暴力犯罪對就業決策之影響：以墨西哥毒品戰爭為例 - 政大學術集成

全文

(1)國立政治大學應用經濟與社會發展英語碩士學位學程 International Master’s Program of Applied Economics and Social Development College of Social Sciences National Chengchi University. 立. 碩士論文 Master’s 治 Thesis 政. 大. ‧. ‧ 國. 學暴力犯罪對就業決策之影響：. y. Nat. er. io. sit. 以墨西哥毒品戰爭為例. n. The Effects a l of Violent Crime oni vLabor Decisions: n Drug War C From Evidence U h e n ThehMexican i. gc. Student: Montes Garc´ıa, Laura Advisor: Huang, Po-Chun. 中華民國110 年1 月 January, 2021. DOI:10.6814/NCCU202100271.

(2) 暴力犯罪對就業決策之影響：以墨西哥毒品戰爭為例 The Effects of Violent Crime on Labor Decisions: Evidence From The Mexican Drug War. 研究生：林蘿拉. Student: Laura Montes Garc´ıa. 指導教授：黃柏鈞 Advisor: Huang, Po-Chun. 立. 政治大國立政治大學. ‧ 國. 學. 應用經濟與社會發展英語碩士學位學程. ‧. 碩士論文. n. er. io. sit. y. Nat. al. Ch. e nAgThesis chi. i Un. v. Submitted to International Master’s Program of Applied Economics and Social Development National Chengchi University. 中華民國110 年1 月 January, 2021. DOI:10.6814/NCCU202100271.

(3) Acknowledgments I would like to acknowledge my sincerest thanks to Professor Po-Chun Huang for made this research possible. His patience, guidance, and assistance have been invaluable throughout all stages of this work. I also owe a great deal of gratitude to my committee members Professor Tzu-ting Yang and Professor Yu-Hsuan Su, for their valuable suggestions and comments. I greatly appreciate the support provided by the Consejo Nacional de Ciencia y Tecnolog´ıa (CONACYT) and to the Fundación Mexicana para la Educación, Ciencia y Tecnolog´ıa (FUNED); without the financing from CONACYT and FUNED, my studies at the National Chengchi University would not have been possible.. 立. 政治大. ‧. ‧ 國. 學. Finally, my warmest thanks are to my friends and family. Words can not express how grateful I am to my parents. Thank you for your unconditional support and love.. n. er. io. sit. y. Nat. al. Ch. engchi. i Un. v. iii DOI:10.6814/NCCU202100271.

(4) Abstract 本篇論文中，我研究墨西哥毒品戰爭期間暴力犯罪率增加，對男女就業決策之影響。我選擇使用MxFLS縱向調查，因為它與墨西哥毒品戰爭同時發生，並使用MxFLS資料以及單個固定效果模型。同時，我也兼顧個人和州際變數下，暴力就業決策的相關內生特性、時不變特性，以及未觀察到的特性。由實驗結果得知，暴力犯罪對就業決策的結果具影響甚重，但是，這種影響對於不同的人也有是不同的影響程度。暴力犯罪的增加並不會影響男性的就業決策；然而對於婦女而言，暴力減少了她們的工作可能性，並減少了她們工作的星期數和小時數。. 立. 政治大. ‧. ‧ 國. 學. This study examines the impact of an unexpected surge in violent crime on the labor behavior of women and men in the context of the Mexican drug war. I exploit the longitudinal nature of the MxFLS survey and its precise timing encircling the implementation of the Mexican drug war. By using the MxFLS information combined with an individual fixedeffects strategy, I consider potentially endogenous, time-invariant, unobserved characteristics of states and individuals that may be correlated with violence exposure and labor decisions. The results suggest that violent crime does have a la powerful impact on thei vlabor market outcomes, though this effect is different n of the population. AlC h for distinct groups U n g c h iin violent crime do not change though findings suggest thateincreases men’s labor decisions, for women, the Mexican drug war-related violence significantly decreases their probability of joining the labor market and reduces the number of weeks and hours they choose to work.. n. er. io. sit. y. Nat. iv DOI:10.6814/NCCU202100271.

(5) TABLE OF CONTENTS LIST OF TABLES. vi. LIST OF FIGURES. vii. 1 Introduction. 1. 2 Literature Review 2.1 Violent Crime and the Mexican Labor Market . . . . . . . . . . . . . . 2.2 Violent Crime and the Labor Market in Other Countries . . . . . . . . .. 4 4 6. 3 Background 3.1 Mexico’s Violence Frame . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Labor Market Structure . . . . . . . . . . . . . . . . . . . . . . . . . .. 7 7 8. 政治大 . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. ‧. ‧ 國. 立. 學. 4 Data and Sample 4.1 Measuring Violence . . . . . . . 4.2 The Mexican Family Life Survey 4.3 Sample Selection . . . . . . . . 4.4 Descriptive Statistics . . . . . .. n. Ch. . . . . .. . . . . .. . . . . .. . . . . .. engchi. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. sit. io. al. . . . . .. . . . . .. . . . . .. er. Nat. 6 Results 6.1 Main Results . . . . . . . . 6.2 Gender and Formality Status 6.3 Age Groups . . . . . . . . . 6.4 Potential Mechanisms . . . . 6.5 Leading-values Test . . . . .. 13. y. 5 Empirical Strategy. i Un. v. . . . . .. . . . . .. 9 10 11 12 12. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. . . . . .. 16 16 17 19 20 21. 7 Conclusion. 22. 8 Bibliography. 24. 9 Tables and Figures. 27. v DOI:10.6814/NCCU202100271.

(6) LIST OF TABLES. Table 7. 立. 政治大. . 30 . 31 . 32 . 33. 學 ‧. Nat. y. Table 6. . 29. io. sit. Table 5. . 27 . 28. n. al. er. Table 4. Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . Impact of Total Crime Rate on Labor Market Outcomes . . . . Impact of Violent Crime on Labor Market Outcomes: By crime type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impacts of Total Crime Rate on Labor Market Outcomes: By gender an formality status . . . . . . . . . . . . . . . . . . . . Impacts of Total Crime Rate on Labor Market Outcomes: By age group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impact of Total Crime Rate on Labor Market Outcomes: By Subjective Measures of Fear . . . . . . . . . . . . . . . . . . . . . Impact of Total Crime Rate on Labor Market Outcomes comparing the same individual in MxFLS1 and MxFLS2 . . . . . . . .. ‧ 國. Table 1 Table 2 Table 3. Ch. engchi. i Un. v. vi DOI:10.6814/NCCU202100271.

(7) LIST OF FIGURES Victims of intentional homicide, rate per 100,000 population . . . . . . 34 Monthly homicide rate per 100,000 population and MxFLS application periods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35. 立. 政治大. 學 ‧. ‧ 國 io. sit. y. Nat. n. al. er. 1 2. Ch. engchi. i Un. v. vii DOI:10.6814/NCCU202100271.

