Policy Research Working Paper 9982 Immigration, Labor Markets and Discrimination Evidence from the Venezuelan Exodus in Perú Andre Groeger Gianmarco León-Ciliotta Steven Stillman Social Sustainability and Inclusion Global Practice March 2022 Policy Research Working Paper 9982 Abstract Venezuela is currently experiencing the biggest crisis in its document a causal relationship between the level of employ- recent history. This has led to a large increase in emigration. ment in the informal sector—where most immigrants are According to recent estimates, there are a total of 5.6 million employed—and reports of discrimination. The second part Venezuelan immigrants worldwide with over one million is focused on studying the impact of Venezuelan migra- now living in Peru, which has led to an over 2 percent tion on local’s labor market outcomes, reported crime rates increase in the country’s population. Unlike in many other and attitudes using a variety of data sources. The results episodes of refugee migration, Venezuelan immigrants are provide evidence that inflows of Venezuelans to particu- not only very similar in cultural terms, but are, on average, lar locations in Peru lead to better labor market outcomes also more skilled than Peruvians. This study first examines for locals, decreased reported crime, as well as improved Venezuelans’ perceptions about being discriminated against reported quality of local services, greater trust in neighbors in Peru. Using an instrumental variable strategy, the results and higher community quality. This paper is a product of the Social Sustainability and Inclusion Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at steven.stillman@unibz.it. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Immigration, Labor Markets and Discrimination: Evidence from the Venezuelan Exodus in Perú∗ Andre Groeger Gianmarco León-Ciliotta Universitat Autonoma de Barcelona Universitat Pompeu Fabra & BSE & MOVE & BSE & IPEG Steven Stillman Free University of Bozen-Bolzano JEL Codes : F22, J15, O15, R23 ∗ Groeger: Universitat Autonoma de Barcelona (UAB), Barcelona School of Economics (BSE) and Markets, Organizations and Votes in Economics (MOVE), andre.groger@uab.es; León-Ciliotta: Universitat Pompeu Fabra (UPF), Barcelona School of Economics (BSE), and Institute of Political Economy and Governance (IPEG), gianmarco.leon@upf.edu; Stillman: Free University of Bozen-Bolzano, steven.stillman@unibz.it. We thank Diego Aguilar for outstanding research assistance. This paper was commissioned by the World Bank Social Sustainability and Inclusion Global Practice as part of the activity "Preventing Social Conflict and Promoting Social Cohesion in Forced Displacement Contexts." The activity is task managed by Audrey Sacks and Susan Wong with assistance from Stephen Winkler. This work is part of the program "Building the Evidence on Protracted Forced Displacement: A Multi-Stakeholder Partnership." The program is funded by UK aid from the United Kingdom’s Foreign, Commonwealth and Development Office (FCDO), it is managed by the World Bank Group (WBG) and was established in partnership with the United Nations High Commissioner for Refugees (UNHCR). The scope of the program is to expand the global knowledge on forced displacement by funding quality research and disseminating results for the use of practitioners and policy makers. This work does not necessarily reflect the views of FCDO, the WBG or UNHCR. 1 Introduction Crisis-driven migration flows have significantly increased in the past two decades (Bhabha, 2018). The arrival of a large number of migrants and refugees have triggered fierce political disputes over its impact on local labor markets and have been the center of much discussion in the academic literature, as well as in the media. While local’s perceptions about the effects of immigration on the labor market determine their behavior and attitudes towards immigrants, the way in which these perceptions are formed are much less well understood. We study the economic underpinnings of hostility and discrimination against immigrants. The current crisis in Venezuela has led about 5.6 million people to emigrate (R4V, 2021). This migration wave intensified in 2017 when the country’s political instability added to the worsening economic situation in the country. While many Venezuelans chose neighboring Colombia, Brazil and Ecuador as their destinations, at the time, the Peruvian economy was experiencing sustained economic growth, thus about one million Venezuelans were attracted to the country. This large inflow of immigrants potentially put pressure on local labor mar- kets, especially in large cities, and increased the negative public discourse against immigrants in the media (Winter, 2020; Freier et al., 2021). Some of these sources claiming, not only that the large wave of newcomers had led to an increase in unemployment, but it had also contributed to an upsurge in urban violence. Our analysis has two parts. First, we use a specialized survey of Venezuelan immigrants in Peru to identify the causal effect of the quality of local labor markets on discrimination against immigrants. Then, we turn to the economic determinants of attitudes towards immigrants, and study the flip-side of the first analysis, namely, how does the presence of immigrants affect Peruvians’ labor market outcomes, overall criminal activity, and their perceptions about crime and their local community. In both analyses, the main identification issue is that Venezuelans are not randomly allocated to specific locations in Peru and hence unobserved characteristics of both the location and the Venezuelans there might be correlated with local economic conditions and individual outcomes for both Venezuelans and Peruvians in the same location. We account for this using different instrumental variables strategies combined with detailed controls on the local economic environment. In our first analysis, we examine whether Venezuelans who live in local areas with a stronger informal sector experience different levels of discrimination. A large majority of Venezuelans arriving to Peru have taken up jobs in the informal sector, directly competing with rela- tively low skilled native workers. To causally identify the relationship, we use a shift-share instrumental variable strategy that exploits local exposure to exogenous national-level ex- port shocks (Jaeger et al., 2018). As informal employment and discrimination could both be related to other area characteristics, we also control for the local industrial structure, household expenditure, population size, distance from the capital and center of economic activity (Lima), and, importantly, the number of Venezuelans based in each location prior to the current immigration wave, which we show to be a significant pull factor for where Venezuelans settle. Our results show that weaker informal labor markets lead to significant increase in the dis- crimination reported by Venezuelans in Peru. Overall, a 10% decrease in the informal em- ployment rate increases discrimination by 2.3-3%. This effect is twice as large for men as for women. The data we use also collects information on where discrimination occurs. We find that weaker informal labor markets lead to more discrimination for men in public places, as well as on public transit and, for women, on public transit exclusively. We do not find evidence of an impact on workplace discrimination for either gender. One interesting pattern is that We find that more educated Venezuelans are more likely to report being discriminated against. A potential explanation for this, consistent with the previous results, is that higher skilled Venezuelans are disappointed with their situation in Peru, especially when they settle in areas with strong labor markets, and this lack of opportunity is either caused by or perceived as discrimination (Guerrero-Ble et al., 2020). In the second part of our analysis, we examine the impact of immigration in terms of changes in the number of Venezuelans as a share of the local population in a province on a wide-variety of outcomes. We rely on administrative data to measure the number of Venezuelans newly registered in each district in Peru on a monthly basis between January 2015 and December 2020. We aggregate this information at the province level, which roughly corresponds to a labor market. Having a time-varying measure of the presence of Venezuelans in each of the 198 provinces allows us to use repeated cross-sectional data on outcomes for Peruvians and control for location and time fixed effects, as well as, location-specific time-trends. Hence, we identify the impact of the presence of Venezuelans by examining how outcomes for Peruvians change when more Venezuelans arrive in a province, conditional on the trend in that outcome. However, it is possible that local shocks impact both the destination choice of Venezuelans and outcomes for Peruvians, hence we also use an instrumental variable strategy where we in- strument for the number of Venezuelans in a location with the presence of Venezuelans in that 3 location in the past, interacted with the year of observation. This is a semi-parametric ver- sion of the traditional migrant network instrument as recommended by Goldsmith-Pinkham et al. (2020) and it allows the strength of the network effect to potentially vary in each year. An overidentification test can be used to examine whether the instrument has a consistent relationship over time. We find robust evidence that increased immigration from Venezuela has a positive impact on labor market outcomes for Peruvians, with increased employment rates, incomes and expenditure in locations that receive more Venezuelans. Additionally, locations that receive more immigrants have lower levels of reported non-violent crime, improved reported quality of local services, greater reported trust in neighbors and higher reported community quality. On the other hand, we find evidence that in locations with more Venezuelans, Peruvians report that their community likes diversity less. There are a number of potential explanations for these findings that we plan to explore in future work.1 The arrival of Venezuelans may have expanded the economic opportunities for Peruvian because of their higher levels of potential productivity, due to higher human capital, and their concentration in low wage jobs. Furthermore, most of these jobs are in the service sector which potentially could have freed up time, especially for Peruvian women, to be more engaged in the labor market, as well as lowered the costs for these types of goods and services. Venezuelans also might have expanded opportunities by increasing the demand for certain goods and services. The paper proceeds as follows. In section 2, we describe the context and institutional back- ground, then turn to present the related literature in section 3. Section 4 describes the data we use in our analysis as well as our empirical model and identification strategy. We then present the results in Section 5, and finally we discuss policy implications and conclude. 