The World Bank Economic Review, 36(1), 2022, 141–170 https://doi.org10.1093/wber/lhab023 Article Unfolding Trade Effect in Two Margins of Informality. Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 The Peruvian Case Camila Cisneros-Acevedo Abstract This paper studies the effect of an increase in import competition on informality along two margins. It considers the extensive margin, where workers are hired by unregistered employers, and the intensive margin, where even though jobs are carried out by registered firms, employees are off the books. Peru’s relentless informal employment and its unprecedented trade-driven growth provides an ideal case study. Using a rich household survey, this study finds that exposure to trade impacts on informality through two competing and contrasting mechanisms. On the one hand, extensive-informal employment declines as unregistered employers shrink or exit due to their low productivity. On the other hand, intensive-informal employment rises as registered employers reduce costs by hiring informal workers. Furthermore, results suggest that the intensive margin drives the overall effect. Hence, trade liberalization increases informality. JEL classification: F16, F14, F66, J46 Keywords: international trade, globalization, informal labor market, Peru 1. Introduction The International Labour Organization (ILO) defines informality as the sum of employment in unregis- tered firms and the portion of employment in registered firms that does not comply with labor legisla- tion.1,2 In this way, the ILO acknowledges the existence of two types of informal workers. Surprisingly, this distinction has been mostly overlooked in the literature. This paper studies the effect of trade lib- eralization on informal labor by distinguishing two crucial margins of informality. Following Ponczek and Ulyssea (2021), the extensive margin refers to workers hired by firms that are not legally registered Camila Cisneros-Acevedo is Assistant Professor at the University of Tübingen, Tübingen, Germany; her email address is camila.cisneros-acevedo@uni-tuebingen.de. The author thanks Alejandro Riaño and Facundo Albornoz for their invaluable guidance. She also thanks seminar participants at the CSAE PhD conference (Oxford), The Shadow Economy, Tax Evasion and Informal Labor conference (Warsaw), 16th GEP Postgraduate Conference (Nottingham), Informality and Development, conference in honor of Elinor Ostrom (Indiana), and 15th ETSG conference (Helsinki). A supplementary online appendix is available with this article at The World Bank Economic Review website. 1 As Kanbur (2009) points out, informality can be described as a “conceptual incoherence” since the literature has not reached consensus regarding its definition. Loosely, informal labor refers to jobs that do not comply with taxation nor the regulations that ensure protection for workers for such as paid holidays, parental leave, and retirement. 2 The informal sector accounts for over 50 percent in Latin America, over 58 percent in Asia, and 76 percent of total employment in Africa (ILO 2018; WB 2018). © The Author(s) 2021. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development / THE WORLD BANK. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com 142 Cisneros-Acevedo with the tax collection agency, and the intensive margin refers to workers employed “off the books” in registered firms. The growing literature addressing the effect of trade liberalization on informality has treated infor- mality as a binary decision to comply or not with taxes and regulations (Goldberg and Pavcnik 2003, 2007; Menezes-Filho and Muendler 2011; Bosch, Goni-Pacchioni, and Maloney 2012; Paz 2014; Dix- Carneiro and Kovak 2015, 2017; McCaig and Pavcnik 2015, 2018; Pavcnik 2017; Dix-Carneiro, Soares, Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 and Ulyssea 2018; Cruces, Porto, and Viollaz 2018; Ulyssea 2020; Ponczek and Ulyssea 2021).3 Papers looking at firm informality focus on the extensive margin of informality and those looking at worker informality combine both margins. This study builds on this literature and takes a novel approach as it disentangles these two margins of informality when studying a reduction in tariffs. In doing so, it can analyze two different economic mechanisms in isolation. Putting together these margins under the tag of “informal labor” obscures the nature of the effect that trade has on informality. A reduction in tariffs affects each margin of informality through two different channels. On the one hand, a reduction in tariffs forces the least productive firms to exit the market. Since unregistered employers are less productive than registered ones (Maloney 2004; Perry et al. 2007; Díaz 2014; Ponczek and Ulyssea 2021), informal firms that hire only these types of workers are not efficient enough to survive and are forced to exit or shrink. In this way, trade liberalization reduces extensive- informal employment in the economy. On the other hand, as a tariff reduction intensifies import compe- titions, firms may look for ways to cut costs. Since informal workers are “cheaper” than formal workers, because of taxes associated with formality, a reduction in tariffs could translate into an increase in the fraction of informal workers as an adjustment mechanism to remain competitive. I argue that before the cut in tariffs, firms restrain themselves from hiring informal workers, despite being cheaper, because hir- ing them comes with a risk of, for example, paying a fine in case of getting caught. This risk is worth taking only after import competition is increased, as the alternative could be bankruptcy. In this way, intensive-informal employment escalates with trade liberalization. Shedding light on the mechanisms at play on the effect that trade liberalization has on informal labor has important policy implications. If a tariff cut reduces on its own the extensive margin of informality, policies that aim to lessen informal employment in unregistered firms are not required. However, if the intensive margin increases when tariffs fall, it would be wise to enforce policies that focus on preventing formal firms from hiring informal workers. In recent decades, the reduction of informality has been one of the core goals for policy makers, especially in developing countries. Due to lack of resources, countries cannot always implement policies that attack informal labor from different angles. This paper provides guidelines on a more efficient allocation of resources. For instance, in the context of trade liberalization, inspection enhancement of labor law compliance among registered firms would be a more appropriate policy than reducing the administrative costs associated with formalization. Using a very rich household survey on the Peruvian manufacturing sector that identifies worker infor- mality, this study distinguishes the intensive and the extensive margins of informality. Because the data set is not a firm survey, I can only speculate as to what drives firms decisions. Regardless, it is possible to examine the outcome of these decisions on employment. Furthermore, since the household survey is representative of the Peruvian population, the data allow us to observe employment in all types of firms, including firms operating under the radar. Peru provides a very suitable context to study the effect of trade liberalization on informality. Approx- imately 70 percent of total employment in manufacturing was informal in 2014, the highest in the region. 3 In the past, the relationship between informality and trade seemed weak. Bosch, Goni-Pacchioni, and Maloney (2012) establish that trade liberalization explains, to a very small extent, the increase in the size of the informal sector in Brazilian metropolitan labor markets. Goldberg and Pavcnik (2003) study the effect that an increase in foreign competition has on Brazil and Colombia. They conclude that trade policy is of second order of relevance to the effect on informal employment. The World Bank Economic Review 143 At the same time, Peru exhibited a practically unprecedented annual growth rate of 6 percent for more than a decade.4 Furthermore, Peru was also the most open country in Latin America in 2014, when its average applied tariff was less than 5 percent while the norm in the region was 15 percent. This paper examines firms’ labor outcomes and studies the effect on informal employment in both margins when tariffs go down. The empirical strategy is two-fold. First, it studies the effect at the individual level on the probability of being hired as an informal employee in each margin. Second, it examines how Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 employment composition shifts at the industry level when tariffs fall. Results are robust to these two different approaches. This study finds that when tariffs fall, both the probability of hiring informally in the extensive margin and the fraction of workers hired as extensive-informal employees at the industry level decrease. Con- versely, as competition tightens, the probability of being employed in the intensive margin of informality, rather than as formal employees, increases in industries where employers, on average, have fewer than 50 employees. Similarly, in those same industries the share of intensive-informal employment on total employment within registered firms increases.5 Crucially, results suggest that the intensive margin of in- formality drives the effect of trade liberalization on informal labor. Namely, the impact on informal labor employed by registered employers is stronger than the effect on labor employed by unregistered employers. Hence, informality as a whole increases with trade liberalization. This paper relates to a vast literature on informality. There are three different views on informality and this paper encompasses all of them (La Porta and Shleifer 2008, 2014). Some authors consider informal firms as a consequence of burdensome regulations that prevent productive firms from formalizing (De Soto 1989, 2000). Others believe that informal firms are parasites that choose to remain informal to avoid paying taxes (Farrell 2004; Levy 2008). Conversely, many believe that informal workers and firms are inherently inefficient and could not operate in the formal sector (Rauch 1991; Maloney 2004; Amaral and Quintin 2006; Perry et al. 2007; De Paula and Scheinkman 2011). Since this study considers the intensive and the extensive margins of informality, it contributes to all of these views. They are not competing frameworks. They are echoing the presence of firm heterogeneity when deciding what taxes to pay and which regulations to obey (Ponczek and Ulyssea 2021). Hence, this paper’s most significant contribution is the distinction between the intensive and the ex- tensive margins when looking at the effect of trade on informality. While this distinction is not new, it is the first one to consider these two margins in an open economy framework.6 This paper also relates to recent work on trade and local labor markets (Dix-Carneiro and Kovak 2015, 2017; Dix-Carneiro, Soares, and Ulyssea 2018; Ponczek and Ulyssea 2021; Dix-Carneiro and Kovak 2019). They show that informality acts like a buffer that absorbs displaced workers from trade liberalization. When studying the Peruvian manufacturing sector, a local labor market approach is not suitable because the economic activity is highly concentrated.7 Thus, this study contributes to the body of literature that takes an industry-level approach (Acosta and Montes-Rojas 2014; Paz 2014; Cruces, Porto, and Viollaz 2018). 4 Peru’s growth only slowed down in the crisis period of 2008–2009, ranking second for growth in Latin America from 2002 to 2013. 5 There are no systematic differences in changes in cuts in tariffs across employer size. See the discussion in Identification Strategy and Results. 6 Ponczek and Ulyssea (2021) uses an estimated model to conduct counterfactual analysis of a reduction in the payroll tax, an increase in law enforcement on hiring off the books, and a change in entry cost for formal and informal firms. He finds that there are winners and losers in all policies and that a reduction on informality does not necessarily mean higher GDP, TFP, or welfare. Samaniego de la Parra (2016) studies empirically an increase in the inspections in the Mexican market. She finds that spouses of informal workers change their labor market participation decision and reservation wage after an inspection. 7 Nuñez (2014) documents that 55 percent of manufacturing firms are in the capital of Peru, Lima. Moreover, he also highlights that most firms that are not in Lima, are located in one of three regions: Arequipa, La Libertad, and Junin. 144 Cisneros-Acevedo The remainder of the paper is organised as follows: The section “Data and Definitions” describes the dataset and the definitions used throughout the paper. The following three sections characterize the Peruvian economy focusing on the manufacturing sector and its informal sector. They discuss the Peruvian liberalisation process and relevant labour policy changes. The section “Endogeneity of Trade Policy” addresses potential concerns regarding the endogeneity of trade policy. The section “Identification Strategy and Results” discusses the estimation strategy and presents the empirical results. The section “Robustness” Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 reviews possible threats to identification. Finally, the last section concludes. 