Sage Journals: Discover world-class research

Abstract

Natural disasters can affect the behavior of criminal organizations. This article analyses the effect of the 2017 earthquakes that hit Mexico on the incidence of crimes related to organized crime. In particular, the outcomes of interest are rates for homicides, extortion, drug crime, and kidnapping. Using a difference-in-differences and an event-study methodologies, the findings indicate an increase in kidnapping rates of 4.0% after the earthquakes in affected areas. Then, we explore what type of criminal organizations modifies its behavior after earthquakes. We find a shift in organizational behavior (e.g. violence and altruism) of local criminal organizations, but no effect on large criminal organizations. In all, earthquake events in Mexico increase kidnapping rates through the organizational behavior of local criminal organizations. The significance of this research stems from its methodological approach, which enables the inference of causality by leveraging geographical variation in earthquake strikes.

Keywords

Earthquakes natural disasters organized crime social altruism

Introduction

There is no consensus on the impact of disasters on crime. In Houston, Texas, burglaries increased during Hurricane Harvey in 2017 (Augusto, 2021). In contrast, Louisiana experienced a significant decrease in crime in areas affected by Hurricane Katrina (Leitner et al., 2011). Following the 2010 Haiti earthquake, physical and sexual assaults increased significantly compared to the summer of 2009 in Port-au-Prince (Kolbe et al., 2010). Conversely, assault, burglary, and arson decreased in Christchurch, New Zealand, following the Canterbury earthquakes (Breetzke et al., 2018). A limitation of quantitative studies analyzing the effect of disasters on crime is that the literature focuses on personal and property crimes. Yet, there remains a knowledge gap for the effect of natural disasters on crimes related to organized crime.

Qualitative evidence suggests that natural disasters affect organized crime activities. Criminal organizations sent emergency supplies after the Kobe earthquake in Japan, reducing criminal activity (Fukumi, 2010). Yet, the proliferation of gangs in Haiti during the 2010 earthquake may explain the increase in sexual crimes (Niño and González, 2022). These activities of organized crime that replace the state are conceptualized as criminal governance (Arias, 2017; Barnes, 2017; Lessing, 2021). However, the relationship between disasters and criminal governance remains unexplored in the broader literature (Mantilla and Feldmann, 2021).

This article examines the effects of the 2017 earthquakes that hit Mexico on crimes related to organized crime. We use four data sources for this study: the first database corresponds to earthquake intensity at each municipality as recorded by the National Center for Disaster Prevention (CENAPRED, 2017); the second source is incidence data from the National Public Safety System (NPSS), which contains crime reports by municipalities; the third source is a machine learning algorithm containing organized criminal violence event data (Osorio and Beltrán, 2019); and the fourth source is a database that maps criminal organizations and altruistic behavior (Sobrino, 2021).

Using a difference-in-differences and an event study methodologies, the results show an increase on kidnapping rates by 4%. Other outcomes analyzed include rates for homicides, extortion, and drug crime, which were not affected by the 2017 earthquakes in Mexico. We then explore what type of criminal organizations changes its organizational behavior (e.g. violence and altruism). The results show a shift in organizational behavior of local criminal organizations, but no effect on organizational behavior of large criminal organizations.

This article contributes to the natural disasters and crime literature by examining earthquakes’ causal effects on organized crime. First, this article infers causality of disasters on organized crime by showing an increase in kidnapping rates in treated areas. Second, this article contributes to the literature that examines the effects of disasters on violence exertion by criminal organizations. Qualitative evidence points to an increase in gang activities after a disaster (Cromwell et al., 1995). We confirm increased violence committed by organized crime after a disaster, with local criminal enterprises mainly driving the spur in violence.

The significance of this research stems from its methodological approach, which enables the inference of causality by leveraging geographical variation in earthquake strikes. Earthquakes are exogenous events that can modify the behavior of criminal enterprises in different geographical locations. As such, scholars and policymakers interested on the effects of climate on crime must build upon the empirical evidence of these events that have clear heterogeneous geographical impacts.

Literature review

The understanding of disaster risk-reduction procedures has continued to expand over time. Initially, government responses to disasters were reactive, focusing on preparedness, planning, and management (Quarantelli, 1988). However, the study of disasters has evolved, shifting from a reactive stance to one emphasizing disaster risk reduction (Oliver-Smith et al., 2017; Shaw et al., 2013). This new stance has focused on vulnerability and resilience factors to minimize the impact of disasters (Shaw et al., 2013; UNISDR, 2009).

Criminology primarily utilizes three theories to explain the effects of earthquakes on crime incidence (Frailing and Harper, 2017; Prelog, 2016). The routine activity theory predicts increased criminal activity because authorities focus on rescue and relief efforts, leaving fewer resources for crime prevention (Cohen and Felson, 1979). The social disorganization theory attributes the rise in crime to changes in factors such as poverty, family disruption, and unemployment that weaken social control mechanisms in affected areas (Shaw and McKay, 1942). Finally, the therapeutic community theory suggests that natural disasters decrease crime due to increased social cohesion within communities experiencing shared human and material losses (Fritz, 1996).

Criminal governance can be another theory to understand the effects of natural disasters on the behavior of organized crime. The activities of organized crime that compete with or replace the state in providing public goods or regulating violence are conceptualized as criminal governance (Arias, 2017; Barnes, 2017; Lessing, 2021; Mantilla and Feldmann, 2021). Recent studies have analyzed the organization of crime in cities (Barnes, 2022; Lessing and Willis, 2019; Magaloni et al., 2020a, 2020b; Trejo and Ley, 2018). However, the relationship between disasters and criminal governance remains understudied (Mantilla and Feldmann, 2021).

Empirical evidence shows mixed effects of disaster on crime. As mentioned above, some studies suggest an increase in assaults (Kolbe et al., 2010), while others, a drop in theft, home burglary, larceny, and robbery (Breetzke et al., 2018; García-Hombrados, 2020; Yates and Mackenzie, 2018). A limitation of the existing empirical studies on earthquakes and crime is their primary focus on property and personal crimes, often neglecting crimes associated with criminal organizations, such as homicides, extortion, drug trafficking, and kidnapping.

To the best of our knowledge, there is no theory that relates disasters and crimes committed by criminal organizations. However, there are efforts in the literature to understand the relationship between unexpected shocks and organized crime. If a shock disrupts the regular financing activities of local criminal groups, then these organizations can look for alternative forms of financing. Specifically, in the case of kidnapping, a crime often associated with organized crime (O’Brien, 2012), it may be employed to stabilize the finances of criminal organizations when facing disruptions in their operational structures (Jones, 2013; O’Brien, 2012).

