction with the NME indicator is negative and significant. This implies that the total effect of Chinese import penetration was initially positive but became negative after the implementation of the NME. The coefficients for ROW import penetration show the exact opposite pattern. The IV estimates for the worker-level specification employing state-industry import penetration are reported in table 32 for the log of the hourly wage and in table 33 for the informal employment status indicator. All the estimated specifications in both tables exhibit a KP statistic below 2.5, which is very low and implies that the estimates are plagued by the weak instrument problem. Table 32 exhibits only three statistically significant coefficients. In column (2), the coefficient of the Chinese market share abroad is negative and significant at the 5% level. In column (3), the coefficient for ROW import penetration is also negative and significant, whereas in column (6), its interaction with the NME indicator is positive and statistically significant at the 10% level. Finally, looking at the informal employment status indicator (table 33), there are only two statistically significant coefficients, namely the effective applied tariffs, which implies that larger tariffs reduce informality and the interaction between ROW import penetration and the manufacturing state indicator has a positive and significant coefficient. In the next subsection I move on to analyze the state-level specification. C. State-level variation Adjustments to changes in the trade environment might occur between different regions of the country within the same industry. This margin of adjustment may not be captured by the industry-level specifications used in this paper. 30 To fill this gap, I estimate state-level specifications based upon equation (6). The first set of OLS estimates using all four measures of state-level trade exposure are set out in table 34. All the dependent variables are expressed in their first difference in equation (6), and the trade exposure measures are calculated using the first difference of the import volumes. Panel A reports the estimates for the trade exposure measures constructed according to ADH’s methodology. Chinese imports per worker showed a positive effect on the log of the employment level and on the log of the hourly wage and were statistically significant at the 10% level. These results can be interpreted as follows. A US$1,000 increase in Chinese imports per worker increases the employment level by 5.4%, whereas a similar increase in ROW imports per worker reduces employment by 0.1%. In panel B, trade exposure is measured in terms of imports per firm. The estimates reveal that Chinese imports per firm have a statistically significant positive effect on employment and a positive effect on average years of schooling. ROW imports per firm exhibit a small positive effect on the log of the hourly wage and a negative effect on the share of informal employment, the average years of schooling, the share of workers with high-school diplomas, and the share of workers with college degrees. In panels C and D, the trade exposure measures are calculated using the actual imports obtained from the Aliceweb system. These estimates in panel C indicate a negative effect of Chinese imports per worker on the share of informal employment and the share of workers with high-school diplomas, whereas the estimated coefficients for ROW imports per worker suggest a positive statistically significant effect on the log of employment, the share of informal employment, and the share of workers with high-school diplomas. It shows a negative effect on the log of the hourly wage, which is significant at the 5% level. In panel D, the estimated coefficients for Chinese imports per firm reveal a negative effect on the share of informal employment and on the share of workers with high-school diplomas, and a positive effect on the average years of schooling. The IV estimates for the state-level specification are reported in table 35. Before discussing the estimated coefficients, it is important to note that the KP test statistics for panels A and B are above 170, which means that under the assumption that the excluded instruments are valid, the weak instrument problem is not affecting the estimates of these panels. Nonetheless, the KP test statistics for panels C and D are below 2.5. This clearly indicates that the weak instrument problem is afflicting the estimates reported in these panels. Focusing the analysis on the estimates that are statistically significant at the 10% level at least, we can see in panel A that ROW imports per worker have a positive impact on the average years of schooling. In panel B, the coefficient for Chinese imports per firm is negative for the share of workers with high-school diplomas. ROW imports per firm show a positive effect on the log of the hourly wage and a negative effect on the share of workers with college degrees and on years of schooling. Note that the latter result is the opposite of that found in panel A for ROW imports per worker. panel C showed no statistically significant estimated coefficients. In panel D, Chinese imports per firm impacted positively the employment level, the employment share within manufacturing, and the log of the hourly wage, whereas it decreased the average years of schooling and the share of workers with college degrees. The ROW imports per firm had exactly the opposite effect, albeit with a smaller magnitude. The null hypothesis of exogeneity of the trade exposure measures was rejected only in some specifications of the share of employment in the population and of years of schooling. The next collection of estimates employs an augmented version of equation (6) that includes interactions between the NME indicator and the trade exposure measures. Table 36 exhibits the OLS estimates of this specification. In panel A, the estimated effect of Chinese imports per worker on the log of the hourly wage is positive, although it turned negative after the implementation of the NME. The impact of ROW imports per worker follows the exact opposite pattern for both the log of the hourly wage and the share of informal employment, albeit with a smaller magnitude. In panel B, ROW imports per firm initially had a positive effect on the share of workers with college degrees. After the implementation of the NME, it had a negative effect on this and also started to negatively impact the share of workers with high-school diplomas. The results in panels C and D imply that after the implementation of the NME, Chinese imports per worker and per firm reduced both the share of informal employment and the share of workers with highschool diplomas, while ROW imports boosted the share of informal employment. 31 Table 37 reports the IV estimates of the augmented version of equation (6). The KP statistics of the estimates in panels A and B are greater than 6, which suggest that the weak instrument problem is not an issue here. Nevertheless, the same cannot be said about the specifications in panels C and D because their KP test statistics are below 0.5. The estimated coefficients in panel A indicate that Chinese imports per worker had a negative impact on the share of employment, which is statistically significant at the 10% level. It also had a negative estimated coefficient for the share of informal employment, which is statistically significant at the 5% level. The coefficients for ROW imports per worker were negative for the log of the hourly wage and positive for the share of informal employment. After the implementation of the NME, Chinese imports exhibited a positive impact on the share of informal employment, while ROW imports showed a negative impact. Moreover, the interaction between ROW imports per worker and the NME coefficient had positive and significant estimated coefficients for the share of employment in the population and for the log of the hourly wage. The IV estimates for panel A did not differ much from the OLS estimates in table 36, panel A. In panel B, we can see that Chinese imports per firm had a negative effect on the hourly wage and on the share of informal employment, and a positive effect on years of schooling and the share of workers with college degrees. After the implementation of the NME, Chinese imports per firm had a positive effect on the hourly wage and a negative effect on the shares of workers with high-school diplomas and college degrees. Moving to panel C, Chinese imports per worker reduced the share of informal employment before the NME was implemented, but after this the effect became positive. In panel D there are no statistically significant estimated coefficients. 