2 Context Venezuela is currently experiencing the biggest crisis in recent history. A deep economic and humanitarian crisis started ramping up in with the fall in oil prices and the death of former president Hugo Chavez in 2013 (Chaves-González and Echevarría Estrada, 2020). This has led to what some authors have called the great Venezuelan exodus (Hausmann et al., 2018; Rozo and Vargas, 2021). In mid-2016, large waves of migrants started to leave the country, with Colombia (1’700,000), Peru (870,000), Ecuador (385,000) and Chile (371,000) being their 1 Additional data which we do not currently have access to on firms and consumer prices is needed to examine these explanations in more detail. 4 main destinations (data reported in Boruchowicz et al. (2021) up to June 2019). According to recent estimates, there are a total of 5.6 million Venezuelan immigrants worldwide, and the number of Venezuelans living in Peru has increased from 6,615 in 2016 to more than 840,000 by June 2019 (see Figure 1), and has gone up to one million by 2021 (R4V, 2021). This wave of immigration has increased Peru’s population by around 2 percent. Travelling from Venezuela to Peru entails a journey of over 4,500 kilometers, and before 2017 Peruvian authorities required all immigrants to be in possession of a passport (without any visa requirements). However, obtaining a passport in Venezuela at the time was difficult, as processing times were extremely long and the overall process was relatively expensive. In light of this situation, in 2017, the Peruvian government implemented a temporary residence permit (permiso temporal de permanencia, henceforth PTP). This permit allowed immigrants to legally work and study in the country, pay taxes and open a bank account. According to the national statistical institute (INEI), 97% of Venezuelan immigrants were able to get a PTP by 2019, implying that the large majority of Venezuelan immigrants having arrived up to that point were legally in Peru and able to work both in the formal and informal sector. Unlike in other episodes of crisis migration, such as that of Syrian refugees during the recent civil war or Central American immigrants in the US, Venezuelan immigrants in Peru are relatively similar to Peruvians in terms of language and religion. Nevertheless, they were, on average, significantly more skilled than Peruvians: as of December 2018, 47.8% of Peruvians had less than secondary education which was true of only 17.2% of Venezuelan immigrants (see Table 1). However, there is anecdotal evidence that Peruvians view Venezuelans as con- tributing little to the economy and that their presence in the country has increased criminal activities (Janetsky, 2019). Furthermore, there is some evidence that this has started to lead to political backlash (Winter, 2020). Before the pandemic, the Peruvian economy was averaging around a 5% annual growth rate, which made it an attractive destination for migrants. Moreover, the labor market is highly informal (and therefore flexible): In 2018, only 21% of Peruvians held a formal job. Boruchowicz et al. (2021) show that the Venezuelan exodus had negligible effects on the Peruvian labor market, and argue that this is precisely due to the flexibility associated with the high levels of informality in the labor market.2 Very little is known about the actual discrimination affecting Venezuelan immigrants in the country. Still, it is plausible that part of the discrimination reported by immigrants is related 2 There is some contrasting evidence that the increase in Venezuelan immigrants in Peru led to small decreases in employment rates and earnings of low skilled (and specially female) Peruvian workers in the informal sector (Morales and Pierola, 2020; Asencios and Castellares, 2020). 5 to the way they are portrayed in the media. Freier et al. (2021) provides a detailed analysis of how the Peruvian written media has refers to Venezuelan immigrants. They show that 46% of articles refer to Venezuelan immigrants in a neutral fashion, while 28% (26%) of them have a negative (positive) tone. Still a high proportion of articles (44%) mention a problem associated with immigrants, with the most prominent ones being crime (26%), the contribution to unemployment (7%), and their effects on wages (4%). 3 Motivation and Related Literature Our paper has two main components. First, we test whether better labor market conditions impact discrimination against immigrants that have arrived as part of a forced displacement. Second, we examine how the presence of Venezuelans influences labor market outcomes and the attitudes of local Peruvians. Hence, this paper relates to different strands of the literature analyzing the effects of immigration at destination on labor market outcomes, discrimination, and popular attitudes. There is a long-standing and rich literature in economics concerned with the effects of im- migration on labor markets in developed countries (Borjas, 1983; Card, 2001). There is a more recent literature on the impact of crisis migration on less developed countries. Most of this literature has focused on the impact of Syrian refugee immigration on natives in Turkey and Jordan and has found that these inflows reduce the employment and wages of low-skilled natives (Del Carpio and Wagner, 2015; Tumen, 2016; Ceritoglu et al., 2017). There is also a growing literature on the effects of the current Venezuelan exodus on neighboring countries in Latin America. Results are more mixed than in the Syrian case and appear to relate to the conditions in the destination country.3 We also contribute to the literature on the effects of immigration on popular opinion and dis- crimination. Most of the literature concentrates on developed countries. For example, Alesina et al. (2018) document strong (negative) misperceptions about the magnitude of immigra- tion and the characteristics of immigrants among natives across major OECD destination 3 For Colombia, existing studies have mostly identified negative effects (Lebow, 2020; Bahar et al., 2021; Delgado-Prieto, 2021; Lebow et al., 2021). One exception is Santamaria (2021) who finds null effects on labor market outcomes of native workers. For Ecuador, Olivieri et al. (2020) do not find any effects on natives’ labor market outcomes on average, but identify a deterioration of employment quality and earnings among young and low-educated natives in high immigration regions. For Peru, Boruchowicz et al. (2021) find null effects, whereas Morales and Pierola (2020) find small positive effects on employment quality for high-skilled workers and a decrease in monthly earnings for secondary educated natives. Unlike this previous literature, we find that inflows of Venezuelans to Peru lead to positive labor market effects for natives in general with increased employment, incomes and expenditures among Peruvians. 6 countries. They also find that these misperceptions lead to a negative bias in preferences for redistribution, which does not disappear with the provision of objective information. Hangartner et al. (2019) find that refugee immigration to Greece increased natives’ hostility toward refugees and support for restrictive immigration policies. There is a small literature on the effects of the Venezuelan exodus on popular opinion in Latin America. For Chile, Ajzenman et al. (2021) find negative effects on natives’ security perceptions, despite null effects on the objective crime rate.4 They identify media to be an important determinant in this context due to disproportionate coverage of immigrant- perpetrated homicides that trigger crime misperceptions. For Colombia, Chatruc and Rozo (2021) find that economic concerns, despite a lack of objective evidence on negative labor market effects, are another important driver of anti-immigrant sentiment. Additionally, Rozo and Vargas (2021) identify strategic electoral misinformation in Colombia as an additional channel. Exploiting rich opinion polls, we contribute to this literature by estimating the causal impact of migration on both attitudes and perceptions of natives and immigrants in the same context. There are also a number of studies conducted within the DFID-WB-UNHCR initiative "Pre- venting social conflict and promoting social cohesion in forced displacement contexts" that relate to our analysis. Relying on data from the Gallup World Poll, Aksoy and Ginn (2021) analyze the effect of refugee inflows on natives’ attitudes towards refugees in a global sample of low- and middle income countries. Interestingly, they do not find evidence of a negative effect on attitudes. They also investigate potential heterogeneity in those effects across camp and non-camp refugee settings as well as across progressive and restrictive right-to-work environments and find little evidence for any differential effects. For the case of Colombia, Allen et al. (2021) conduct a survey experiment investigating the preferred policy preferences of native Colombians in response Venezuelan mass immigration. On the one hand, their results suggest that natives favor the use of numerical immigration limits and limiting the length of the residency permit. On the other hand, natives support granting conditional labor market access to Venezuelan immigrants (i.e., only in certain occupations) as well as unrestricted location choice and complete access to public services. In terms of effect heterogeneity, natives who have less contact with Venezuelans, who put more emphasis on economic matters, and who perceive the situation in Venezuela as mainly an economic problem, have more restrictive policy preferences. Additionally, Albarosa and Elsner (2021) study the example of refugee immigration into Ger- 4 On the effects of immigration on crime, see also: Bianchi et al. (2012) and Bell et al. (2013). 