Background

The September 2017 earthquakes in Mexico

Mexico’s location on multiple tectonic plates makes it prone to earthquakes. In September 2017, two earthquakes struck central and southeastern Mexico within a time lapse of 2 weeks between each of the earthquakes. These earthquakes impacted 689 of over 2457 municipalities (see Figure 1). Namely, the earthquakes hit the states of Oaxaca, Chiapas, Veracruz, Mexico City, Morelos, Puebla, Estado de Mexico, Tlaxcala, and Guerrero. The first earthquake occurred on September 7 with a magnitude of 8.2 on the Richter scale and a depth of 45.9 km in Chiapas (lower blue region in Figure 1). This earthquake is considered the most powerful in 100 years. The second earthquake occurred on September 19 with a magnitude of 7.1 on the Richter scale and a depth of 57 km in Morelos (upper blue region in Figure 1). The earthquakes damaged 182,797 homes, caused the deaths of 468 people, and resulted in 731,188 people being affected across 10 states (GOBMEX, 2019). The damages amounted to over 4 billion USD or 0.5% of the country’s GDP (CENAPRED, 2017).

Figure 1.

Municipalities impacted by the 2017 earthquakes and intensity recorded: Modified Mercalli scale.

During the first of the earthquakes, damage assessment of buildings—mainly houses—shows that poor construction quality and design were the main reasons for the damage observed in masonry houses and apartment buildings (Godínez-Domínguez et al., 2021). This earthquake caused the damage of more than 110,000 houses, with over 40,000 of these being total losses (Godínez-Domínguez et al., 2021). A similar conclusion is drawn for the second of the earthquakes that hit several buildings in Mexico City (Tena-Colunga et al., 2021). For both of these assessments, there is a high correlation between the local acceleration of the earthquakes and the damage to the buildings (Godínez-Domínguez et al., 2021; Tena-Colunga et al., 2021). This allows us to approximate damage with the recorded intensity of the earthquakes through the recorded Modified Mercalli scale intensity in each municipality. This approximation remains imperfect as higher seismic activity and very soft soil deposits “greatly amplify the intensity of ground motions” (Galvis et al., 2017). Still, we rest assure that both earthquakes had a significant effect in the lives of local residents, as reflected in the coverage of local news stories, which peaked right after the earthquakes and then dropped by half every 8 days (Curiel et al., 2019).

During and after the earthquakes, there were anecdotal reports indicating that some criminal organizations changed their behavior in response to the emergency situation. In several areas, rival gangs temporarily established truces to avoid conflicts during relief efforts. In addition, young cartel members, who are usually tasked with transporting drugs on motorcycles, were redirected to deliver supplies to disaster-stricken regions. These groups reportedly purchased water and energy drinks for rescue workers, especially in areas like the center of Mexico City, where volunteers and emergency responders were actively engaged. This behavior shows how local criminal organizations adapted to the post-disaster context, using it to gain favor with communities and expand their influence while the state’s focus was on relief operations (Balderas, 2019).

Organized crime in Mexico

During 2000–2006, Mexico maintained relatively low levels of violence (around 10 homicides per 100,000 inhabitants). However, for the 2007–2012 period, homicide rates increased drastically to 22 homicides per 100,000 inhabitants (Brown and Velasquez, 2017). For the 2013–2015 period, the rate decreased to 17 per 100,000 inhabitants. However, in 2017, the number of homicides reached its highest level. The reasons for the increase in homicide rates and violence, more generally, can be grouped into the following: (1) a stricter internal policy in the fight against drugs, (2) heightened conflict between cartels, (3) crime diversification among criminal organization groups, (4) external factors, and (5) socioeconomic aspects.

First, President Calderon’s strategy (2006–2012) to capture leaders of a drug trafficking organization increased violence (Castillo et al., 2020; Lindo and Padilla-Romo, 2018; Massa Roldan et al., 2021). However, these captures only account for 31.5% of the increase in the homicide rate during the period 2006–2010 (Lindo and Padilla-Romo, 2018). Second, either due to opportunistic behavior to further weaken rival criminal groups already debilitated by the government or due to long-standing violent rivalries, evidence shows an increase in homicides linked to conflicts between rival drug trafficking organizations (Brown and Velasquez, 2017; Corcoran, 2013; Dell, 2015; Herrera and Martinez-Alvarez, 2022; Osorio, 2015). The harsher governmental policies and the heightened violence between drug trafficking organizations led large criminal groups to split and diversify into other criminal activities (Balmori de la Miyar, 2016; Bunker, 2013; Durán-Martínez, 2017; Shirk and Wallman, 2015), such as kidnapping, oil theft, extortion, and human trafficking, to recover economic revenues lost due to the war on drugs (Atuesta, 2020; Jones, 2013; Shirk and Wallman, 2015). Fourth, other factors that accelerated the homicide rates in Mexico include external shocks such as a reduction in the supply of cocaine from Colombia (Mexico’s main supplier), which explains between 10% and 14% of the increase in violence in Mexico (Castillo et al., 2014). Finally, socioeconomic factors, such as an increase of one point in the Gini coefficient (indicating greater inequality), correspond to an increase of approximately 6 homicides per 100,000 inhabitants from 2007 to 2010 (Enamorado et al., 2016). Indeed, these factors have sparked a profound change within Mexico’s criminal organizations and, consequently, in state governance.

Drug trafficking and crime diversification may motivate criminal organizations to exert territorial control in Mexico (Barnes, 2022). Territorial control facilitates the transportation of drugs (Dell, 2015), and the oversight of other local markets (Reitano and Shaw, 2018). Importantly, for this study, an earthquake may provide criminal groups opportunities to assert or increase control over a region, as police and security forces may be preoccupied with addressing the damages caused by the earthquake.

It is within this context of violence that the 2017 September earthquakes took place. Given the potential effects on both altruistic behavior and an increase in opportunistic crime, we now employ novel data within a causal framework to disentangle these effects.

Data and empirical strategy

Data

We use four data sources for this study. The first source is the intensity of each earthquake at each municipality, which comes from the National Center for Disaster Prevention (CENAPRED, 2017). The second database contains municipal criminal incidence records and comes from the NPSS. The third source is a machine learning algorithm containing organized criminal violence event data (Osorio and Beltrán, 2019). Finally, the fourth source is also a machine learning database that measures criminal organization activities, including altruistic activities, in Mexico (Sobrino, 2021).