5. DISCUSSION OF THE RESULTS The first round of estimates employed an industry-level identification strategy. Very few IV estimates, regardless of the excluded instruments used, did not suffer from the weak instrument problem, and the null of exogeneity of import penetration measures could not be rejected in these estimated specifications. Consequently, I will focus my analysis on the results of the OLS specification. The industry-level results suggest that the employment level was negatively impacted by ROW import penetration and Chinese import penetration, albeit to a lesser extent. The implementation of the NME strengthened the negative effect of ROW import penetration. As predicted by hypothesis 1, labor-intensive industries were more negatively impacted by Chinese imports, even though the coefficient was not statistically significant. The share of employment in the population was not affected by changes in trade exposure. However, the employment share within manufacturing was negatively impacted by ROW import penetration. Interestingly, Chinese import penetration exhibits a negative impact once additional controls are present in the specification. More precisely, the implementation of the NME worsened this negative impact, though not statistically significant, and labor-intensive industries suffered a larger impact, as predicted by hypothesis 1. Upstream Chinese import penetration showed a positive effect. The estimated effects on job characteristics indicate that both Chinese and ROW import penetrations have a negative effect on the hourly wage but that upstream ROW import penetration has a positive effect on this. The interaction of Chinese import penetration with the labor-intensive industry indicator is positive and not statistically significant, which does not support hypothesis 2. The interindustry wage premium was positively affected by both import penetration measures. Again, the sign of the effect of Chinese import penetration is sensitive to the additional controls included. The share of informal employment is negatively affected by the import penetration measures, more so for the labor-intensive industries. These negative impacts became stronger after the implementation of the NME. The average years of schooling were only positively affected by Chinese import penetration in the labor-intensive industries, as was also the case with upstream ROW import penetration. The share of workers with high-school diplomas seems to be positively affected by Chinese import penetration, with a stronger impact on labor-intensive industries. And this effect declined after the implementation of the NME, but the net effect remained positive. Upstream Chinese import penetration had a negative effect on the share of workers with high-school diplomas. There is not much evidence that the share of workers with college degrees was impacted by trade exposure measures, except for 32 upstream ROW import penetration, which had a positive effect. At the end of the day, there seems to be almost no support for hypothesis 2. The results discussed so far were obtained without controlling for the workers’ observable characteristics. The use of worker-level data enabled the estimation of several additional econometric specifications, which resulted in several interesting new findings. Unlike the industry-level results, both ROW and Chinese import penetration led to an increase in wages, although the latter effect was small and only statistically significant in specifications with state-industry fixed effects. The effect of Chinese import penetration on wages is stronger for workers employed in labor-intensive industries. No evidence was found to suggest that the wages of workers residing in manufacturing states experienced the impact of import penetration differently. Workers living in coastal states were found to be more positively affected by ROW import penetration. When state-industry effects are used, the estimates reveal that the implementation of the NME magnified the positive effects of Chinese import penetration on wages and at the same time reduced the effect of ROW import penetration. Additionally, upstream ROW import penetration positively impacted wages. Chinese import penetration showed a larger positive effect on the workers of labor-intensive industries, and a smaller effect on the wages of workers living in coastal states. Moving on to the impacts on the likelihood of having an informal job, both import penetration measures negatively impacted the likelihood of formal employment. The coefficient for ROW import penetration before the implementation of the NME was larger than its net effect after the NME was put into effect. Chinese market share abroad also exhibited a negative effect. The effect of Chinese import penetration on the log of the hourly wage was stronger for labor-intensive industries. When state-industry fixed effects are used, I also find that the effects of import penetration on the manufacturing states were considerably smaller. A rise in ROW import penetration reduced the likelihood of informal employment less in coastal states. The estimates using state-industry-level trade exposure measures led to some results that differ from those using industry-level import penetration. In the log of the hourly wage estimates, the positive effect of ROW import penetration persists, although it is negative before the NME and its net effect becomes positive after the NME. The positive effect of Chinese import penetration found earlier becomes negative after the NME, although this is not statistically significant. Its effect on the wages of workers in labor-intensive industries is negative, as predicted by hypothesis 2. When the NME interactions are included in the econometric model, the before-NME effect of Chinese import penetration is positive but becomes negative after the NME. For the informal employment indicator, the use of the specification with the state-industry trade exposure measures leads to a sign reversal for the ROW import penetration measure, which becomes positive. The net effect of ROW import penetration is negative before the implementation of the NME and positive after this, while Chinese import penetration shows the exact opposite pattern. The last set of estimates looks at the effects of the state-level variation in trade exposure measures on the statelevel outcomes. These estimates exhibited few statistically significant coefficients and they varied considerably according to the trade exposure measure used. The effects of ADH’s imports-per-worker trade exposure measure changed over time due to the implementation of the NME. Before the NME, both the Chinese and ROW imports per worker reduced employment, and their impact became positive after the NME, even though none of the coefficients were statistically significant. Chinese imports per worker reduced the share of employment in the population and the employment share in manufacturing, but solely the former is statistically significant. Both effects were also smaller after the implementation of the NME. The net effect of ROW imports per worker initially led to a reduction in the log of the hourly wage, but after the NME its net effect became positive. In the pre-NME period, the share of informal workers was negatively affected by Chinese imports per worker. After 2008, its net impact became positive. An exact opposite pattern was displayed by ROW imports from China. None of the three worker skill-level-related outcomes were affected by ADH’s