(8) 1. Introduction. Violent crime has been a phenomenon recently enlarged in Mexico. After a decade of steadily declining rates of violent crime, in the mid-2000s, homicides per capita nearly tripled within just three years (2007-2010). According to numbers from the National Institute of Statistics and Geography (INEGI), by the end of 2011, the number of homicides increased to almost 23 deaths per 100,000 people (See Figure 1), approximately four times the U.S. homicide rate. Moreover, the Institute for Economics and Peace (IEP) estimates that the cost of violence in Mexico corresponds to USD 233,000 million or 17% of the country’s GDP, not including intangible costs difficult to quantify. Academics have widely discussed this sudden increase in violent crime, creating a broad debate about its possible causes. While different hypotheses prevailed, the preferred explanation attributes violence increases to the Government’s military strategy against Drug Trafficking Organizations (DTOs) launched in late 2006, commonly called the Mexican drug war.1. 立. 政治大. ‧. ‧ 國. 學. Correctly measuring the effects of violent crime on labor market outcomes is critical to understanding the social and economic costs of violence. Moreover, the escalation of drug-related violence in Mexico provides an appropriate framework to analyze the impact of violent crime on labor market decisions.2 By identifying the direction and magnitude of the violent crime impact on labor decisions, this paper contributes to the literature on the effects of the unexpected and heterogenous drug-related crime surge in Mexico on the labor market.. er. io. sit. y. Nat. Nevertheless, studying the impact of violence on individuals’ labor decisions imposes some empirical challenges. First, the anticipated and non-random nature of violence can create a potential selection bias. Second, the intensification of violence may be correlated to unobserved trends in regional characteristics, causing omitted variable bias. Lastly, increases in violent crime usually happen concurrently with other events that also influence the labor market, confounding most analyses of the relationship between violence and labor behavior.. n. al. Ch. engchi. i Un. v. 1 Until. 2006, the Mexican Government relied on the eradication of marijuana and opium crops as a strategy to combat drug trafficking. However, after late 2006, the approach to confront DTOs changed when the Government implemented the Mexican drug war strategy. The Mexican drug war primary purpose was to capture or kill DTOs’ leaders by executing a wide-reaching military deployment across the country. However, the most notorious consequence of the Mexican drug war was a breakdown of the DTOs’ structure that brought fragmentation and a subsequent increase in the number of existing DTOs that began fighting among each other for territorial control. The war within DTOs pushed them to look for additional funding sources to fight against their rivals; thus, the criminal organizations diversified their range of activities to include extortion, kidnapping, human trafficking, and fuel theft, among other crimes (Guerreo-Gutierrez, 2011). 2 The new dynamic of violence in Mexico had a much more direct effect on society at large, and, therefore on the economic behavior of individuals and households. The fear of being a target of DTOs would induce behavioral changes that may impact individuals’ economic and labor decisions.. 1 DOI:10.6814/NCCU202100271.

(9) To overcome these empirical challenges and effectively estimate the consequences of violence on labor market decisions, I first rely on the unexpected nature of the surge in violence caused by the Mexican drug war to counter the typically non-random aspect of violent crime. Then, by using the MxFLS individual-level data collected before and after the sudden increases of drug-related crimes in 2007 combined with state-level data on violent crime rates for the same time period, I can adopt an individual fixed-effects model as the primary empirical strategy. This approach allows individuals to act as their own counterfactual by comparing their labor decisions during a low-violence period to their labor decisions while experiencing high-violence levels. Furthermore, I take into account potentially endogenous, time-invariant, unobserved characteristics of states and individuals that may be correlated with violence exposure and labor decisions. Estimates from the fixed-effects model show that violence increases have a negative impact on individuals’ labor decisions. Exposure to a violent environment significantly decreases an individuals’ probability to join the labor market and reduces the number of hours they choose to work. From 2000 until the drug war implementation in December 2006, there was a reduction in the total crime rate of 14%. After 2006, however, Mexico experienced an increase in the total crime rate of approximately 90%. Consequently, my results suggest that an individual experiencing the drug war violence surge would decrease their probability of joining the labor market by 2.25 percentage points, contrary to the increase of 0.35 percentage points in their likelihood of being employed before the drug war implementation.. 立. 政治大. ‧. ‧ 國. 學. sit. y. Nat. n. al. er. io. Besides the direction and magnitude of the effects of violence on labor decisions, it is also relevant to understand the possible channels through which violent crime changes individuals’ labor choices. Brown and Velásquez (2017) argue that the fear of being a victim of violent crime is the main mechanism affecting a person’s choice of participating or not in the labor market. To confirm this idea, I estimate an individual fixed-effects model using the respondent’s subjective measure of fear reported in the MxFLS. The findings suggest that fear of victimization is, as previous studies proposed, a channel that triggers the decrease in the probability to join the labor market as a response to increases in violent crime.. Ch. engchi. i Un. v. Furthermore, according to the theory proposed by Becker and Rubenstein (2011), economic incentives shape the degree to which fear of victimization distorts choices. Since people can learn to control their fear when it is in their long-run interest, the authors argue that the willingness to control the fear depends on the final economic costs and benefits. For this study, the theory implies that people with a weak attachment to the labor market should experience a greater effect on their labor decisions during a violence surge. Since the level of attachment to the labor market mainly differs by gender and working formality status of individuals, I separately analyze women and 2 DOI:10.6814/NCCU202100271.

(10) men, and formal and informal workers. Consistent with the theory, I find that individuals more weakly attached to the labor market are the most responsive to violent crime. Specifically, exposure to a violent environment significantly decreases females’ probability of being employed and reduces the number of weeks and hours they choose to work. The magnitude of this effect is even greater for women working in the informal sector. Women experiencing the drug war violence surge – a 90% increase in the total crime rate, would decrease their probability to participate in the labor market by 4.14 percentage points; and informal female workers would reduce their likelihood by 9.81 percentage points. Conversely, violent crime changes do not affect males’ decision to participate in the labor market nor affect the number of hours or weeks they choose to work, regardless of whether they are formal or informal workers. This difference highlights the disparity of the impacts of violence between men and women and suggests that not differentiating by gender and formality status may mask the different reactions among these groups.. 政治大 The impact of violent立 crime caused by the Mexican drug war on individuals’ labor ‧. ‧ 國. 學. behavior remains understudied. Moreover, most of the work related to the effects of violence on labor outcomes uses aggregated and cross-sectional data and cannot control for potential selection problems. This study directly addresses this challenge by exploiting the exogenous variation in the location, timing, and magnitude of the violent crime increase and by using longitudinal data to capture individuals’ decisions before and after the onset of the Mexican drug war. Relatedly, Velásquez (2020) makes one of the few contributions to the literature on the labor market effects of the Mexican drug war by examining the impact of intentional homicides at a municipal-level on labor market outcomes of Mexican workers. In my analysis, I explore this same relationship using different violence measurements and definitions, extend Velásquez’s sample to all working and non-working individuals, focus on the state-level total crime rate impact on men and women, and differentiate by formality status.. n. er. io. sit. y. Nat. al. Ch. engchi. i Un. v. Furthermore, current research takes gender neutrality for granted, often considering a male-centric analysis. Understanding the particular effects of violence by gender is imperative in an already unequal society where any shock that disproportionally affects women’s labor decisions should be considered a contributing factor to the gender gap widening. Moreover, a country in which informality levels surpass Latin America’s average by 11 percentage points needs to recognize the impact of violent crime on informally-employed individuals’ labor decisions to design better public policies. It becomes even more critical when we consider the effects of violence in addition to the vulnerability of informal sector workers.3 3 For. many workers, informality means a lack of social protection, labor rights, and decent working conditions; while for informal businesses, it means low productivity and lack of ac-. 3 DOI:10.6814/NCCU202100271.