2. Data and Definitions The empirical analysis in this paper uses two main data sources. The data on informality comes from the Peruvian National Household Survey (ENAHO) provided by the Peruvian National Institute of Statistics and Informatics (INEI), and the information on tariffs comes from the World Trade Organization’s Tariff Analysis Online (WTO’s TAO) Database. The study also controls for input tariffs constructed using the 2007 Peruvian Input-Output Table provided by the INEI. The ENAHO is a continuous survey that began in May 2003. However, since the questions used to classify workers’ informality status are only present from 2007, the analysis starts in 2007. Also, the question used to identify the intensive margin of informality changes in 2015. Hence, the period of analysis is 2007–2014 to minimize differences in definitions of an informal worker. The survey is representative of the Peruvian population. It comprises information regarding all households and their occupants surveyed in all 24 Peruvian regions.8 This paper is only interested in the population that is employed.9 Hence, the two main sources of data are the sections regarding the independent worker (ENAHO.04) and labor and income (ENAHO.500), which contain information on individuals who are at least 14 years old. All individuals in the data set are classified as either intensive-informal, extensive-informal, or formal employees. To do this, the study follows the methodology proposed by the INEI and the ILO’s agency for Formalization in Latin America and the Caribbean (FORLAC). First, it distinguishes the extensive margin of informality. When workers declare that their employer does not keep books in a way that agrees with the Peruvian Tax Collection Agency (SUNAT), the employer is classified as an “informal employer.”10 Self-employed workers are considered to have a registered employer if they registered as a legal person or as a legal entity (with a Tax Identity Number of RUC, RUS, or RER).11 Otherwise, they are considered to have an unregistered employer and are classified as an informal worker in the extensive margin. Second, the study characterizes the intensive margin of informality. An intensive-informal worker can be employed in a family firm or as a salaried employee. An individual employed by a registered family firm but as a non-paid family worker is considered to be an intensive-informal worker. Due to a change in the questionnaire, salaried workers are classified as intensive-informal using different criteria before and after 2011. Salaried workers surveyed between 2007 and 2011 are classified as informal workers if they declare that the tax collection agency does not deduct their income in any way. Individuals surveyed between 2012 and 2014 are considered informal workers if, contrary to Peruvian legislation, their employer does not pay health insurance on their behalf.12 To ensure that the change in the classification criteria for 8 Peru’s first-level administrative subdivisions are called “departamentos”; these are the regions used as a geographic indicator. 9 The unemployment rate in Peru during the period of study dropped from 4.8 in 2007 to 3.7 in 2014 (INEI 2018). 10 SUNAT requires all firms to keep books using either an online platform or with specific software. 11 RUC comes from the acronym in Spanish for “registro unico del contribuyente”; it uniquely registers all taxpayers, firms, or individuals. RUS is a tax system that was created to motivate formalization. It targets small firms and taxes them based on their sales. RER is a system that targets only people whose annual income does not exceed 61,200 US dollars. 12 The question regarding income deductions only exists between 2007 and 2011. The question regarding health insurance only exists between 2012 and 2014. The World Bank Economic Review 145 intensive-informal employment does not affect results, estimations are also done for the subperiod 2007– 2014 as a robustness check. Throughout this paper, extensive-informal workers are defined as those employed by an unregistered employer and all the self-employed individuals who do not pay taxes on their income. Moreover, intensive- informal workers are those that are defined as informal workers and are employed by an employer regis- tered with SUNAT.13 Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 As with all self-reported surveys, the reliability of individuals’ answers is a concern. However, since this paper’s interest is mostly on how informality changes across time, rather than measuring it at one specific point in time, the measurement error would arguably not alter results. Moreover, both the data set and the method of calculating informality are widely used by academic papers and policy reports from the ILO and OECD (Chong, Galdo, and Saavedra-Chanduvi 2008; FORLAC 2014a,b; Chacaltana 2017; OECD 2019; Villagomez and Chafloque 2019). The survey contains information on individuals’ sociodemographic characteristics, such as age, gender, civil status, race, education level, and region of residence. This information is valuable as it provides a profile of the informal worker in each margin of informality. Crucial to the empirical strategy is that, even though it is a household survey, individuals are also asked specific questions about their employer. They are asked how many people work in their place of work and the employer’s industry at the International Standard Industrial Classification of All Economic Activities (ISIC) 4-digit level revision 3. This informa- tion is fundamental to the analysis as it allows to link worker’s data with trade data through the worker’s employer industry affiliation.14 The chosen measure of trade openness is the average Most Favoured Nation (MFN) tariff, which is what countries promise to impose on imports from other members of the WTO unless the country is part of a preferential trade agreement (such as a free trade area or customs union). Thus, MFN rates are the highest (most restrictive) that WTO members charge one another. As shown in fig. 1, the MFN tariff is higher than the applied tariff because it does not take into account trade agreements. Even though the MFN tariff is a conservative measure of the country’s openness, to my knowledge it is the only tariff available for Peru at the HS 6-digit level of disaggregation.15 The MFN tariff data is sourced from the WTO’s TAO for the period 2007–2014 without information for 2012. For unknown reasons, information for 2012 was not published by the WTO, but it was available on the Peruvian Central Bank’s (BCRP) website.16 The WTO publishes the data using the HS commodity classification at a 6-digit level of aggregation in three editions.17 Thanks to correspondence tables from the World Integration Trade Solution (WITS) the informality data is linked to the tariffs data. Data on imports and exports at the HS 6-digit level of aggregation is obtained from Comtrade. Information on these trade flows are included as a robustness check. 3. Trade Liberalization Peru underwent a major trade liberalization in the 1990s that focused on dismantling the import sub- stitution industrialization policy, a protectionist scheme set in the 1970s. An additional opening process 13 Note that self-employed workers can only be classified as extensive-informal or formal employees. As a robustness check, self-employed workers are excluded from the empirical analysis and similar results are found. 14 The Most Favoured Nation (MFN) tariffs published by the WTO are at the Harmonized System (HS) code 6-digit level. They are converted to 4-digit ISIC codes to relate them to the informality data from ENAHO. 15 The World Development Indicators database (source of the data in fig. 1) only provides the average for the manufacturing sector. 16 Data published by the BCRP for other years is consistent with WTO datasets. Nevertheless, the section Robustness reports the analysis for the period 2007–2011 to avoid the change in the data source. Results are very similar. 17 For 2007 and 2008 the information is expressed in the 2007 edition (HS07), and for the following periods it is in the 2012 edition (HS12). 146 Cisneros-Acevedo Figure 1. Tariffs in Manufacturing 14 12 Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 10 8 % 6 4 2 2000 2005 2010 2015 Applied Tariff (simple mean) Applied Tariff (weighted mean) MFN Tariff (simple mean) MFN Tariff (weighted mean) Source: World Development Indicators (WDI). Note: The WDI solely provides the average for the manufacturing sector. happened in the 2000s during one of the fastest economic growth periods recorded in Peru. In this way, Peru concluded its liberalization process by reducing its average tariffs from 14 percent to 5 percent. While the first wave is more drastic, it was also more homogeneous. The reduction in tariffs in the 2000s shows much more heterogeneous tariff reduction across industries. This paper studies the second wave of liberalization to take advantage of this heterogeneity. Moreover, during this period, informality varies across industries both in terms of importance and in terms of the margin of informality. As shown in table 1, the level of protection varies significantly from one industry to another. All tariffs are falling after 2007, and they do so at different rates in different industries. For example, food, textiles, clothing, wood, and furniture experienced a stronger liberalization than chemicals and machinery. In any case, it is evident by looking at the “average in manufacturing” that Peru experienced a continuous liberalization process in terms of average output tariffs during the period of study. When countries cut down tariffs across industries, they might also reduce tariffs on intermediate inputs. As a result, the increase in import competition that firms face might be offset by a reduction in the cost of their imported inputs. To ensure that firms are indeed facing an increase in import competition as a result of a reduction in tariffs, the empirical analysis also controls for input tariffs. Following Topalova and Khandelwal (2011), input tariffs are estimated as IT jt = α js ∗ OTst , s where ITjt is the input tariff in industry j in year t, α js is the share of input s in the value of output j, and OTst is the output tariff in industry s in year t. The α ’s are calculated using the Peruvian Input-Output Table for 2007 made available by the INEI. The World Bank Economic Review 147 Table 1. Tariffs in Manufacturing (Percentage) 2007 2008 2009 2010 2011 2012 2013 2014 Food products and beverages 17.04 10.37 4.23 4.09 2.4 2.13 2.31 2.31 Textiles 18.21 15.33 14.62 14.41 10.95 9.18 9.45 9.39 Wearing apparel 19.55 16.53 16.52 16.53 12.61 10.71 10.72 10.73 Tanning and dressing of leather 17.36 15.56 15.32 14.14 10.68 9.74 9.29 9.53 Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 Wood products, except furniture 11.44 6.56 5.75 5.4 3.64 3.82 3.89 3.67 Publishing, printing, and media 11.18 7.32 6.39 5.27 4.87 4.82 4.69 4.4 Chemicals and chemical products 6.8 5.18 4.75 4.69 2.96 3.37 2.66 2.82 Rubber and plastics products 8.47 4.27 4.96 5.23 3.41 3.44 3.38 3.61 Other non-metallic mineral products 4.08 2.39 1.32 2.51 1.84 2.25 1.36 1.19 Basic metals 9.36 5.6 4.95 5.55 2.93 3.33 3.12 2.08 Fabricated metal products 11.44 2.9 1.85 2.39 0.89 1.08 1.05 1.01 Machinery and equipment 2.41 0.6 1.42 0.86 0.75 0.85 0.47 0.71 Other transport equipment 1.12 0.22 0.65 0 0.1 0.01 0.05 0.04 Furniture 11.27 8.04 7.97 8.04 5.36 5.36 5.37 5.34 Average in manufacturing 14.72 10.55 7.99 8.15 5.56 5.15 5.17 5.14 Source: Author’s elaboration based on Most Favoured Nation (MFN) tariff data from the World Trade Organization’s Tariff Analysis Online. Note: Average MFN tariff at the 2-digit International Standard Industrial Classification (ISIC) level within the manufacturing sector. As shown in fig. 2, not only were output tariffs falling during the period of study, but input tariffs were also significantly cut.18 It could be argued that firms’ costs might have declined enough to offset the increase in import competition. Hence, all specifications discussed in the section Identification Strategy and Results control for input tariffs. Coupled with a reduction in tariffs, Peru’s manufacturing sector experienced unprecedented growth and an increase in domestic demand during the period of analysis (Chacaltana 2017). The section Ro- bustness discusses in more detail this potential threat to identification as a favorable economical context may play a role in the evolution of informality. Furthermore, during the period of analysis, Peru signed free trade agreements (FTAs) with the United States and with China. In February 2009, the FTA with the United States came into force, and in March 2010 the one with China. These FTAs meant that China and the United States might have lower tariffs than the MFN tariff in some products. Since both China and the United States are important partners for Peru, the section on Robustness also presents a specification that controls for these FTAs. 4. Informal Labor Taking advantage of a rich data set, this study distinguishes three types of labor: (a) informal labor in the extensive margin (jobs carried out in unregistered firms), (b) informal labor in the intensive margin (workers hired by registered firms but in an employment relationship that does not comply with labor legislation), and (c) formal labor in registered firms. As fig. 3 shows, informal employment accounted for almost 80 percent of total employment in 2007 and, even though it decreased in importance, by 2014 it still accounted for 70 percent. It is easier to understand the underlying mechanisms through which trade affects informal labor when opening up the informal sector in these two margins of informality. The analysis in the section Identfication 18 Figure 2 shows a box plot for input tariffs in manufacturing. The box upper limit is the third quartile (Q3) and the lower limit is the first quartile (Q1). The line in the middle of the box is the median. The ends of the whiskers are the most extreme values within Q3 + 1.5(Q3 − Q1) and Q1 − 1.5(Q3 − Q1), respectively. The points on top of the whiskers are outliers. The figure shows that the dispersion of input tariffs has decreased significantly during the period of analysis. 