The National Center for Disaster Prevention (CENAPRED, 2017) provides information regarding the municipalities impacted by earthquakes. There are 2457 municipalities in Mexico, and the earthquakes impacted 689 municipalities. The municipalities impacted by the earthquakes are the treatment group, and the rest are the control group. These data also contain the earthquake intensity by municipality, using the Modified Mercalli scale for each of the earthquakes. The lower blue region in Figure 1 contains the intensity for the 7 September earthquake, while the upper blue region in Figure 1 is for the 19 September earthquake. The darker color of a municipality corresponds to a higher earthquake local intensity. Showily, there is no geographical overlap between the treatment of the earthquakes.

Data from the NPSS is used to obtain information regarding crimes related to organized crime. The NPSS collects monthly data for crimes reported to the police at the municipality level. We analyze the following crimes related to organized crime: homicides, extortion, petty drug crime, and kidnapping. We use rates per 100,000 inhabitants for each month per municipality from January 2017 to September 2018 (8 months before and 12 months after the earthquake). The NPSS data covers the period from January 2015 to December 2022. Due to incomplete data before 2017, especially for Oaxaca, which was heavily affected by the 2017 earthquakes, we limited the pre-earthquake analysis to 8 months before September 2017. The post-earthquake analysis focuses on the year following the earthquakes to minimize the impact of other factors on our variables of interest. Our final sample comprises 51,597 observations (2457 municipalities × 21 months).

The Organized Criminal Violence Event (OCVED) database records daily data at the municipal level. These data contain detailed information on violent incidents such as homicides, kidnapping, extortion, torture, and other crimes committed by members of criminal organizations. The data include information about nine large criminal cartels (Beltran Leyva, Cartel de Jalisco Nueva Generación, Cartel de Juarez, Cartel de Sinaloa, Cartel de Tijuana, Cartel del Golfo, La Barbie, La Familia Michoacana, and Los Zetas), and 71 local criminal organizations. The classification of large and local cartels comes directly from the dataset, and the nine large criminal groups are the same considered by the United States Drug Enforcement Administration (DEA) as major criminal organizations active in Mexico.

OCVED data utilizes 105 sources of information, including national and local newspapers and government agencies, and machine learning algorithms to classify news related to violence committed by criminal organizations from January 2000 to December 2018. These categories are distinct and do not overlap, as each news article is assigned to a specific organization using a set of fixed dictionaries. Thus, the algorithm classifies each news article based on the organization, action, location, and time. Based on the OCVED data, we generate two dichotomous variables (large criminal organizations and local criminal organizations) that measure the incidence of violence per month per municipality from January 2017 to September 2018 (8 months before and 12 months after the earthquakes).

The Mapping Criminal Organizations (MCO) database contains information for providing services, public goods, or gifts to the population (a proxy for altruism) by criminal organizations. A panel of 79 criminal organizations was constructed from 1990 through July 2021. We separate these organizations into two groups: nine large criminal organizations (Beltran Leyva, Cartel de Jalisco Nueva Generación, Cartel de Juarez, Cartel de Sinaloa, Cartel de Tijuana, Cartel del Golfo, La Barbie, La Familia Michoacana, and Los Zetas), and the rest of criminal organizations as local criminal organizations. Using the MCO data, we generate two dichotomous variables (large criminal organization and local criminal organizations) that measure whether a municipality per year reported at least one act of altruism by a criminal organization from 2015 to 2018. Our sample consists of 9808 observations (2452 municipalities × 4 years).

Table 1 presents summary statistics using the NPSS incidence data. The panel reports the average level of crime incidence at the municipality and month levels for the period of study. For instance, on average, within a month, we observe 2.37 homicides per 100,000 inhabitants at the municipality level. To infer the causal effect of the earthquakes on homicides, extortion, drug crime, and kidnappings, we estimate a causal difference-in-differences and an event-study specification, both described below.

Table 1.

Descriptive statistics.

	Mean	SD	N
Rates per 100,000 INH
Homicides	2.375	7.645	51,597
Extortion	0.204	1.426	51,597
Drug crime	0.993	4.456	51,597
Kidnapping	0.059	0.680	51,597

Source: National Public Safety System (NPSS). The data is grouped at the municipality and month level.

Empirical strategy

Difference-in-differences

In this study, we apply a difference-in-differences (DiD) methodology and an event-study design to estimate the causal effect of the 2017 earthquakes on organized crime outcomes. The identification strategy relies on the parallel trends assumption, which requires that the trends in the outcome variables would have evolved similarly for both treatment and control municipalities in the absence of the earthquakes. While balancing tests are often used to ensure comparability between treatment and control groups, they are not a strict requirement for valid causal inference in DiD (Abadie, 2005; Bertrand et al., 2004). Instead, the primary concern is ensuring that the control group serves as a reliable counterfactual by exhibiting similar pre-treatment trends as the treated group. In addition, we account for potential heterogeneity across municipalities by including municipality-fixed effects and controlling for time-varying factors through month and year-fixed effects. This approach mitigates the influence of unobserved, time-invariant characteristics that might otherwise bias the estimates.

First, we estimate a difference-in-differences model. The difference-in-differences provides the average effect of the earthquakes on organized crime rates. The specification is given as follows

$Y_{m t y} = α + β Earthquakemty + a_{m} + Y_{t} + V_{y} + e_{m t y}$ (1)

where Y_mty is the outcome of interest for municipality m in month t and year y. Earthquakemty is a dummy variable that equals one from September 2017 through September 2018 for the municipalities impacted by the earthquakes and zero otherwise. a_m are municipality- fixed effects, γ_t are month fixed-effects, and ν_y are year fixed effects. In order to take into account the population heterogeneity at the municipal level, the specification is weighted by the population at that level. We cluster the standard errors at the municipality level. Furthermore, in a second estimation, we employ a continuous treatment variable that interacts the Earthquakemty dummy variable with the local intensity as measured by the Modified Mercalli scale.

We employ a log-transformation of the dependent variable as it allows us to interpret the results as approximations of percentage changes. In particular, we estimate the transformation log(y + 1). A log-transformation of the dependent variable permits addressing the issue of overdispersion in the distribution. In many municipalities, our dependent variables tend to report zero incidents, while certain municipalities experience large incidences at specific times, resulting in a right-tailed distribution. A log-transformation helps to alleviate this overdispersion. Nevertheless, in the robustness section, we run our results using the original dichotomous variable (a linear-linear model) to show that the results are not driven by the functional form selected.