7 many in the years following 2015 and its effects on different proxies of social cohesion such as trust, perceived fairness, and attitudes towards immigrants. Similarly, they find no evidence of any effect, neither negative nor positive. In contrast, they identify an increased incidence of anti-immigrant violence in the short-term which was larger in areas with higher unem- ployment and greater support for right-wing parties. Such null effects of refugee migration on native attitudes in host communities are also identified by Zhou et al. (2021) for the case of Sudanese refugee immigration to Uganda. 4 Research Design and Data 4.1 Data Our analysis relies on data from the following sources: Encuesta Dirigida a la Población Venezolana que Reside en El País (ENPOVE) is a special- ized survey of Venezuelans living in Peru conducted by the National Institute of Statistics (INEI) in December 2018. The sample covers five main urban areas in the country where Venezuelan immigrants were most likely to be present. The survey collects data on the immi- grant’s origin, migration date, and details on their current employment. Importantly, a full module asks about the immigrant’s experiences with locals, which includes questions about discrimination and hostile attitudes towards them. The respondent’s current location is iden- tified down to the centro poblado level, which roughly corresponds to an urban neighborhood or a rural town. Encuesta Nacional de Hogares (ENAHO) is the Peruvian version of the Living Standards Measurement Survey, e.g. a nationally representative household survey collected monthly on a continuous basis. For our analysis, we use data from January 2007 to December 2020. The survey covers a wide variety of topics, including basic demographics, educational back- ground, labor market conditions, crime victimization, and a module on respondent’s percep- tions about the main problems in the country and trust on different local and national level institutions. Observations are also spatially identified at the municipality level, but here we focus on variation in the Venezuelan share of the population at the province level, of which there are 196, as these are best representative of local labor markets. Latin American Public Opinion Project (LAPOP) is a opinion survey conducted bi-annually in all countries in Latin America and designed to be representative of urban populations. This was fielded in Peru in 2010, 2012, 2014, 2017 and 2019 and consists of about 2,000 observations from mostly urban areas. The survey questions are centered around politics, 8 governance and opinions on current events. Observations are also spatially identified at the municipality and again we focus on variation in the Venezuelan share of the population at the province level. Gallup World Poll (GWP) is a nationally representative opinion survey and has been con- ducted annually since 2006 in a wide range of countries around the world. The sample collected in Peru is a repeated cross-section of about approximately 1,000 observations each year. For our analysis, we use data from 2013 to 2020. The survey questions are centered around politics, governance and opinions on current events. We make use of several opinion indices provided by Gallup that measure individual opinions on various domains. Observa- tions are spatially identified at the region level for Peru, which is our level of analysis in this case (there are 25 regions in Peru). PTP We measure the location of Venezuelan immigrants on a monthly basis from January 2015 to December 2020 using administrative data on the district Venezuelan immigrants register at with the Peruvian authorities to obtain access to social services. There are strong incentives to register as this is also a prerequisite for applying to obtain the PTP. This data only records monthly gross arrivals so we do not know the outflows of Venezuelans to other locations within Peru or out of the country entirely. However, in ENPOVE, 84% of Venezuelan immigrants in Peru report having lived in the same district during their entire time since arriving in the country. The data shows the arrival of 511,223 Venezuelans as of December 2020, which, while somewhat lower than estimates of the actual number of Venezuelans living in Peru, is quite substantial. We also use data from the National Census 2007 and 2017. We use the 2017 Census data to measure the share of workers in the formal and informal sector in each centro poblado as well as the total local population in each centro poblado, province and region. We use the 2007 data to construct both of our instruments discussed in more detail below as well as to create additional controls for the local economic environment. More specifically, in the first part of our analysis, we use information on the industrial distribution (using detailed four-digit codes) in each centro poblado, while in the second part, we use information on the total number of Venezuelans in each province in Peru. To construct the Trade shock instrument for the first part of our analysis, we also use trade data from the reports of TradeMap. From this website, we are able to identify export and import values for Peru on a monthly basis since 2006 at the HS 6-digit product revision. In addition, correspondence tables of HS 6-digit product revision to ISIC 3.1 revision (United Nations) are used to harmonize products with their corresponding industry sector in order 9 to be matched with census data. This allows us to create a year-ISIC panel with information about exports and imports on 86 industry sectors in Peru. 4.2 Outcome Variables In the first part of our analysis, we examine the impact of local labor market conditions on self-reported information on experiencing discrimination as reported by Venezuelans surveyed in ENPOVE. Overall, 36.4% of Venezuelans report having experienced discrimination, with this being slightly more common among women (38.1%) than men (35.0%). Figure 2 shows the distribution of reported discrimination in different municipalities in Peru. There is clearly variation both across and within regions. Report discrimination is least common in Tumbes (23.4%), which is the typical entry point to Peru for Venezuelans and currently hosts 5% of ENPOVE sample, while Cusco and Lima, where 7% and 48% of Venezuelans are located, show the highest (47.8%) and median levels (37.1%) of reported discrimination, respectively. Individuals who experienced discrimination are then asked in which locations did the episode took place. We examine reports for the three most common locations, at work (20.0%), on the streets/in public places (25.0%), and on public transit (9.8%). In the second part of our analysis, we examine the impact of Venezuelans on a wide variety of outcomes for Peruvians. First, we examine impact on labor market outcomes, specifi- cally employment, formal employment, log wages if employed, log household income and log household expenditure. Second, we examine the impact on crime and opinions about per- sonal safety. Specifically, we look at the reported (log) number of crime in each district from administrative data split into non-violent and violent crimes (data starting in 2011, means 3.54 for log violent crime and 3.31 for log non-violent crime), from ENAHO whether crime is a major national problem (12.7%), from LAPOP whether they have been a crime victim in the last two months (32.0%) and standardized variables from LAPOP on opinions about neighborhood safety and from Gallup on personal security. Lastly, we examine the impact on community outcomes. Specifically, we look at standardized indexes measuring quality of local services and trust in neighbors from LAPOP and indexes from Gallup on community attachment, the quality of the local community and whether the community likes diversity. 