Violent Crime and Risk Aversion

A Look at the paper by Ryan et al. on the impact of violent crime on risk aversion with evidence from the Mexican Drug War

By Ken Varghese



With the election of President Calderon in 2006, Mexico experienced a brief decrease in crime followed by a much more significant rise in violent crime. President Calderon’s efforts in the new War on Drugs removed the heads of the prominent Mexican cartels, however, this only led to more chaos. Groups fractured into smaller cells to fill in the vacancies left by the original cartels. The number of cartels in Mexico rose from six in 2006 to sixteen in 2011 and violence spread over the country as a result. The nature of crime in Mexico also changed as gangs sought more ways to find funds, including methods that impacted the citizenry more directly, such as extortion, kidnapping and auto theft. Citizens easily became targets for more brutal crime such as executions if they refused to cooperate.

In their paper “Impact of Violent Crime on Risk Aversion: Evidence from the Mexican Drug War,” Ryan Brown, Verónica Montalva, Duncan Thomas and Andrea Velásquez, analyze data from this period to gain insight into the ways exposure to violent local crime can effect risk preferences in individuals. The authors believe that, given the increasing brutality of crimes in Mexico and the heightened visibility of violent crime due to media coverage, the population living in affected areas would have been significantly psychologically impacted. They state that this may have a number of possible outcomes such as changing an individual’s perception of how risky their environment and future are or leading individuals to make risk averse decisions as a result of fear of victimization.

The Data

The authors used the Mexican Family Life Survey (MxFLS) to obtain data about local populations. The baseline survey collected data from a period in 2002 and included 8,440 households and 35,600 individuals in 16 states throughout the country. The first follow up survey (MxFLS2) in 2005 and 2006 occurred during a period of relatively stable levels of violent crime, and the second follow-up survey (MxFLS3) in 2009 and 2010 occurred after a major rise in violent crime. Using these three surveys allowed the authors to examine the impacts of different levels of violence on the same individuals over an extended period of time. A trait of the MxFLS that makes this analysis possible, is its low levels of attrition (loss of participants), with 89% of the original respondents being contacted again for the first follow up and 87% for the second follow up. The surveys allowed for the gauging of risk preferences by having respondents decide between sure outcomes and gambles that resulted in either an attractive or unattractive outcome. The differences between the expected values of the gambles and the values of the sure options were used to rank the level of risk aversion. This individual-level data from the MxFLS is combined with monthly municipality-specific homicide data from the National Institute of Statistics and Geography (INEGI) in order to measure how risk preferences vary as the level of local violent crime changes over time. The monthly homicide rate (the primary data used from INEGI) for the period studied can be seen in figure 1 below.

Figure 1 blog

Brown et al. 2017

Analysis and Results

The authors’ primary calculations suggest that an increase of 1 homicide per 10,000 people resulted in a 5% increase in risk aversion in MxFLS3 (compared to the average per capita expenditure from MxFLS2). Table 1 shows how the overall risk aversion distribution changed from MxFLS2 to MxFLS3. Surprisingly, the authors also found that the risk attitudes of households in the lowest quartile of per capita expenditure are not effected by homicide rate like those in other quartiles. This could be because of a variety of reasons, such as the possibility that the change in violence was greater in high income neighborhoods than in low income neighborhoods. The authors also state that it is possible (if fear as a result of violence is what drives the risk preference change) that the worst off individuals may already be past the point where increased local violence no longer significantly impacts their risk preferences. It is also possible that the used measure is noisier for these individuals.

table 1 blog

Brown et al. 2017

It also noteworthy that the influence of violence crimes significantly reduced the earnings of self-employed men and the labor market participation of self-employed women. This finding led the authors to explore whether the primary fear impacting risk preference was financial in nature. In looking at the data, they found that the risk attitudes of the self-employed are not more strongly impacted by local violence than other participants. However, further analysis revealed that risk aversion was more than double the size among participants in the population who were fearful of victimization during the escalation of violence. This provides evidence suggesting that local violent crime is impacting risk attitudes through the fear of victimization rather than increased financial hardship.

How Does This Relate to Development Economics?

As the authors state, there is evidence that increased risk aversion is negatively correlated with involvement in riskier but more profitable investment decisions, occupational choices and migration. This suggests that increased levels of risk aversion as a result of exposure to violent crime can negatively impact household wealth accumulation and can hurt a country’s economic development in the long-term. As such, level of exposure to violent crime and statistics such as local homicide rates should be seen as closely tied to inequality and economic growth. The impact on household wealth accumulation and on willingness to take riskier but more profitable financial decisions can deepen inequality and diminish growth in the long term. Similarly, the authors also mention that the violent crimes impacted self-owned businesses negatively. This is somewhat relevant to the field of microfinance, as some programs find effectiveness primarily in enabling individuals to have more choice by allowing them the option of self-employment. If self-employment becomes less profitable as a result of violent crime, these forms of microfinance will be less successful in affected communities.


Brown, Ryan, Verónica Montalva, Duncan Thomas and Andrea Velásquezv (2017). Impact of Violent Crime on Risk Aversion: Evidence from the Mexican Drug War, NBER Working Paper No. 23181 (2017): retrieved from

Schaffner, Julie (2014). Development Economics: Theory, Empirical Research, and Policy Analysis. Hoboken, NJ: Wiley.