(11) This study offers two main contributions to the literature on the consequences of the Mexican drug war caused violent crime on the labor market. First, it adds to the existing literature by analyzing the causal effects of violent crime on individuals’ labor decisions, which has not been a subject of proper attention. To date, the literature mostly suggests that violent crime has a negative impact on labor market activity (Dell, 2015; Basu and Pearlman, 2018; Velásquez, 2020); however, there is not a clear consensus about its magnitude and direction. Second, to shed light on violent crime’s specific effects on distinct groups of the population. Few studies have analyzed its effects on labor choices differentiating by gender, job type and age; in this study, I stratify the sample by gender, age, and formality status to capture crucial differences in deciding whether to join the labor market. The next section covers the existing literature regarding the diverse effects of violence in the Mexican and other countries’ labor markets. Section 3 provides an overview of Mexican drug war-related violent crime and Mexico’s labor market characteristics. Section 4 describes the data and sample used in this study. Section 5 discusses the empirical strategy. Section 6 presents the results and the leading-values test outcomes. Section 7 concludes.. 政治大. 立. ‧ 國. ‧. Literature Review. y. Nat. io. sit. Violent Crime and the Mexican Labor Market. al. er. 2.1. 學. 2. v. n. A novel body of literature has analyzed the consequences of violent crime on the Mexican labor market supply and demand, yet the effects remain unclear. On the supply side, individuals’ perceived fear and concerns result in a reluctance to join the labor market (Dell, 2015; Velásquez, 2020) or other general activities of their environment (Becker and Rubinstein, 2011). Additionally, an increase in illegal but profitable jobs may reduce the individual’s willingness to participate in the legal labor market (Basu and Pearlman, 2018). In contrast, labor supply can rise if members of a household join the labor force to offset a decline in family income, i.e., if the earnings of the household head decrease due to a shock resulting from violence (Cunningham, 2001; Brown and Velásquez, 2017).. Ch. engchi. i Un. Labor demand can also change due to violent crime, and the direction is likewise inconclusive. On the one hand, it may decline if businesses harmed by violence reduce their investments, downsize, go out of business, or never enter the market. The reduction of pedestrian traffic due to fear may also affect businesses in previously prolific cess to financing (Bonnet et al., 2019).. 4 DOI:10.6814/NCCU202100271.

(12) shopping areas. On the other hand, labor demand may increase due to an economic boost caused by excessive spending of disposable income obtained from illegal activities (Velásquez, 2020). Among the literature on the consequences of the growing violence in Mexico, few studies address the impact on individual labor outcomes. Although the economic costs of violence are just briefly discussed in her work, Dell (2015) made the first contribution. By using a regression discontinuity design, she identifies a causal relationship between the sudden increases in violent crime and the electoral victories of the National Action Party (PAN) candidates for municipality mayors. Moreover, she finds that increases in violence reduce female labor participation and male earnings in the informal sector. By contrast, Dell identifies an insignificant impact on workers in the formal sector.. 政治大. In a more extensive study on the impact of drug trafficking violence on the labor market, Velásquez (2020) finds similar results. By using longitudinal data from the MxFLS, she spots that the surge in violent crime has negative labor consequences for self-employed workers. Her results suggest that self-employed men experience a significant reduction in their earnings while self-employed women decrease their worked hours or leave their jobs entirely. In contrast, outcomes for waged-employed workers show insignificant changes. Furthermore, the negative impact on the self-employed women supports the idea that self-employed individuals, who have a weaker attachment to the labor market, are the most responsive to increases in violence. Since selfemployed workers are the direct victims of extortion and, at the same time, the group with the highest chance to participate in the informal sector, Velásquez proposes that the mechanism behind the negative impact on labor participation is the sense of victimization caused by fear.. 立. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i Un. v. The repercussions of excessive violence are differently perceived among diverse groups. Gender, level of education, marital status, and household role are some of the characteristics that define the direction and magnitude of violent crime on an individual’s labor response. Taking this into consideration, Basu and Pearlman (2018) examine the impact of homicides on employment for the Mexican case. They use quarterly data from the Mexican National Survey of Occupation and Employment (ENOE) and apply an instrumental variables strategy by using the length of federal toll-highways as an instrument of homicides interacted with the amount of cocaine seized in Colombia. Their results suggest that homicides reduce employment only for young males and females, or non-household-heads. Namely, violence exclusively modifies the behavior of individuals with a weak attachment to the labor market. Additionally, outcomes reveal that employed people increase their worked hours, but their monthly income remains unchanged. 5 DOI:10.6814/NCCU202100271.

(13) 2.2. Violent Crime and the Labor Market in Other Countries. Besides the abovementioned studies focused on the Mexican case, scholars targeting diverse countries have also made relevant contributions to the literature. For instance, Shemyakina (2011) analyzes the effects of the Tajikistan civil war on education and labor market outcomes for women and men. Her research uses difference-in-difference to measure the long-term conflict’s economic impact after Tajikistan’s independence in 1991. The results suggest a decline in female educational attainment combined with an increase in their labor participation. Nonetheless, the conflict had insignificant effects on men’s outcomes. Hamermesh (1999) shows evidence from the U.S. by studying the impacts of crime on work trends. Connecting household data on work timing from 1973 to 1991 to FBI crime reports, he finds that high homicide rates in metropolitan areas discourage individuals from working evening and night shifts and encourage them to work morning shifts instead. His main argument is that criminal activity inflicts negative externalities on the labor market, mainly because the individual’s fear of crime generates deviations from the optimal working time patterns.. 立. 政治大. ‧. ‧ 國. 學. Continuing on to the impact of violence on firms, Collier and Duponchel (2013) use the World Bank 2007 Employers’ Survey to analyze the case of Sierra Leone, which experienced a violent civil war from 1991 to 2002. Their hypothesis suggests that, during a conflict, violence negatively impacts production through technical regress; and demand due to a decrease in income. The results show that by 2006, the conflict’s intensity reduced firms’ size, confirming the authors’ assumption.. sit. y. Nat. n. al. er. io. Colombia has been a subject of extensive research related to violence literature. For example, Bozzoli, Brück, and Wald (2013) examine the impact of violent conflict on self-employment. Their results show that increments in the displacement rate due to violence increase the self-employment rate and sharply reduce the hourly income for self-employed workers. Furthermore, Fernandez, Ibanez, and Peña (2014) find that as a response to violent shocks, households in rural Colombia decrease the time spent on agricultural work and increase their supply of labor to non-agricultural activities. However, the labor market cannot completely assimilate the additional labor supply making women unable to find jobs in the formal market and increasing the time men dedicate to household chores and leisure.. Ch. engchi. i Un. v. 6 DOI:10.6814/NCCU202100271.

(14) 3. Background. 3.1. Mexico’s Violence Frame. In Mexico, criminal organizations’ primary role is to transport illicit drugs into the United States, the largest cocaine market in the world (UNODC, 2010). Due to the geographical proximity between the two countries, Mexico has been a strategic location for drug trafficking. As the demand for cocaine increased in the United States in the 1970s, criminal organizations started to grow and represent a powerful influence in Mexico. Furthermore, since the Andean states are the largest cultivators of cocaine,4 Mexico nowadays primarily plays the intermediary role connecting South American cocaine suppliers with consumers in the United States. The Mexican drug war broke out in December 2006, when the conservative National Action Party (PAN) assumed command of the Government and initiated a war against DTOs as a measure to mitigate illegal-drug related activities. The main distinction of the new approach to eradicating drug trafficking compared to former initiatives was the use of militarized force to capture DTOs’ leaders (Calderón et al., 2015). To accomplish this strategy, the Government deployed thousands of military, navy, and police troops to different states in late 2006. By 2011, there were approximately 45,000 troops involved.5 While the Mexican drug war aimed to reduce violent crime and stabilize the country, it fortuitously increased the total crime rate, especially in states intervened by military forces. This unanticipated effect resulted from the capturing of DTOs’ leaders since it created a power void and encouraged competition among rival blocks over the control of specific regions.. 立. 政治大. ‧. ‧ 國. 學. er. io. sit. y. Nat. al. n. iv n C As the number of rival blockshcompeting among e n g c h i Ueach other increased, the violence within DTOs got stronger. More DTOs were fighting to obtain control of the same. critical territory. Consequently, numerous states with low crime rates and no presence of organized crime became affected by DTOs’ incidence and increases in drug-related violence. Moreover, the war within DTOs required them to look for additional funding sources to combat their rivals. Thus, drug trafficking organizations expanded their criminal activities to include non-drug-related offenses such as assault, robbery, and kidnapping. By doing so, the violence landscape in Mexico changed; the gruesome nature of these crimes, the lack of trust in the state’s institutions, and police corruption triggered unusual levels of fear and concern among the society (Velásquez, 2020). 4 Colombia. being the largest producer in the world (UNODC, 2006). annual cost of the Mexican drug war was about 9 billion USD, nearly as high as the Government’s expenditure on social development (Dell, 2015). 5 The. 7 DOI:10.6814/NCCU202100271.