Event study

In addition, we estimate an event-study specification to test for the lack of pre-parallel trends in our main outcomes of interest and to analyze the monthly evolution of the effects of the September 2017 earthquakes. The event-study specification is as follows

$Y_{m t y} = α + \sum_{\begin{array}{l} q = - 8, \\ q \neq - 1 \end{array}}^{12} β_{q} {Earthquake}_{mqy} + a_{m} + Y_{t} + V_{y} + e_{m t y}$ (2)

where Y_mty stand for the outcome of interest. Earthquakemqy is a binary variable that equates 1 if it is observed q months away from the September 2017 earthquakes. As such, q ranges from q = −8 (January 2017) to q = 12 (September 2018). We consider eight lags due to data restrictions, as data availability for the state of Oaxaca at the municipality level is limited before 2017. The horizon of q = 12 was selected to study the dynamic monthly effects up to 1 year after the earthquakes. We use q = −1 as our reference period (August 2017). Finally, a_m, γ_t, and ν_y stand for municipality, month, and year fixed-effects, respectively.

Our parameters of interests are the β_q, ranging from q = −8 to q = 12. To infer causality, we must satisfy the parallel-trends assumption, which requires that there be no statistically significant differences in outcome trends between the treatment and control groups before the event. Hence, before the earthquake (q = −8 to q = −2), we expect the parameters to not be statistically significant in order to satisfy the parallel-trends assumption. If this condition is met, we can then explore the dynamic monthly effects by analyzing the estimates from q = 0 to q = 12.

Results

Main findings

Table 2 contains the difference-in-differences results, as specified formally in equation (1). The main findings of this article indicate no statistically significant effects of the 2017 earthquakes on homicides and petty drug crime rates. However, the results show that extortion rates increased by 3.0% and kidnapping rates by 4.0% after the earthquakes. It should be noted that the former coefficient is only significant at the 10% level, while the latter does meet the statistically significance threshold at the 1% level.

Table 2.

Difference-in-differences: earthquakes and organized crime.

Rates per 100,000 INH	Homicides	Extortion	Drug Crime	Kidnapping
	(1)	(2)	(3)	(4)
Earthquake	–0.020	0.029*	0.065	0.039***
	(0.026)	(0.016)	(0.054)	(0.012)
R ²	0.54	0.47	0.81	0.21
Observations	51,597	51,597	51,597	51,597
Baseline FE	X	X	X	X

Source: National Public Safety System (NPSS).

Baseline fixed effects are included at the municipality, month, and year. Robust standard errors are clustered at the municipal level.

Significance levels: *p < 0.1; **p < 0.05; ***p < 0.01.

Next, we use a continuous treatment effect by iterating the earthquake dummy variable with a Modified Mercalli intensity matrix by municipality. Table 3 has three different panels. The top panel presents the main continuous treatment variable. Results show that the coefficients for extortion and kidnapping rates remain statistically significant at the 10% and 1% levels, with a very similar magnitude than the dichotomous treatment specification, once the mean intensity interacts with the estimated effects. The middle panel of Table 3 takes as treated municipalities experiencing intensities greater than four on the Modified Mercalli scale. Municipalities that did have a recorded intensity below four are not considered in this sample. Again, only the coefficients for extortion and kidnapping rates remain statistically significant and similar to the general specification. Finally, the bottom panel of Table 3 takes five as the minimum threshold of intensity to be considered as treated. All those municipalities with a recorded earthquake intensity below five are excluded from the sample. In this case, only the coefficient of kidnapping rates remains statistically significant and of a similar magnitude.

Table 3.

Difference-in-differences: earthquakes intensity and organized crime.

Rates per 100,000 INH	(1) Homicides	(2) Extortion	(3) Drug crime	(4) Kidnapping
Panel A: intensity as main treatment variable
Earthquake x Intensity	–0.004	0.004*	0.011	0.007***
	(0.004)	(0.003)	(0.009)	(0.002)
R ²	0.54	0.47	0.81	0.21
Observations	51,597	51,597	51,597	51,597
Mean intensity	6.14	6.14	6.14	6.14
Baseline FE	X	X	X	X
Panel B: intensity equal or greater to 4
Intensity >= 4	–0.021	0.029*	0.065	0.039***
	(0.026)	(0.016)	(0.054)	(0.012)
R ²	0.54	0.47	0.81	0.21
Observations	51,584	51,584	51,584	51,584
Mean Intensity	6.14	6.14	6.14	6.14
Baseline FE	X	X	X	X
Panel C: intensity equal or greater to 5
Intensity >= 5	–0.019	0.026	0.063	0.040***
	(0.026)	(0.016)	(0.055)	(0.013)
R ²	0.54	0.47	0.81	0.21
Observations	50,947	50,947	50,947	50,947
Mean Intensity	6.14	6.14	6.14	6.14
Baseline FE	X	X	X	X

Source: National Public Safety System (NPSS).

Baseline fixed effects are included at the municipality, month, and year. Panel A presents the difference-in-differences model using the iteration of the binary earthquake (treatment) variable with the Modified Mercalli intensity scale values. Panel B estimates the difference-in-differences model setting the treatment cutoff at a minimum intensity of 4; treated municipalities with an intensity lower than 4 are excluded from the sample. Panel C does the same as Panel B, at 5, based on the Modified Mercalli scale; treated municipalities with an intensity lower than 5 are excluded from the sample. Robust standard errors are clustered at the municipal level.

Significance levels: *p < 0.1; **p < 0.05; ***p < 0.01.

Finally, we estimate an event-study specification to check the assumption of parallel trends and to study the earthquakes’ dynamic effects. Figure 2 displays the results. To rule out confounding factors related to different pre-trends between the treatment and control groups before the earthquake, we expect no significant monthly trends for any particular crime. We use August 2017 (1 month before the earthquake) as our reference period. In favor of the parallel trends assumption, we do not observe pre-trends in our main outcomes of interest, extortion and kidnapping rates. Homicide rates also exhibit stable pre-trends, with the exception of t = −2. However, petty drug crimes show deviations from parallel pre-trends.

Figure 2.

Event studies: earthquakes and organized crime: (a) homicides, (b) extortion, (c) drug crime, and (d) kidnapping.