148 Cisneros-Acevedo Figure 2. Input Tariffs in Manufacturing 10 Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 8 Input Tariff (%) 6 4 2 0 2007 2008 2009 2010 2011 2012 2013 2014 Source: Author’s elaboration based on Most Favoured Nation (MFN) tariffs from the World Trade Organization’s Tariff Analysis Online and the Peruvian Input-Output Table for 2007. Note: Outliers from the following industries are not included in the figure: manufacture of food products and beverages, manufacture of textiles, manufacture of wearing apparel, tanning and dressing of leather, and manufacture of wood products and cork. Strategy and Results recognizes that the reaction of informal employment to an increase in import com- petition might be different for each margin of informality. On the one hand, extensive-informal workers might struggle to retain their jobs as they tend to be the least efficient workers in the economy (Maloney 2004; Ponczek and Ulyssea 2021). On the other hand, intensive-informal workers might benefit from being cheaper than formal workers and manage to preserve their position. As fig. 4 shows, extensive-informal workers consistently earn a lower income than all other workers. Moreover, extensive-informal employees’ average income is lower than the minimum wage.19 Also, from fig. 4 it is evident that there is a big difference between wages for workers in the intensive margin of informality and formal employees. Intensive-informal employees earn on average half of formal workers’ monthly salary. Registered and unregistered employers are different, even if both hire informally. It is widely docu- mented in the literature that firms that employ extensive-informal workers are smaller than those hiring intensive-informal workers (Maloney 2004; Perry et al. 2007; Paz 2014). This is often driven by Tax and Labor legislation that favors smaller firms. In the Peruvian case, in order to access those benefits, firms had to have fewer than 10 employees up to 2008 when the law changed and small enterprises were defined as those with fewer than 100 workers.20 Throughout the period 2007–2014, approximately 60 percent of individuals in Peru were hired by employers with fewer than 100 workers, while only 10 percent were hired by employers with more than 500 workers. Moreover, firms’ size distribution is skewed to the right. 19 The minimum wage in 2007 was equivalent to 160 US dollars, and in 2014, 264 US dollars. 20 Section Informal Labour provides more details on the Promotion and Formalization of Micro and Small Enterprises Act. The World Bank Economic Review 149 Figure 3. Total Employment: Manufacturing Sector 100 0.22 0.21 0.23 0.22 0.23 0.28 0.28 0.30 Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 75 % Employment 0.19 0.20 0.20 0.20 0.19 0.22 0.22 0.22 50 0.59 0.59 0.57 0.58 0.58 25 0.50 0.50 0.48 0 2007 2008 2009 2010 2011 2012 2013 2014 Extensive-Informal (%) Intensive-Informal (%) Formal (%) Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014. Note: The figure shows total employment in manufacturing in Peru. Extensive-informal refers to jobs carried out in firms that are not registered with the Tax Collection Agency. Intensive-informal considers jobs in registered firms but in an employment relationship that does not comply with labor legislation. Figure 5 shows that extensive-informal workers are employed in industries where employers tend to have fewer than 100 workers. In the same way, intensive-informal jobs are more likely to be in industries where employers, on average, have between 200 and 300 workers. Finally, formal workers are, on average, hired in industries where employers tend to have more than 400 workers. While the literature often describes informal workers as single women with low education, table 2 suggests that such a description is only somewhat accurate when equating the extensive margin of infor- mality with the totality of informal employment. The intensive margin of informality is not necessarily composed by that demographic. Columns (1–3) in table 2 show that the individual characteristics of extensive-informal workers are significantly different from those of workers hired by registered employ- ers. Similarly, columns (4–6) in table 2 show that workers employed in the intensive margin of informality are different from those formally hired. Columns (1–3) in table 2 confirm that married women younger than 45 years old and with no education or primary education are more likely to be hired in the extensive margin of informality than in any other type of contract. Also, individuals identified with the Quechua ethnicity are more likely to be in the extensive margin of informality. As expected, extensive-informal workers tend to be the most vulnerable in the workforce as they are also those with very little bargaining power. Hence, they take the job that is offered. Columns (4–6) in table 2 show that single men with a secondary or lower level of education are more likely to be hired in the intensive margin of informality than formally. Moreover, older workers employed by registered employers are less likely to be informally hired. As I do not have more information, I can only speculate. However, it appears that informal workers are entry-level posts within registered employers. 150 Cisneros-Acevedo Figure 4. Average Monthly Income by Informality Status 700 600 Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 500 USD 400 300 200 100 2006 2008 2010 2012 2014 wave Extensive-Informal Intensive-Informal Formal Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014. Note: The figure shows average monthly income calculated for the individual’s primary occupation in the manufacturing sector in Peru. Extensive-informal refers to jobs carried out in firms that are not registered with the Tax Collection Agency. Intensive-informal considers jobs in registered firms but in an employment relationship that does not comply with labor legislation. Since they are probably young and healthy, they are willing to renounce the benefits that come with a formal job, such as a pension, health insurance, and parental pay. Labor Market Policies Informality and its unremitting rise in past decades has been a long-lasting challenge for the Peruvian economy. Hence, the trend towards formalization in recent years is the most striking development in the Peruvian labor market. This paper focuses on the effect that falling tariffs had on informal employment. However, the reduction in tariffs was accompanied by major institutional changes. The two most im- portant ones for the labor market were the Promotion and Formalization Act, which reduced costs for smaller firms, and the implementation of an electronic payroll system, which might have increased the state’s ability to enforce labor laws (Chacaltana 2017). The Promotion and Formalization of Micro and Small Enterprises Act passed in 2003. It reduced non- wage costs such as holiday pay significantly, and also cut dismissal costs to a third for micro-enterprises (firms with fewer than 10 workers). In 2008, this special regime was extended for firms with up to 100 workers, and it came into effect in 2009. Chacaltana (2017) documents that micro-enterprises ac- count for 70 percent of wage employment in Peru and that the reduction in non-wage labor costs fell from 54 percent to 17 percent in 2003 in weighted terms. Even though the decrease in costs was significant in 2003, according to Chacaltana (2017)’s own calculations, the index of labor costs remained fairly stable during the period 2007–2014. Since the focus of this paper is precisely that period, this new policy would not have affected the results discussed in the section Identification Strategy and Results. The World Bank Economic Review 151 Figure 5. Average Industry Size by Informality Status 2007 2008 Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 2009 2010 2011 2012 2013 2014 0 200 400 600 800 number of workers Extensive-Informal Intensive-Informal Formal Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014. Note: All surveyed individuals are asked how many workers are employed in their workplace. With this information, an industry-year average at the 4-digit level is calculated. This figure presents the average industry size per informality status in each year of the sample. The figure considers solely employment in manufacturing in Peru. Extensive-informal refers to jobs carried out in firms that are not registered with the Tax Collection Agency. Intensive-informal considers jobs in registered firms but in an employment relationship that does not comply with labor legislation. Furthermore, in 2006 the Ministry of Labor and Employment Promotion (MTPE) and the Office of the Tax Collection Agency (SUNAT) put in place an electronic payroll system. Before 2006, all registered firms had to annually submit hard copies of their payrolls to the MTPE, including information on work- ers, wages, and type of contracts. After the electronic payroll system was set up, they had to submit it monthly to SUNAT together with the firm’s tax return. Chacaltana (2017) states that since SUNAT has demonstrated a more reliable inspection capability than MTPE, the shift towards the electronic payroll system might have meant an increase in inspections and in the likelihood of infractions being detected, at least for firms registered with SUNAT. Chacaltana (2017) finds that despite the scale of the institutional changes, the variables associated with the labor market reforms and inspections strength did not have a significant effect on the formalization process. He states that these findings are not surprising as at least two of every three workers with an informal job in Peru are employed by economic units that are not registered for tax purposes. In other words, due to the size of the extensive margin of informality, policies that target formal employers do not have a significant effect on the evolution of informality. Results might be different after 2013 when the National Labor Inspection Authority (SUNAFIL) started to operate to strengthen the inspection service. To ensure that the creation of SUNAFIL does not influence results in the last two years of the sample, the empirical analysis is also conducted for the period 2007–2011 and results are very similar. 152 Table 2. Individual Characteristics Extensive-informal Registered employer Intensive-informal Formal workers for reg. workers employees Diff: 1−2 workers employer Diff: 4−5 (1) (2) (3) (4) (5) (6) Mean SD Mean SD b Mean SD Mean SD b Married 0.32 0.47 0.29 0.45 −0.03*** 0.15 0.36 0.40 0.49 0.25*** Male 0.43 0.50 0.72 0.45 0.29*** 0.68 0.46 0.76 0.43 0.07*** Age 14–29 0.31 0.46 0.39 0.49 0.07*** 0.57 0.50 0.24 0.43 −0.33*** Age 30–44 0.30 0.46 0.35 0.48 0.05*** 0.28 0.45 0.41 0.49 0.12*** Age 45–64 0.30 0.46 0.24 0.43 −0.06*** 0.13 0.34 0.33 0.47 0.20*** Age >65 0.09 0.28 0.02 0.15 −0.06*** 0.02 0.13 0.03 0.17 0.01*** No education 0.08 0.27 0.00 0.07 −0.07*** 0.01 0.07 0.00 0.06 −0.00* Primary education 0.32 0.46 0.10 0.30 −0.21*** 0.12 0.32 0.09 0.29 −0.02*** Secondary education 0.46 0.50 0.53 0.50 0.07*** 0.61 0.49 0.47 0.50 −0.14*** Technical education 0.10 0.29 0.21 0.41 0.12*** 0.17 0.37 0.25 0.43 0.08*** UG-PG degree 0.05 0.22 0.15 0.36 0.10*** 0.10 0.30 0.19 0.39 0.09*** Quechua ethnicity 0.16 0.36 0.11 0.31 −0.05*** 0.10 0.30 0.11 0.32 0.02*** Observations 16,883 14,126 31,009 6,380 7,746 14,126 Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014. Note: Extensive-informal (column 1) refers to workers employed in firms that are not registered with the Tax Collection Agency. Intensive-informal (column 4) considers workers employed in registered firms but in an employment relationship that does not comply with labor legislation. Columns (1), (2), (4), and (5) present the means and standard deviations for a specific group of workers. Column (1) takes into account only employees in the extensive margin of informality and, in column (2), all employees hired by registered employers (both formal and intensive-informal workers). Column (4) refers only to intensive-informal workers and Column (5) takes into account solely formal employees. “b” in Columns (3) and (6) stands for the mean difference between two groups and *p < 0.05, **p < 0.01, ***p < 0.001. Column (3) presents the results from comparing the difference in the means of employees in a registered firm (intensive-informal and formal) and extensive-informal workers. Column (6) reports the results from comparing the difference in the means of informal and formal employees hired by a registered employer. Cisneros-Acevedo Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 The World Bank Economic Review 153 5. Manufacturing Sector Díaz (2014) finds that over 40 percent of the reduction of informality in the