4.3 Control Variables Table 1 show the descriptive statistics for the control variables used in each analysis. The information we have available from each dataset varies, but we can always control for age, gender, education, marital status, whether employed and household size. ENPOVE col- lects additional relevant data about Venezuelans including how long they have been in Peru, 10 whether they work in the formal sector, their labor income, their occupation and the so- cioeconomic status of their household. ENAHO collects very similar data from Peruvians. Neither LAPOP nor Gallup collect detailed data on employment and occupation. We first show the information for our main explanatory variable. In the first analysis, this is the informal employment rate measured in the 2017 census in the centro poblado in which Venezuelans reside which has a mean of 31.2%.5 This is noticeably below the overall informal employment rate of 59.9% among Peruvians surveyed in ENAHO, indicating the Venezuelans are generally settling in areas in Peru with less formal unemployment. Only 8.0% of Venezuelan immigrants surveyed in ENPOVE are employed in the formal sector (i.e., they have an employment contract with social security benefits), hence the employment rate in the informal sector among Peruvians in a particular location is a good measure of the availability of job opportunities for Venezuelans and the competition with Peruvians for these jobs. We hypothesize that locations with higher informal sector employment rates have more opportunities and less competition with Peruvians for jobs. For this reason, we sometimes describe locations with high informal employment rates as having ’strong’ informal labor markets. In the second analysis, this is the number of Venezuelan immigrants in a particular month and province (measured in the administrative data) as a share of the total local population measured in the 2017 census.6 Over the full sample period of our analysis, the mean share of Venezuelans in the population is very low, 0.4% in the ENAHO sample, 0.3% in the LAPOP sample and 0.6% in the LAPOP sample. However, if we just look at ENAHO in December 2018, we see that the share has risen to 1.4%. The remainder of the table shows the means and standard deviation for the control variables for Venezuelan immigrants in December 2018, as captured in the ENPOVE (Columns 1 and 2), and for the average Peruvian respondents in ENAHO in two periods: December 2018 and 2007-2020 (Columns 3 and 4, and 5 and 6, respectively). Additionally, we also provide descriptive data for the LAPOP and Gallup opinion surveys. Venezuelan immigrants were slightly less likely to be female and are younger than their local 5 Centro poblado is the smallest level of geographical disaggregation. In urban areas, they are equivalent to neighborhoods, while in rural areas they correspond to small towns. 6 There are 198 provinces in Peru which generally correspond to labor market areas. In our regressions, we take the log of share variable. In order to include provinces with no Venezuelans and data prior to 2015, we add 1 to both the number of Venezuelans and the total population of each province. The median province has 55,000 inhabitants and the smallest nearly 3,000 hence this transformation should be immaterial. As the Venezuelan share of the population is very low in most provinces but highly skewed, it is important to measure this variable in logs. 11 counterparts (46.9% vs. 52.7%, and 31 vs. 42 years old). As mentioned above, Venezue- lans are more educated than Peruvians: only 17.2% had less than secondary education and 38.8% had a university education compared to 47.8% and 13.4% only, respectively, among Peruvians. Despite the differences in human capital, immigrants had worst labor market outcomes: 13% were unemployed, 8% had a formal job and the average income was of S/ 941. On the other hand, Peruvians had a slightly higher unemployment rate (16.1%), but were almost three times as likely to have a formal employment (21.4%) and earned 50% more than immigrants (S/ 1,482). Despite being more educated, Venezuelans work in less skilled jobs than Peruvians, especially in sales and services and elementary occupations. The big exception is agriculture and fishing which is the occupation for 16.2% of Peruvians but almost no Venezuelans in Peru. 4.4 Empirical Model and Identification We first examine the impact of local labor market conditions on immigrants’ reports of experiencing discrimination. More specifically, we estimate the following regression model: yij = α + β1 lnEmpj + δXij + θZj + αo + εij (1) where yij equals one if individual i in centro poblado j reports having experienced discrimina- tion in the 2018 in the ENPOVE (in general, or in a particular location) and zero otherwise. lnEmpj is the log informal employment rate in the same centro poblado measured in the 2017 census. We control for a variety of individual (X ) and centro poblado (Z ) level con- trols.7 We also include origin municipality (in Venezuela) fixed effects (αo ) to control for any origin-specific factors that could affect perceptions of discrimination (e.g. skin color, accent). εij is an error term clustered at the centro poblado level as this is the level of aggregation of our main explanatory variable. Among our centro poblado (Z ) level controls, we include the (log) number of Venezuelans who lived in centro poblado j in 2007 (as identified in the census). Clearly, the number of immigrants in a certain location is an important determinant of discrimination, but including the current number of immigrants in the regression would introduce additional endogeneity 7 Individual level controls in the regression include gender, age, education, marital status, months living in Peru, household socioeconomic strata, household size and number of people sharing one’s bedroom. Employ- ment and occupation controls include total income, whether in formal employment, and occupation including not working. Centro poblado level controls include log population in 2017, the log number of Venezuelans in 2007, log mean household expenditure in 2013, log agricultural rate in 2007, log manufacturing rate in 2007 and log travel distance to Lima. 12 problems. Previous literature has shown that immigrants are more likely to move to locations where they have a network of peers from the same country. We show below that this is true among Venezuelans in Peru as well. Our main interest is on β1 , which represent the impact of the labor market conditions in centro poblado j on the discrimination experienced by Venezuelans. Venezuelans who arrive to the country clearly evaluate where to settle based on the labor market opportunities (among other reasons), and therefore to causally identify β1 we need a source of exogenous variation for the labor market at the local level. We use an instrumental variable strategy that exploits variation in the share of workers employed in different industries in 2007, along with national level shocks to trade in specific industries between Oct 2016 and Oct 2017 when the census was collected. More precisely, the first stage regression is given by: lnEmpj = α + νSharejk(2007) × ∆lnExportk + ηXij + νZj + αo + j (2) where Sharejk(t−1) is the share of workers in centro poblado j employed in industry k in 2007, and ∆Exportk represents the log change in national level exports in industry k between 2016 and 2017. The remaining control variables are similar to those in Equation 1. The identifying assumption in this instrumental variable regression is that the change in trade in specific sectors at the national level affects the local labor market conditions without directly having effects on the the discrimination and hostilities reported by immigrants in a specific location. Importantly, we also control for other local level economic characteristics, such as the importance of agriculture and manufacturing which could be related to both exposure to export shocks and experiencing discrimination. We then turn to examine the impact of receiving a larger population of Venezuelan immigrants on natives’ labor market outcomes and different dimensions of locals’ perceptions. To do this, we estimate the following regression: yipt = α + βln(ImmigrantShare)pt + δXipt + αt + αp + time ∗ αp + εipt (3) where yipt represent a particular outcome for individual i living in province p interviewed at time t, ln(ImmigrantShare)pt is the (log) of the number of Venezuelan immigrants in province p at time t as a share of the total population of province p in December 2017 and Xipt include a series of individual level controls, including age, education, marital status, and household size for all models, and employment status and occupation for non-labor market 13 outcomes measured in ENAHO. εipt is an error term clustered at the province level as we measure the number of Venezuelan immigrants at this level and suspect there is strong serial correlation in many of our outcomes. β identifies the effect of the number of Venezuelan immigrants in year t in region p. Impor- tantly, we also control in all models for time (either year or month*year) fixed effect (αt ) and province fixed effects (αp ). Hence, we control for any time-invariant differences in outcomes across provinces and aggregate changes in outcomes, both of which may be related to the location choice decisions of Venezuelans. In our preferred specification, we also control for province-specific time-trends (time ∗ αp ) which account for any local trends in the outcome variable. In this model, the impact of the presence of Venezuelans is identified by examining how outcomes for Peruvians change when more Venezuelans arrive in an area conditional on the trend in that outcome.8 It is possible that local shocks impact both the destination choice of Venezuelans and out- comes for Peruvians, hence we also use an instrumental variable strategy where we exploit the intuition that immigrants are more likely to move to localities where immigrants from the same nationality are located. We therefore instrument our measure of the number of Venezuelans in a province p with the presence of Venezuelans in that province as recorded in the 2007 census interacted with year dummy variables. This is a semi-parametric version of the traditional immigrant network instrument as recommended by Goldsmith-Pinkham et al. (2020) as it allows the strength of the network effect to potentially vary in each year. An overidentification test can be used to examine whether the instrument has a consistent relationship over time. 5 Results 5.1 Labor Market Conditions and Discrimination Table 2 shows our main results on the effects of local labor market conditions on self reported discrimination. We first present the OLS results, and