(15) The consequences of this new violence’s dynamic shaped many possible channels through which the economy and, therefore, the labor market could be disturbed. Given that the impact of violence on emotions is pernicious and the influence of fear distorts the perception, altering the decision-making processes of individuals (Becker and Rubinstein, 2011), the increased rate of violent crime may influence individuals to modify their labor decisions due to the fear of being physically harmed or financially damaged by criminal groups operating in their community. Victimization surveys in Mexico corroborate this behavior by revealing that the percentage of adults feeling their state of residence is unsafe increased from 5.4% in 2004 to 65% in 2009, while the percentage of people considering their workplace unsafe also rose from 13.7% to 19% (Benyishay and Pearlman, 2013). These increases in the sentiment of victimization could increase the Mexican drug war costs if it triggers behavioral changes like reducing worked hours or school attendance. For instance, the National Survey on Insecurity (ENSI) for the year 2009 unveils that half of the individuals stopped going out during nighttime due to increased violence, while 15% stopped using public transportation, eating out, and attending public events (Benyishay and Pearlman, 2013).. 立. ‧ 國. 學. Labor Market Structure. ‧. 3.2. 政治大. Nat. y. sit. n. al. er. io. The Mexican labor market is defined by a stable and relatively low labor force participation compared to other Latin American countries. Just in 2020, 63.6% of the total population between 15 and 64 years-old were participating in the labor force (STPS, 2020). Moreover, the unemployment rate has been somewhat low during the last two decades, reaching 3.4% by the end of 2019, 1.9% lower than the world’s average, and 4.6% lower than Latin American’s average (ILO, 2020).. Ch. engchi. i Un. v. Furthermore, employment and labor force participation differ significantly between men and women and among age groups. Labor participation for women between 20 and 64 years-old has increased over the last decades, from 66.52% in 2000 to 71.5% in 2019. However, it remains below the men’s participation rate of 82% in 2019. The employment rate for women within the same age group is 44.64%, also lower than the 77.1% for men in 2019 (Henry and Fraga, 2019), while men’s earnings were 18.8% higher than women’s (ILO, 2020). Additionally, in 2018, the gender wage gap was more pronounced for self-employed workers at 44%, and highly educated individuals with a 33% differential (OECD, 2020). Despite the relatively low unemployment rate, Mexico does not offer a national system of unemployment insurance. Many scholars recognize this factor as a potential trigger of informality. In this sense, informality has been a significant issue in Mexico 8 DOI:10.6814/NCCU202100271.

(16) over the last two decades. In 2019, approximately 57% of the labor force worked in the informal sector, just slightly above Latin America’s average and considerably exceeding the informality levels in developed countries. In Mexico, employment in the informal sector is extensive among the general society; while it is most prevalent in low-skilled workers and vulnerable people, educated or well-payed workers also participate in the informal labor market (OECD, 2019). Moreover, Mexican states present notable differences in employment and informality rates. In 2017, the unemployment rate fluctuated between 7.3% in Tabasco to 1.4% in Guerrero and other southern states. Informality follows the same pattern. It varies considerably from low rates in the northern states of Chihuahua and Coahuila (37% each) to higher rates in the central and southern states of Guerrero and Chiapas (78% each) (Romero and Ortiz, 2019).. 政治大. Hence, informality is widely found among people of different educational levels, in all geographical areas, and across jobs with low and high wages. This wide-reaching predominance displays a duality in the labor market that permeates the whole Mexican economy by reducing productivity and economic growth (Alvarez and Ruane, 2019). Informal employment can be recognized as a problem since it also aggravates inequality and encourages social segregation (ILO, 2015). For instance, a person informally working does not have sufficient job security nor access to social benefits and does not receive proper training opportunities at their workplace (UNDESA, 2020).. 立. ‧. ‧ 國. 學. y. Nat. sit. n. al. er. io. Despite specific arrangements or employment contracts, people in Mexico tend to work long hours. Mexican workers invest around 2,137 hours per year working, 385 hours more than the OECD average in 2019. Still, the annual average wage is comparatively low (around USD 17,000 PPP in 2019) and has remained constant during the last two decades (OECD, 2019). Additionally, the average wage varies notably among states. Workers in Mexico City and northern states perceive monthly wages ranging between MXN 7,500 and 8,500 (Mexican pesos), while the average monthly wage for southern states like Chiapas and Oaxaca remains under MXN 4,500 (OECD, 2019). Differences between wages in rural and urban areas also prevail, where average wages are three to four times higher for individuals working in cities (Romero and Ortiz, 2019).. 4. Ch. engchi. i Un. v. Data and Sample. For this study, I collect data from four sources. First, I use criminal activity information from the National Institute of Statistics and Geography (INEGI) and the Office of the Mexican Attorney-General (PGR). I use the INEGI monthly reports on the number of intentional homicides in Mexico at a state level, and the PGR data for the number of 9 DOI:10.6814/NCCU202100271.

(17) violent burglary, robbery, and aggravated assault cases, all of them recorded annually at a state level. I use the total number of the abovementioned crimes per state in conjunction with population data from the National Population Council (CONAPO) to create an annual violent crime rate per 100,000 people. Then, I link the previously constructed violent crime rate to the MxFLS, a rich longitudinal survey representative of the Mexican population that compiles broad demographic and socioeconomic characteristics started in 2002. The MxFLS ideally captures a general picture of Mexico’s environment by recording information surrounding the sudden increase in violent crime. The first follow-up of the survey occurred between 2005 and 2006, a time with relatively low levels of violence, while the second follow-up was conducted from 2009 to 2013 when the levels of crime spread exponentially.. 政治大 For this empirical exercise, 立I use data regarding four types of violent crime. First, I build 4.1. Measuring Violence. ‧. ‧ 國. 學. a state-level rate of intentional homicides per 100,000 people by collecting the reported annual numbers of intentional homicides from INEGI and using the total population figures from CONAPO. Then, using data from PGR and CONAPO, I constructed a robbery rate per 100,000 people by combining annual violent burglary cases and robbery cases in each state as a supportive measure of violence. Similarly to the robbery rate, I used annual aggravated assault cases and population data to create an aggravated assault rate per 100,000 people for each state. Lastly, I combined the three previously constructed violence rates to create a total crime rate to serve as the principal indicator of violent crime.. n. er. io. sit. y. Nat. al. Ch. engchi. i Un. v. In this study, I use the definition of violent crime haped by the U.S. Department of Justice.6 Thus, violent crime refers to those offenses which involve force or threat of force against a victim and is composed of four offenses: murder and nonnegligent manslaughter, forcible rape, robbery, and aggravated assault (Douglas et al., 2013). Three out of these four criminal offenses are used to construct the total crime rate mentioned before,7 resulting in a more comprehensive indicator of the drug violence behavior.8 6 Although. the way of classifying the concept of violent crime varies among countries. this study, I do not include the forcible rape rate, although it is part of the definition of violent crime. I omit it because the triggers and mechanisms of this type of offense are different from those in drug-related violent crime. 8 The homicide rate expressed as the number of intentional deaths per 100,000 habitants has been the preferred measure of violence in the literature. The finality of death and the difficulty of “hiding a body” secure the homicide rate as an indicator with a small underestimation problem (Shrader, 1999). However, a measurement error is still possible, and the homicide rate could bias any empirical estimation if homicide cases are not accurately reported to the authorities. 7 In. 10 DOI:10.6814/NCCU202100271.