period 2002–2011 in Peru is due to a change in the structure of employment by firm size. Moreover, Infante and Chacaltana (2014) study the same period in Peru and show that medium-size firms are more dynamic in terms of output employment and productivity. Furthermore, Kleven, Kreiner, and Saez (2016) state that the proportion of informal workers in a firm’s labor force diminishes with the size of the firm as larger firms are more likely Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 to face an inspection. Putting Kleven, Kreiner, and Saez (2016)’s result, together with Levy (2008)’s result that unregistered firms tend to be smaller than registered ones, it seems that firm size plays an important role when studying the evolution of informality. Moreover, size is especially relevant when distinguishing between the intensive and the extensive margins of informality. It would be ideal to have firm-level data that reveal both the informality status of the firm itself and the informality status of its workers. Due to the illegal nature of informality, that type of data set is very hard to come by. Nevertheless, the Peruvian household survey ENAHO asks all individuals in the survey how many workers are employed in their workplace. I acknowledge that it is quite unlikely for an individual to know the exact number of workers employed. However, I claim that most workers have a rough idea of how many people work with them. Intending to correct for this measurement error, I calculate the average size of each 4-digit industry in every given year. In other words, based on individuals’ declaration on how big their employer is in term of number of workers, I calculate an average employer size per industry and year. I prefer to take this approach instead of mixing sources. If I were to replace the industry size measure obtained using the household survey with another one constructed from a firm-level survey, I would not be able to observe workers in informal firms as they are not present in firm-level surveys. As a result, the industry size measure for industries abundant in informal firms would be underestimating the size of the average employer significantly. Nevertheless, as a check, in the section Industry Size in the supplementary online appendix, the average industry size calculated using household data is compared with a similar measure made using firm-level data for industries where formal employers are predominant. The analysis suggests that the industry size distribution constructed with data from ENAHO is akin to the one obtained when using firm-level data. Hence, I argue that any inaccuracies in workers’ knowledge of the size of their employer cancel out when averaged up. Table 3 groups manufacturing industries at the ISIC 2-digit level. It shows that most employers in “food products and beverages” tend to have over 50 employees. In contrast, in “textiles” employers tend to have from 16 to 50 workers. Also, while in industries such as “basic metals” and “furniture” the average industry size does not vary much over time, in industries such as “food products” and “other transport equipment” there is quite a change in employers’ size between 2007 and 2014. In the same way as some industries seem to consistently have smaller employers, some industries tend to be characterized by one the informality margins. As table 4 shows, in “furniture,” “textiles,’ and “wearing apparel” workers are mostly hired as extensive-informal employees. Conversely, workers employed in “basic metals” are mostly formally employed. Since the main data source is a household survey, it is not possible to understand the underlying reason why firms choose to register or why registered firms decide to hire informal workers. Only the consequence on employment of their decisions is observed. Nevertheless, it might be helpful to have at least a vague idea of what motivates firms to make these choices. Studying data from the World Bank Enterprise Surveys (WB-ES) provides some insights on the matter.21 21 Table S3.1 in the supplementary online appendix provides more detailed information regarding how many firms were surveyed and their industry within manufacturing. Since many of the relevant questions are only available in the 2006 survey, the analysis is conducted solely for that wave of the WB-ES. 154 Cisneros-Acevedo Table 3. Employer Size in Manufacturing (Percentage) 2007 2014 Average Average Average Average Average Average employer size employer size employer size employer size employer size employer size <16 16–50 >50 <16 16–50 >50 (1) (2) (3) (4) (5) (6) Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 Food products and beverages 7.81 4.85 87.34 8.58 36.1 55.32 Textiles 27.5 53.53 18.97 27.65 58.53 13.82 Wearing apparel — — 100 — — 100 Tanning and dressing of leather 30.08 69.92 — 25.95 74.05 — Wood products, except furniture 35.61 — 64.39 35.11 59.92 4.96 Publishing and printing — 6.8 93.2 4.27 0.85 94.87 Chemicals and its products — — 100 — 2.3 97.7 Rubber and plastics products 11.11 — 88.89 — — 100 Other non-metallic mineral products 61.97 20.42 17.61 6.81 49.74 43.46 Basic metals — — 100 — — 100 Fabricated metal products — 20.66 79.34 — 12.38 87.62 Machinery and equipment 15.22 20.65 64.13 — 12.39 87.61 Other transport equipment 48.65 2.7 48.65 2 — 98 Furniture 0.84 94.14 5.02 6.22 93.78 — Total 13.29 28.01 58.71 10.64 39.11 50.25 Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014. Note: This table shows the distribution within industries at the 2-digit level within manufacturing in Peru. Columns (1–3) show how total employment in a given industry was distributed in 2007. Similarly, columns (4–6) show total employment distribution in 2014. From an employee’s declaration of their workplace size in terms of the number of workers, the average industry size at the 4-digit level for every year in the sample is calculated. Table 4. Two Margins of Informality in Manufacturing (Percentage) 2007 2014 Extensive Intensive Extensive Intensive informal informal Formal worker informal informal Formal worker (1) (2) (3) (4) (5) (6) Food products and beverages 47.78 23 29.22 40.75 24.04 35.21 Textiles 82.35 6.47 11.18 82.08 7.34 10.58 Wearing apparel 59.57 20.21 20.21 51.47 22.96 25.57 Tanning and dressing of leather 54.47 28.46 17.07 46.49 36.76 16.76 Wood products, except furniture 52.2 34.15 13.66 50 26.72 23.28 Publishing and printing 18.45 44.66 36.89 14.53 40.17 45.3 Chemicals and its products 26.58 26.58 46.84 6.9 17.24 75.86 Rubber and plastics products 13.89 30.56 55.56 12.5 22.5 65 Other non-metallic mineral products 75.35 8.45 16.2 56.54 12.04 31.41 Basic metals 7.41 18.52 74.07 10.42 18.75 70.83 Fabricated metal products 57.85 21.07 21.07 33.88 25.73 40.39 Machinery and equipment 21.74 33.7 44.57 15.04 27.43 57.52 Other transport equipment 51.35 16.22 32.43 38 30 32 Furniture 73.43 13.39 13.18 53.53 23.03 23.44 Total 58.76 19.38 21.86 47.82 22.14 30.04 Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014. Note: This table shows how total employment is composed of intensive-informal, extensive-informal, and formal workers within industries at the 2-digit level within manufacturing in Peru. Extensive-informal (columns 1 and 4) refers to workers employed in firms that are not registered with the Tax Collection Agency. Intensive- informal (columns 2 and 5) considers workers employed in registered firms but in an employment relationship that does not comply with labor legislation. Columns (1–3) show the labor composition in a given industry in 2007 and columns (4–6) in 2014. The World Bank Economic Review 155 In 2006, 2010, and 2017, firms were asked whether they were registered when they started operations; above 90 percent of them answered that they were. However, in 2006, only 71 percent confirmed reporting 100 percent of their sales for tax purposes. Moreover, only 12 percent declared up to 50 percent of their sales. Interestingly, when firms were asked why they decided to register, 65 percent confirmed that they wanted to comply with the law, 16 percent affirmed that customers and suppliers only deal with registered entities, and 15 percent stressed the need to be registered to access finance. On the other hand, more than Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 50 percent of firms argued that tax administration and tax rates are moderate or major obstacles to their daily operations. Furthermore, 25 percent of firms singled out competing with informal firms as their biggest obstacle, even above corruption and access to finance. Still studying data from the 2006 WB-ES, only 65 percent of firms declare 100 percent of their work- force for tax purposes. Furthermore, 17 percent of firms declared up to 50 percent of their workforce. Additionally, 30 percent of firms claimed that their decision to hire or fire a permanent worker was af- fected by labor regulations. Moreover, 60 percent of firms considered labor regulations as a moderate, major, or severe obstacle to their daily operations. From firms’ answers to the WB-ES, it seems that firms would like to register to comply with the law. However, when they do so, they are inclined to not comply fully either by not declaring total sales or not declaring their entire workforce. This is possibly because they still have to compete with informal firms. Moreover, firms seem to hire in the intensive margin of informality to avoid having to comply with labor regulations. 6. Endogeneity of Trade Policy This section addresses possible concerns regarding the endogeneity of trade policy. First, certain industries might have higher lobby power and be able to get more favorable treatment. Second, relatively more informal industries may enjoy greater protection if the Peruvian authorities consider they need it due to lower efficiency or for political reasons. If more informal industries were not liberalized as intensely as other industries, small reductions in tar- iffs might be associated with a decline in informality and wrongfully infer that trade liberalization reduces informality. Hence, following Topalova (2007), this study examines whether tariffs moved together by an- alyzing the changes in tariffs for 102 ISIC 4-digit products in manufacturing over the period 2008–2014. Figure 6 shows that most of the tariff changes across products are in the same direction.22 Moreover, the supplementary online appendix S5 examines the change in tariffs in 2-digit products and shows that the majority of the products move with a very similar trend to the average manufacturing tariff. Additionally, table S2.1 in the supplementary online appendix reports the empirical analysis controlling for possible pre-existing trends in trade policy and finds that results are not very different.23 In the spirit of Topalova and Khandelwal (2011), this paper tests whether measures of trade protection are correlated with industry characteristics associated with political importance. It regresses changes in output tariffs and changes in input tariffs on the average wage, the share of women, the share of skilled workers, the share of married workers, and average workers’ age. Policy makers may choose to protect industries where income is lower, where there are more women, where workers are less skilled, where there are fewer married workers, or where workers are younger. Results are presented in table 5. The table reveals no statistical correlation between changes in output tariffs and any of the industry charac- teristics. Similar to Topalova and Khandelwal (2011)’s findings, except for average wages, none of the other industry characteristics are correlated with changes in input tariffs. 22 Figure 6 does not feature products for which tariffs remained unchanged. 23 Since identification of these two margins of informality is only possible after 2007, it is not possible to examine the effect of tariffs at a prior date. Hence, it is not possible to conduct placebo tests. 