then turn to provide the estimates from our IV specification. Importantly, given that the types of jobs in which men and women work differ, in Table 2 we show the main results for the full sample of immigrants who responded the survey, and split the sample by gender. Columns (1)-(3) show the OLS relationship between the (log) local informal employment 8 Our results are robust to controlling for district fixed effects and time-trends as well, but we believe this is over-fitting the model as many individuals commute across district boundaries for work. 14 rate and the reports of discrimination. In the three panels, this relationship shows small and insignificant coefficients, and further, as we include individual level controls and munic- ipality of origin fixed effects, the coefficients become even smaller (and still not statistically significant). As discussed above, the OLS results presented in Columns (1)-(3) cannot be interpreted as causal due to an endogeneity issue that arises from the fact that immigrants make their location decisions based on the local labor market conditions. To overcome this problem, we use an instrumental variable strategy in which we exploit exogenous variation in the impact of national level export shocks on local (informal) employment rates depending on detailed measure of the local industry composition. Spatial variation in both the informal employment rate in 2017 and the instrument are show in Figure 3 and 4, respectively. Importantly, there is clear variation in both across and within regions. The first stage relationship is shown graphically in Figure 5. An increase in predicted local exports has a positive relationship with formal employment rates. Consequently, higher predicted exports implies that a lower share of people work in the informal sector, where most Venezuelan immigrants are employed. The full results of the first stage relationship between export shocks and informal employment rates are provided in Appendix Table A.1. Using a linear and a quadratic specification for the first stage yields a very strong instrument, with an F-stat for the excluded instrument that ranges between 22 and 27 even while controlling for a number of other measures of the local economic environment. Columns (5) and (6) in Table 2 show the second stage results from our IV strategy. The first thing to note is that the IV coefficients in all instances are much larger than the OLS coefficients. This difference indicates that Venezuelan immigrants are selecting themselves into labor markets where there is more discrimination. This is consistent with the idea that immigrants are willing to deal with more discrimination as long as the labor market offers better opportunities and higher wages. The selection seems more relevant for men than for women. Higher employment rates in the informal sector causes a reduction in the level of discrimi- nation reported by Venezuelan immigrants. On average, a 10 percent increase in informal employment in a centro poblado causes a 2.3-3% reduction in discrimination depending on whether we use a linear or quadratic specification of the instrument. Furthermore, the results show that men are more likely to suffer from discrimination due to changes in the informal employment rate: a 10% increase in employment reduces discrimination against men by al- most 4%, while for women, the effect ranges between 1.2 and 1.9% and is not statistically 15 significant. Importantly, the reduction in discrimination as a response is not explained by having more exposure in the labor market: as we show in Appendix Table A.2, variation in local informal employment is unrelated to the Venezuelan immigrants’ probability of being employed. While this may be a bit surprising, we have to take into account that 94% of Venezuelan immigrants are employed, so there is little margin for improvement. Additionally, for men we do not see any effects on wages, while we see that higher employment rates in the informal sector do have a positive effect on wages of Venezuelan women in Peru. The fact that discrimination against men shows a stronger response to labor market condi- tions is potentially related to the types of interactions immigrants have with locals. In Table 3, we present the full results from the quadratic IV specification. For both men and women, reported discrimination increases with time spent in Peru, education and household socioeco- nomic status. On the other hand, it also increases with lower skilled occupations, particularly for women. This is consistent with reported discrimination reflecting a lack of progress in the labor market for high-skilled Venezuelans, whether this reflects being disappointed or actually being discriminated against is difficult to quantify. To explore further the mechanisms underlying these effects, in Table 4 we exploit the fact that ENPOVE collects detailed information on where discrimination episodes took place. We show the OLS and IV results for our preferred specification, the one that includes all controls and fixed effects, and for the IV, the specification that uses the linear instrument. Discrimination at work seems to respond the least to local employment, with a coefficient that implies that a 10% increase in informal employment leads discrimination to decrease by 1.4%, although the relationship is not statistically significant. Interestingly, there is a clear gender split on whether discrimination in streets and public spaces. A 10% increase in informal employment causes a decrease in discrimination against men in streets and public spaces of about 3.7%, with no significant change in discrimination against women in these spaces. Finally, discrimination in public transit responds similar regardless of the gender, with an effect of about 2% for reductions in employment of 10%. 5.2 Immigration, Local’s Labor Market Outcomes and Perceptions In the previous section, we established that labor market conditions have a causal effect on the way Venezuelan immigrants perceive that are treated by locals: lower unemployment in the informal labor market leads to a decrease in discrimination. We now turn to study the flip-side, namely, the way in which the presence of Venezuelans affect Peruvians’ labor 16 market outcomes and their perceptions about crime, corruption and public good provision at the local level. To conduct this analysis we follow Equation 3 above. Table 5 shows the estimates of the impact of the presence of Venezuelan immigrants on Peruvian’s labor market outcomes. In the table, we show our OLS estimates with different sets of controls (Panels A and B). We then introduce our IV estimates, separated by men and women (Panels C, D and E). A higher number of Venezuelan immigrants in a given province correlates with higher levels of employment (but not formal employment) for Peruvians once we include control for province and time fixed effects as well as province-specific trends (Panel B). Additionally, households in provinces with more immigrants report higher incomes and expenditures. As mentioned above, there is a clear endogeneity problem, and the correlations observed in Panels A and B could be caused by the fact that immigrants sort into more dynamic labor markets. We introduce our instrumental variable estimates in Panel C of Table 5. Recall that in these regressions we instrument the log share of Venezuelans at the province-year level with the log share of Venezuelans in the same province in 2007 interacted with year dummies. The relationship between our instrument and the endogenous regressor is depicted in Figure 6, where is clear that immigrants are more likely to move to locations where there is an established network of compatriots and that this relationship is stable over time even though there has been a large increase in the Venezuelan immigrant share over time. This is true even though the number of Venezuelans in Peru in 2007 was quite small. The F-stat for the excluded instrument is 2300, showing the strong relationship in the first-stage robustness of the instrument. We also typically fail to reject that our model is over-identified which is an indication that the shift-share instrument is truly picking up the impact of increasing Venezuelans being pulled to locations where Venezuelans previously settled. Our IV results tell the same qualitative story. A doubling in the share of Venezuelans in a province increases the probability of a Peruvian being employed by 0.6%, increases household income raise by 2.2% and expenditures by 1.4%. The effects on income and expenditure are nearly twice as large for women as for men. These positive impacts on labor market outcomes are sizeable, given the large overall increase in the Venezuelan share of the population.9 There are a number of potential explanations for these findings that we plan to explore in future work. The arrival of Venezuelans may have expanded the economic opportunities for Peruvian because of their higher levels of potential productivity, due to higher human 9 Previous studies have shown that Venezuelan immigration caused either small but significant losses in the labor market for low education women (Morales and Pierola, 2020) or null effects (Boruchowicz et al., 2021). 