(18) 4.2. The Mexican Family Life Survey. The MxFLS includes information on approximately 8,440 households and 35,600 individuals from 16 of the 32 Mexican states. It has been developed and managed by researchers from the Iberoamerican University (UIA, per its name in Spanish) and the Center for Economic Research and Teaching (CIDE, per its name in Spanish) in collaboration with researchers from Duke University in the United States. The year 2002 is the MxFLS baseline, and the main goal of the survey is to keep track of all the baseline respondents and their children born after 2002. This group of individuals is considered the panel respondents and will participate in the two consecutive follow-ups. Furthermore, the MxFLS holds many useful characteristics for analyzing the consequences of the Mexican drug war among the population. As I pointed before, the first follow-up of the survey (MxFLS2) occurred in 2005-2006, and the second one (MxFLS3) in 2009-2013. The MxFLS1 and MxFLS2 display data from respondents living during a period of “normal” levels of violence, while the MxFLS3 shows information of respondents living during a period of high levels of violent crime (See Figure 2). The set-up of the MxFLS allows me to compare the information provided by the same respondent in periods of both low and high violence levels, while the longitudinal nature of the survey permits the control for unobserved time-fixed characteristics.. 立. 政治大. ‧. ‧ 國. 學. n. al. er. io. sit. y. Nat. Another helpful aspect of the MxFLS is the abundance of information about its respondents. The data includes questions about the social, economic, and health status of the surveyed individuals. The questions cover segments related to labor history, educational level, mental and general health history, fertility history, migration, and marriage history for adult respondents. Additionally, the MxFLS includes householdlevel data. One family member reports the socio-demographic information of each household member, household expenditures, and household economy.. Ch. engchi. i Un. v. Although the nature of the MxFLS and its rich contents are quite suitable for this study, the benefits of using longitudinal data depend on the extent to which the respondent’s baseline is successfully re-interviewed in the two successive follow-ups. Moreover, attrition can be a critical problem in longitudinal studies. Fortunately, the MxFLS has low attrition levels. For the first and second follow-up, about 89% of the respondents were successfully re-interviewed.. 11 DOI:10.6814/NCCU202100271.

(19) 4.3. Sample Selection. My sample consists of all individual between 15 and 80 years old, successfully contacted in the three MxFLS waves.9 For this reason, the usual age range of 15 to 65 has been extended to include those up to 80 years old. I allow working and non-working respondents in the sample considering that information about their decision to eventually join the labor market can be relevant for the results.10 From the MxFLS1 baseline sample of 35,000 respondents, I excluded individuals who died between the survey waves, individuals who migrated to the United States, and respondents that did not continue the follow-ups for the subsequent surveys. In MxFLS2 and MxFLS3, new household members joined the surveys. I discarded these new members from the sample since they were non-panel respondents in the MxFLS1 or MxFLS2. After deleting observations with missing or dubious information, 10,928 male and female individuals and a total of 32,784 observations remained.. 治政大 and aggravated assault) were State-level crime rates (total crime, homicide, robbery, 立 assigned to each individual in the MxFLS sample, subject to the individual’s state of ‧. ‧ 國. 學. residence and the MxFLS interview date. Interviews were carried from mid to end 2002, late 2005 to mid-2006, and mid-2009 to late 2012. Due to expected delays in the response to violence, I assign crime rates from a year before the MxFLS interview date. From a total of 32 states in Mexico, the MxFLS baseline was conducted in 16 of them.11 Including respondents that changed their state of residence for subsequent MxFLS follow-ups, a total of 22 states are covered in my sample. Conducting this study at a more disaggregated level of analysis, such as at the municipality-level, would provide more accuracy about local variations of the impact of violent crime on labor decisions; however, criminal activity data is only available at the state-level of analysis.. n. er. io. sit. y. Nat. al. 4.4. Ch. engchi. i Un. v. Descriptive Statistics. Table 1 shows the descriptive statistics for the sample. I examine three primary labor dependent variables: worked last year, which is a dummy variable equal to 1 if the respondent worked during the 12 months before the interview, and zero otherwise; the average number of weeks worked during the 12 months before the interview; and the average number of hours worked per week during the 12 months before the in9 Financial. incentives that allow worker’s retirement are an important determinant on the decision to stop participating in the labor force, however, in Mexico the pension system is less generous than in developed countries. Because of the lack of income and financial resources, elderly individuals are more likely to continue working (Van Gameren, 2008). 10 In correspondence to the labor force participation definition. 11 The 16 states included in the MxFLS baseline are Baja California Sur, Ciudad de Mexico, Coahuila, Durango, Guanajuato, Jalisco, Mexico, Michoacan, Morelos, Nuevo Leon, Oaxaca, Puebla, Sinaloa, Sonora, Veracruz, and Yucatan.. 12 DOI:10.6814/NCCU202100271.

(20) terview. Here, the variable worked last year does not denote labor force participation since respondents who said they did not work in the past 12 months are not necessarily looking for work. From the total sample, about 51% of the individuals living in states with below-average crime rates have worked during the 12 months before the interview. Likewise, 50% of the respondents living in states with crime rates higher than the mean have been working during the year prior to the survey. The average number of weeks worked during the last year is 24 for individuals in low-crime states and 25 weeks for individuals in high-crime states. The average number of hours worked per week in low-crime states is 22, while in high-crime states is 23. I include individual, household, and state-level controls in the model. Individuallevel controls are age, gender, marital status, years of education, whether or not the respondent is the head of the household, whether or not the head of the household head is female, and whether or not the individual has relatives in the United States. Table 1 suggests that individuals living in states with a higher total crime rate are, on average, more educated, more likely to have relatives in the U.S. and, less likely to be informal workers (Velásquez, 2020). This pattern emphasizes the importance of using longitudinal data that evidences the diverse characteristics of individuals living across different locations. As a macroeconomic independent variable, I use the annual GDP per capita at a state-level to control for any possible effects of the 2007-2008 financial crises on the labor outcomes. Data suggests that states with higher GDP are also states with higher total crime rates.. 立. 政治大. ‧. ‧ 國. 學. sit. y. Nat. n. al. er. io. Additionally, the individual’s perception of violence is registered in a series of questions in the MxFLS. I use as central variables whether or not the respondent feels fear of being assaulted, whether or not they feel fear of being assaulted at night, and whether or not they feel more unsafe going out at night than five years ago. Table 1 shows that individuals’ fear dramatically increased between the second and third waves of the MxFLS in specific states, supporting the notion that the increases in violent crime were unanticipated and unrelated to previous trends.. 5. Ch. engchi. i Un. v. Empirical Strategy. This study aims to understand individuals’ labor responses due to exponential increases in violent crime caused by the Mexican drug war. By using a longitudinal survey comparing the same respondent over periods of low and high violence intensity, I adopt an individual fixed-effects model as the main empirical strategy and include temporal and geographical crime rate variations between 2002 and 2012. The underlying intuition for using an individual fixed-effects strategy is that respondents can act as their. 13 DOI:10.6814/NCCU202100271.