The event-study specification is also helpful in estimating the dynamic effects of the earthquake every month. The results, displayed in Figure 2, explore the dynamic impacts of the earthquake in each of the crimes being studied. The findings indicate that the earthquakes had varying effects on different types of crime. In particular, kidnapping shows a noticeable uptick post-earthquake, and there are no pre-trends. Extortion shows a slight increase post-earthquake, marked by a statistically significant positive coefficient shortly after the event, even though the standard errors are too large to conclude a positive effect. Importantly, neither homicides nor drug crime show any changes after the earthquake, providing a sense of stability in these criminal activities. In summary, extortion and kidnapping satisfy the assumption of parallel trends, but only kidnapping observes statistically significant increases.

Robustness checks

We conduct five robustness checks to test the validity of the difference-in-difference results: (1) using a linear-linear form, (2) applying a bounding methodology to test the sensibility of the results to omitted variable bias, (3) using a different control group, (4) using a placebo test, and (5) testing the sensitivity of the results to deviations of the parallel trends assumption.

In our preferred model, we employ a log-linear specification to interpret the results as a percentage change. However, there is a possibility that the results are influenced by this specification. To assess the robustness of our findings, we also utilize a linear-linear specification. As shown in the Panel A of Table 4, the conclusions remain consistent regardless of the functional form being used.

Table 4.

Robustness checks.

Rates per 100,000 INH	Homicides(1)	Extortion(2)	Drug Crime(3)	Kidnapping(4)
Panel A: linear function
Earthquake	–0.109	0.064**	0.255	0.058***
	(0.099)	(0.028)	(0.307)	(0.017)
R2	0.46	0.31	0.86	0.13
Observations	51,597	51,597	51,597	51,597
Baseline FE	X	X	X	X
Panel B: Oster’s bounds
Earthquake	–0.020	0.029*	0.065	0.039***
	[–0.073, 0.188]	[0.011, 0.088]	[0.0004, 0.223]	[0.033, 0.061]
R ²	0.41	0.37	0.73	0.58
Observations	51,597	51,597	51,597	51,597
Baseline FE	X	X	X	X
Panel C: using an alternative control group
Earthquake	–0.011	0.065***	–0.007	0.032**
	(0.029)	(0.018)	(0.066)	(0.013)
R ²	0.53	0.50	0.83	0.27
Observations	20,265	20,265	20,265	20,265
Baseline FE	X	X	X	X

Source: National Public Safety System (NPSS).

Baseline fixed effects are included at the municipality, month, and year. Robust standard errors are clustered at the municipal level. Panel A presents the difference-in-difference model using a linear model. Panel B calculates Oster’s bounds, which are presented in brackets. Panel C presents the difference-in-difference model using as a control group municipalities belonging to states on the border with the United States.

Significance levels: *p < 0.1; **p < 0.05; ***p < 0.01.

Then, we check the sensibility of our results to omitted variable bias. The event-study results support the parallel trends assumption and that the earthquakes drive our results. To confirm that other omitted variables do not bias our findings, we conduct a robustness check based on a bounding methodology (Oster, 2019). This methodology generates a bound around the parameter of interest. This bound is based on assumptions regarding the R2. The standard practice is to use 1.3 times the value of the R2 in the main specification (Oster, 2019). If the bound excludes zero, then it implies that the results are robust to the problem of omitted variable bias. Reassuringly, the bounds for kidnapping rates [0.033, 0.061], reported in Panel B of Table 4, are robust to Oster’s methodology for omitted variable bias.

Third, it is possible that some municipalities used as control are contaminated for two reasons: (a) individuals relocated after the earthquakes to nearby municipalities and (b) criminal activities displaced to neighboring municipalities (Maheshri and Mastrobuoni, 2020). Thus, we estimate the difference-in-differences model using as a control group distant municipalities that belong to states that are on the border with the United States (Baja California, Sonora, Coahuila, Chihuahua, Nuevo Leon, and Tamaulipas). Panel C of Table 4 presents the results. The main results are maintained.

In particular, there are no effects on homicides and petty drug crimes. And, using this new strategy, the coefficients associated with extortion and kidnapping continue being statistically significant.

Fourth, we present some placebo tests to rule out alternative explanations. The first placebo test assumes that the earthquakes happened in 2016. Since this did not happen, we should expect no effects on the variables tested. Figure I in the Supplemental Appendix presents the results of the Event Study using this placebo test. As expected, there is no effect on any of the four variables analyzed (homicides, extortion, drug crime, and kidnapping). The second placebo test examines the effect of earthquakes on domestic violence. Domestic violence is primarily committed by intimate partners and this crime is not associated with organized crime. Figure I in the Supplemental Appendix shows the result for this placebo. Although an increase in domestic violence is observed after earthquakes, an increasing trend in domestic violence is also observed prior to earthquakes. This suggests that the assumption of parallel trends does not hold, and therefore, earthquakes are not necessarily responsible for the observed increase in domestic violence.

Finally, following Rambachan and Roth (2023), we test the sensitivity of the results to deviations from the parallel trends assumption. Rambachan and Roth (2023) generate a series of bounds based on a coefficient M, which measures the extent of the deviation from the parallel trends assumption. For instance, if the bounds do not include zero when M = 0.05, this indicates that the results remain robust up to a 5% deviation from the worst pre-treatment deviation of the parallel trends assumption. The results are shown in Figure II in the Supplemental Appendix. Overall, the findings regarding kidnapping suggest that are robust up to a 5%, but not up to 10%. It should be noted that Rambachan and Roth (2023) make no suggestions regarding the optimal value of M. Thus, this methodology should only be taken as a sensitivity analysis of possible omitted variables affecting the parallel trends. However, the evidence provided by Oster’s methodology (robustness check 2) and the placebo tests (robustness check 4) provide evidence that the results regarding kidnapping are not driven by omitted variables.

Criminal organizational behavior

The main results suggest a clear effect of earthquakes on organized crime, particularly on kidnapping rates. However, these results do not explain how criminal organizations adapt to new circumstance after a disaster. To further explore how earthquakes affect organized crime, we explore two organizational behaviors: violence and altruism. We use OCVED data to count the number of violent events related to a particular cartel, year pair, and MCO data to count the number of altruistic-related news to a particular cartel year pair. Moreover, we examine how these behaviors vary by organization type (large vs local), highlighting the heterogeneity in adaptive strategies.

Table 5 shows the difference-in-differences results for changes in the exertion of violent and altruistic activities by the size of the criminal organization. Columns (1) and (2) explore incidences of violence by type of criminal organizations after the earth quakes. We find no increase in incidences of violence for large criminal organizations. Conversely, local criminal organizations increased the level of violence by 5.3%. Then, Columns (3) and (4) explore incidences of altruism by type of criminal organizations after the earthquakes. Large criminal organizations behave non-altruistically, whereas local criminal organizations increased their social altruism activities by 7.0%.