imports-per-worker measure. New estimates employing trade exposure measures that replace the number of workers with the number of firms using ADH’s methodology led to statistically significant results but only for the share of informal employment and for the skill-level-related outcomes. Chinese imports per firm had a negative impact on the share of informal employment 33 and a positive one on average years of schooling. Before the implementation of the NME, Chinese imports per firm increased the share of workers with high-school diplomas and of those with college degrees, but only the former was statistically significant, which contradicts hypothesis 2. The net impact on both these shares became negative and statistically significant after the NME policies came into effect. At first, ROW imports per firm had a negative effect on the share of workers with high-school diplomas and college degrees, but this became positive after the NME. Interestingly, the estimates using actual imports per worker and per firm did not have much of an impact on state-level outcomes. Overall, the results revealed in this study provide support for hypothesis 1, in that they find trade exposure to have had a positive effect on the interindustry wage premium and, in most cases, a negative impact on the share of informal employment. The results on the skill level in manufacturing were mixed, which implies little support for hypothesis 2. The fact that trade exposure reduces the share of informal employment may seem counterintuitive since greater import competition might seem to push firms toward informal employment terms. Under the assumption that firms employing informal workers tend to be smaller and less productive, as discussed in Paz (2014a), it is conceivable that the strong competitive edge of Chinese producers—magnified by the similarity between their products and Brazilian-made ones— may have resulted in more firms leaving the market altogether (including ones that originally employed informal workers), rather than pushing firms toward higher levels of informal employment. These factors lead to a lower employment level (as found in my estimates) and a lower share of informal employment. The effects of Chinese market share in destinations served by Brazilian exporters were usually not statistically significant. The small magnitude of these estimated coefficients may come from Brazil’s limited participation in the world trade system and is perhaps due to the fact that Brazilian exports would not be much greater in the absence of Chinese competition in foreign markets. This conjecture is supported by several studies. Batista (2008) used a constant market approach methodology and found that for 1990–2004, Latin American countries lost approximately 1.7% of their total manufacturing exports to China. Using an industry-level gravity equation, Hanson and Robertson (2010) estimate that in the absence of the China shock, export demand would have been higher by between 1% and 2% for countries like Mexico. Using a methodology similar to that of Batista (2008), Moreira (2006) also found that Brazil lost only about 2.5% of its manufacturing exports to China in 1990–2004. The myriad results revealed in this project are highly dependent on the type of data variation exploited and on the econometric specification used. This underscores the difficulty of identifying the effects of Chinese imports due to the following five issues. First, Brazil’s experience is not one of a simple substitution of ROW imports by Chinese products. Second, the estimated effects differ according to the labor intensity of the industry in question. Third, states with a large share of manufacturing or with a sea harbor in their territory are impacted differently. Fourth, increased access to imported intermediate inputs has nonnegligible effects on labor market’s outcomes, and these effects differ not only according to the origin of the inputs but also often have the opposite sign to that of import penetration of final goods. Last but not least, the implementation of the NME had a deep impact on how labor-market outcomes respond to changes in trade exposure. In sum, these issues are evidence that Chinese imports have a very nuanced impact on the Brazilian manufacturing industry, in which the effects of the China shock depend on the year, state, and industry in question. 6. CONCLUSIONS China, one of the most populous countries in the world, entered the 21st century not only as one of the largest and fastest-growing economies but also as a major player in world trade. This quick ascent and its cost advantage in manufacturing production prompted several concerns in developing countries as to whether they would still be able to sustain a dynamic manufacturing sector in view of China’s competitive edge. This concern is built on the fact that many observers perceive a strong manufacturing sector to be a key driver of economic growth and a provider of higher-wage jobs relative to those available in agriculture and services. 34 The changes undergone by the Brazilian economy in 2000–2012 provide a good case study through which to assess such concerns. Besides being the most populous country and largest economy in Latin America, Brazil also has a large, diversified manufacturing sector. In this period, import penetration in Brazil increased by more than 25% and China’s share of such imports increased from 3% to 20%. Equally importantly, Chinese exporters’ market share increased vigorously in markets that are also served by Brazilian exporters. In the same period, the share of manufacturing in the gross domestic product in Brazil declined by more than 20%. This study employed Brazilian census and household survey data to examine the impacts of the increasing trade exposure experienced by the Brazilian economy on several labor-market outcomes of the manufacturing sector in 2000–2012. This data is notable for encompassing both formal and informal workers, which is significant given that the latter represent more than 20% of the workforce employed in the manufacturing sector. This study presents an empirical methodology that decomposes the effects of import penetration into that generated by Chinese and non-Chinese (or ROW) imports. This analysis is conducted at the industry, state, and worker levels. The results at the industry level indicate that Chinese and ROW import penetration have a negative effect on the employment level, and that the impact of Chinese imports on labor-intensive industries was considerably stronger than on the other industries. Interestingly, upstream Chinese import penetration positively impacted the employment level. Both ROW and Chinese import penetration reduced the industry-level hourly wage and share of informal employment, while boosting the interindustry wage premium. The implementation of the NME policy in 2008 attenuated these effects. The worker-level results suggest that after controlling for the worker’s observable characteristics, Chinese and ROW import penetration at the industry level increased the workers’ wages, with a larger effect for ROW import penetration in coastal states and a bigger effect for Chinese import penetration on labor-intensive industries. The implementation of the NME magnified the effects of Chinese import penetration and dampened that of ROW import penetration. Import penetration from both China and the ROW had a negative effect on the worker’s likelihood of having an informal job. These effects are smaller in states with a large manufacturing sector, and Chinese import penetration had a larger impact on labor-intensive industries. However, the use of state-by-industry trade exposure measures led to some different results. In fact, the effect of ROW import penetration went from negative to positive after the implementation of the NME, while the exact opposite was true of Chinese import penetration. Analogously for the informal job indicator, the sign of the ROW import penetration measure also went from negative to positive in response to the implementation of the NME, while Chinese import penetration again showed the exact