(21) own counterfactual by comparing their labor decisions during a low-violence period to their labor decisions while experiencing high violence levels. Hence, the premise is that each respondent’s labor behavior before the Mexican drug war implementation is a fair counterfactual for what their behavior would have been after 2007 in the absence of an increase in violent crime. Moreover, identifying the impact of violent crime on labor market outcomes is not straightforward since confounding factors can interfere in a simple analysis of the relationship. For instance, increases in violent crime rates usually coexist with other events that also affect the local economy and the labor market behavior; these events may alter an individual’s labor decisions even in the absence of any actual causal effect of violent crime on labor choices. Furthermore, an important factor to consider is the omitted variable bias that may arise from not considering the effect of unobserved, heterogeneous characteristics of states and individuals on labor decisions and their correlation with the violence trend. These unobserved characteristics, such as the economic performance of a state (and other confounding factors), may directly affect an individuals’ labor decisions. At the same time, a states’ violence trend may be dependent on its economic performance. Such omitted variable bias is an inconvenience since it can lead to underestimating or overestimating the impact of violent crime on labor decisions.. 立. 政治大. ‧. ‧ 國. 學. n. al. er. io. sit. y. Nat. To correctly identify the effects of violent crime on labor outcomes, I use a fixedeffects model as an empirical specification. This strategy ensures cleaner findings by eliminating unobservable state and individual’s time-invariant characteristics that may create endogeneity. I estimate the individual fixed-effects strategy in the following regression:. Ch. engchi. yi jt = βV jt + γ 0 Xi jt. i Un. v. + δ 0 Z jt + θt + λ j + µi + εi jt (1). Individuals are indexed i = 1, . . . , N. They are observed once per period t = 1, ..., Ti in state j = 1, ..., J. yi jt is the outcome of interest, and it refers to the labor market behavior of individuals who already were of working-age (15-80) during the MxFLS1, expressed as a binary variable which indicates whether or not the respondent worked during the 12 months before the interview. The other two outcomes refer to the number of weeks worked by an individual over the last 12 months, and the average hours worked during a week over the last 12 months. The violence variable is represented by V jt which denotes the logarithm of the total violent crime rate at a state-level 12 months before the MxFLS interview date. Xit is a vector of individual time-varying characteristics that includes marital status, years of education, household status, and whether or not the respondent has relatives in the 14 DOI:10.6814/NCCU202100271.

(22) United States. Z jt is a vector of state time-varying characteristics conformed by the annual GDP at a state level. θt captures the year fixed-effects, λ j captures the state fixed-effects, and µi captures the individual fixed-effects. If the time, state or individual fixed-effects are correlated with the violence covariates V jt , then omitting θt , λ j or µi will lead to a biased estimate of the violence effects, β . The proposed fixed-effects model allows to eliminate bias from unobservables that change over time but are constant in states and individuals (θt ), and it controls for factors that differ across states and individuals but are constant over time (λ j and µi , respectively). The year fixed-effects θt captures unobserved temporal situations, such as economic events, that occurred during some but not all of the studied years; these unobservables that change over time would affect all individuals in all states. The state fixed-effects λ j captures the unobserved factors in a given state that affect its residents’ labor decisions; these factors are time-invariant during the studied years. The individual fixed-effects µi captures unobserved inherent characteristics of an individual that influence their labor decisions; these do not change over time and also do not change with the state of residence.. 立. 政治大. ‧ 國. 學. ‧. The abrupt change in the violence trend after the Mexican drug war across states is crucial to establishing a relationship between violent crime and labor decisions. My identifying assumption proposes that in the absence of the Mexican drug war strategy, each individual’s labor decisions would be constant over time (common trends assumption). Therefore, I assume the crime rate trend to be random and not correlated to any unobserved events; following the fixed-effects assumptions, I consider extreme exogeneity so that the idiosyncratic error εi jt is not correlated with changes in the crime rate or with any of the explanatory variables.. n. er. io. sit. y. Nat. al. Ch. engchi. i Un. v. The fixed-effects model appears to be a useful tool if we think that there are unobserved variables such as individuals’ inherent abilities or the states’ characteristics that affect all its residents, and unobserved variables such as economic events on the same date as the MxFLS year. After controlling for individual, state, and year fixed-effects, and observable time-varying individual and state characteristics, I can accurately measure violent crime’s causal effect on labor decisions. Finally, a factor to consider when estimating the effects of violence on labor outcomes is the possible systematic responses to violent crime. Given the increase in violence in certain states, it is reasonable to think that some individuals will choose to migrate to less dangerous regions. In this sense, non-random migration can be a potential problem since it would affect the estimation of violent crime’s effect on labor decisions. To deal with potential selective migration, I use an intent-to-treat approach.12 12 It is difficult to assume that selective migration can be completely ruled out from the analysis.. Even if violence escalation affects migration decisions, destination choices may be related to unobserved factors.. 15 DOI:10.6814/NCCU202100271.

(23) For the cases where respondents migrate after the violence escalation in 2007, I assigned the total violence rate based on the state of residence in the first wave of the MxFLS rather than the current state. By fixing respondents to their location before the violence increase, the correlation between increases in the crime rate and migration should not impact the respondent’s assigned total crime rate.. 6. Results. In this section, I first estimate the effects of violent crime on labor decisions using an individual fixed-effects model and differentiating by gender and formality status using the same strategy. After that, I repeat the analysis by age group. Following this, I explore the potential mechanisms triggering the causal effect of violent crime on labor market outcomes. Finally, I present a leading-values test to support my empirical design.. 立. Main Results. ‧ 國. 學. 6.1. 政治大. ‧. Table 2 displays the results of analyzing the impact of the total crime rate on labor market outcomes.13 Panel A includes as a dependent variable the participation of the respondent on the labor market for the 12 months before the interview, Panel B displays the number of weeks worked for the 12 months before the interview, and Panel C shows the average number of hours worked per week for the 12 months before the interview.. n. er. io. sit. y. Nat. al. i Un. v. To illustrate the relevance of controlling for individual and regional heterogeneity, I first present the results of a model without fixed-effects. Column (1) contains a simple OLS estimates, where the estimate treats λ j and µi are part of the error term. Column (2) contains both time and state fixed-effects, and Column (3) includes the full set of state, time, and individual fixed-effects. Estimates for the model with the full set of fixed-effects and socio-demographic controls are shown in Column (4).. Ch. engchi. The OLS estimates in Column (1) suggest an insignificant effect of the total crime rate on the labor market outcomes for the three panels. Since the strict exogeneity assumption for the OLS estimator is different from that on the fixed-effects estimator, these findings are unsurprising. The model probably fails to account for individual and regional heterogeneity, inaccurately estimating the impact of the total crime rate on laIf this is the case, it will lead to assigning an improper total violence rate to individuals that migrated; this situation could bias my analysis if I do not use an intent-to-treat approach. 13 Estimates for the impact of violent crime on labor market outcomes disaggregated by crime type are shown in Table 3. Overall, results from the effects of homicide, robbery, and aggravated assault rates separately do not show significant impacts on labor outcomes.. 16 DOI:10.6814/NCCU202100271.

(24) bor market outcomes. Columns (2) to (4) include fixed-effects estimations in stages, first adding state and time fixed-effects and then individual fixed-effects. Columns (3) and (4) that include individual fixed-effects share similar results. Both models indicate a negative impact of the total crime rate on the labor market outcomes. However, the intensity of the effects is slightly higher in Column (3), a model without the socioeconomic controls. Column (4) shows the estimates for Equation 1. Results from Panel A indicate that the decision to participate in the labor market is negatively affected by increases in the total crime rate. From 2000 until the drug war implementation in December 2006, there was a reduction in the total crime rate of 14%. After 2006, however, Mexico experienced an increase in the total crime rate of approximately 90%. Column (4) estimates suggest that an individual experiencing the drug war violence would decrease their probability of joining the labor market by 2.25 percentage points, contrary to the increase of 0.35 percentage points in their likelihood of being employed before the drug war implementation. Similarly, Panel C shows a negative impact of violence on the average hours worked per week. Considering the actual increase of 90% in the total crime rate, individuals will experience a reduction of 1.6 hours worked per week, contrary to the addition of 0.25 hours worked per week that an individual would experience before the drug war.. 立. 政治大. ‧. ‧ 國. 學. Nat. y. Gender and Formality Status. io. sit. 6.2. n. al. er. Violence may affect an individual’s decision to join the labor market through diverse mechanisms. The effects are likely to differ for women and men if violence dynamics produce gender-heterogeneous responses. For example, violent crime against women is more brutal and personal. While most homicides involving male victims are caused by guns, female deaths are further hideous – suffocation and stabbing are the most common types of female homicide (Velásquez, 2020). Thus, in addition to the general model that includes all individuals, I estimate Equation 1 separately for men and women.. Ch. engchi. i Un. v. Table 4 includes the individual fixed-effects estimates by gender and informality status to identify an individual’s level of attachment to the labor market. It is important to make this differentiation since the response to violence may be conditional on the level of attachment. Results in Columns (1) and (2) indicate the participation in the labor market of men and women, respectively. From Column (1), I find no evidence that violence leads to a change in men’s decision to participate in the labor market. This result implies that men’s level of attachment to the labor market is stronger and not easily affected by changes in the crime rate. Column (2) in Table 4 shows the results for women. Panel A suggests that exposure 17 DOI:10.6814/NCCU202100271.