17 capital, and concentration in low wage jobs. Furthermore, most of these jobs are in the service sector which potentially could have freed up time, especially for Peruvian women to be more engaged in the labor market, as well as lowered the costs for these types of goods and services. Venezuelans also might have expanded opportunities by increasing the demand for certain goods and services. One widespread claim mentioned in some media reports is that Venezuelan migration led to an increase in crime (Freier et al., 2021). We test whether this claim is supported by the data in Table 6, where we use administrative information on the number of non-violent and violent crimes reported in each municipality, the personal security index from Gallup, and reports on whether crime is perceived as a major problem in ENAHO. The structure of this table is the same as the previous with Panel C our preferred specification. Consistent with the idea that Venezuelan inflow lead to labor market conditions improving, we observe that locations that received a larger number of immigrants have lower number of reported non-violent crimes (columns 2). This effect is large with a double of Venezuelans in a province leading to a 42% decline in reported non-violent crimes. Individuals are also less likely to report that crime is a major national problem, this is true for both men and women. Finally, in Table 7 we examine the effects of immigration on local communities. Consistent with our previous findings, Peruvians living in areas with a higher share of Venezuelan im- migrants report that the quality of local services and community quality are higher and that they have a greater trust in their neighbors. However, in these locations they also report that the community is less likely to value diversity. In general, these findings are stronger for men than for women. It is important to note that these findings could be driven by the positive impact that Venezuelans have on the labor market outcomes of Peruvians. Overall, we find that increased Venezuelan migration leads to improvements in both objective and subjective measures of the lives of Peruvians living in the same locations. This may explain why, even though some media has discussed Venezuelans in a negative light, there has been little political backlash against them in Peru up to this point in time. 6 Conclusion and Policy Implications In this paper, we study the economic underpinnings of hostility and discrimination against immigrants. In our analysis, we first use a specialized survey of Venezuelan immigrants in Peru to identify the causal effect of local labor markets on discrimination against immigrants. Then, we studying the flip-side of this analysis, namely, how does the presence of Venezuelan 18 immigrants affect Peruvians’ labor market outcomes, local crime, and their perceptions about crime and their local community. We find that a higher local share of employment in the informal labor market leads to signifi- cant decreases in the discrimination reported by Venezuelans in Peru. However, Venezuelans report the discrimination in public places and on public transit is a fairly common occur- rence. Hence, reinforcing policies that prevent the stigmatization of foreigners should be a first order concern for policymakers, especially in areas with low informal employment rates. On the other hand, we show that increased migration from Venezuela leads to better labor market outcomes for Peruvians, less reported crime and better community outcomes. This evidence provides clear avenues for future research. High levels of informality in the labor market might be important for providing sufficient flexibility that allows large shocks (such as a massive inflow of workers) to be quickly absorbed. Beyond this, there are a number of potential explanations for why Venezuelan inflows improve outcomes for Peruvians. More research is needed to understand more precisely the extent to which the positive effects of larger immigration is attributable to the high levels of informality in the Peruvian labor market. This positive outcome is, however, at odds with the objective of increasing social security coverage and the tax base in Peru. An important policy issues is to incorporate Venezuelan immigrants (along with Peruvians) into the formal labor market, ensuring their basic worker rights are respected. One challenge to implementing this is the current regulation that formal firms can only hire foreigners if they are less than 20% of the workforce and their salaries are less than 30% of the total wage bill. Additionally, foreign formal workers face higher tax rates: beyond the standard rate of personal income tax of 13%, foreigners have to pay 30% more. This clearly disincentivizes formalization for both Peruvian firms and Venezuelan workers. Policymakers in Peru need to carefully consider the trade-offs presented with this policy. Most importantly, our results suggest that recent policy changes by Peruvian authorities to limit immigration from Venezuela (e.g., since June 2019, Venezuelans who want to legally enter Peru have been required to apply for a humanitarian visa in specific Peruvian con- sulates abroad and provide documentation that is difficult to obtain, such as passports) are unnecessary as the arrival of Venezuelans up until the Covid-19 crisis had led to general improvements in labor market outcomes for Peruvians as well as reduced levels of crime and better community outcomes. A better approach would be to help facilitate their arrival as was the previous policy and focus on reducing perceptions around their negative impacts. 19 References Ajzenman, Nicolas, Patricio Dominguez-Rivera, and Raimundo Undurraga, “Im- migration, Crime, and Crime (Mis)Perceptions,” IZA Discussion Papers 14087, Institute of Labor Economics (IZA) February 2021. Aksoy, Cevat Giray and Thomas Ginn, “Host Country Policies and Attitudes toward Refugees: A Global Evaluation,” 2021. Albarosa, Emanuele and Benjamin Elsner, “Forced Migration, Social Cohesion and Conflict: The 2015 Refugee Inflow in Germany,” 2021. Unpublished Working Paper. Com- missioned as part of the âPreventing Social Conflict and Promoting Social Cohesion in Forced Displacement Contextsâ Series. Washington, DC: World Bank Group. Alesina, Alberto, Armando Miano, and Stefanie Stantcheva, “Immigration and Re- distribution,” Working Paper 24733, National Bureau of Economic Research June 2018. Allen, William, Isabel Ruiz, and Carlos Vargas-Silva, “Policy Preferences in Response to Large Migration Inflows,” 2021. Asencios, Roger and Renzo Castellares, “The Impact of Venezuelan Immigration on Employment and Wages: the Peruvian Case,” Working Paper 2020-002, Peruvian Central Bank 2020. Bahar, Dany, Ana María Ibáñez, and Sandra V. Rozo, “Give me your tired and your poor: Impact of a large-scale amnesty program for undocumented refugees,” Journal of Development Economics, 2021, 151, 102652. Bell, Brian, Francesco Fasani, and Stephen Machin, “Crime and Immigration: Evi- dence from Large Immigrant Waves,” The Review of Economics and Statistics, 2013, 21 (3), 1278–1290. Bhabha, Jacqueline, Can we solve the migration crisis?, John Wiley & Sons, 2018. Bianchi, Milo, Paolo Buonanno, and Paolo Pinotti, “Do Immigrants Cause Crime?,” Journal of the European Economic Association, 2012, 10 (6), 1318–1347. Borjas, George J, “The Substitutability of Black, Hispanic, and White Labor,” Economic Inquiry, January 1983, 21 (1), 93–106. Boruchowicz, Cynthia, Cesar Martinelli, and Susan Parker, “Economic Consequences of Mass Migration: The Venezuelan Exodus in Peru,” Working Papers 1080, George Mason University, Interdisciplinary Center for Economic Science 2021. 20 Card, David, “Immigrant Inflows, Native Outflows, and the Local Labor Market Impacts of Higher Immigration,” Journal of Labor Economics, 2001, 19 (1), 22–64. Carpio, Ximena Vanessa Del and Mathis Christoph Wagner, “The impact of Syrian refugees on the Turkish labor market,” Policy Research Working Paper Series 7402, The World Bank August 2015. Ceritoglu, Evren, H. Burcu Gurcihan Yunculer, Huzeyfe Torun, and Semih Tu- men, “The impact of Syrian refugees on natives’ labor market outcomes in Turkey: evi- dence from a quasi-experimental design,” IZA Journal of Labor Policy, December 2017, 6 (1), 1–28. Chatruc, Marisol Rodríguez and Sandra V. Rozo, “Attitudes Towards Migrants during Crisis Times,” IZA Discussion Papers 14319, Institute of Labor Economics (IZA) April 2021. Chaves-González, Diego and Carlos Echevarría Estrada, “Venezuelan Migrants and Refugees in Latin America and the Caribbean: A Regional Profile,” Fact Sheet ROBUE- OIM-035, International Organization for Migration 2020. Delgado-Prieto, Lukas, “Dynamics of Local Wages and Employment: Evidence from the Venezuelan Immigration in Colombia,” 2021. Freier, Luisa Feline, Matthew Bird, Gustav Brauckmeyer, Andrea Kvietok, David Licheri, Edicson Luna Román, José Javier Olivas Osuna, and Lucía Ponce, “Diagnostico de la Cobertura Mediática de las Personas Refugiadas y Migrantes,” Technical Report March 2021. Goldsmith-Pinkham, Paul, Isaac Sorkin, and Henry Swift, “Bartik Instruments: What, When, Why, and How,” American Economic Review, August 2020, 110 (8), 2586– 2624. Guerrero-Ble, Martha, Izza Leghtas, and Jimmy Graham, “From Displacement to Development: How Peru Can Transform Venezuelan Displacement into Shared Growth,” Technical Report, Refugees International and Center for Global Development 2020. Hangartner, Dominik, Elias Dinas, Moritz Marbach, Konstantinos Matakos, and Dimitrios Xefteris, “Does Exposure to the Refugee Crisis Make Natives More Hostile?,” American Political Science Review, 2019, 113 (2), 442–455. 