Table 5.

Criminal organization behavior: difference-in-differences specification.

Rates per 100,000 INH	Large CO violence	Local CO violence	Large CO altruism	Local CO altruism
Rates per 100,000 INH	(1)	(2)	(3)	(4)
Earthquake	0.022	0.053***	0.017	0.070**
	(0.015)	(0.017)	(0.011)	(0.028)
R2	0.41	0.37	0.73	0.58
Observations	51,597	51,597	9808	9808
Baseline FE	X	X	X	X

Source: Columns (1) and (2) use the Organized Criminal Violence Event (OCVED), and Columns (3) and (4) use the Mapping Criminal Organizations (MCO) data.

Baseline fixed effects are included at the municipality, month, and year in Columns (1) and (2). Baseline fixed effects are included at the municipality and year in Columns (3) and (4). Robust standard errors are clustered at the municipal level.

Significance levels: *p < 0.1; **p < 0.05; ***p < 0.01.

The results show heterogeneity in how different types of organizations adapt to post-disaster conditions. Large organization shows limited responsiveness, with no significant change in either violence or altruism. On the other hand, local organizations adapt by intensifying both violence and altruistic activities. This behavioral probably reflects the need of local groups to strengthen their control over communities while capitalizing the chaos to expand their criminal operations.

Similarly, as before, we conduct a couple of robustness checks for the organizational behavior mechanisms explored: (1) using a linear-linear form to test the sensibility of the results to a different functional form and (2) using a bounding methodology to check the sensibility of the results to omitted variable bias. Table A.1 in the Supplemental Appendix confirms that organizational behavior changes such as violence and altruism activities only increased for local criminal organizations.

In the case of Mexico, earthquakes do not affect the organizational behavior of large criminal organizations regarding violence and altruism. On the contrary, local criminal organizations change their behavior after the earthquakes. These types of criminal enterprises had to strengthen ties with local communities, but at the same time, they had to intensify their criminal activities, such as kidnapping.

Discussion

We analyze the earthquakes’ effects on violence committed by organized crime. Using a difference-in-difference methodology, the results show significant increases in extortion and kidnapping rates in the order of 3.0% and 4.0%. Nonetheless, only the latter coefficient is significant at the 1% level. Furthermore, we find no statistical effects on homicides and drug crimes.

We then explore organizational behavior through which earthquakes may impact organized crime. Again, using a difference-in-differences methodology, the results show a significant increase in violence committed by local criminal organizations but no effect on violence committed by large criminal organizations. In addition, there is a significant increase in the incidences of altruism by local criminal organizations but no effect on large criminal organizations.

Theoretically speaking, the findings in this research relate to the routine activity theory and to the social disorganization theory, which forecast increased criminal activity due to lack of security forces and socioeconomic shifts, respectively (Cohen and Felson, 1979; Shaw and McKay, 1942). However, due to data limitations, it is not possible to pinpoint what exactly is driving the increase in crime activity after earthquakes. One important conclusion that this article is able to draw is a shift in the violent and altruistic behavior exerted by local criminal groups, in comparison to large criminal organizations. This behavior relates to the criminal governance theory, which provides explanations for shifts in public goods and violence (Arias, 2017; Barnes, 2017; Lessing, 2021; Mantilla and Feldmann, 2021).

There is qualitative evidence that after a disaster, gang activity increases (Cromwell et al., 1995). We find similar results in the Mexican context, where local criminal organizations increase violence after a disaster. The findings are consistent with studies on the diversification of organized crime in Mexico (Atuesta, 2020; Bunker, 2013; Durán-Martínez, 2017; Jones, 2013; Shirk and Wallman, 2015). The economic income of local criminal organizations depends on local markets. If the earthquake disrupts economic activities in the area, the income of smaller groups is affected. Consequently, they may resort to kidnapping as a temporary means of financing while the illegal markets recover.

There is also qualitative evidence that organized crime supports communities impacted by disasters (Rankin, 2012). However, we did not find significant declines in homicides and drug dealing after the earthquake, as suggested by literature supporting the hypothesis of criminal governance (Arias, 2017; Barnes, 2017; Lessing, 2021; Mantilla and Feldmann, 2021; Trejo and Ley, 2018). This article finds that local criminal organizations behave in an altruistic form while increasing their incidence on kidnapping.

A limitation of this study is the measurement error in the earthquake’s reported damages at the municipality level. The intensity of the earthquake, as recorded by the Modified Mercalli, approximates the level of damage exerted by the earthquake, but it is only an approximation. Furthermore, the damage caused by the earthquakes also depends on the level of soft soil deposits and the municipality’s governance capacity. Moreover, a limitation of the differentiation between local groups and large criminal organizations is that local groups may be associated with larger organization. Nevertheless, to the best of our knowledge, the data used is the best measure available regarding the presence of criminal organization groups at the municipality level.

A question that remains open is why heterogeneous effects are observed in crimes related to organized crime: increases in kidnapping but no effects in homicides. One potential explanation is that large criminal organizations are more involved in homicide rates. Thus, if earthquakes do not impact large criminal organizations, we should observe no effects on homicide rates. In the case of the local criminal organizations, they have more information regarding wealthy individuals in their localities. Moreover, after being affected by the disasters, these wealthy individuals can be kidnapping targets. Thus, the rise in kidnappings may indicate that this crime is a crime of opportunity that local gangs can exploit during the chaos resulting from an environmental catastrophe.

Another issue that deserves to be explored in future research is the geographical distribution of the earthquakes’ effect on violence. Special attention should be given to municipalities with high youth unemployment (Shaw and McKay, 1942), as well as with shifts in the supply of young people to criminal enterprises due to school infrastructure damages (Cromwell et al., 1995), and high rates of cartel recruitment in recent years (Prieto-Curiel et al., 2023). In addition, future studies may analyze the effects of land type or municipal governance when these data are available.

Conclusion

This article provides evidence of the causal effect of the September 2017 earthquakes on organized crime. Namely, the estimates suggest a causal effect of the earthquakes on kidnapping rates equal to 4.0%. Our article further contributes to the literature studying the organizational behavior mechanisms through which earthquakes impact the activities of altruism and violence criminal organizations exert. Specifically, we identify an increase of 5% in the incidence of violence by local criminal organizations and an increase of 7% in altruism activities by this same type of criminal groups. Conversely, findings suggest no organizational behavior changes for large criminal organizations after earthquakes.