(25) to increases in violence negatively affects women’s decision to participate in the labor market. Women living in a state facing a 10% rise in the total crime rate would decrease their probability of being employed by 0.46 percentage points. Considering the actual increase of 90% in the total crime rate, women will experience a reduction of 4.14 percentage points in their probability of joining the labor market. Contrastingly, before the drug war implementation, when the total crime rate decreased by 14% on average, women would increase their likelihood of being employed by 0.64 percentage points. Likewise, Panel B indicates that the number of weeks worked by women drop by 2.5 weeks worked per year when experiencing a 90% increase in violent crime, compared to the addition of 0.38 weeks worked per year that women would experience in the absence of the drug war. Supporting these results, Panel C shows that a 90% rise in the total crime rate is associated with an average decline of 3.45 hours worked per week, opposite to the increase of 0.54 hours worked in the absence of the drug war.. 政治大. From the results in Columns (1) and (2), we can identify that the escalation of violence after 2006 had a distinct impact on women’s labor market outcomes. While male labor participation was not affected, the female population experienced a decrease in the likelihood of joining the labor market and a reduced number of hours and weeks worked. This discrepancy between men and women may be explained by the traditional roles assigned to each gender. For instance, the opportunity cost of leaving the labor market might be lower for women since, generally, they are not the head of the household or the primary income earner. The level of attachment to the labor market of men seems to be stronger than that of women. Therefore, I can infer that violence enlarges the labor gender gap in Mexico.. 立. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. i Un. v. Informal workers may also perceive violence differently than formal workers; consider the elevated exposure to violence and socioeconomic vulnerability associated with individuals working in the informal market. These workers are at an increased risk of being victims of violent crime due to their occupation being performed on the streets (i.e., street vendors, waste pickers). Additionally, plenty of non-formal workers in Mexico exhibit a low socioeconomic status; their risk of victimization may be further increased by a higher reliance on public transportation or the inability to hire homeprotection services (Women UN, 2019). For this reason, I develop two separate models based on the formality status of the respondents categorized by gender. The results are displayed in Table 4.. Ch. engchi. In 2003, the ILO defined the concept of informal employment as all remunerative work that is not registered, regulated, or protected by existing legal or regulatory frameworks. According to the ILO, informal workers do not have secure employment contracts, workers’ benefits, social protection, or workers’ representation. Following this concept, any respondent in the sample will be considered an informal worker if they 18 DOI:10.6814/NCCU202100271.

(26) directly report being one, if they don’t have employer-sponsored health insurance, or in the absence of an employment contract. Columns (3) and (5) in Table 4 provide the results for informal male and formal male workers, respectively. Here, we see consistent results to those in Column (1). Regardless of whether men are formal or informal workers, changes in violent crime do not affect their decision to participate in the labor market, nor affect the number of hours or weeks they choose to work. Also agreeing with earlier results, estimates in Column (4) suggest that changes in violence have a negative and significant effect on informal female workers’ labor participation decisions. Considering the actual increase of 90% in the crime rate, informal female workers will experience a reduction of 9.81 percentage points in their probability of joining the labor market, compared to the increase of 1.52 percentage points in their likelihood of being employed in the absence of the drug war. Overall, the outcomes of Columns (3) to (5) in Table 4 highlight the heterogeneity in the labor responses to violence by gender and formality status. Noticeably, women in general, and those specifically working in the informal economy, are the most adversely impacted by increased violent crime. While men working in the non-formal sector do not show any effects on their labor behavior under increased violence levels, informally-employed women experienced a non-irrelevant decrease in their probability to stay employed. Furthermore, women working in the informal market are negatively affected by a larger magnitude than women working in the formal sector. Two main reasons can explain this difference in the outcomes between formal and informal female workers. First, women working in the informal economy tend to have more flexibility to adjust their work schedules than formal workers; this facilitates changes in their labor decisions in response to safety concerns. Second, informal workers are more exposed to crime. Many women working in the informal market have less access to prevention and deterrence tools and may work in areas more frequently attacked by criminals. For example, street vendors are more likely to be affected by violent crime while they are working than office workers. Though I do not know the exact channels through which violent crime affects labor decisions, the results suggest that increases in violent crime have significant and negative effects on women’s labor market behavior, particularly among the informal workers.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. 6.3. Ch. engchi. i Un. v. Age Groups. Table 5 provides estimates for Equation 1, differentiating by age group. Column (1) shows the results for younger individuals (15 to 35 years old) and indicates that violent crime does not affect their decision to join the labor market or the hours and weeks they choose to work. When analyzing the findings for older age groups in Columns. 19 DOI:10.6814/NCCU202100271.

(27) (2)-(5), however, changes in violence appear to have a significant and negative effect on the labor market behavior. For instance, individuals between 35 and 50 years old experienced a decrease in their probability of joining the labor market by 0.49 percentage points when the total crime rate increases by 10%. Considering that the average increase in the violent crime rate between 2002 and 2013 was approximately 90%, the results suggest that individuals between 35 and 50 years old experiencing the drug war violence surge would decrease their probability of joining the labor market by 4.41 percentage points. Furthermore, the 50-65 age group shows a decline of 0.97 percentage points in the probability of joining the labor market when the total crime rate increases by 10% and, a reduction of 8.7 percentage points in the probability of joining the labor market considering the actual 90% increase in the total crime rate. The oldest age group appears to be the most negatively affected by changes in violent crime. Again, under a 10% increase in the total crime rate, individuals between 65 and 80 years old reduce their likelihood of participating in the labor market by 1.06 percentage points. When experiencing an increase of 90% in the total crime rate, older individuals decrease their probability of joining the labor market by 9.5 percentage points.. 立. 政治大. ‧. ‧ 國. 學. These discrepancies between young and old adults confirm that age is an important factor in the perception of violent crime. As the literature suggests, older individuals are more afraid of becoming victims of crime than younger individuals (Weinrath and Gartrell, 1996; Lamnek, 1992; Salem and Lewis, 2016). Namely, older people who feel more fear show a negative impact of violent crime on their probability of joining the labor market. Contrastingly, younger people who feel less fear show no effects of violent crime on their probability of joining the labor market.. er. io. sit. y. Nat. n. al. 6.4. i n C U hengchi Potential Mechanisms. v. The fear of being a victim of violent crime may be a potential mechanism modifying individuals’ labor behavior. To analyze if the fear of victimization is a mechanism driving the negative impact of violent crime on individuals’ labor decisions, I estimate Equation 1 differentiating by gender and self-reported indicators of fear included in the MxFLS. Results are displayed in Table 6. Columns (1) to (12) show the estimates for women and men for all three violence indicators, sequentially. Columns (1) and (2) display the results for individuals who reported feeling fear of being assaulted. Columns (3) and (4) display the results for individuals who reported feeling fear of being assaulted during the night. Columns (5) and (6) show the results for individuals who reported going out less than five years ago. Lastly, columns (7) to (12) show the victimization estimates for individuals reporting not feeling fear. Results from Panel A of Table 6 suggest that women feeling fear of victimization 20 DOI:10.6814/NCCU202100271.