opposite pattern. The last set of estimates were obtained using state-level identification using the methodology of Autor, Dorn, and Hanson (2013). These estimates showed that before the NME was put into effect, both Chinese and ROW imports per worker reduced employment levels and these effects became positive after 2008. In the pre-NME period, the net effect of ROW and Chinese imports per worker decreased the log of the hourly wage and the share of informal employment, respectively. After the NME went into effect, both effects became positive. Moving on to trade exposure measured as imports per firms, this study finds that Chinese imports per firm negatively impacted the share of informal employment and positively affected the average years of schooling. Before the NME was implemented, Chinese imports per firm increased the shares of workers with high-school diplomas and those with college degrees, but the net impact on both shares became negative after the NME policies came into effect. The effects of ROW imports per firm on these two outcomes behaved entirely to the contrary. Finally, the estimates using actual imports per worker and per firm did not impact the state-level labor-market outcomes. At the end of the day, this paper provides evidence suggesting that the effects of Chinese imports are different than those of imports from other countries. These effects also differ according to the characteristics of both the industry and the state. Moreover, the new policies introduced by the implementation of the NME policies altered the effects of the changes in the trade exposure measures on labor-market outcomes, and in some cases, the NME even reversed them. Since the China shock affected manufacturing in Brazil in many ways, it is not possible to offer a conclusion that the shrinkage in the participation of manufacturing in Brazil’s GDP is the result of the China shock. These heterogeneous 35 effects also suggest that a deeper investigation at the firm level may shed some light on the adjustment mechanisms used to cope with increased trade exposure and to the changes in policy caused by the implementation of the NME. 36 REFERENCES Abreu, M. 1996. “Trade in Manufactures: The Outcome of the Uruguay Round and Developing Countries Interests.” In: The Uruguay Round and Developing Countries, edited by W. Martin and L.A. Winters. Washington, DC: World Bank. Acemoglu, D., D. Autor, D. Dorn, G. Hanson, and B. Price. 2015. “Import Competition and the Great US Employment Sag of the 2000s.” Journal of Labor Economics 34(S1). Acosta, P., and L. Gasparini. 2007. “Capital Accumulation, Trade Liberalization, and Rising Wage Inequality: The Case of Argentina.” Economic Development and Cultural Change 55(4): 793–812. Araújo, B., and L. Paz. 2014. “The Effects of Exporting on Wages: An Evaluation Using the 1999 Brazilian Exchange Rate Devaluation.” Journal of Development Economics 111: 1–16. Autor, D., D. Dorn, and G. Hanson 2013. “The China Syndrome: Local Labor Market Effects of Import Competition in the United States.” American Economic Review 103(6): 2121–2168. Bartel, A., Sicherman, N., 1999. “Technological Change and Wages: An Interindustry Analysis.” The Journal of Political Economy 107: 285–325. Batista, J.C. 2008. “Competition Between Brazil and Other Exporting Countries in the US Import Market: A New Extension of Constant-Market-Shares Analysis.” Applied Economics 40(19): 2477–2487. Baumann, R. 2001. “Brazil in the 1990s: An Economy in Transition.” CEPAL Review 73: 147–169. Corseuil, C.H., and M.N. Foguel. 2002. “Uma sugestão de deflatores para rendas obtidas a partir de algumas pesquisas domiciliares do IBGE.” Textos Para Discussão, IPEA TD 0897. Costa, F., J. Garred, and J.P. Pessoa. 2014. “Winners and Losers from a Commodities-for-Manufactures Trade Boom.” CEP Discussion Paper No. 1269. Facchini, G., Olarreaga, M., Silva, O., and G. Willmann. 2010. “Substitutability and Protectionism: Latin America’s Trade Policy and Imports from China and India.” The World Bank Economic Review 24(3): 446–473. Fernandes, A. 2007. “Trade Policy, Trade Volumes and Plant Level Productivity in Colombian Manufacturing Industries.” Journal of International Economics 71(1): 52–71. Ferreira, P.C., and J.L. Rossi. 2003. “New Evidence from Brazil on Trade Liberalization and Productivity Growth.” International Economic Review 44(4): 1383–1405. Hanson, G.H., and R. Robertson. 2010. “China and the Manufacturing Exports of Other Developing Countries.” In China's Growing Role in World Trade, edited by Robert C. Feenstra and Shang-Jin Wei. Chicago: University of Chicago Press. Iacovone, L., F. Rauch, and L.A. Winters. 2013. “Trade as an Engine of Creative Destruction: Mexican Experience with Chinese Competition.” Journal of International Economics 89. IBGE. 2015. Banco de Dados Agregados. Sistema IBGE de Recuperação Automática – SIDRA, 2015. IBGE. 2016a. Tabelas Sinóticas Retropoladas 2000–2013. IBGE. 2016b. CONCLA-Comissão Nacional de Classificação. Correspondência de classificações de atividade econômica. Available at http://concla.ibge.gov.br/classificacoes/correspondencias/atividades-economicas International Monetary Fund. 2015. IFS-International Financial Statistics. IPEADATA 2017. Instituto de Pesquisa Economica Aplicada. Kovak, B. 2013. “Local Labor Market Effects of Trade Policy: Evidence from Brazilian Liberalization.” American Economic Review 103(5): 1960–1976. 37 Kume, H., Piani, G., and Souza, C.F. 2003. “Instrumentos de politica comercial no periodo 1987–1998.” In A abertura comercial brasileira nos anos 1990: impactos sobre emprego e salario, edited by C.H. Corseuil and H. Kume. Rio de Janeiro, Brazil: Ipea. Kume, H., Piani, G., and P. Miranda. 2008. “Política comercial, instituições e crescimento economico no Brasil.” In: Crecimiento económico, instituciones, política comercial y defensa de la competencia en el MERCOSUR, edited by H. Kume. Montevideo: Red Mercosur. Volume 11. Lu, Y., and T. Ng. 2013. “Import Competition and Skill Content in US Manufacturing Industries.” Review of Economics and Statistics 95(4): 1404–1417. Mesquita Moreira, M. 2006. “Fear of China: Is There a Future for Manufacturing in Latin America?” INTAL–ITD, Occasional Paper 36. Paz, L. 2014a. “The Impacts of Trade Liberalization on Informal Labor Markets: A Theoretical and Empirical Evaluation of the Brazilian Case.” Journal of International Economics 92(2): 330–348. Paz, L. 2014b. “Trade Liberalization and Industry Wage Premia: The Missing Role of Productivity.” Applied Economics 46(4): 408–419. United Nations. Statistical Division. UN Comtrade. New York: United Nations; 2003. World Bank 2017. World Development Indicators Database. 44 TABLE 7. SIMPLE CORRELATIONS BETWEEN ENDOGENOUS REGRESSORS AND EXCLUDED INSTRUMENTS PANEL A. IMPORT PENETRATION AT THE INDUSTRY LEVEL. Endogenous regressor\Excluded instrument China’s share of imports in Latin American countriest-1 high-income countries’ share of imports in Latin American countries t-1 ROW imp. penetrationt-1 -0.129 0.299 Chinese imp. penetrationt-1 0.574 0.316 ROW alt. imp. penetrationt-1 -0.120 0.297 Chinese alt. imp. penetrationt-1 0.580 0.320 ROW adj. imp. penetrationt-1 -0.051 0.184 Chinese adj. imp. penetrationt-1 0.616 0.352 ROW adj. alt. imp. penetrationt-1 -0.056 0.220 Chinese adj. alt. imp. penetrationt-1 0.625 0.356 Number of observations: 286. PANEL B. ALTERNATIVE INDUSTRY-LEVEL EXCLUDED INSTRUMENTS Endogenous regressor\Excluded instrument Chinese imp. penet.1998  Real exch. ratet-1 ROW imp. penet.1998  Real exch. ratet-1 ΔChinese imp. penetrationt 0.550 0.458 ΔROW imp. penetrationt -0.011 0.748 ΔChinese alt. imp. penetrationt-1 0.557 0.441 ΔROW alt. imp. penetrationt-1 0.031 0.741 ΔChinese adj. imp. penetrationt-1 0.561 0.451 ΔROW adj. imp. penetrationt-1 -0.029 0.775 ΔChinese adj. alt. imp. penetrationt-1 0.576 0.422 ΔROW adj. alt. imp. penetrationt-1 0.034 0.788 Number of observations: 286. PANEL C. IMPORT PENETRATION AT THE STATE-BY-INDUSTRY LEVEL. Endogenous regressor\Excluded instrument ΔLatin American countries’ Chinese share ΔLatin American countries’ high-income countries share ΔChinese imports per worker 0.881 0.702 ΔROW imports per worker 0.629 0.888 ΔChinese imports per firm 0.760 0.845 ΔROW imports per firm 0.880 0.683 Number of observations: 6,057. 