156 Table 5. Change in Tariffs and Industrial Characteristics Output Output Output Output Output tariffs Input tariffs tariffs Input tariffs tariffs Input tariffs tariffs Input tariffs tariffs Input tariffs (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Log wage 0.0056 −0.5333*** — — — — — — — — (0.143) (0.187) Share of women — — -0.1680 0.7246 — — — — — — (0.518) (0.471) Primary education (share) — — — — −0.3653 −1.8004 — — — — (1.212) (2.571) Secondary education (share) — — — — 0.2620 −1.0793 — — — — (1.148) (2.438) Technical education (share) — — — — 0.2790 −1.1753 — — — — (1.096) (2.512) UG-PG degree (share) — — — — 0.2802 −1.1806 — — — — (1.126) (2.394) Share of married workers — — — — — — −0.1855 −0.8851 — — (0.273) (0.541) Workers age — — — — — — — — −0.0083 −0.0195 (0.010) (0.013) Observationsa 595 595 596 596 596 596 596 596 596 596 R2 0.51 0.15 0.51 0.14 0.52 0.14 0.51 0.15 0.51 0.14 Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014, Most Favoured Nation (MFN) tariff data from the World Trade Organization’s Tariff Analysis Online, and Peruvian Input-Output Table for 2007. Notes: This Table tests if measures of trade protection (input and output tariffs) are correlated with industry characteristics that might influence government’s wish to protect a specific industry. Non-significant coefficients suggest that protectionism is not linked to any of these characteristics. All specifications include a constant, year, and industry fixed effects. Robust standard errors for industrial clusters in parentheses. *p < 0.10, **p < 0.05, ***p < 0.01. a Observations in columns (1) and (2) are fewer than in the other columns because there is no information on monthly income for ISIC code 2912 (Manufacture of pumps, compressors, and valves) in 2011. Cisneros-Acevedo Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 The World Bank Economic Review 157 Figure 6. Tariff Changes 2008–2014 (in Percent of Total ISIC Codes) 100 90 80 Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 70 60 % 50 40 30 20 10 0 2008 2009 2010 2011 2012 2013 2014 Tariffs Increases Tariff Reductions Source: Author’s elaboration based on Most Favoured Nation (MFN) tariff data from the World Trade Organization’s Tariff Analysis Online. Note: International Standard Industrial Classification (ISIC) codes where tariffs remained unchanged are not featured in the figure. Finally, this paper examines whether policy makers adjusted tariffs in response to industries’ infor- mality level or due to industries’ size. If the former was correct, one should expect that past informality predicts current trade policy. If the latter was true, past industry size would have a significant relationship with current trade policy. Therefore, table 6 presents results from regressing changes in output tariffs and changes in input tariffs in t, on the share of informal employment at industry level and on industry size in t – 1. Columns (1) and (2) report the correlation between changes in tariffs and the share of informal employment on total employment without distinguishing margins. Columns (3) and (4) present the re- lationship between the proportion of intensive-informal employment on total employment in registered firms and changes in tariffs. Columns (5) and (6) measure the correlation between the change in tariffs and the proportion of extensive-informal employment on total employment. Lastly, columns (7) and (8) show the results when regressing the average industry size calculated with data from ENAHO and the change in tariffs. Table 6 reveals that the correlation between past share of informal employment and current trade protection is statistically insignificant for all measures of informality. Similarly, there is no correlation between past industry size and current change in tariffs. 7. Identification Strategy and Results This section estimates the impact of an increase in import competition on informal employment. First, it studies informal labor as a binary phenomenon as it does not discriminate between different types of infor- mal worker. It presents these baseline estimations. Second, it examines the intensive margin of informality. Third, it focuses on the extensive margin. Since the two margins of informality are not independent, the analysis on each margin is done separately. 158 Cisneros-Acevedo Table 6. Change in Tariffs, Share of Informal Employment, and Industry Size Output Input Output Input Output Input Output Input tariffs tariffs tariffs tariffs tariffs tariffs tariffs tariffs (1) (2) (3) (4) (5) (6) (7) (8) Share informal employment (−1) −0.1363 0.1024 — — — — — — (0.189) (0.162) Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 Share intensive-informal (−1) — — 0.1269 0.1193 — — — — (0.210) (0.224) Share extensive-informal (−1) — — — — −0.2500 0.0615 — — (0.168) (0.153) Industry size (−1) — — — — — — 0.0001 −0.0001 (0.000) (0.000) Observations 574 574 547 547 574 574 574 574 R2 0.41 0.09 0.41 0.10 0.42 0.09 0.53 0.14 Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014, Most Favoured Nation (MFN) tariff data from the World Trade Organization’s Tariff Analysis Online, and the Peruvian Input-Output Table for 2007. Note: “Share of informal employment” does not distinguish between margins of informality and refers to the proportion of informal employment on total employment. “Share of extensive-informal” refers to the fraction of informal workers in the extensive margin on total employment. “Share of intensive-informal” refers to the proportion of intensive-informal employment on total employment on registered firms. Extensive-informal refers to workers employed in firms that are not registered with the Tax Collection Agency. Intensive informal considers workers employed in registered firms but in an employment relationship that does not comply with labor legislation. Industry size is calculated as the average industry-year number of workers individuals declare to work with them. Standard errors for industrial clusters in parentheses. All specifications include a constant and year fixed effects. *p < 0.10, **p < 0.05, ***p < 0.01. These three studies are two-fold. On the one hand, they study the effect of a reduction in tariffs on the probability of being hired as an informal worker. In this way, the analysis is conducted at the individual level, and it controls for individual characteristics that might influence workers’ probability of being informal.24 On the other hand, they calculate the impact of an increase in import competition at the industry level by estimating the effect on the importance of informal employment on total employment. 7.1. Baseline Estimations Even though the data allows to distinguish between informal workers in the intensive and in the exten- sive margin of informality, baseline estimations do not make this distinction. In this way, results from estimating the linear probability model described by equation (1) are comparable to those obtained in the literature (Goldberg and Pavcnik 2003; Acosta and Montes-Rojas 2014; Ponczek and Ulyssea 2021; Cruces, Porto, and Viollaz 2018): INF jt = α0 + ατ OT jt + αgSg jt ∗ OT jt + αs Sg jt + αi IT jt + αH H t + F j + F t + ε jt , (1) where INF jt = 1 if individual working in industry j in year t is informal either in the extensive margin or in the intensive margin, and 0 otherwise. The variable OTjt is the output tariff in industry j in year t and its coefficient provides the effect of import competition on informal employment. Since an increase in import competition is identified as a reduction in tariffs, the distinction between input and output tariffs is imperative. Input tariffs are the tariffs that a domestic firm pays for all imported inputs, while output tariffs are the entry cost paid by foreign competitors when selling at home. Then a cut in input tariffs is a reduction in domestic firms’ costs. As such, it could compensate for an increase in competition coming from abroad, i.e., a decrease in output tariffs. For example, a domestic firm might import leather to manufacture shoes. If tariffs for leather go down, the firm’s costs also go down. On the 24 These individual characteristics are included as proxy variables of unobservable characteristics, such as ability or bar- gaining power, that might influence workers’ probability of accepting an informal job. The World Bank Economic Review 159 other hand, if tariffs for shoes produced abroad are also cut, a firm’s import competition increases. While the reduction in input tariffs is a positive shock for the domestic firm, the cut in output tariffs is a negative shock. Hence, equation (1) also controls for input tariffs with ITjt , which is the input tariff in industry j in year t. In this way, it accounts for the possibility that effects do not cancel each other out when all tariffs diminish. Equation (1) also includes an industry-size indicator calculated each year (Sgjt ) and interaction terms Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 with output tariffs (Sgjt * OTjt ). Since the average industry size is constructed each year based on workers’ answers regarding the size of their employer, it changes with time for a given industry.25 Then it is possible to control for size variation within industries over time. Change in industry size is controlled with the index Sgjt , where g = s, m, l. When g = s, employers in industry j at time t have on average less than 16 employees, when g = m they have more than 15 and less than 51 workers, and when g = l they have more than 50 workers. Since Sgjt is an industry-size indicator variable, each of the coefficients of the interaction term, α g , are the differential effect of output tariffs on informality on industries with different average size. Henceforth, small, medium, and large industries are classified according to the definition given by Sgjt . It can be argued that informality is the result of a bargaining process between the worker and the employer. Hence, the individual’s attributes are critical in determining the outcome of the bargain. For example, it is expected that more educated individuals are less likely to be hired as informal workers (table 2). The vector H t controls for observable attributes of worker in year t such as age, education, gender, civil status, ethnic background, and region of residency.26 Finally, equation (1) controls for industry time-invariant characteristics and aggregate shocks by adding industry fixed effects Fj and year fixed effects Ft . This study uses a linear probability model to estimate equation (1). It accounts for heteroskedasticity and serial correlation in the error term by using robust (Huber–White) standard errors clustered by in- dustry at the 4-digit level. The results are in table 7 in columns (1–4). Column (1) suggests that there is no significant relationship between the reduction of tariffs and the probability of being hired as an informal worker. Column (2) also controls for average industry size, and output tariffs remain statistically insignifi- cant. Column (3) adds industry size interacted with output tariffs. It finds that a reduction in output tariffs increases the probability of being hired as an informal worker in industries where on average employers have fewer than 16 employees. In particular, a decrease of 1 percentage point in output tariffs generates an increase of 0.55 percentage points in the probability of being hired as an informal worker in such an industry. To a lesser extent, this is also true for medium-size industries as a reduction of 1 percentage point in output tariffs translates into a boost of 0.23 percentage points in the probability of being hired as an informal worker. Moreover, these results are robust to controlling for input tariffs in column (4). However, in this case, results suggest that there is a positive significant relation between output tariffs and informality. Then, when controlling for input tariffs, the effect on small and medium industries is coun- teracted, eliminating the effect on medium industries and reducing the effect on small firms significantly. Furthermore, the impact on large industries goes in the opposite direction. When output tariffs are cut, informal employment in these industries is also reduced. The reduction in tariffs is not systematically different across industry size. For instance, in 2008 the average tariff reduction took place among large industries. Conversely, in 2011 medium industries ex- perienced the most significant tariff cut and large industries the lowest. During the period of study, all industries experienced a reduction in tariffs and the extent of the contraction is not correlated with the 25 Note that this industry-size indicator does not capture industry-specific trends as it can change with time for any given industry if the industry’s average size changes. Moreover, in the robustness section, controls for industry trends are included and results are very similar. 26 Region refers to Peru’s first-level administrative subdivision, “departamento.” 160 Cisneros-Acevedo Table 7. Informal Employment without Distinguishing Margins Informality indicator Share of informal employment (1) (2) (3) (4) (5) (6) (7) (8) Output tariff 0.0014 0.0015 0.0023 0.0027* 0.0044 0.0043 0.0075** 0.0077** (0.001) (0.001) (0.001) (0.001) (0.003) (0.003) (0.003) (0.003) Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 Input tariff — — — −0.0017 — — —0.0011 (0.002) (0.007) S * output tariff — — −0.0055*** −0.0055*** — — −0.0130* −0.0130* (0.001) (0.001) (0.007) (0.007) M * output tariff — — −0.0023* −0.0025* — — −0.0109* −0.0109* (0.001) (0.001) (0.006) (0.006) Small industry (S) — 0.0201 0.0593*** 0.0588*** — 0.2143*** 0.2685*** 0.2686*** (0.013) (0.017) (0.017) (0.043) (0.061) (0.061) Medium industry (M) — 0.0232** 0.0333** 0.0337** — 0.1235*** 0.1691*** 0.1688*** (0.011) (0.014) (0.014) (0.039) (0.056) (0.056) Observations 31,009 31,009 31,009 31,009 695 695 695 695 R2 0.31 0.31 0.31 0.31 0.61 0.65 0.65 0.65 Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014, Most Favoured Nation (MFN) tariff data from the World Trade Organization’s Tariff Analysis Online, and the Peruvian Input-Output Table for 2007. Note: Specifications in columns (1–4) include region of residence fixed effects. All specifications include controls for individual characteristics (civil status, education, ethnicity, and age), industry and year fixed effects. Average industry-year size is classified as S: ≤15 workers, M: >15 and ≤50 workers, L: >50 workers. Robust standard errors for industrial clusters in parentheses. *p < 0.10, **p < 0.05, ***p < 0.01. average employer size of the industry.27 Hence, these results are consistent with the hypothesis that an increase in import competition would motivate small employers to hire more informal workers in the extensive margin of informality as they are less likely to be detected transgressing the law. On the other hand, the probability of being informally hired in industries where employers have more than 15 employ- ees goes down as they are more notorious, and are more likely to be audited. The reduction in output tariffs not only has an impact on the decision to hire an informal worker but also on the proportion of informal workers employed in a given industry and year. Thus, after estimating the effect on the probability of hiring informal workers, this study estimates equation (2): ShareINF jt = ψ0 + ψτ OT jt + ψgSg jt ∗ OT jt + ψs Sg jt + ψi IT jt + Fj + Ft + η jt , (2) where total informal employment ShareINF jt = in industry j in year t . total employment This identification strategy is somewhat comparable with the two-stage approach used by Goldberg and Pavcnik (2003) and Acosta and Montes-Rojas (2014). Instead of estimating the propensity to be informal, equation (2) directly calculates the share of informal labor in total employment. Table 7 reports the OLS estimation of equation (2) in columns (5–8). The regressors are the same as those described in equation (1), but in this case the dependent variable is the proportion of informal labor in total employment. Findings in columns (5) and (6) suggest that there is no significant effect on the importance of informal employment when output tariffs are cut. However, when including the average industry-year size indica- tors interacted with output tariffs in column (7), there is a statistically significant relationship between the share of informal employment on total employment and output tariffs.28 Namely, results suggest that a 27 See supplementary online appendix S5 for more details. 