(28) suffer a significant and negative effect on their willingness to participate in the labor market due to violent crime.14 However, estimates show that men do not experience a significant effect of violence on their labor decisions, whether or not they reported feeling fear of victimization. These results suggest that one possible channel through which violent crime affects women’s labor decisions is the fear of being a victim of violent crime.. 6.5. Leading-values Test. Given that my empirical strategy uses a fixed-effects approach, the most severe threat to assert causal identification is the existence of some unobserved state or individuallevel trends correlated with the increases in violent crime rates. The influence of these unobserved factors would be captured by the coefficients of the independent variables, distorting the estimation of the actual impact of violent crime on labor outcomes. While I assume that this scenario is improbable given the abruptness of the increase in violence and its policy origins, I believe it is relevant to conduct a proper test to support this claim.. 立. 政治大. ‧ 國. 學. ‧. To examine if the effects estimated by Equation 1 are biased due to unobserved state or individual-level factors correlated with the increases in violent crime rates, I conduct a leading-values test, similar to what Velásquez (2020) does in her work. Namely, I test if the labor outcomes of earlier periods are correlated with the high levels of violent crime of later periods; in the absence of unobserved trends, we should see nonsignificant effects of violent crime on labor outcomes.. n. er. io. sit. y. Nat. al. i Un. v. Given that the increases in violence started in early 2007, I estimate Equation 1 limiting my observations of labor outcomes to the MxFLS1 (2002) and MxFLS2 (2005); that is, the two waves before the sudden increase in violent crime. Then, I assigned them the state-level total crime rate of the subsequent waves, MxFLS2 (2005) and MxFLS3 (2009), respectively; this includes a period of low violence and another one of high violence. Table 5 shows the results of this analysis. Assuming that unobserved trends did not cause the violence increase, the change in future crime rate between MxFLS2 and MxFLS3 should not predict decreases in the labor market participation between 2002 and 2005.. Ch. engchi. Column (1) in Table 7 shows the general results of the leading-values test. Note that the relationship between the total crime rate and the probability to join the labor market in Panel A is positive but insignificant and 32% smaller than the impact found in the 14 The effect of the total crime rate on labor outcomes (Columns 7 and 9) is still negative and significant. for women who do not report fear of being assaulted during the day or night; however, the magnitude is smaller.. 21 DOI:10.6814/NCCU202100271.

(29) main specification. Similarly, Panel B and Panel C show non-significant effects of violent crime on the number of weeks or hours worked. Column (2) shows the test results for women. Contrary to the findings in Colum (2) of Table 4, the leading-values test outcomes show insignificant effects of violence on female labor decisions; moreover, the test shows an estimate 15% smaller than that found in Table 4. Accordingly, results for weeks and hours worked by women are also insignificant. Overall, the estimates of the leading-values test are insignificant at the 10% level. These results confirm that the outcomes in Table 3 and Table 4 are not caused by unobserved trends correlated with the increase in violence.. 7. Conclusion. 政治大. This paper intends to examine the impact of the unexpected escalation in violence caused by the Mexican drug war on individuals’ labor decisions. Empirically, I exploit the longitudinal nature of the MxFLS survey and its precise timing encircling the implementation of the Mexican drug war combining it with state-level data on violent crime. This strategy allows me to capture the respondents’ exposure to violent crime before and after the violence onset. By using the MxFLS data combined with an individual fixed-effects strategy, I take into account potentially endogenous state’s and individual’s time-invariant unobserved characteristics that may be correlated with violence exposure and labor decisions.. 立. ‧. ‧ 國. 學. sit. y. Nat. n. al. er. io. My results suggest that violent crime has a powerful impact on the labor market outcomes of women. I find that exposure to the violence caused by the Mexican drug war significantly decreases women’s probability to join the labor market and reduces the number of weeks and hours they work. The magnitude of this impact is even greater for women working in the informal economy. Specifically, women experiencing the drug war violence surge would decrease their probability to participate in the labor market by 4.14 percentage points; and women working in the informal economy would reduce their likelihood of being employed by 9.81 percentage points. However, for men, findings suggest that increases in violent crime do not have a significant effect on their labor decisions. Furthermore, young individuals between 15 and 34 are not affected by increases in violence, while 35 and older respondents experienced a significant and negative effect on the labor market behavior due to increases in the total crime rate. These findings underline how the costs of violence are unequally distributed between women and men, and the old and young.. Ch. engchi. i Un. v. This study adds to the growing literature on the labor market consequences of violent crime, further providing evidence of violence’s adverse economic shocks. Though. 22 DOI:10.6814/NCCU202100271.

(30) many spillover effects of violent crime on the labor market have been identified before, this study shows that the fear of victimization leads certain individuals living in a violent environment to decrease their labor participation and total time worked. For the Mexican case, increases in violent crime triggered by the Mexican drug war not only affected the lives of DTO members, the army, and police but also disproportionally harmed particular sectors of the Mexican population. Specifically, understanding how violent crime asymmetrically affects women employed in the formal and informal economy should be relevant to governments for better resource planning, more informed policymaking, and ultimately improving the labor outcomes of a large proportion of the Mexican workforce.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i Un. v. 23 DOI:10.6814/NCCU202100271.

(31) 8. Bibliography. Alvarez, J. and Ruane, C. (2019). Informality and aggregate productivity: The case of Mexico. IMF Working Paper No. 257. Barron, M. A. (2020). The rural labor gap in Mexico: An invisible crack of unemployment. Economia UNAM, 50(9), 182–202. Basu, S. and Pearlman, S. (2018). Violence and labor market activity: Evidence from Mexico’s drug war. Unpublished. Becker, G. S. and Rubinstein, Y. (2011). Fear and the response to terrorism: An economic analysis. CEP Discussion Paper No. 1079. Benyishay, A. and Pearlman, S. (2013). Homicide and work: The impact of Mexico’s drug war on labor market participation. SSRN Electronic Journal.. 政治大. Bonnet, F., Vanek, J., and Chen, M. (2019). Women and men in the Informal economy: A statistical brief. Manchester: WIEGO.. 立. ‧ 國. 學. Bozzoli, C., Tilman B., and Nina W. (2013). Self-employment and conflict in Colombia. Journal of Conflict Resolution 57(1), 117-142.. ‧. Brown, R. and Velásquez, A. (2017). The effect of violent crime on the human capital accumulation of young adults. Journal of Development Economics 127(7), 1-12.. Nat. y. sit. n. al. er. io. Calderón, G., Robles, G., Diaz-Cayeros, A., and Magaloni, B. (2015). The beheading of criminal organizations and the dynamics of violence in Mexico. Journal of Conflict Resolution 59(8), 1455–1485.. i Un. v. Collier, P. and Duponchel M. (2013). The economic legacy of civil war: firm-level evidence from Sierra Leone. Journal of Conflict Resolution 57(1), 65-88.. Ch. engchi. Cunningham, W. (2001). Breadwinner or caregiver? How household role affects labor choices in Mexico. The World Bank. Dell, M. (2015). Trafficking Networks and the Mexican Drug War. American Economic Review 105(6), 1738–1779. Douglas, J. E., Burgess A. W., Burgess, A. G., and Ressler R. K. (2013). Crime classification manual: A standard system for investigating and classifying violent crime. John Wiley & Sons. Fernandez, M., Ibanez, A. M., and Peña, X. (2014). Adjusting the labor supply to mitigate violent shocks: evidence from rural Colombia. Journal of Development Studies 50(8), 1135–1155.. 24 DOI:10.6814/NCCU202100271.