45 TABLE 8. INDUSTRY-LEVEL OLS REGRESSION OF TRADE EXPOSURE MEASURES ON LABOR-MARKET OUTCOMES USING EQUATION (2) (1) (2) (3) (4) (5) (6) (7) (8) (9) Outcome Log (employment level) Share of employment in the population (%) Emp. share of manuf. Log (hourly wage) Wage premium Informal share Years of schooling High-school share College share A. Chinese import penetrationt-1 -0.002 0.002** 0.013 -0.055* 0.070*** -2.890 -0.001 -2.317 -2.026 (0.011) (0.001) (0.012) (0.030) (0.024) (2.033) (0.103) (1.672) (1.366) B. Chinese import penetrationt-1 0.001 0.002* 0.017 -0.048* 0.070*** -2.590 -0.059 -2.161 -1.869 (0.014) (0.001) (0.014) (0.028) (0.022) (1.869) (0.099) (1.585) (1.158) ROW import penetrationt-1 -0.002 -0.000 -0.002 -0.003 0.010** -0.152 0.030*** -0.079 -0.080 (0.002) (0.000) (0.002) (0.004) (0.004) (0.124) (0.004) (0.146) (0.140) C. Alternative import penetration Chinese alt. import penetrationt-1 -0.005 0.003* 0.015 -0.036 0.072*** -2.919 -0.045 -1.913 -1.922 (0.013) (0.001) (0.015) (0.031) (0.026) (1.881) (0.125) (1.883) (1.166) ROW alt. import penetrationt-1 -0.004* -0.000 -0.004* -0.005 0.013** -0.546** 0.067*** 0.056 -0.228 (0.002) (0.000) (0.002) (0.005) (0.005) (0.234) (0.018) (0.170) (0.154) D. Adjusted import penetration Chinese adj. import penetrationt-1 -0.004 0.001 0.009 -0.040* 0.043*** -1.742 -0.009 -0.917 -1.006** (0.011) (0.001) (0.008) (0.019) (0.011) (1.323) (0.042) (0.776) (0.484) ROW adj. import penetrationt-1 -0.000 -0.000 -0.000 -0.003** 0.006*** 0.002 0.012*** -0.097 -0.031 (0.001) (0.000) (0.001) (0.001) (0.002) (0.069) (0.004) (0.078) (0.059) E. Adjusted alternative imp. penet. Chinese adj. alt. import penetrationt-1 -0.004 0.002* 0.011 -0.024 0.044*** -1.909 -0.001 -0.848 -1.051** (0.010) (0.001) (0.011) (0.024) (0.013) (1.345) (0.059) (1.038) (0.497) ROW adj. alt. import penetrationt-1 -0.003** -0.000 -0.003* -0.006 0.007** -0.420** 0.053*** 0.061 -0.207 (0.001) (0.000) (0.002) (0.005) (0.003) (0.161) (0.016) (0.160) (0.135) Notes: Number of observations: 286. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Year and industry fixed effects included in the specification. Standard errors clustered at the industry level. Regressions for wage premium uses the inverse of the estimated wage premium variance as weights, and for the remaining outcomes, the weights are the inverse of the number of observations used to compute the industry-level variable. 46 TABLE 9. INDUSTRY-LEVEL IV REGRESSION OF TRADE EXPOSURE MEASURES ON LABOR-MARKET OUTCOMES USING EQUATION (2) (1) (2) (3) (4) (5) (6) (7) (8) (9) Outcome Log (employment level) Share of employment in the population (%) Emp. share of manuf. Log (hourly wage) Wage premium Informal share Years of schooling High-school share College share A. Chinese import penetrationt-1 0.047 0.014 0.074 -0.041 0.013 -0.089 0.076 0.842 -0.498 (0.058) (0.009) (0.088) (0.075) (0.015) (0.717) (0.082) (1.145) (1.322) B. Chinese import penetrationt-1 0.037 0.018 -0.020 0.168 0.011 0.765 1.359 7.722 3.112 (0.403) (0.023) (0.876) (1.901) (0.020) (8.394) (9.221) (46.893) (33.944) ROW import penetrationt-1 -0.014 0.006 -0.138 0.306 -0.019 1.254 1.883 10.101 5.299 (0.586) (0.035) (1.279) (2.715) (0.025) (13.092) (13.315) (68.214) (47.672) C. Alternative import penetration Chinese alt. import penetrationt-1 0.053 0.013 0.113 -0.122 0.011 -0.421 -0.412 -1.761 -1.904 (0.145) (0.014) (0.140) (0.191) (0.019) (1.929) (1.835) (10.562) (4.097) ROW alt. import penetrationt-1 0.009 -0.003 0.075 -0.165 -0.023 -0.675 -1.008 -5.399 -2.852 (0.254) (0.020) (0.187) (0.321) (0.027) (3.698) (3.476) (20.515) (5.694) D. Adjusted import penetration Chinese adj. import penetrationt-1 0.034 0.002 0.137 -0.244 -0.037 -0.966 -1.347 -7.013 -4.139 (0.418) (0.025) (0.810) (1.470) (0.131) (6.858) (8.052) (45.458) (27.126) ROW adj. import penetrationt-1 0.002 -0.003 0.039 -0.094 -0.083 -0.389 -0.595 -3.215 -1.637 (0.175) (0.010) (0.314) (0.585) (0.252) (2.721) (3.087) (17.183) (10.912) E. Adjusted alternative imp. penet. Chinese adj. alt. import penetrationt-1 0.032 0.007 0.075 -0.093 -0.011 -0.337 -0.380 -1.776 -1.496 (0.123) (0.006) (0.139) (0.146) (0.048) (1.779) (0.647) (5.067) (2.851) ROW adj. alt. import penetrationt-1 0.001 -0.001 0.020 -0.048 -0.072 -0.200 -0.307 -1.663 -0.842 (0.082) (0.003) (0.066) (0.078) (0.170) (1.112) (0.286) (2.549) (1.709) Notes: Number of observations: 286. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Year and industry fixed effects included in the specification. Standard errors clustered at the industry level. Regressions for wage premium uses the inverse of the estimated wage premium variance as weights, and for the remaining outcomes, the weights are the inverse of the number of observations used to compute the industry-level variable. The excluded instruments used are China’s share of imports in Latin American countries and the high-income countries’ share of imports in Latin American countries. 47 TABLE 10. INDUSTRY-LEVEL FIRST DIFFERENCE IV REGRESSION OF TRADE EXPOSURE MEASURES ON LABOR-MARKET OUTCOMES USING EQUATION (2) (1) (2) (3) (4) (5) (6) (7) (8) (9) Outcome Log (employment level) Share of employment in the population (%) Emp. share of manuf. Log (hourly wage) Wage premium Informal share Years of schooling High-school share College share A. Chinese import penetrationt-1 0.003 0.010 -0.023 -0.472 0.305 -13.415 -2.242 -47.147 4.096 (0.134) (0.009) (0.119) (0.462) (0.290) (15.834) (2.902) (45.152) (4.156) B. Chinese import penetrationt-1 -0.037 0.010 -0.034 -0.153 0.266 -10.639 -1.191 -36.017 2.700 (0.126) (0.008) (0.106) (0.606) (0.310) (9.785) (1.755) (28.428) (5.789) ROW import penetrationt-1 0.032 0.000 0.008 -0.255 -0.258 -2.222 -0.841 -8.911 1.117 (0.039) (0.002) (0.020) (0.228) (0.560) (2.425) (0.593) (6.092) (1.329) C. Alternative import penetration Chinese alt. import penetrationt-1 -0.150 0.009 -0.064 0.722 0.312 -3.438 1.659 -6.847 -1.037 (0.307) (0.010) (0.118) (2.256) (0.522) (20.343) (6.894) (81.090) (15.603) ROW alt. import penetrationt-1 -0.126 -0.002 -0.029 1.059 -0.517 10.495 3.584 41.174 -4.914 (0.254) (0.010) (0.078) (2.264) (1.431) (30.191) (8.433) (105.145) (14.693) D. Adjusted import penetration Chinese adj. import penetrationt-1 -0.019 0.005 -0.017 -0.077 0.223 -5.455 -0.608 -18.461 1.382 (0.076) (0.004) (0.057) (0.287) (0.385) (5.663) (1.212) (17.746) (2.702) ROW adj. import penetrationt-1 0.011 0.000 0.003 -0.089 -0.095 -0.724 -0.288*** -2.938** 0.377 (0.009) (0.000) (0.006) (0.055) (0.208) (0.695) (0.108) (1.236) (0.501) E. Adjusted alternative imp. penet. Chinese adj. alt. import penetrationt-1 -0.100 0.005 -0.040 0.535 0.375 -0.624 1.372 1.269 -1.189 (0.079) (0.004) (0.043) (0.870) (0.917) (11.186) (2.777) (39.655) (7.214) ROW adj. alt. import penetrationt-1 -0.040 -0.001 -0.009 0.339 -0.268 3.428 1.151 13.403 -1.587 (0.043) (0.002) (0.026) (0.254) (0.782) (5.150) (1.060) (14.828) (2.719) Notes: Number of observations: 286. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Year fixed effects included in the specification. Standard errors clustered at the industry level. Regressions for wage premium uses the inverse of the estimated wage premium variance as weights, and for the remaining outcomes, the weights are the inverse of the number of observations used to compute the industrylevel variable. The excluded instruments used in IV estimates are the first difference of China’s share of imports in Latin American countries and of the high-income countries’ share of imports in Latin American countries. 