21 Hausmann, Ricardo, Julian Hinz, and Muhammed A. Yildirim, “Measuring Venezuelan Emigration with Twitter,” CID Faculty Working Paper 342, Harvard Uni- versity 2018. Jaeger, David A, Joakim Ruist, and Jan Stuhler, “Shift-Share Instruments and the Impact of Immigration,” Working Paper 24285, National Bureau of Economic Research February 2018. Janetsky, Megan, “Fears Stoke Backlash Against Venezuelans in Peru,” BBC NEWS, Aug 2019. Lebow, Jeremy, “Refugees in the Colombian Labor Market: The Consequences of Occupa- tional Downgrading,” SSRN Electronic Journal, 01 2020. , Jonathan Moreno Medina, and Horacio Coral, “Immigration and Trust: The Case of Venezuelans in Colombia,” Technical Report 2021. Morales, Fernando and Martha Denisse Pierola, “Venezuelan Migration in Peru: Short-term Adjustments in the Labor Market,” IDB Working Paper 1146, Inter-American Development Bank, Migration Unit 2020. Olivieri, Sergio, Francesc Ortega, Ana Rivadeneira, and Eliana Carranza, “The Labor Market Effects of Venezuelan Migration in Ecuador,” IZA Discussion Papers 13501, Institute of Labor Economics (IZA) July 2020. R4V, “R4V América Latina y el Caribe, Refugiados y Migrantes Venezolanos en la Región - Julio 2021,” Technical Report, Plataforma de Coordinacion Interagencial Para Refugiados y Migrantes de Venezuela July 2021. Rozo, Sandra V. and Juan F. Vargas, “Brothers or invaders? How crisis-driven migrants shape voting behavior,” Journal of Development Economics, 2021, 150, 102636. Santamaria, Julieth, “When a Stranger Shall Sojourn with Thee’: The Impact of the Venezuelan Exodus on Colombian Labor Markets,” 2021. Tumen, Semih, “The Economic Impact of Syrian Refugees on Host Countries: Quasi- experimental Evidence from Turkey,” American Economic Review, May 2016, 106 (5), 456–60. Winter, Brian, “The Backlash to Venezuelan Migration Is Here,” Americas Quarterly, Apr 2020. 22 Zhou, Yang-Yang, Guy Grossman, and Shuning Ge, “When Refugee Exposure Im- proves Local Development and Public Goods Provision: Evidence from Uganda,” 2021. 23 Figures and Tables Figure 1: Evolution of Venezuelan immigrant stock in Peru Source: Peruvian National Superintendence of Migrations. 24 Figure 2: Spatial Variation in Reported Discrimination in ENPOVE 25 Source: Own calculations ENPOVE. Figure 3: Spatial Variation in Informal Employment in 2017 in ENPOVE 26 Source: Own calculations ENPOVE. Figure 4: Spatial Variation in 2016-2017 Log Export Shock in ENPOVE 27 The instrument allocates national level changes in exports at the industry level to different locations based on local industry structure in 2007 Log Informal Employment Rate 2017 Figure 5: First Stage: Local Trade Shocks and Employment -2.5 -2 -1.5 -1 -.5 -.35 -.3 -.25 -.2 -.15 -.1 -.05 0 Log Formal Employment Rate 2017 -3 -2.5 -2 -1.5 -1 -.5 -.35 -.3 -.25 -.2 -.15 -.1 -.05 0 Export Shock Year Prior Oct 2017 (Log Change) 28 Figure 6: First Stage: Stocks of Venezuelans and Immigration 5% 5% 5% Percent Registered VZs 2015 Percent Registered VZs 2016 Percent Registered VZs 2017 0.25% 0.25% 0.25% 0.01% 0.01% 0.01% 0% 0% 0% 0% 0.002%0.005% 0.01% 0.03% 0% 0.002%0.005% 0.01% 0.03% 0% 0.002%0.005% 0.01% 0.03% Percent Venezuelan-Born 2007 Percent Venezuelan-Born 2007 Percent Venezuelan-Born 2007 5% 5% 5% Percent Registered VZs 2018 Percent Registered VZs 2019 Percent Registered VZs 2020 0.25% 0.25% 0.25% 0.01% 0.01% 0.01% 0% 0% 0% 0% 0.002%0.005% 0.01% 0.03% 0% 0.002%0.005% 0.01% 0.03% 0% 0.002%0.005% 0.01% 0.03% Percent Venezuelan-Born 2007 Percent Venezuelan-Born 2007 Percent Venezuelan-Born 2007 29 Table 1: Descriptive Statistics ENPOVE ENAHO ENAHO LAPOP Gallup Dec 2018 Dec 2018 2007-2020 2010-2019 2013-2020 Mean SD Mean SD Mean SD Mean SD Mean SD Community characteristics: Informal employment rate (2017) 0.312 0.064 District Population (2017) 298,000 242,000 Share Venezuelan immigrants 0.014 0.011 0.004 0.009 0.003 0.009 0.006 0.120 Individual characteristics: Female 0.469 0.499 0.527 0.499 0.524 0.499 0.510 0.500 0.574 0.495 Age 30.8 9.7 42.4 13.0 40.9 13.0 36.6 12.9 40.1 17.4 Months in Peru 8.19 6.90 Education: Less than secondary 0.172 0.382 0.478 0.499 0.474 0.499 0.250 0.499 0.179 0.384 Education: Complete secondary 0.256 0.437 0.265 0.441 0.247 0.432 0.337 0.473 0.697 0.459 Education: Technical 0.186 0.389 0.123 0.328 0.132 0.339 0.085 0.279 Education: University 0.385 0.487 0.134 0.341 0.146 0.353 0.328 0.470 0.123 0.329 Marital status: Married/Cohabitation 0.576 0.494 0.626 0.484 0.627 0.484 0.592 0.492 0.499 0.500 Marital status: Formerly Married 0.042 0.202 0.199 0.399 0.176 0.381 0.074 0.262 0.113 0.316 Marital status: Never Married 0.382 0.486 0.175 0.380 0.197 0.398 0.334 0.472 0.386 0.487 Formal Employment 0.080 0.271 0.214 0.410 0.205 0.404 Labor Income 941 633 1482 1346 1265 1306 Occupation Not Working 0.133 0.340 0.161 0.368 0.200 0.400 0.413 0.400 0.349 0.477 Military/Police 0.000 0.000 0.004 0.060 0.005 0.073 Managers 0.000 0.016 0.004 0.060 0.005 0.068 Professionals 0.017 0.127 0.058 0.233 0.058 0.234 Technicians and Ass Professionals 0.061 0.239 0.046 0.209 0.043 0.204 Clerical Support Workers 0.048 0.214 0.037 0.189 0.036 0.187 Services and Sales Workers 0.266 0.442 0.139 0.346 0.120 0.325 Skilled Agricultural and Fishery 0.001 0.032 0.162 0.368 0.169 0.375 Craft and Related Trades Workers 0.094 0.292 0.052 0.222 0.055 0.227 Operators, Assemblers, Construction 0.068 0.252 0.072 0.259 0.064 0.246 Elementary Occupations 0.312 0.463 0.266 0.442 0.245 0.430 HH characteristics: Low Socioeconomic Status 0.098 0.297 Medium Socioeconomic Status 0.446 0.497 High Socioeconomic Status 0.456 0.498 Household Size 3.28 1.90 3.81 1.86 4.03 1.95 4.15 2.15 3.95 2.04 Number of People Who Share Bedroom 2.46 0.89 Individuals 7,869 2,201 336,109 8,049 8,005 Descriptive statistics are presented from four surveys used in the paper, ENPOVE, ENAHO, LAPOP and Gallup. More details are available in the paper. 30 Table 2: Impact of Local Labor Market Conditions on Reported Discrimination by Venezue- lans OLS IV - Linear IV - Quad Overall: Have felt discriminated - Mean Outcome = 0.364 Log Local Informal Emp Rate -0.014 -0.021 -0.043 -0.234 -0.301* (0.071) (0.071) (0.067) (0.167) (0.173) F-Stat Weak Identification 25,4 24,7 Overidentification P-Value 0,156 R-squared 0.023 0.032 0.072 Observations 7,869 7,869 7,869 7,869 7,869 Men: Have felt discriminated - Mean Outcome = 0.350 Log Local Informal Emp Rate 0.012 0.011 0.000 -0.374** -0.416** (0.078) (0.077) (0.072) (0.162) (0.171) F-Stat Weak Identification 21,9 22,0 Overidentification P-Value 0,305 R-squared 0.022 0.028 0.093 Observations 4,176 4,176 4,176 4,176 4,176 Women: Have felt discriminated - Mean Outcome = 0.381 Log Local Informal Emp Rate -0.044 -0.055 -0.095 -0.117 -0.198 (0.085) (0.084) (0.083) (0.182) (0.190) F-Stat Weak Identification 27,3 25,7 Overidentification P-Value 0,166 R-squared 0.027 0.042 0.100 Observations 3,693 3,693 3,693 3,693 3,693 Sociodemographic Characteristics Yes Yes Yes Yes Yes Employment and Occupation No Yes Yes Yes Yes Origin Municipality FE No No Yes Yes Yes *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors clustered at centro poblado level in parentheses. Local informal employment rate is measured in the 2017 census. The predicted export shock in each centro poblado in year prior to Oct 2017 is used to instrument for the informal employment rate. Sociodemographic controls include gender, age, education, marital status, months living in Peru, household socioeconomic strata, household size and number of people sharing one's bedroom. Employment and occupation controls include total income, whether in formal employment, and occupation including not working. All regression also control for the following variables measured at the centro poblado level: log population in 2017, the log number of Venezuelans in 2007, log mean household expenditure in 2013, log agricultural rate in 2007, log manufacturing rate in 2007 and log travel distance to Lima. 31 Table 3: Correlates of Reported Discrimination by Venezuelans Men Women Age -0.001 (0.001) -0.002*** (0.001) Log Months in Peru 0.050*** (0.012) 0.083*** (0.013) Education: Complete secondary 0.012 (0.022) 0.039 (0.028) Education: Technical 0.061*** (0.023) 0.035 (0.029) Education: University 0.076*** (0.023) 0.062** (0.028) Married/Cohabitation 0.022 (0.017) 0.019 (0.019) Formal Employment -0.018 (0.026) 0.011 (0.035) Labor Income (Thousands) -0.009 (0.013) -0.011 (0.021) Managers -0.358** (0.154) -0.354*** (0.061) Professionals 0.053 (0.085) 0.039 (0.056) Technicians and Ass Professionals -0.022 (0.049) -0.001 (0.038) Clerical Support Workers -0.011 (0.053) 0.067* (0.039) Services and Sales Workers 0.011 (0.038) 0.111*** (0.023) Skilled Agricultural and Fishery 0.244** (0.101) Craft and Related Trades Workers -0.008 (0.042) 0.133*** (0.050) Operators, Assemblers, Construction 0.016 (0.039) 0.209* (0.111) Elementary Occupations 0.066* (0.036) 0.155*** (0.028) Medium Socioeconomic Status 0.095** (0.037) 0.078* (0.042) High Socioeconomic Status 0.040 (0.043) 0.059 (0.041) Household Size -0.007 (0.006) -0.008* (0.005) Number of People Who Share Bedroom 0.012 (0.014) 0.003 (0.009) Log Local Informal Employment Rate in 2017 -0.416** (0.171) -0.198 (0.190) Log Local Population in 2019 -0.025 (0.026) -0.003 (0.030) Log Local Venezuelans in 2007 0.004 (0.021) -0.014 (0.020) Log Local Household Expenditure PC 2013 -0.222 (0.146) -0.088 (0.160) Log Proportion in Agriculture in 2007 -0.051 (0.032) -0.035 (0.024) Log Proportion in Manufacturing in 2007 0.103* (0.061) 0.076 (0.053) Log Travel Duration to Lima 0.033*** (0.012) 0.020 (0.013) R-squared 0.014 0.044 Mean dep. var 0.350 0.381 Observations 4,176 3,693 *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors clustered at centro poblado level in parentheses. Local informal employment rate is measured in the 2017 census. A quadratic in the predicted export shock in each centro poblado in year prior to Oct 2017 is used to instrument for the informal employment rate. The default category for the occupation variables is not working. All regressions also control for origin municipality in Venezuela fixed effects. 