28 Supplementary online appendix S5 shows that the distribution of cuts in tariffs is not systematically different across industry sizes. The World Bank Economic Review 161 decrease of 1 percentage point in output tariffs cuts the share of informal labor by 0.77 percentage points. However, this is only true for industries where employers on average are large, as in industries with mostly small and medium firms this effect is counteracted. The results suggest that when import competition increases because tariffs are cut by 1 percentage point, industries that on average have small firms in- crease their share of informal employment by 1.30 percentage points and industries with mostly medium firms by 1.09 percentage points. This result remains almost unchanged when controlling for input tariffs Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 in column (8). From the results presented in table 7, it is possible to infer that two mechanisms go in opposite direc- tions when studying the impact of an increase in import competition on informal labor. On the one hand, when tariffs fall, the probability of being hired as an informal worker increases only in industries that on average have fewer than 50 workers. In all other industries, the probability of working as an informal employee goes down. Interestingly, this result is slightly different when discussing the effect of a reduction of tariffs on the proportion of informal employment. Notably, in medium-size industries, the proportion of informal employment increases. Hence, findings suggest that there are two mechanisms at play. One that increases informality and one that decreases informality when tariffs go down. In the following sec- tions, this study shows that when unfolding informality in two margins, these mechanisms are revealed in more detail. 7.2. Intensive Margin To gain some insight into what is driving the effect of the reduction in tariffs on informality suggested in table 7, this study takes advantage of the detailed individual Peruvian data and distinguishes between informal employees in the intensive and in the extensive margin. This section focuses on the intensive margin of informality. First, it conducts an analysis at the indi- vidual level and estimates equation 3. The dependent variable in equation 3, INT jt , takes the value 1 if individual is informal in the intensive margin in industry j in year t and 0 if the individual is a formal worker. Note that the regression of equation (3) is conditional on individuals being hired by a registered firm:29 INT jt = β0 + βτ OT jt + βgSg jt ∗ OT jt + βs Sg jt + βi IT jt +βH H t + Fj + Ft + ε jt . (3) Equation (3) is estimated with a linear probability model. It accounts for heteroskedasticity and serial correlation in the error term by computing Huber–White standard errors clustered by industry at the 4-digit level of aggregation. Results are in columns (1–4) in table 8. All specifications in columns (1–4) in table 8 control for individual characteristics and include industry, year, and region of residence fixed effects. Results show that married men with above technical education and older than 30 years are less likely to be hired as intensive-informal workers instead of formal ones. Column (1) only controls for individual characteristics. Column (2) also controls for average firm size in each industry-year. Both specifications find that a reduction in output tariffs does not have a statistically significant impact on the probability of being hired as an intensive-informal worker. However, when also controlling for industry-year size in column (3), results show that in industries that on average have 29 Table S1.1 in the supplementary online appendix reports results for a multinomial logit model where the dependent variable has three categories: extensive-informal, intensive-informal, and formal. When testing for the violation of the independence of irrelevant alternatives (IIA) assumption, results are ambiguous, hinting that being hired by a registered employer (as a formal or as an intensive-informal worker) and being intensive-informal are not independent. Hence, equation (3) is estimated conditional on being hired by a registered employer. 162 Cisneros-Acevedo Table 8. Intensive Margin of Informality Informality indicator Share of informal employment (1) (2) (3) (4) (5) (6) (7) (8) Output tariff 0.0005 0.0006 0.0014 0.0018 −0.0013 −0.0013 0.0030 0.0027 (0.002) (0.002) (0.002) (0.002) (0.003) (0.003) (0.003) (0.003) Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 Input tariff — — — −0.0025 — — — 0.0021 (0.003) (0.007) S * output tariff — — −0.0129*** −0.0130*** — — −0.0196** −0.0196** (0.004) (0.004) (0.009) (0.009) M * output tariff — — −0.0057** -0.0059** — — −0.0133** −0.0132** (0.002) (0.002) (0.006) (0.007) Small industry (S) 0.0184 0.0991*** 0.0988*** 0.0893* 0.1755** 0.1757** (0.021) (0.037) (0.037) (0.049) (0.074) (0.074) Medium industry (M) 0.0112 0.0367** 0.0369** 0.1040** 0.1583** 0.1577** (0.017) (0.018) (0.018) (0.043) (0.060) (0.060) Observationsa 14,126 14,126 14,126 14,126 660 660 660 660 R2 0.25 0.25 0.25 0.25 0.37 0.39 0.40 0.40 Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014, Most Favoured Nation (MFN) tariff data from the World Trade Organization’s Tariff Analysis Online, and the Peruvian Input-Output Table for 2007. Note: Specifications in columns (1–4) include region fixed effects. All specifications include controls for individual characteristics (civil status, education, ethnicity, and age), industry and year fixed effects. Average industry-year size is classified as S: ≤15 workers, M: >15 and ≤50 workers, L: >50 workers. Robust standard errors for industrial clusters in parentheses. *p < 0.10, **p < 0.05, ***p < 0.01. a All specifications are conditional on individuals being hired by a registered firm. employers with fewer than 50 employees, a reduction in output tariffs translates into an increase in the probability of hiring informal-intensive workers.30 Interestingly, this result remains the same when controlling for input tariffs in column (4). A reduction of 1 percentage point in output tariffs generates an increment of 1.3 percentage points in the probability of being hired as an intensive-informal worker in an industry where, on average, firms are small. The effect is quantitatively lesser in industries that have mostly medium-size firms as in these industries a decrease of 1 percentage point in tariffs increases the likelihood of being hired as an intensive-informal worker by 0.59 percentage points.31 This result is consistent with the hypothesis that smaller registered employers tend to hire informal workers as a mean to reduce costs and remain competitive when there is an increase in import competition. Findings suggest that this is the case in industries where firms mostly have fewer than 50 employees. These firms are less likely to get caught when avoiding the payroll taxes that come with hiring formal workers. To get a more detailed picture at the industry level, the analysis also studies the effect that a reduction in output tariffs has on the importance of informal labor within registered firms by estimating equation (4). In this case, the dependent variable is the proportion of informal employment on registered employers: ShareINT jt = δ0 + δτ OT jt + δgSg jt ∗ OT jt + δs Sg jt + δi IT jt + Fj + Ft + η jt , (4) 30 Table S2.1 in the supplementary online appendix reports results for the intensive and extensive margin including an industry trend to make sure that the industry-year size indicator is not capturing any industry-specific trends. Results are qualitatively similar. 31 When testing the joint significance of the industry size indicators for the regression reported in column (4), the F-statistic is 4.21 and the corresponding p-value is 0.0176. Thus, the null hypothesis that both coefficients are zero is rejected. When testing whether the coefficients of the industry size indicators are different from each other, the F-statistic is 3.08 and the corresponding p-value is 0.0824. Hence, the null hypothesis that coefficients for the indicators for small and medium industries are the same is rejected. The World Bank Economic Review 163 where total intensive-informal employment ShareINT jt = in industry j in year t . total employment in registered firms Columns (5–8) in table 8 present results from the OLS estimation of equation (4).32 The estimation computes Huber–White standard errors clustered by industry at the 4-digit level of aggregation. All Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 specifications also control for industry and year fixed effects. Results in columns (5) and (6) find no statistically significant relation between the share of informal employment on total employment by registered employers and output tariffs. However, when controlling for average industry-year size in column (7), the findings show that a reduction in output tariffs changes the labor composition towards intensive-informal employment in industries where employers have fewer than 50 workers. Moreover, results remain unchanged when controlling for input tariffs in column (8). The coefficients remain almost the same. A reduction of 1 percentage point in output tariffs increases the share of intensive- informal workers by 1.9 percentage points in industries that, on average, have firms with fewer than 16 employees. Similarly, the same reduction in output tariffs translates into an increase of 1.3 percentage points in the share of intensive-informal employment by registered employers in medium-size industries.33 Results are consistent with the hypothesis that when tariffs fall, registered employers might be moti- vated to increase their share of informal employment. Since intensive-informal workers are cheaper than formal ones, and they might be facing a substantial reduction in profits, employers would be willing to risk getting caught when hiring informal workers only after import competition increases. Moreover, the findings suggest that this is true for industries where employers are, on average, small or medium size. Assuming that smaller firms are less likely to be audited, this result is not surprising. This analysis strengthens the validity of policies that would enhance inspection of labor regulation compliance by registered firms such as the one implemented by SUNAFIL.34 Importantly, it provides a guideline to such institutions as it would advise shifting the focus from larger firms towards small- and medium-size firms. In particular, when resources are of the essence, these sort of directions could translate into a more efficient policy implementation. 7.3. Extensive Margin This section studies the impact of an increase in import competition on the extensive margin of informality. First, it explores the effect on the probability of being hired as an extensive-informal worker when output tariffs fall. Second, it analyzes the impact on the employment composition between labor in the extensive margin of informality and formal or extensive-informal employment. Hence, equation (5) is estimated: EXT jt = γ0 + γτ OT jt + γgSg jt ∗ OT jt + γs Sg jt + γi IT jt +γH H t + Fj + Ft + ε jt , (5) where the dependent variable is EXT jt = 1 if individual is informal in the extensive margin and 0 otherwise. In other words, EXT jt = 0 if the individual is an intensive-informal worker or a formal worker. As in the previously estimated models, the chosen measure of import competition is output tariffs, OTjt , while controlling for input tariffs, ITjt , and individuals’ demographic characteristics, H t . Furthermore, equation (5) also controls for the average size in industry j at time t in which the individual is employed 32 Since the dependent variable is a proportion, table S1.3 in the supplementary online appendix presents results for a fractional logit model estimation. Results are very similar. 