48 TABLE 11. INDUSTRY-LEVEL FIRST DIFFERENCE IV REGRESSION OF TRADE EXPOSURE MEASURES ON LABOR-MARKET OUTCOMES USING EQUATION (2) AND SECOND SET OF INSTRUMENTS (1) (2) (3) (4) (5) (6) (7) (8) (9) Outcome Log (employment level) Share of employment in the population (%) Emp. share of manuf. Log (hourly wage) Wage premium Informal share Years of schooling High-school share College share A. Chinese import penetrationt-1 0.088 0.011 0.167* -0.023 0.053 0.261 -0.049 0.392 0.189 (0.064) (0.007) (0.099) (0.086) (0.042) (1.122) (0.174) (1.755) (1.274) B. Chinese import penetrationt-1 0.080 0.009 0.113 -0.076 0.135*** 1.877 0.028 -5.465 -0.614 (0.096) (0.009) (0.117) (0.119) (0.036) (3.525) (0.432) (6.108) (1.825) ROW import penetrationt-1 -0.007 -0.001 -0.013 -0.028 0.053* 0.352 0.009 -2.224 -0.291 (0.031) (0.003) (0.039) (0.063) (0.027) (1.527) (0.205) (2.850) (0.838) C. Alternative import penetration Chinese alt. import penetrationt-1 0.096 0.011 0.114 0.048 0.119*** 1.321 0.287 -0.284 -1.273 (0.074) (0.007) (0.093) (0.081) (0.045) (1.976) (0.213) (3.139) (2.420) ROW alt. import penetrationt-1 -0.010 -0.001 -0.034 0.069 0.063 -0.108 0.316 0.336 -1.582 (0.031) (0.004) (0.062) (0.081) (0.041) (1.929) (0.202) (3.418) (1.474) D. Adjusted import penetration Chinese adj. import penetrationt-1 0.061* 0.007* 0.084 0.001 0.064*** 0.764 0.035 -0.161 -0.076 (0.036) (0.004) (0.054) (0.053) (0.017) (0.995) (0.116) (0.868) (0.610) ROW adj. import penetrationt-1 -0.006 -0.000 0.005 -0.006 -0.027 -0.291 -0.013 0.008 -0.029 (0.006) (0.001) (0.013) (0.007) (0.042) (0.202) (0.027) (0.528) (0.129) E. Adjusted alternative imp. penet. Chinese adj. alt. import penetrationt-1 0.068 0.008* 0.088 0.000 0.068*** 0.981 0.043 -0.384 -0.089 (0.044) (0.004) (0.065) (0.050) (0.019) (1.093) (0.106) (0.906) (0.696) ROW adj. alt. import penetrationt-1 -0.003 0.000 0.024 -0.002 -0.024 -0.624 0.010 0.364 -0.223 (0.018) (0.001) (0.031) (0.024) (0.035) (0.617) (0.066) (1.111) (0.415) Notes: Number of observations: 286. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Year fixed effects included in the specification. Standard errors clustered at the industry level. Regressions for wage premium uses the inverse of the estimated wage premium variance as weights, and for the remaining outcomes, the weights are the inverse of the number of observations used to compute the industrylevel variable. The excluded instruments used in all IV estimates are Chinese import penet.1998Real exch. ratet-1 and ROW import penet.1998Real exch. ratet-1. 49 TABLE 12. INDUSTRY-LEVEL REGRESSION OF IMPORT PENETRATION MEASURES ON THE LOG (EMPLOYMENT) USING EQUATION (3) Regressors OLS-1 OLS-2 OLS-3 OLS-4 OLS-5 IV-1 IV-2 IV-3 IV-4 IV-5 Chinese import penetrationt-1 -0.031 -0.026 -0.065* -0.032 -0.025 0.056 0.076 0.225 0.083 0.216 (0.033) (0.023) (0.032) (0.027) (0.024) (0.189) (0.332) (2.054) (0.199) (3.781) ROW import penetrationt-1 -0.010** -0.010** -0.008** -0.007 -0.010** 0.070* 0.037 0.056 -0.056 0.184 (0.004) (0.004) (0.004) (0.004) (0.005) (0.040) (0.371) (0.342) (0.076) (4.570) Chinese market share abroadt-1 0.004 0.009 (0.018) (0.048) Effective applied tarifft-1 -0.005 0.005 (0.008) (0.081) Upstream Chinese imp. penet.t-1 0.109 -0.145 (0.068) (2.072) Upstream ROW imp. penet.t-1 0.005 0.023 (0.008) (0.094) Chinese imp. penet.t-1 × NMEt 0.021 -0.056 (0.013) (0.156) ROW imp. penet.t-1 × NMEt -0.007** 0.005 (0.003) (0.058) Chinese imp. penet.t-1 × L. intensivetj -0.028 -0.446 (0.040) (7.502) ROW imp. penet.t-1 × L. intensivetj 0.031 -0.775 (0.037) (14.028) Notes: Number of observations: 286. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Year and industry fixed effects included in the specification. Standard errors clustered at the industry level. Regressions weights are the inverse of the number of observations used to compute the industry-level variable. The excluded instruments used in all IV estimates are China’s share of imports in Latin American countries and the high-income countries’ share of imports in Latin American countries. 50 TABLE 13. INDUSTRY-LEVEL REGRESSION OF IMPORT PENETRATION MEASURES ON THE SHARE OF EMPLOYMENT IN THE POPULATION USING EQUATION (3) Regressors OLS-1 OLS-2 OLS-3 OLS-4 OLS-5 IV-1 IV-2 IV-3 IV-4 IV-5 Chinese import penetrationt-1 -0.001 0.003 -0.004 0.004 0.003 0.017 0.006 0.017 0.023 0.031 (0.002) (0.002) (0.003) (0.003) (0.002) (0.011) (0.071) (0.136) (0.024) (0.450) ROW import penetrationt-1 -0.000 -0.000* -0.000 -0.000 -0.000 0.004 -0.009 0.006 -0.006 0.021 (0.000) (0.000) (0.000) (0.000) (0.000) (0.004) (0.084) (0.022) (0.008) (0.541) Chinese market share abroadt-1 0.003* -0.000 (0.002) (0.003) Effective applied tarifft-1 0.000 -0.002 (0.001) (0.017) Upstream Chinese imp. penet.t-1 0.017*** 0.004 (0.004) (0.142) Upstream ROW imp. penet.t-1 0.000 0.002 (0.001) (0.006) Chinese imp. penet.t-1 × NMEt -0.001 -0.010 (0.001) (0.014) ROW imp. penet.t-1 × NMEt -0.000 0.001 (0.000) (0.005) Chinese imp. penet.t-1 × L. intensivetj 0.002 -0.032 (0.006) (0.882) ROW imp. penet.t-1 × L. intensivetj -0.012 -0.057 (0.010) (1.645) Notes: Number of observations: 286. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Year and industry fixed effects included in the specification. Standard errors clustered at the industry level. Regressions weights are the inverse of the number of observations used to compute the industry-level variable. The excluded instruments used in all IV estimates are China’s share of imports in Latin American countries and the high-income countries’ share of imports in Latin American countries. 51 TABLE 14. INDUSTRY-LEVEL REGRESSION OF IMPORT PENETRATION MEASURES ON THE MANUFACTURING SHARE OF EMPLOYMENT USING EQUATION (3) Regressors OLS-1 OLS-2 OLS-3 OLS-4 OLS-5 IV-1 IV-2 IV-3 IV-4 IV-5 Chinese import penetrationt-1 -0.006 0.008 -0.065** 0.019 0.027 0.015 0.293 0.544 0.160 0.535 (0.023) (0.023) (0.029) (0.032) (0.027) (0.137) (1.462) (3.505) (0.230) (10.629) ROW import penetrationt-1 -0.008** -0.010** -0.005 -0.009* -0.008** 0.024 0.276 0.061 -0.056 0.463 (0.004) (0.004) (0.003) (0.005) (0.003) (0.060) (1.782) (0.624) (0.069) (12.760) Chinese market share abroadt-1 0.012 0.017 (0.020) (0.035) Effective applied tarifft-1 -0.016 0.042 (0.012) (0.353) Upstream Chinese imp. penet.t-1 0.201*** -0.498 (0.071) (3.403) Upstream ROW imp. penet.t-1 0.005 0.028 (0.017) (0.179) Chinese imp. penet.t-1 × NMEt -0.008 -0.108 (0.015) (0.143) ROW imp. penet.t-1 × NMEt -0.000 -0.001 (0.004) (0.051) Chinese imp. penet.t-1 × L. intensivetj -0.330*** -1.409 (0.076) (20.874) ROW imp. penet.t-1 × L. intensivetj 0.047 -2.177 (0.158) (39.107) Notes: Number of observations: 286. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Year and industry fixed effects included in the specification. Standard errors clustered at the industry level. Regressions weights are the inverse of the number of observations used to compute the industry-level variable. The excluded instruments used in all IV estimates are China’s share of imports in Latin American countries and the high-income countries’ share of imports in Latin American countries. 52 TABLE 15. INDUSTRY-LEVEL REGRESSION OF IMPORT PENETRATION MEASURES ON THE LOG (HOURLY WAGE) USING EQUATION (3) Regressors OLS-1 OLS-2 OLS-3 OLS-4 OLS-5 IV-1 IV-2 IV-3 IV-4 IV-5 Chinese import penetrationt-1 0.009 -0.004 0.027 0.016 -0.005 0.101 -0.187 -1.345 -0.132 0.824 (0.027) (0.023) (0.030) (0.051) (0.024) (0.221) (0.633) (8.091) (0.142) (26.806) ROW import penetrationt-1 0.000 0.000 0.000 -0.000 0.001 -0.006 -0.163 -0.205 -0.035 1.049 (0.005) (0.005) (0.004) (0.004) (0.005) (0.040) (0.750) (1.391) (0.069) (32.138) Chinese market share abroadt-1 -0.010 -0.034 (0.010) (0.045) Effective applied tarifft-1 -0.014 -0.048 (0.013) (0.136) Upstream Chinese imp. penet.t-1 0.008 1.447 (0.091) (8.030) Upstream ROW imp. penet.t-1 0.064*** 0.002 (0.023) (0.405) Chinese imp. penet.t-1 × NMEt -0.021 0.086 (0.036) (0.129) ROW imp. penet.t-1 × NMEt 0.002 0.019 (0.003) (0.030) Chinese imp. penet.t-1 × L. intensivetj 0.032 -1.620 (0.025) (52.221) ROW imp. penet.t-1 × L. intensivetj 0.018 -3.037 (0.031) (97.509) Notes: Number of observations: 286. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Year and industry fixed effects included in the specification. Standard errors clustered at the industry level. Regressions weights are the inverse of the number of observations used to compute the industry-level variable. The excluded instruments used in all IV estimates are China’s share of imports in Latin American countries and the high-income countries’ share of imports in Latin American countries. 