32 Table 4: Impact of Local Labor Market Conditions on Discrimination in Different Locations At Work Streets/Public Places Public Transit OLS IV OLS IV OLS IV Overall Log Local Informal Emp Rate -0.035 -0.131 0.020 -0.174 -0.046 -0.204* (0.055) (0.130) (0.066) (0.158) (0.049) (0.112) Observations 6,810 6,810 7,869 7,869 7,869 7,869 R-squared 0.064 0.063 0.063 0.060 0.076 0.072 Mean dep. var 0,201 0,250 0,098 Men Log Local Informal Emp Rate -0.016 -0.148 0.061 -0.371** -0.016 -0.210* (0.055) (0.115) (0.080) (0.176) (0.053) (0.121) Observations 3,923 3,923 4,176 4,176 4,176 4,176 R-squared 0.084 0.083 0.082 0.068 0.109 0.103 Mean dep. var 0.191 0.233 0,095 Women Log Local Informal Emp Rate -0.087 -0.141 -0.021 0.015 -0.073 -0.206* (0.081) (0.176) (0.069) (0.155) (0.054) (0.110) Observations 2,887 2,887 3,693 3,693 3,693 3,693 R-squared 0.100 0.100 0.095 0.095 0.104 0.101 Mean dep. var 0.215 0.269 0.100 *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors clustered at centro poblado level in parentheses. Local informal employment rate is measured in the 2017 census. The predicted export shock in each centro poblado in year prior to Oct 2017 is used to instrument for the informal employment rate. All regressions control for gender, age, education, marital status, months living in Peru, household socioeconomic strata, household size, number of people sharing one's bedroom, total income, whether in formal employment, occupation including not working and origin municipality in Venezuela fixed effects and the following variables measured at the centro poblado level: log population in 2017, the log number of Venezuelans in 2007, log mean household expenditure in 2013, log agricultural rate in 2007, log manufacturing rate in 2007 and log travel distance to Lima. 33 Table 5: Impact of Venezuelans on the Labor Market Outcomes of Peruvians Formal Log Wages if Log Household Log Household Employment Employment Employed Income Expenditure a) OLS: Month*Year and Province Fixed Effects Log Share Venezuelans -0.001 0.003*** -0.002 -0.008*** -0.014*** (0.001) (0.001) (0.003) (0.003) (0.003) R-squared 0.145 0.236 0.434 0.469 0.528 b) OLS: Month*Year and Province Fixed Effects and Province Time-Trends Log Share Venezuelans 0.003** -0.001 0.005 0.014*** 0.013*** (0.001) (0.001) (0.004) (0.004) (0.003) R-squared 0.149 0.237 0.438 0.474 0.534 c) IV: Month*Year and Province Fixed Effects and Province Time-Trends Log Share Venezuelans 0.006*** -0.000 0.001 0.022*** 0.014** (0.002) (0.002) (0.007) (0.005) (0.006) Men: IV: Month*Year and Province Fixed Effects and Province Time-Trends Log Share Venezuelans 0.006*** 0.003 -0.002 0.017*** 0.010 (0.001) (0.002) (0.010) (0.006) (0.007) Women: IV: Month*Year and Province Fixed Effects and Province Time-Trends Log Share Venezuelans 0.006 -0.003 0.005 0.027*** 0.019*** (0.004) (0.003) (0.007) (0.006) (0.007) Mean Outcome 0.801 0.205 6.75 9.93 9.44 Men 0.898 0.245 6.92 9.47 9.97 Women 0.712 0.169 6.50 9.41 9.88 Individuals 337,725 337,725 106,744 337,718 337,680 Men 160,699 160,699 63,833 160,694 160,682 Women 177,006 177,006 42,696 177,004 176,978 *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors clustered at province level in parentheses. Log Share Venezuelans is relative to the 2017 population of the province. Log Share VZs in the province in 2007 interacted with year fixed effects are used as instruments for Log Share of Venezuelans. All regressions include controls for a quadratic in age, education, marital status, and household size. 34 Table 6: Impact of Venezuelans on Crime and Safety Log Reported Log Reported Non- Neighborhood Crime Major Crime Victim Personal Security Violent Crimes in Violent Crimes in Safety (STD): National Problem (0/1): LAPOP (STD): Gallup Municipality Municipality LAPOP (0/1): ENAHO a) OLS: Month*Year (ENAHO) or Year (LAPOP/Gallup) and Province (ENAHO/LAPOP) or Region (Gallup) Fixed Effects Log Share VZs 0.012 -0.003 0.001 0.009 -0.017 -0.001 (0.023) (0.214) (0.003) (0.009) (0.010) (0.002) R-squared 0.586 0.714 0.065 0.085 0.09 0.088 b) OLS: Month*Year (ENAHO) or Year (LAPOP/Gallup) and Province (ENAHO/LAPOP) or Region (Gallup) FEs and Time-Trends Log Share VZs -0.009 -0.373*** -0.010 0.009 -0.010 -0.012*** (0.018) (0.097) (0.007) (0.020) (0.011) (0.002) R-squared 0.596 0.794 0.073 0.101 0.098 0.094 c) IV: Month*Year (ENAHO) or Year (LAPOP/Gallup) and Province (ENAHO/LAPOP) or Region (Gallup) FEs and Time-Trends Log Share VZs -0.048 -0.420** -0.008 0.016 -0.001 -0.013*** (0.031) (0.172) (0.007) (0.023) (0.016) (0.003) Men: IV: Month*Year (ENAHO) or Year (LAPOP/Gallup) and Province (ENAHO/LAPOP) or Region (Gallup) FEs and Time-Trends Log Share VZs -0.011 0.023 0.013 -0.011** (0.010) (0.024) (0.022) (0.005) Women: IV: Month*Year (ENAHO) or Year (LAPOP/Gallup) and Province (ENAHO/LAPOP) or Region (Gallup) FEs and Time-Trends Log Share VZs -0.006 0.008 -0.005 -0.014*** (0.009) (0.027) (0.018) (0.003) Individuals 234,002 213,093 7,998 7,962 7,997 308,993 Men 3,916 3,908 3,405 141,010 Women 4,077 4,049 4,592 167,957 *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors clustered at province level in parentheses. Log Share Venezuelans is relative to the 2017 population of the province. Outcome variables noted as STD are standardized. Log Share VZs in the province (region for Gallup outcomes) in 2007 interacted with year fixed effects are used as instruments for Log Share of Venezuelans. All regressions include controls for a quadratic in age, education, marital status, household size and whether employed. Outcomes from ENAHO include additional controls for whether in formal employment and one's occupation. 35 Table 7: Impact of Venezuelans on Local Communities Quality of Local Community Community Likes Trust Neighbors Community Quality Services (STD): Attachment (STD): Diversity (STD): (STD): LAPOP (STD): Gallup LAPOP Gallup Gallup a) OLS: Year and Province (LAPOP) or Region (Gallup) Fixed Effects Log Share VZs -0.007 0.017** -0.012 0.014 -0.011 (0.008) (0.008) (0.011) (0.012) (0.011) R-squared 0.070 0.061 0.027 0.066 0.053 b) OLS: Year and Province (LAPOP) or Region (Gallup) Fixed Effects and Time-Trends Log Share VZs 0.024 0.016 0.002 0.037*** -0.018* (0.017) (0.013) (0.013) (0.013) (0.009) R-squared 0.085 0.076 0.033 0.075 0.062 c) IV: Year and Province (LAPOP) or Region (Gallup) Fixed Effects and Time-Trends Log Share VZs 0.033* 0.031* 0.011 0.043* -0.018** (0.019) (0.017) (0.016) (0.022) (0.008) Men: IV: Year and Province (LAPOP) or Region (Gallup) Fixed Effects and Time-Trends Log Share VZs 0.059** 0.059** 0.014 0.058* -0.022* (0.023) (0.024) (0.018) (0.029) (0.011) Women: IV: Year and Province (LAPOP) or Region (Gallup) Fixed Effects and Time-Trends Log Share VZs 0.006 0.006 0.011 0.028 -0.017 (0.021) (0.020) (0.020) (0.022) (0.012) Individuals 7,717 7,910 7,997 7,997 7,997 Men 3,807 3,876 3,405 3,405 3,405 Women 3,900 4,029 4,592 4,592 4,592 *** p<0.01, ** p<0.05, * p<0.1.Robust standard errors clustered at province level in parentheses. Log Share Venezuelans is relative to the 2017 population of the province. Outcome variables noted as STD are also standardized. Log Share VZs in the province (region for Gallup outcomes) in 2007 interacted with year fixed effects are used as instruments for Log Share of Venezuelans. All regressions include controls for a quadratic in age, education, marital status, household size and whether employed. 36 Appendix Figures and Tables Table A.1: First-Stage Regression for Quality of Local Labor Markets Overall Men Women Outcome: Log Local Informal Emp Rate Export Shock in Year Prior Oct 2017 -3.607*** -7.294** -3.609*** -6.845* -3.611*** -7.813** (0.716) (3.374) (0.771) (3.433) (0.691) (3.400) Export Shock Squared -10.005 -9.027 -11.144 (8.021) (8.204) (8.036) Log Local Total Population 2017 0.028 0.011 0.025 0.010 0.031 0.012 (0.034) (0.041) (0.036) (0.044) (0.033) (0.039) Log Local Venezuelans 2007 -0.010 0.001 -0.008 0.001 -0.012 0.001 (0.036) (0.039) (0.037) (0.041) (0.036) (0.039) Log Mean Local Expenditure 2013 -0.482*** -0.430*** -0.480*** -0.437*** -0.486*** -0.423*** (0.126) (0.131) (0.127) (0.131) (0.127) (0.132) Log Agriculture Rate 2007 -0.091** -0.117** -0.090** -0.112** -0.092** -0.122** (0.042) (0.049) (0.042) (0.049) (0.043) (0.049) Log Manufactoring Rate 2007 0.060 -0.009 0.072 0.011 0.048 -0.030 (0.089) (0.108) (0.090) (0.110) (0.091) (0.108) Log Travel Duration to Lima 0.044* 0.045* 0.047* 0.048** 0.040* 0.042* (0.024) (0.023) (0.025) (0.024) (0.023) (0.022) R-squared 0.762 0.766 0.762 0.766 0.763 0.770 Observations 4,176 4,176 4,176 4,176 3,693 3,693 Robust standard errors clustered at centro poblado level in parentheses. Local informal employment rate is measured in the 2017 census. All regressions also control for gender, age, education, marital status, months living in Peru, household socioeconomic strata, household size, number of people sharing one's bedroom, total income, whether in formal employment, and occupation including not working and original municipality in Venezuela fixed effects. 37 Table A.2: Impact of Local Labor Market Conditions on the Labor Market Outcomes of Venezuelans Employment Log Wages if Employed Men Log Local Informal Employment Rate 0.055 0.061 0.003 -0.083 -0.098 -0.171 (0.040) (0.039) (0.066) (0.089) (0.083) (0.172) R-squared 0.056 0.118 0.076 0.124 Observations 4,176 4,176 4,176 3,909 3,909 3,909 Mean Outcome 0,939 6,97 Women Log Local Informal Employment Rate 0.066 0.074 -0.043 -0.076 -0.076 0.357* (0.053) (0.059) (0.118) (0.112) (0.104) (0.196) R-squared 0.056 0.118 0.068 0.145 Observations 4,176 4,176 4,176 2,869 2,869 2,869 Mean Outcome 0,782 6,81 Origin Municipality FE No Yes Yes No Yes Yes OLS/IV OLS OLS IV OLS OLS IV *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors clustered at centro poblado level in parentheses. Local informal employment rate is measured in the 2017 census. The predicted export shock in each centro poblado in year prior to Oct 2017 is used to instrument for the informal employment rate. All regressions control for gender, age, education, marital status, months living in Peru, household socioeconomic strata, household size, and number of people sharing one's bedroom and the following variables measured at the centro poblado level: log population in 2017, the log number of Venezuelans in 2007, log mean household expenditure in 2013, log agricultural rate in 2007, log manufacturing rate in 2007 and log travel distance to Lima. 38