33 When testing the joint significance of the industry size indicators for the regression reported in column (8), the F-statistic is 4.44 and the corresponding p-value is 0.0142. Thus, the null hypothesis that both coefficients are zero can be rejected. Also, when testing whether the coefficients are different from each other, the F-statistic is 0.06 and the corresponding p-value is 0.0809. Hence, the null hypothesis that they are the same is also rejected. 34 See the discussion on labor market policies in the section entitled “Informal Labor”. 164 Cisneros-Acevedo Table 9. Extensive Margin of Informality Informality indicator Share of informal employment (1) (2) (3) (4) (5) (6) (7) (8) Output tariff 0.0034** 0.0035** 0.0035* 0.0035* 0.0054** 0.0054** 0.0058** 0.0055* (0.002) (0.002) (0.002) (0.002) (0.003) (0.003) (0.003) (0.003) Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 Input tariff — — — −0.0002 — — — 0.0021 (0.003) (0.007) S * output tariff — — −0.0035 −0.0035 — — −0.0042 −0.0042 (0.003) (0.003) (0.008) (0.008) M * output tariff — — 0.0017 0.0017 — — 0.0008 0.0009 (0.003) (0.003) (0.005) (0.005) Small industry (S) — 0.0171 0.0503** 0.0503** — 0.2475*** 0.2667*** 0.2669*** (0.021) (0.021) (0.022) (0.039) (0.053) (0.053) Medium industry (M) — 0.0166 0.0052 0.0052 — 0.0316 0.0254 0.0248 (0.015) (0.017) (0.017) (0.027) (0.032) (0.032) Observations 31,009 31,009 31,009 31,009 695 695 695 695 R2 0.38 0.38 0.38 0.38 0.71 0.76 0.76 0.76 Source: Author’s elaboration based on data from the Peruvian Household Survey (ENAHO) for 2007–2014, Most Favoured Nation (MFN) tariff data from the World Trade Organization’s Tariff Analysis Online, and the Peruvian Input-Output Table for 2007. Note: Specifications in columns (1–4) include region fixed effects. All specifications include controls for individual characteristics (civil status, education, ethnicity, and age), industry and year fixed effects. Average industry-year size is classified as S: ≤15 workers, M: >15 and ≤50 workers, L: >50 workers. Robust standard errors for industrial clusters in parentheses. *p < 0.10, **p < 0.05, ***p < 0.01. by indexing size with Sgjt , where g = s, m, l. When g = s, firms in the industry have on average fewer than 16 employees, when g = m they have more than 15 workers and less than 51 worker, when g = l they have more than 50 workers. Unlike when studying the intensive margin of informality, the sample of analysis is not constrained. It takes into account all individuals as it did when estimating equation (1). However, the dependent variable, in this case, takes the value 1 only for extensive-informal individuals rather than informal in any margin. Equation (5) is estimated with a linear probability model and it computes Huber–White robust standard errors clustered by industry at the 4-digit level of aggregation. It includes industry, Fj , year, Ft , and region fixed effects in all specifications. Results for equation (5) are in columns (1–4) in table 9.35 Column (1) only controls for individual characteristics. Findings show that young married men with above primary education are less likely to be hired as extensive-informal workers. These results are consistent in all specifications. Furthermore, when output tariffs are cut, the probability of being hired as an extensive-informal worker also falls. This result is robust to the inclusion of industry-year size indicator in column (2). In the same way, results are not altered by the addition of interaction terms between industry-year size indicator and output tariffs in column (3). From columns (2) and (3) is evident that the average industry size does not impact the probability of being hired as an extensive-informal worker when import competition increases. More interestingly, results remain unchanged when including input tariffs in column (4). Findings in column (4) suggest that a reduction of 1 percentage point in output tariffs decreases the probability of being hired as an extensive-informal worker by 0.35 percentage points. In other words, a decline of 50 percentage points in tariffs would diminish the probability of being employed in the intensive margin of informality by 17.5 percentage points.36 Thus, the magnitude of the effect is less than that of the intensive margin. Nevertheless, the impact on the extensive margin is on all industries, while the effect 35 Results are similar when estimating logit models (see table S1.2 in the supplementary online appendix). 36 When testing the joint significance of the industry size indicators for the regression reported in column (4), the F-statistic is 3.39 and the corresponding p-value is 0.0377. Thus, the null hypothesis that both coefficients are zero is rejected. Also, The World Bank Economic Review 165 in the intensive margin discussed in the previous section, is only on industries that, on average, have small- and medium-size firms. This result is consistent with the hypothesis that unregistered employers are the least productive ones. Consequently, the probability of being hired in the extensive margin of informality falls when import competition increases. Next, this study estimates equation (6) to gain insight on the effect that a reduction in output tariffs has on the proportion of employment by unregistered employers in the manufacturing sector. Columns Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 (5–8) in table 9 report the results from the OLS estimation. It computes Huber–White robust standard errors clustered by industry at the 4-digit level of aggregation. It includes industry and year fixed effects in all specifications. Table S1.4 in the supplementary online appendix presents results when estimating equation (6) as a fractional logit response model and the findings are similar: ShareEXT jt = φ0 + φτ OT jt + φgSg jt ∗ OT jt + φs Sg jt + φi IT jt + Fj + Ft + η jt , (6) where total extensive-informal employment ShareEXT jt = in industry j in year t . total employment Columns (5–8) in table 9 show that a reduction in output tariffs diminishes the share of extensive- informal employment on total employment. This result holds when adding industry-year size indicators and when including interaction terms between these size indicators and output tariffs. Moreover, the relationship between import competition and the relative importance of extensive-informal employment remains unchanged both quantitatively and qualitatively when controlling for input tariffs in column (8). From column (8), the results suggest that a decrease of 1 percentage point in output tariffs translates into a reduction of 0.55 percentage points in the share of extensive-informal employment in the manufacturing sector.37 Additionally, columns (6–8) show that industries with average firm size below 16 employees are more likely to show an increase in the proportion of extensive-informal employment. Results from table 9 are consistent with findings that the least productive firms are the first ones to leave the market when a trade liberalization process takes place.38 Since the literature tends to describe unregistered firms as unproductive (La Porta and Shleifer 2014; Ponczek and Ulyssea 2021), it makes sense that when import competition increases, the probability of being hired in the extensive margin of informality decreases and the share of extensive-informal employment on total employment also falls. This finding contradicts those that advocate for a reduction on entry regulations in the formal market as a necessary measure to reduce informality (De Soto 1989, 2000). In fact, it suggests that, in the context of globalization, it is not necessary for policy makers to implement regulations that would help informal firms formalize as these unproductive firms will naturally disappear and their workers would have to seek employment in registered (more productive) firms. Importantly, this study finds that informality in both margins will not decrease as a natural consequence of globalization. As discussed in the section Identification Strategy and Results when analyzing the intensive margin of informality, trade liberalization should be coupled with strong policies that ensure registered firms compliance with labor law regulations. when testing whether the coefficients of the industry size indicators are different from each other, the F-statistic is 5.89 and the corresponding p-value is 0.0170. Hence, the null hypothesis that coefficients for the indicators for small and medium industries are the same is rejected. 37 When testing the joint significance of the industry size indicators for the regression reported in column (8), the F-statistic is 12.87 and the corresponding p-value is 0.0000. Thus, the null hypothesis that both coefficients are zero is rejected. When testing whether the coefficients are different from each other, the F-statistic is 15.23 and the corresponding p-value is 0.0002. Hence, the null hypothesis that they are the same is rejected. 38 This finding is widely accepted in the literature following Melitz (2003). 166 Cisneros-Acevedo 8. Robustness Peru recorded high rates of growth in the 2000s and an increase in domestic demand. Chacaltana (2017) studies the effect of sectoral growth on formalization. He finds that growth reduces informality on employment-intensive sectors such as farming, commerce, and services. Moreover, he highlights the im- portance of growth composition. He points out that regional value-added per worker and regional output share of labor-intensive sectors are crucial in the formalization process. Chacaltana (2017) concludes that Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 growth did not have an effect on formalization in manufacturing. Even though the focus of this paper is the manufacturing sector, this study performs a robustness check on this matter. Due to data constraints, GDP or domestic demand at the ISIC 4-digit level is not available. However, it is possible to control for import and export growth. If imports increase while the GDP is also increasing, import growth would proxy an increase in domestic demand. Given that national production is growing, an increase in imports would mean that more consumers at home have increased their demand for manu- facturing goods. Hence, table S2.2 in the supplementary online appendix reports results when controlling for import growth. Performing a similar analysis to the one discussed in the section on Identification Strategy and Results for the extensive and intensive margins of informality, results are almost identical for the intensive margin. Table S2.2 also presents results when controlling for export growth as it could be that the excess of supply in the domestic market is sent abroad. Results remain almost unchanged. Findings are slightly different for one of the specifications concerning the extensive margin of informality. Nevertheless, the overall conclusions are qualitatively similar. Results in the section Identification Strategy and Results control for input tariffs but not for the interac- tion input tariffs-industry size. If workers in the sample are employed by exporting firms, which are known to be larger, it might be the case that the reduction in input tariffs benefits them more than the increase in import competition harms them.39 With that in mind, table S2.3 in the supplementary online appendix reports results controlling for said interaction. The findings remain very similar to those reported in the section Identification Strategy and Results for the extensive and intensive margins of informality. During the period of analysis, China and the United States signed FTAs with Peru. Both countries are important trade partners for Peru. Since the measure of trade liberalization of choice is MFN tariffs, by definition it does not reflect the preferential treatment granted to China and the United States. To my knowledge, there is no availability of applied tariffs disaggregated at the 4-digit level within manufactur- ing as there is of MFN tariffs. Hence, to alleviate concerns regarding the potential effect that the FTAs with China and the United States might have had on informal labor in Peru, tables S2.4 and S2.5 in the supplementary online appendix report results when controlling for the share of imports coming from China and the United States on total Peruvian imports. An increase in import competition due to an increase in trade coming from China or the United States might be the cause of the effects on informality documented in this study. Thus, it is important to control for the importance of imports coming from the United States and China in regressions similar to those presented in the section Identification Strategy and Results for the extensive and intensive margins of informality. Findings show that even when controlling for these potential confounding factors, results remain unchanged. Conclusions are almost the same, both qualitatively and quantitatively.40 Tables S2.6 and S2.7 in the supplementary online appendix report results from restricting the analysis to the period 2007–2011. There are three sources of concern. First, due to a change in the questionnaire in 2012, the definition of intensive-informal workers changed. Second, for some unknown reason, Peru did not report tariffs to the WTO in 2012. Hence, the tariffs data for that specific year comes from the 39 Nataraj (2011) finds that larger-formal firms’ productivity increases when input tariffs fall. 40 Pierola, Sanchez-Navarro, and Mercado (2019) find that greater import competition due to the surge of imports from China in 2001–2010 increased the chances of working in the informal sector among workers with only elementary education. This result does not necessarily contradict this paper’s findings as they put together the intensive