53 TABLE 16. INDUSTRY-LEVEL REGRESSION OF IMPORT PENETRATION MEASURES ON THE INTERINDUSTRY WAGE PREMIUM USING EQUATION (3) Regressors OLS-1 OLS-2 OLS-3 OLS-4 OLS-5 IV-1 IV-2 IV-3 IV-4 IV-5 Chinese import penetrationt-1 -0.011* 0.004 0.014 -0.002 -0.000 -0.005 0.005 0.036 -0.033 -0.010 (0.006) (0.007) (0.014) (0.007) (0.004) (0.130) (0.029) (0.029) (0.067) (0.026) ROW import penetrationt-1 0.002 0.004* 0.001 0.004 0.002 -0.017 -0.028 -0.030 0.008 -0.017 (0.003) (0.002) (0.002) (0.003) (0.002) (0.042) (0.039) (0.029) (0.065) (0.028) Chinese market share abroadt-1 0.010*** 0.006 (0.003) (0.045) Effective applied tarifft-1 -0.001 0.003 (0.002) (0.007) Upstream Chinese imp. penet.t-1 -0.026 -0.091** (0.027) (0.046) Upstream ROW imp. penet.t-1 0.012** 0.044 (0.005) (0.031) Chinese imp. penet.t-1 × NMEt 0.003 0.030 (0.007) (0.048) ROW imp. penet.t-1 × NMEt 0.000 0.010 (0.002) (0.020) Chinese imp. penet.t-1 × L. intensivetj 0.016 0.032 (0.013) (0.035) ROW imp. penet.t-1 × L. intensivetj -0.011 -0.024 (0.010) (0.049) Notes: Number of observations: 286. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Year and industry fixed effects included in the specification. Standard errors clustered at the industry level. Regressions weights are the inverse of the estimated wage premium variance. The excluded instruments used in all IV estimates are China’s share of imports in Latin American countries and the high-income countries’ share of imports in Latin American countries. 60 TABLE 23. WORKER-LEVEL OLS ESTIMATES OF THE EFFECTS OF INDUSTRY-LEVEL IMPORT PENETRATION ON THE INFORMAL EMPLOYMENT STATUS INDICATOR USING EQUATION (5) Regressors (1) (2) (3) (4) (5) (6) (7) Chinese import penetrationt-1 0.010 -0.013 -0.046 -0.026 -0.002 -0.040** -0.031* (0.016) (0.011) (0.036) (0.022) (0.007) (0.016) (0.015) ROW import penetrationt-1 -0.020*** -0.013** -0.015** -0.016** -0.007 -0.026*** -0.026*** (0.006) (0.005) (0.007) (0.007) (0.005) (0.008) (0.009) Chinese market share abroadt-1 -0.029** (0.013) Effective applied tarifft-1 -0.014*** (0.004) Upstream Chinese imp. penet.t-1 0.024 (0.070) Upstream ROW imp. penet.t-1 -0.046** (0.021) Chinese imp. penet.t-1 × NMEt -0.019 (0.017) ROW imp. penet.t-1 × NMEt 0.012*** (0.003) Chinese imp. penet.t-1 × L. int.j -0.095*** (0.029) ROW imp. penet.t-1 × L. int.j 0.006 (0.034) Chinese imp. penet.t-1 × Manuf.s 0.007 (0.004) ROW imp. penet.t-1 × Manuf.s 0.001 (0.001) Chinese imp. penet.t-1 × Coastals -0.001 (0.003) ROW imp. penet.t-1 × Coastals 0.000 (0.001) Notes: Number of observations: 669,9664. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors clustered at the industry level. Sample weights from PNAD/Census used. Workers’ observable characteristics; and industry, state, and year fixed effects included in the estimated model. 61 TABLE 24. WORKER-LEVEL IV ESTIMATES OF THE EFFECTS OF INDUSTRY-LEVEL IMPORT PENETRATION ON THE LOG (HOURLY WAGE) USING EQUATION (5) Regressors (1) (2) (3) (4) (5) (6) (7) (8) Chinese import penetrationt-1 0.016 -0.038 0.007 0.029 -0.312 -0.015 0.010 0.008 (0.015) (0.085) (0.024) (0.033) (0.816) (0.029) (0.015) (0.020) ROW import penetrationt-1 0.006 0.010 -0.006 0.004 0.230 -0.015 0.012 0.005 (0.023) (0.029) (0.049) (0.034) (0.558) (0.041) (0.023) (0.023) Chinese market share abroadt-1 0.021 (0.033) Effective applied tarifft-1 0.004 (0.009) Upstream Chinese imp. penet.t-1 -0.043 (0.058) Upstream ROW imp. penet.t-1 0.014 (0.039) Chinese imp. penet.t-1 × NMEt 0.228 (0.581) ROW imp. penet.t-1 × NMEt 0.066 (0.159) Chinese imp. penet.t-1 × L. int.j 0.046 (0.055) ROW imp. penet.t-1 × L. int.j 0.013 (0.052) Chinese imp. penet.t-1 × Manuf.s 0.008 (0.025) ROW imp. penet.t-1 × Manuf.s 0.021 (0.018) Chinese imp. penet.t-1 × Coastals 0.008 (0.009) ROW imp. penet.t-1 × Coastals 0.001 (0.008) Endogeneity test 0.0707 0.541 0.149 0.0668 4.611 3.775 8.578* 1.970 [0.790] [0.763] [0.928] [0.967] [0.330] [0.437] [0.072] [0.741] Notes: Number of observations: 669,9664. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors clustered at the industry level. Sample weights from PNAD/Census used. Workers’ observable characteristics; and industry, state, and year fixed effects included in the estimated model. The excluded instruments used in all IV estimates are China’s share of imports in Latin American countries and the high-income countries’ share of imports in Latin American countries. 62 TABLE 25. WORKER-LEVEL IV ESTIMATES OF THE EFFECTS OF INDUSTRY-LEVEL IMPORT PENETRATION ON THE INFORMAL EMPLOYMENT STATUS INDICATOR USING EQUATION (5) Regressors (1) (2) (3) (4) (5) (6) (7) (8) Chinese import penetrationt-1 -0.062* 0.066 -0.022 -0.080 0.292 0.013 -0.068** -0.055 (0.037) (0.130) (0.052) (0.098) (0.688) (0.056) (0.030) (0.038) ROW import penetrationt-1 -0.034 -0.044 0.021 -0.023 -0.205 0.016 -0.035 -0.034 (0.060) (0.041) (0.121) (0.067) (0.445) (0.078) (0.060) (0.058) Chinese market share abroadt-1 -0.050 (0.044) Effective applied tarifft-1 -0.019 (0.020) Upstream Chinese imp. penet.t-1 0.073 (0.206) Upstream ROW imp. penet.t-1 -0.045 (0.078) Chinese imp. penet.t-1 × NMEt -0.236 (0.490) ROW imp. penet.t-1 × NMEt -0.041 (0.127) Chinese imp. penet.t-1 × L. int.j -0.111 (0.117) ROW imp. penet.t-1 × L. int.j -0.030 (0.100) Chinese imp. penet.t-1 × Manuf.s 0.006 (0.012) ROW imp. penet.t-1 × Manuf.s -0.007 (0.010) Chinese imp. penet.t-1 × Coastals -0.007 (0.006) ROW imp. penet.t-1 × Coastals -0.001 (0.004) Endogeneity test 0.151 0.243 0.312 0.478 6.517 1.685 6.297 7.083 [0.698] [0.886] [0.856] [0.787] [0.164] [0.793] [0.178] [0.132] Notes: Number of observations: 669,9664. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors clustered at the industry level. Sample weights from PNAD/Census used. Workers’ observable characteristics; and industry, state, and year fixed effects included in the estimated model. The excluded instruments used in all IV estimates are China’s share of imports in Latin American countries and the high-income countries’ share of imports in Latin American countries. 63 TABLE 26. WORKER-LEVEL OLS ESTIMATES OF THE EFFECTS OF INDUSTRY-LEVEL IMPORT PENETRATION ON THE LOG (HOURLY WAGE) USING EQUATION (5) AND STATE-INDUSTRY AND YEAR FIXED EFFECTS Regressors (1) (2) (3) (4) (5) (6) (7) Chinese import penetrationt-1 0.001 0.010* 0.019 0.006 0.002 0.020*** 0.025** (0.005) (0.005) (0.016) (0.006) (0.003) (0.005) (0.010) ROW import penetrationt-1 0.008** 0.007** 0.005* 0.007** 0.003 0.009*** 0.011** (0.003) (0.003) (0.003) (0.003) (0.002) (0.003) (0.005) Chinese market share abroadt-1 0.008** (0.004) Effective applied tarifft-1 0.002* (0.001) Upstream Chinese imp. penet.t-1 -0.012 (0.029) Upstream ROW imp. penet.t-1 0.017** (0.006) Chinese imp. penet.t-1 × NMEt 0.008* (0.004) ROW imp. penet.t-1 × NMEt -0.003** (0.001) Chinese imp. penet.t-1 × L. int.j 0.029** (0.010) ROW imp. penet.t-1 × L. int.j 0.004 (0.009) Chinese imp. penet.t-1 × Manuf.s -0.014 (0.009) ROW imp. penet.t-1 × Manuf.s -0.000 (0.005) Chinese imp. penet.t-1 × Coastals -0.018* (0.011) ROW imp. penet.t-1 × Coastals -0.002 (0.006) Notes: Number of observations: 669,9664. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors clustered at the industry level. Sample weights from PNAD/Census used. Workers’ observable characteristics; and state-industry, and year fixed effects included in the estimated model. 64 TABLE 27. WORKER-LEVEL OLS ESTIMATES OF THE EFFECTS OF INDUSTRY-LEVEL IMPORT PENETRATION ON THE INFORMAL EMPLOYMENT STATUS INDICATOR USING EQUATION (5) AND STATE-INDUSTRY AND YEAR FIXED EFFECTS Regressors (1) (2) (3) (4) (5) (6) (7) Chinese import penetrationt-1 0.003 -0.018 -0.051 -0.029 -0.007 -0.053*** -0.045** (0.017) (0.011) (0.034) (0.022) (0.006) (0.018) (0.020) ROW import penetrationt-1 -0.021*** -0.013** -0.015** -0.017** -0.006 -0.032*** -0.032*** (0.006) (0.005) (0.007) (0.007) (0.005) (0.009) (0.009) Chinese market share abroadt-1 -0.029** (0.012) Effective applied tarifft-1 -0.015*** (0.004) Upstream Chinese imp. penet.t-1 0.019 (0.065) Upstream ROW imp. penet.t-1 -0.049** (0.020) Chinese imp. penet.t-1 × NMEt -0.022 (0.017) ROW imp. penet.t-1 × NMEt 0.012*** (0.003) Chinese imp. penet.t-1 × L. int.j -0.095*** (0.028) ROW imp. penet.t-1 × L. int.j 0.001 (0.034) Chinese imp. penet.t-1 × Manuf.s 0.028*** (0.007) ROW imp. penet.t-1 × Manuf.s 0.010** (0.005) Chinese imp. penet.t-1 × Coastals 0.009 (0.007) ROW imp. penet.t-1 × Coastals 0.007** (0.003) Notes: Number of observations: 669,9664. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors clustered at the industry level. Sample weights from PNAD/Census used. Workers’ observable characteristics; and state-industry, and year fixed effects included in the estimated model. 65 TABLE 28. WORKER-LEVEL IV ESTIMATES OF THE EFFECTS OF INDUSTRY-LEVEL IMPORT PENETRATION ON THE LOG (HOURLY WAGE) USING EQUATION (5) AND STATE-INDUSTRY AND YEAR FIXED EFFECTS Regressors (1) (2) (3) (4) (5) (6) (7) (8) Chinese import penetrationt-1 0.010 0.217 0.010 0.005 -0.026 -0.059 -0.027 0.000 (0.031) (0.454) (0.030) (0.009) (0.033) (0.209) (0.019) (0.031) ROW import penetrationt-1 0.036* 0.022 0.036* 0.031*** -0.000 0.086 -0.004 0.033 (0.021) (0.035) (0.019) (0.010) (0.017) (0.324) (0.021) (0.022) Chinese market share abroadt-1 -0.024 (0.048) Effective applied tarifft-1 -0.000 (0.004) Upstream Chinese imp. penet.t-1 -0.081*** (0.031) Upstream ROW imp. penet.t-1 -0.020* (0.010) Chinese imp. penet.t-1 × NMEt 0.036 (0.043) ROW imp. penet.t-1 × NMEt 0.038*** (0.012) Chinese imp. penet.t-1 × L. int.j 0.091 (0.338) ROW imp. penet.t-1 × L. int.j 0.084 (0.549) Chinese imp. penet.t-1 × Manuf.s 0.043 (0.031) ROW imp. penet.t-1 × Manuf.s 0.043*** (0.016) Chinese imp. penet.t-1 × Coastals 0.012 (0.008) ROW imp. penet.t-1 × Coastals 0.004 (0.005) Endogeneity test 2.945 4.790 3.493 3.722 11.09 6.845 6.894 5.606 [0.086] [0.091] [0.174] [0.156] [0.026] [0.144] [0.142] [0.231] Notes: Number of observations: 669,9664. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors clustered at the industry level. Sample weights from PNAD/Census used. Workers’ observable characteristics; and state-industry, and year fixed effects included in the estimated model. The excluded instruments used in all IV estimates are China’s share of imports in Latin American countries and the high-income countries’ share of imports in Latin American countries. 66 TABLE 29. WORKER-LEVEL IV ESTIMATES OF THE EFFECTS OF INDUSTRY-LEVEL IMPORT PENETRATION ON THE INFORMAL EMPLOYMENT STATUS INDICATOR AND STATE-INDUSTRY AND YEAR FIXED EFFECTS Regressors (1) (2) (3) (4) (5) (6) (7) (8) Chinese import penetrationt-1 -0.000 -0.073 -0.000 0.001 0.089 0.044 0.011 0.010 (0.019) (0.195) (0.022) (0.009) (0.058) (0.175) (0.013) (0.020) ROW import penetrationt-1 -0.015 -0.010 -0.017 -0.010 0.008 -0.062 0.004 -0.013 (0.012) (0.013) (0.011) (0.006) (0.012) (0.279) (0.012) (0.012) Chinese market share abroadt-1 0.008 (0.021) Effective applied tarifft-1 -0.001 (0.003) Upstream Chinese imp. penet.t-1 0.030 (0.021) Upstream ROW imp. penet.t-1 0.003 (0.005) Chinese imp. penet.t-1 × NMEt -0.088 (0.070) ROW imp. penet.t-1 × NMEt -0.023** (0.012) Chinese imp. penet.t-1 × L. int.j -0.059 (0.290) ROW imp. penet.t-1 × L. int.j -0.093 (0.476) Chinese imp. penet.t-1 × Manuf.s -0.012 (0.016) ROW imp. penet.t-1 × Manuf.s -0.020*** (0.008) Chinese imp. penet.t-1 × Coastals -0.012** (0.005) ROW imp. penet.t-1 × Coastals -0.001 (0.001) Endogeneity test 0.806 1.856 2.996 2.637 6.499 5.127 5.302 6.086 [0.369] [0.395] [0.224] [0.267] [0.165] [0.274] [0.258] [0.193] Notes: Number of observations: 669,9664. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors clustered at the industry level. Sample weights from PNAD/Census used. Workers’ observable characteristics; and state-industry, and year fixed effects included in the estimated model. The excluded instruments used in all IV estimates are China’s share of imports in Latin American countries and the high-income countries’ share of imports in Latin American countries. 67 TABLE 30. WORKER-LEVEL OLS ESTIMATES OF THE EFFECTS OF STATE-INDUSTRY IMPORT PENETRATION ON THE LOG (HOURLY WAGE) USING EQUATION (5) Regressors (1) (2) (3) (4) (5) (6) (7) (8) Chinese import penetrationt-1 -0.006 0.000 -0.011 -0.013 0.015 0.004 -0.010 -0.037 (0.017) (0.021) (0.024) (0.017) (0.028) (0.042) (0.018) (0.066) ROW import penetrationt-1 0.742*** 0.748*** 0.139 0.769*** -0.089 0.091 0.686*** 0.861*** (0.102) (0.101) (0.249) (0.109) (0.186) (0.222) (0.186) (0.226) Chinese market share abroadt-1 -0.001 (0.002) Effective applied tarifft-1 0.004 (0.003) Upstream Chinese imp. penet.t-1 -0.015** (0.006) Upstream ROW imp. penet.t-1 -0.004** (0.002) Chinese imp. penet.t-1 × NMEt -0.023 (0.304) ROW imp. penet.t-1 × NMEt 0.844*** (0.240) Chinese imp. penet.t-1 × L. int.j -0.050 (0.481) ROW imp. penet.t-1 × L. int.j 0.932*** (0.198) Chinese imp. penet.t-1 × Manuf.s 0.062 (0.119) ROW imp. penet.t-1 × Manuf.s 0.057 (0.184) Chinese imp. penet.t-1 × Coastals 0.004 (0.074) ROW imp. penet.t-1 × Coastals -0.159 (0.289) Notes: Number of observations: 669,9664. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors clustered at the industry level. Sample weights from PNAD/Census used. Workers’ observable characteristics; and state, industry, and year fixed effects included in the estimated model. 68 TABLE 31. WORKER-LEVEL OLS ESTIMATES OF THE EFFECTS OF STATE-INDUSTRY-LEVEL IMPORT PENETRATION ON THE INFORMAL EMPLOYMENT STATUS INDICATOR USING EQUATION (5) Regressors (1) (2) (3) (4) (5) (6) (7) (8) Chinese import penetrationt-1 -0.028 -0.033** -0.028 -0.030 0.188** 0.013 -0.036* -0.029 (0.204) (0.014) (0.036) (0.019) (0.081) (0.050) (0.019) (0.038) ROW import penetrationt-1 0.167** 0.163** 0.278 0.164** -0.137 0.141 0.459* 0.252** (0.074) (0.072) (0.188) (0.077) (0.184) (0.091) (0.229) (0.107) Chinese market share abroadt-1 0.001 (0.002) Effective applied tarifft-1 -0.017*** (0.004) Upstream Chinese imp. penet.t-1 0.010 (0.008) Upstream ROW imp. penet.t-1 -0.000 (0.002) Chinese imp. penet.t-1 × NMEt -0.224** (0.091) ROW imp. penet.t-1 × NMEt 0.314 (0.184) Chinese imp. penet.t-1 × L. int.j -0.046 (0.052) ROW imp. penet.t-1 × L. int.j 0.037 (0.103) Chinese imp. penet.t-1 × Manuf.s -0.062 (0.062) ROW imp. penet.t-1 × Manuf.s -0.380 (0.241) Chinese imp. penet.t-1 × Coastals 0.002 (0.031) ROW imp. penet.t-1 × Coastals -0.113 (0.145) Notes: Number of observations: 669,9664. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors clustered at the industry level. Sample weights from PNAD/Census used. Workers’ observable characteristics; and state, industry, and year fixed effects included in the estimated model. 69 TABLE 32. WORKER-LEVEL IV ESTIMATES OF THE EFFECTS OF STATE-INDUSTRY-LEVEL IMPORT PENETRATION ON THE LOG (HOURLY WAGE) USING EQUATION (5) Regressors (1) (2) (3) (4) (5) (6) (7) (8) Chinese import penetrationt-1 2.430 2.043 0.731 2.402 1.462 0.173 0.896 5.281 (2.008) (1.597) (0.582) (1.780) (2.836) (0.121) (1.049) (4.216) ROW import penetrationt-1 -0.021 -1.512 -3.619* -1.229 -2.341 -1.102 1.962 -1.272 (2.657) (2.793) (2.112) (1.856) (4.113) (3.418) (1.698) (2.432) Chinese market share abroadt-1 -0.009** (0.005) Effective applied tarifft-1 0.004 (0.004) Upstream Chinese imp. penet.t-1 -0.025 (0.016) Upstream ROW imp. penet.t-1 0.005 (0.006) Chinese imp. penet.t-1 × NMEt -1.475 (2.898) ROW imp. penet.t-1 × NMEt 2.387 (4.175) Chinese imp. penet.t-1 × L. int.j -0.184 (0.120) ROW imp. penet.t-1 × L. int.j 0.052* (0.027) Chinese imp. penet.t-1 × Manuf.s 3.360 (3.487) ROW imp. penet.t-1 × Manuf.s -1.807 (2.986) Chinese imp. penet.t-1 × Coastals -3.864 (4.013) ROW imp. penet.t-1 × Coastals 2.022 (3.414) Endogeneity test 3.671* 6.680** 5.026* 5.920* 10.08** 7.926* 6.859 7.511 [0.0554] [0.035] [0.081] [0.052] [0.039] [0.094] [0.144] [0.111] Notes: Number of observations: 669,9664. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. Standard errors clustered at the industry level. Sample weights from PNAD/Census used. Workers’ observable characteristics; and state, industry, and year fixed effects included in the estimated model. The excluded instruments used in all IV estimates are the state-industry level Chinese share of imports in Latin American countries and the state-industry level high-income countries’ share of imports in Latin American countries.