In terms of public policy, this study provides information on which organized crimes (kidnapping) increased after the earthquakes. This article also points out which type of criminal organizations (local) react to disasters. To better direct security efforts, future research should address the identification of the localities with the greatest increase in the levels of violence by local criminal organizations.

Supplemental Material

sj-docx-1-crj-10.1177_17488958241302274 – Supplemental material for Organized crime after earthquakes

Supplemental material, sj-docx-1-crj-10.1177_17488958241302274 for Organized crime after earthquakes by Adan Silverio-Murillo, Enrique García-Tejeda, Fernanda Sobrino, Jose Roberto Balmori-de-la-Miyar and Daniel Prudencio in Criminology & Criminal Justice

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) received no financial support to the research,authorship,and/or publication of this article.

ORCID iD

Adan Silverio-Murillo

Supplemental Material

Supplemental material for this article is available online.

Author biographies

Adan Silverio-Murillo is an Assistant Professor in the School of Government at Tecnologico de Monterrey. His research focuses on applied microeconomics and economics of crime.

Enrique García-Tejeda is an Associate Professor in the Legal Studies Division at the Center for Economic Research and Teaching (CIDE). His research focuses on spatial crime and urban crime.

Fernanda Sobrino is an Assistant Professor in the School of Government at Tecnologico de Monterrey. His research focuses on development,crime,and natural language processing.

Jose Roberto Balmori-de-la-Miyar is an Associate Professor in the Business and Economics School,Universidad Anahuac. His areas of research are law and economics.

Daniel Prudencio is an Adjunct Professor in Economics at Tecnologico de Monterrey. His research interest includes development economics and economics of crime.

References

Abadie

(2005) Semiparametric difference-in-differences estimators. The Review of Economic Studies 72: 1–19.

Arias

(2017) Criminal Enterprises and Governance in Latin America and the Caribbean. Cambridge: Cambridge University Press.

Atuesta

(2020) The use of violence and the evolution of organized crime: Evidence from Mexico. In: Lopez

Ireland

, et al. (eds) The Handbook of Collective Violence. New York: Routledge, pp. 232–245.

Augusto

(2021) Change-point analysis of Houston crime during Hurricane Harvey. Journal of Applied Security Research 16: 1–18.

Balderas

(2019) Cuál fue el rol de la “Unión Tepito” el 19-S del 2017? Available at: https://mvsnoticias.com/nacional/2019/9/25/cual-fue-el-rol-de-la-union-tepito-el-19-s-del-2017-420529.html (accessed 27 May 2024).

Balmori de la Miyar

(2016) The economic consequences of the Mexican drug war. Peace Economics, Peace Science and Public Policy 22: 213–246.

Barnes

(2017) Criminal politics: An integrated approach to the study of organized crime, politics, and violence. Perspectives on Politics 15: 967–987.

Barnes

(2022) The logic of criminal territorial control: Military intervention in Rio de Janeiro. Comparative Political Studies 55: 789–831.

Bertrand

Duflo

Mullainathan

(2004) How much should we trust differences-in-differences estimates? The Quarterly Journal of Economics 119: 249–275.

10.

Breetzke

King

Fabris-Rotelli

(2018) The impact of the Canterbury earthquakes on the temporal and spatial patterning of crime in Christchurch, New Zealand. Australian & New Zealand Journal of Criminology 51: 135–156.

11.

Brown

Velasquez

(2017) The effect of violent crime on the human capital accumulation of young adults. Journal of Development Economics 127: 1–12.

12.

Bunker

(2013) Introduction: The Mexican cartels, organized crime vs. criminal insurgency. Trends in Organized Crime 16: 129–137.

13.

Castillo

Mejía

Restrepo

(2014) Scarcity without Leviathan: The violent effects of cocaine supply shortages in the Mexican drug war. Working papers 356. Washington, DC: Center for Global Development.

14.

Castillo

Mejía

Restrepo

(2020) Scarcity without leviathan: The violent effects of cocaine supply shortages in the Mexican drug war. Review of Economics and Statistics 102: 269–286.

15.

CENAPRED (2017) Impacto Socioeconómico de los Desastres en México durante 2017. Secretaría de Seguridad y Protección Ciudadana. Available at: https://www.cenapred.unam.mx/es/Publicaciones/archivos/403-NO.19-RESUMENEJECUTIVOIMPACTO2017.PDF

16.

Cohen

Felson

(1979) Social change and crime rate trends: A routine activity approach. American Sociological Review 44: 588–608.

17.

Corcoran

(2013) Mexico’s shifting criminal landscape: Changes in gang operation and structure during the past century. Trends in Organized Crime 16: 306–328.

18.

Cromwell

Dunham

Akers

, et al. (1995) Routine activities and social control in the aftermath of a natural catastrophe. European Journal on Criminal Policy and Research 3: 56–69.

19.

Curiel

Arnau

Pinedo

, et al. (2019) Temporal and spatial analysis of the media spotlight. Computers, Environment and Urban Systems 75: 254–263.

20.

Dell

(2015) Trafficking networks and the Mexican drug war. American Economic Review 105: 1738–1779.

21.

Durán-Martínez

(2017) Drug trafficking and drug policies in the Americas: Change, continuity, and challenges. Latin American Politics and Society 59: 145–153.

22.

Enamorado

Lopez-Calva

Rodríguez-Castelán

, et al. (2016) Income inequality and violent crime: Evidence from Mexico’s drug war. Journal of Development Economics 120: 128–143.

23.

Frailing

Harper

(2017) Toward a Criminology of Disaster. London: Palgrave Macmillan.

24.

Fritz

(1996) Disasters and mental health: Therapeutic principles drawn from disaster studies. Disaster Research Center. Available at: https://udspace.udel.edu/items/0e4bf49b-f7a0-4feb-916d-8ada6367431b

25.

Fukumi

(2010) The Yakuza and its perceived threat. In: Allum

Longo

Panos

(eds) Defining and Defying Organised Crime. New York: Routledge, pp. 119–132.

26.

Galvis

Miranda

Heresi

, et al. (2017) Preliminary statistics of collapsed buildings in Mexico City in the September 19, 2017 Puebla-Morelos Earthquake. John A. Blume Earthquake Engineering Center and Department of Civil and Environmental Engineering Stanford University. Available at: http://learningfromearthquakes.org

27.