and the extensive margins of informality. The World Bank Economic Review 167 BCRP. Third, based on Chacaltana (2017), I claim that the introduction of the electronic payroll system’s effect is negligible on the evolution of informality. However, I do acknowledge that this may no longer be true after SUNAFIL started to operate in 2013. Main conclusions reported in the section Identification Strategy and Results are very similar to those for the period 2007–2011. Finally, results in the section Identification Strategy and Results include both self-employed workers and employees. Self-employed workers predominantly declare themselves to be in the extensive margin of Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 informality and to be employed mostly in “textiles,” “wearing apparel,” “furniture,” and “food products and beverages.” The section Identification Strategy and Results includes self-employed workers as I think that they are essentially employed by small informal undertakings and it is not entirely accurate to classify them as self-employed. However, if they do not work for an employer, the mechanisms I claim in previous sections would not apply. Hence, tables S2.8 and S2.9 in the supplementary online appendix present results when equations (3)–(6) are estimated excluding self-employed workers. Results are very similar.41 9. Conclusions This paper highlights the importance of distinguishing informal workers employed by a registered firm from those working in an unregistered one. Specifically, this study shows that disentangling these two margins of informality is vital when studying the effect of trade liberalization on informal labor. First, a descriptive analysis of employees in the Peruvian manufacturing sector shows that the demographic characteristics of informal workers in the intensive margin differ from those in the extensive margin of informality. Hence, it is essential to control for individual features when relating trade to informal employment. Moreover, this study confirms that the Peruvian informal sector bears several features described by the literature when describing the informal sector, namely, low wages and small firm size. Moreover, the descriptive analysis highlights how some industries within the manufacturing sector are more prone to one margin of informality than others. Hence, when studying informality at the industry level it is also important to distinguish these two margins of informality. Furthermore, the empirical analysis demonstrates that an increase in import competition impacts the two margins of informality through entirely different channels. A reduction in tariffs triggers a decrease of extensive-informal employment and an enlargement of intensive-informal employment. Even though it is not possible to observe firms’ decisions due to data constraints, it is possible to observe the labor outcomes of those decisions in the context of globalization. On the one hand, registered employers respond to the increase in import competition by hiring cheaper (informal) workers. Moreover, since small- and medium-size firms are less likely to be audited by the Tax Agency, they are more inclined to hire informal workers than larger firms. Hence, results suggest that the share of intensive-informal employment expands in industries that, on average, have small- and medium-size employers. In this way, trade liberalization increases informal labor in the intensive margin. On the other hand, extensive-informal workers are characterized by very low productivity. As a result, their employers are not able to cope with stronger competition and are less likely to survive when tariffs fall. Thus, the findings suggest that an increase in import competition reduces both the probability of being hired in the extensive margin of informality and the share of extensive-informal employment on total employment. Importantly, these results hold for all industries; they are not dependent on average industry size. 41 It is worth noting that 1 percent of self-employed workers declared themselves to be formal. Formal workers excluded from the analysis are mostly employed in the following industries: “wearing apparel,” “fabricated metal products,” “food products and beverages,” and, to a lesser extent, “publishing and printing.” 168 Cisneros-Acevedo Since the two margins of informality are not independent, the estimations for each margin of infor- mality are performed separately. By comparing results presented in the section Identification Strategy and Results, it is evident that the tariff reduction effect on the intensive margin is stronger than the impact on the extensive margin. Hence, I infer that the intensive margin of informality drives the trade liberalization effect on informality. In this way, trade increases informality. This result is disconcerting as many anticipated globalization would reduce informality. This paper Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 provides valuable insight as it acknowledges that trade reduces informality but only on the extensive margin. When considering both the intensive and the extensive margins, trade no longer weakens informal labor overall. Nevertheless, this finding is not entirely pessimistic as it can contribute to a more targeted policy aiming to eradicate informality. Given this paper’s findings, policy makers might obtain better results when implementing a policy that focuses on encouraging registered firms to hire workers formally during a trade liberalization process. Even though red tape reduction and tax cuts for small (informal) firms are potentially helpful in the fight against the extensive margin of informality, they might not be the most adequate use of resources as trade liberalization shrinks this portion of informal employment on its own. Data Availability The data underlying this article is available in its online supplementary material at the The World Bank Economic review website. References Acosta, P., and G. Montes-Rojas. 2014. “Informal Jobs and Trade Liberalisation in Argentina.” Journal of Develop- ment Studies 50 (8): 1104–18. Amaral, P., and E. Quintin. 2006. “A Competitive Model of the Informal Sector.” Journal of Monetary Economics 53 (7): 1541–53. Bosch, M., E. Goni-Pacchioni, and W. Maloney. 2012. “Trade Liberalization, Labor Reforms and Formal–Informal Employment Dynamics.” Labor Economics 19 (5): 653–67. Chacaltana, J. 2017. “Peru, 2002-2012: Growth, Structural Change and Formalization.” CEPAL Review 2016 (119): 45–64. Chong, A., J. Galdo, and J. Saavedra-Chanduvi. 2008. “Informality and Productivity in the Labor Market.” Journal of Economic Policy Reform 11 (4): 229–45. Cruces, G., G. Porto, and M. Viollaz. 2018. “Trade Liberalization and Informality in Argentina: Exploring the Ad- justment Mechanisms.” Latin American Economic Review 27 (1): 13. De Paula, A., and J. A. Scheinkman. 2011. “The Informal Sector: An Equilibrium Model and Some Empirical Evidence from Brazil.” Review of Income and Wealth 57 (s1): S8–S26. De Soto, H. 1989. The Other Path: The Invisible Revolution in the Third World. New York: Harper and Row. ———. 2000. The Mystery of Capital: Why Capitalism Triumphs in the West and Fails Everywhere Else. New York: Basic Books. Díaz, J. J. 2014. Formalización empresarial y laboral. Number 40741 in Coediciones. Naciones Unidas Comisión Económica para América Latina y el Caribe (CEPAL). https://econpapers.repec.org/bookchap/ ecrcol013/40741.htm Dix-Carneiro, R., and B. K. Kovak. 2015. “Trade Liberalization and the Skill Premium: A Local Labor Markets Approach.” American Economic Review 105 (5): 551–57. ———. 2017. “Trade Liberalization and Regional Dynamics.” American Economic Review 107 (10): 2908–46. ———. 2019. “Margins of Labor Market Adjustment to Trade.” Journal of International Economics 117: 125– 42. Dix-Carneiro, R., R. R. Soares, and G. Ulyssea. 2018. “Economic Shocks and Crime: Evidence from the Brazilian Trade Liberalization.” American Economic Journal: Applied Economics 10 (4): 158–95. Farrell, D. 2004. “The Hidden Dangers of the Informal Economy.” McKinsey Quarterly 3 (3): 26–37. The World Bank Economic Review 169 FORLAC. 2014a. “Recent Experiences of Formalization in Latin America and the Caribbean.” Notes on Formaliza- tion. https://www.ilo.org/wcmsp5/groups/public/—americas/—ro-lima/documents/publication/wcms_245882.pdf ———. 2014b. “Trends in Informal Employment in Peru: 2004-2012.” Notes on Formalization. http://ciaris.ilo.org/global/docs/WCMS_245891/lang–en/index.htm Goldberg, P. K., and N. Pavcnik. 2003. “The Response of the Informal Sector to Trade Liberalization.” Journal of Development Economics 72 (2): 436–96. Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 ———. 2007. “Distributional Effects of Globalization in Developing Countries.” Journal of Economic Literature 45 (1): 39–82. ILO. 2018. Women and Men in the Informal Economy: A Statistical Picture. https://www.ilo.org/global/ publications/books/WCMS_626831/lang–en/index.htm INEI. 2018. Peru: Indicadores de Empleo e Ingresos por departamento 2007–2017. https://www.inei.gob.pe/ media/MenuRecursivo/publicaciones_digitales/Est/Lib1537/index.html Infante, R., and J. Chacaltana. 2014. Hacia un desarrollo inclusivo: el caso del Perú. Number 40343 in Coediciones. Naciones Unidas Comisión Económica para América Latina y el Caribe (CEPAL). https://ideas.repec.org/b/ecr/col013/40343.html Kanbur, R. 2009. “Conceptualising Informality: Regulation and Enforcement.” Indian Journal of Labour Eco- nomics 52. https://www.researchgate.net/publication/45142672_Conceptualising_Informality_Regulation_and_ Enforcement/citation/download Kleven, H. J., C. T. Kreiner, and E. Saez. 2016. “Why Can Modern Governments Tax So Much? An Agency Model of Firms as Fiscal Intermediaries.” Economica 83 (330): 219–46. La Porta, R., and A. Shleifer. 2008. “The Unofficial Economy and Economic Development.” Brooking Papers on Economic Activity 105 (3): 473–522. ———. 2014. “Informality and Development.” Journal of Economic Perspectives 28 (3): 109–26. Levy, S. 2008. “Good Intentions, Bad Outcomes: Social Policy, Informality, and Economic Growth in Mexico”. Brook- ings Institution Press. https://www.semanticscholar.org/paper/Good-Intentions%2C-Bad-Outcomes%3A-Social- Policy%2C-and-Levy/be6222b87a7605da5e10674e2a5861197f6a9822 Maloney, W. 2004. “Informality Revisited.” World Development 32 (2): 27–302. McCaig, B., and N. Pavcnik. 2015. “Informal Employment in a Growing and Globalizing Low-Income Country.” American Economic Review 105 (5): 545–50. ———. 2018. “Export Markets and Labor Allocation in a Low-Income Country.” American Economic Review 108 (7): 1899–941. Melitz, M. 2003. “The Impact of Trade on Industry Reallocations and Aggregate Industry Productivity.” Econometrica 71 (6): 1695–725. Menezes-Filho, N. A., and M.-A. Muendler. 2011. “Labor Reallocation in Response to Trade Reform.” Working Paper 17372, National Bureau of Economic Research. https://scholar.google.de/citations?view_ op=view_citation&hl=en&user=ehYoE6UAAAAJ&citation_for_view=ehYoE6UAAAAJ:d1gkVwhDpl0C Nataraj, S. 2011. “The Impact of Trade Liberalization on Productivity: Evidence from India’s Formal and Informal Manufacturing Sectors.” Journal of International Economics 85 (2): 292–301. Nuñez, G. 2014, 09. “Existe aun la Industria Manufacturera en el Peru.” Pensamiento Critico 11, 011. https://www.researchgate.net/publication/322163771_Existe_aun_la_industria_manufacturera_en_el_Peru OECD. 2019. Multi-dimensional Review of Peru: Volume 3. From Analysis to Action, OECD Development Pathways. OECD Publishing, Paris. Pavcnik, N. 2017. “The Impact of Trade on Inequality in Developing Countries.” Proceedings of Jackson Hole Eco- nomic Symposium. Paz, L. S. 2014. “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. Perry, G. E., W. F. Maloney, O. S. Arias, P. Fajnzylber, A. D. Mason, and J. Saavedra-Chanduvi. 2007. “Informality: Exit and Exclusion.” Number 6730 in World Bank Publications. The World Bank. https://books.google.de/books/about/Informality.html?id=JXIQHXi_crAC&redir_esc=y Pierola, M. D., D. Sanchez-Navarro, and E. Mercado. 2019. “Import Competition in the Manufacturing Sector in Peru: Its Impact on Informality and Wages.” IDB Working Paper Series No IDB- WP -01093. 170 Cisneros-Acevedo Ponczek, V, and G Ulyssea. 2021. “Enforcement of Labor Regulation and the Labor Market Effects of Trade: Evidence from Brazil,” No 15960 CEPR Discussion Papers. https://EconPapers.repec.org/RePEc:cpr:ceprdp:15960 Rauch, J. E. 1991. “Modelling the Informal Sector Formally.” Journal of Development Economics 35 (1): 33–47. Samaniego de la Parra, B. 2016. “Formal firms, Informal Workers and Household Labor Supply in Mexico.” Job Market Paper, University of Chicago. Topalova, P. 2007. “Trade Liberalization, Poverty and Inequality: Evidence from Indian Districts”. NBER Chapters. National Bureau of Economic Research, Inc. Downloaded from https://academic.oup.com/wber/article/36/1/141/6406720 by LEGVP Law Library user on 08 December 2023 Topalova, P., and A. Khandelwal. 2011. “Trade Liberalisation and Firm Productivity: The Case of India.” Review of Economics and Statistics 93 (3): 995–1009. Ulyssea, G. 2020. “Informality: Causes and Consequences for Development.” Annual Review of Economics 12 (1): 525–46. Villagomez, M., and R. Chafloque. 2019, 04. La Informalidad Laboral en el Peru. Un Mapa Nacional Basado En ENAHO. Universidad de San Martin de Porres. WB. 2018. “The World Bank Annual Report 2018.” Number 6730 in World Bank Publications. Washington, DC. World Bank.