García-Hombrados

(2020) The lasting effects of natural disasters on property crime: Evidence from the 2010 Chilean earthquake. Journal of Economic Behavior & Organization 175: 114–154.

28.

GOBMEX (2019) Impacto Socioeconómico de los Principales Desastres ocurridos en la República Mexicana. Secretaría de Seguridad y Protección Ciudadana. Available at: https://www.cenapred.unam.mx/es/Publicaciones/archivos/429-RESUMENEJECUTIVOIMPACTO2019.PDF

29.

Godínez-Domínguez

Tena-Colunga

Pérez-Rocha

, et al. (2021) The September 7, 2017 Tehuantepec, Mexico, earthquake: Damage assessment in masonry structures for housing. International Journal of Disaster Risk Reduction 56: 102123.

30.

Herrera

Martinez-Alvarez

(2022) Diversifying violence: Mining, export-agriculture, and criminal governance in Mexico. World Development 151: 106142.

31.

Jones

(2013) The unintended consequences of kingpin strategies: Kidnap rates and the Arellano-Felix Organization. Trends in Organized Crime 16: 156–176.

32.

Kolbe

Hutson

Shannon

, et al. (2010) Mortality, crime and access to basic needs before and after the Haiti earthquake: A random survey of Port-au-Prince households. Medicine, Conflict, and Survival 26: 281–297.

33.

Leitner

Barnett

Kent

, et al. (2011) The impact of Hurricane Katrina on reported crimes in Louisiana: A spatial and temporal analysis. The Professional Geographer 63: 244–261.

34.

Lessing

(2021) Conceptualizing criminal governance. Perspectives on Politics 19: 854–873.

35.

Lessing

Willis

(2019) Legitimacy in criminal governance: Managing a drug empire from behind bars. American Political Science Review 113: 584–606.

36.

Lindo

Padilla-Romo

(2018) Kingpin approaches to fighting crime and community violence: Evidence from Mexico’s Drug War. Journal of Health Economics 58: 253–268.

37.

Magaloni

Franco-Vivanco

Melo

(2020a) Killing in the slums: Social order, criminal governance, and police violence in Rio de Janeiro. American Political Science Review 114: 552–572.

38.

Magaloni

Robles

Matanock

, et al. (2020b) Living in fear: The dynamics of extortion in Mexico’s Drug War. Comparative Political Studies 53: 1124–1174.

39.

Maheshri

Mastrobuoni

(2020) The race between deterrence and displacement: Theory and evidence from bank robberies. The Review of Economics and Statistics 103: 1–45.

40.

Mantilla

Feldmann

(2021) Criminal governance in Latin America. In: Ibarra

Erez

(eds) Oxford Research Encyclopedia of Criminology and Criminal Justice. Oxford: Oxford University Press, pp. 1–25.

41.

Massa Roldan

Fondevila

García-Tejeda

(2021) Female homicide victimization in Mexico: A group-based trajectory and spatial study. Global Crime 22: 123–142.

42.

Niño

González

(2022) Phantom state in Haiti: Criminal sovereignty and the mercenary remedy. Trends in Organized Crime. Epub ahead of print 2 August. DOI: 10.1007/s12117-022-09460-3.

43.

O’Brien

(2012) Fluctuations between crime and terror: The case of Abu Sayyaf’s kidnapping activities. Terrorism and Political Violence 24: 320–336.

44.

Oliver-Smith

Alcántar-Ayala

Burton

, et al. (2017) The social construction of disaster risk: Seeking root causes. International Journal of Disaster Risk Reduction 22: 469–474.

45.

Osorio

(2015) The contagion of drug violence: Spatiotemporal dynamics of the Mexican war on drugs. Journal of Conflict Resolution 59: 1403–1432.

46.

Osorio

Beltrán

(2019) Enhancing the detection of criminal organizations in Mexico using Ml and Nl. In: 2020 international joint conference on neural networks, IJCNN 2020, Virtual, Glasgow, 19–24 July, pp. 1–7. New York: IEEE.

47.

Oster

(2019) Unobservable selection and coefficient stability: Theory and evidence. Journal of Business & Economic Statistics 37: 187–204.

48.

Prelog

(2016) Modeling the relationship between natural disasters and crime in the United States. Natural Hazards Review 17: 04015011.

49.

Prieto-Curiel

Campedelli

Hope

(2023) Reducing cartel recruitment is the only way to lower violence in Mexico. Science 381: 1312–1316.

50.

Quarantelli

(1988) Disaster crisis management: A summary of research findings. Journal of Management Studies 25: 373–385.

51.

Rambachan

Roth

(2023) A more credible approach to parallel trends. Review of Economic Studies 90: 2555–2591.

52.

Rankin

(2012) 21st century yakuza: Recent trends in organized crime in Japan. The Asia-Pacific Journal 10: 1–27.

53.

Reitano

Shaw

(2018) Pay up or get hurt: What extortion says about urban governance, and how it might be curbed. In: Wennmann

Jütersonke

(eds) Urban Safety and Peacebuilding. New York: Routledge, pp. 95–117.

54.

Shaw

McKay

(1942) Juvenile Delinquency and Urban Areas. Chicago, IL: University of Chicago Press.

55.

Shaw

Islam

Mallick

(2013) Disaster, risk and evolution of the concept. In: Shaw

Mallick

Islam

(eds) Disaster Risk Reduction Approaches in Bangladesh. London: Springer, pp. 3–21.

56.

Shirk

Wallman

(2015) Understanding Mexico’s drug violence. Journal of Conflict Resolution 59: 1348–1376.

57.

Sobrino

(2021) Mapping criminal organizations: Automated municipal. Working Paper. Available at: https://www.fersobrino.com/dataproject/mco/

58.

Tena-Colunga

Hernández-Ramírez

Godínez-Domínguez

, et al. (2021) Mexico City during and after the September 19, 2017 earthquake: Assessment of seismic resilience and ongoing recovery process. Journal of Civil Structural Health Monitoring 11: 1275–1299.

59.

Trejo

Ley

(2018) Why did drug cartels go to war in Mexico? Subnational party alternation, the breakdown of criminal protection, and the onset of large-scale violence. Comparative Political Studies 51: 900–937.

60.

UNISDR (2009) Terminology on disaster risk reduction. Available at: https://www.unisdr.org/files/7817_UnisdrterminologyEnglish.pdf (accessed 27 May 2024).

61.

Yates

Mackenzie

(2018) Heritage, crisis, and community crime prevention in Nepal. International Journal of Cultural Property 25: 203–221.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.23 MB