Banking on Trust: An Analysis

An in depth analysis of the study entitled “Banking on Trust: How Debit Cards Enable The Poor to Save More”.
Written By: Kaushik Nagarur


Financial Institutions have long been the backbone to many countries’ economies. Institutions such as JP Morgan and Bank of America are critical to the economy of America. Economic development is related closely to these financial institutions. The growth of a developing country can directly be linked to these financial institutions. Yet, what is it that drives and causes people to invest in these financial institutions? A study done by Pierre Bachas, Paul Gertler, Sean Higgins, and Enrique Siera entitled “Banking on Trust: How Debit Cards Enable The Poor to Save More”, attempts to answer this question. The study looks into whether or not trust in a financial institution is needed in order for people to use formal financial services (Bachas et al 1).


In order to examine this question, Bachas used a natural experiment that was occurring in Mexico. A natural experiment is an “observational study in which an event or situation that allows for the random assignment of study subjects to different groups is exploited to answer a question” (Messer). In this case the natural experiment was to answer whether or not there was a connection between trust and the use of formal financial services. The experiment had to do with the handing out of debit cards through a Mexican cash transfer program known as Oportunidades. Oportunidades was a cash transfer program that provided bimonthly cash transfers to poor families. The objectives of Oportunidades was to “Increase capacities in health, nutrition, and education of poor families” (Projects). Essentially the program put money into a savings account and gave poor families money conditional on the fact that the families have to send their children to school and must have regular health checkups.

The program was being conducted in three different time periods (waves), and the authors used these “waves” in order to create their natural experiment. There were three major groups based on the roll out of these debit cards. There were two treatment groups that received the debit cards one year apart, and one control group that received the debit cards at the end of the study period. The authors used a combination of high frequency administrative data on bank transactions as well as survey data of the beneficiaries. The survey data was to explore whether or not people were learning to trust the bank, people use the banks, and people understand how savings work. The authors analyzed 343,204 accounts at 380 Bansefi branches over a four-year period from November 2007 to October 2011(Bachas et al 8).

Analysis and Results

The first analysis that was done in the paper by Bachas was the effect of the debit cards on the stock of savings. In order to estimate the effect of the debit cards on savings, a difference in difference regression was done. The reason that a difference in difference regression was needed was to control for other shocks including “time effects”, as well as control for general individual differences that might affect the outcome. The following formula (Figure 1) was utilized in order to understand the balance effect and see whether debit cards affect the overall balance in an account (Bachas et al 9).

Screen Shot 2017-04-23 at 6.24.03 PM

Source: Bachas, Pierre, Paul Gertler, Sean Higgins, and Enrique Seira. “Banking on Trust: How Debit Cards Enable the Poor to Save More.” NBER Working Paper Series 23252 (2017): n. pag. Web. 24 Apr. 2017. Formula 1

Balance was the average balance in the account over time period t. The  lI is the account level fixed effect, while d is the fixed time period effect. The Tj is a dummy variable that is 1 if there is treatment, and the last term is a time dummy variable. The most important variable is  fk which measures the average difference in balances between treatment and control group in period K. The following figure V is the result of this regression (Bachas et al 42).

Screen Shot 2017-04-23 at 7.36.45 PMSource: Bachas, Pierre, Paul Gertler, Sean Higgins, and Enrique Seira. “Banking on Trust: How Debit Cards Enable the Poor to Save More.” NBER Working Paper Series 23252 (2017): n. pag. Web. 24 Apr. 2017. Fugure 5 page 42

The figure plots the coefficient  fk and the confidence interval. There are two important conclusions in the figure. The first one is that there is no difference in the pretreatment levels between treatment and control groups. Second is that around 8 months after receiving a debit card, balances start to change. After around 2 years with the card, balances are 1,400 pesos higher for the treatment group over the control group. In wave 2, savings begin to rise much more quickly after receiving the card. This is likely due to information spillovers from wave one (Bachas et al 10).

The second major question is what effect trust has on the flow of savings at the beneficiary level. Whether or not someone trusts the bank can be calculated using a dummy variable. In the surveys, they ask “do you leave part of the monetary support from Oportunidades in your bank?”(Bachas et al). After they answer, the authors ask why they do or do not leave money behind. Based on that they capture lack of trust.  Since trust is endogenous, it is possible to use it as an instrument variable with the date of debit card assignment. The main reason to do this is that it isolates variation in trust that can be explained exogenously by time with the card. What results is the following formula( Bachas et al 10).

Screen Shot 2017-04-23 at 8.02.28 PM.png

Source: Bachas, Pierre, Paul Gertler, Sean Higgins, and Enrique Seira. “Banking on Trust: How Debit Cards Enable the Poor to Save More.” NBER Working Paper Series 23252 (2017): n. pag. Web. 24 Apr. 2017. Formula Pg 56

Essentially this formula is just trying to see what effect trust has on the proportion of average income. When running the 2SLS regression test, a statistical test to see the correlation between two items, the authors came up with the following table.

Screen Shot 2017-04-23 at 8.05.38 PM

Source: Bachas, Pierre, Paul Gertler, Sean Higgins, and Enrique Seira. “Banking on Trust: How Debit Cards Enable the Poor to Save More.” NBER Working Paper Series 23252 (2017): n. pag. Web. 24 Apr. 2017. Table II PG 56

This table shows two major conclusions. The first stage shows that an average of six additional months with the card leads to a 10.3 percent increase in trusting the bank. The IV coefficient in column 2 states that those who trust the bank as a result of this six additional months save an extra 2.8% on their income (Bachas et al 56).

The paper as a whole has two major implications. The first is that the people who received debit cards save more. The second major conclusion is that trust plays an important role in the savings effect. Once people who received these cards started trusting the bank, they started saving much more.

Why does it matter?

The conclusion reached by the authors have policy implications that can be used in order to help financial institutions. First and foremost, building savings has been shown to have positive effects on business investment. This is shown in the experiment Saving Constraints and Microenterprise Development: Evidence from a Field Experiment in Kenya, done by Dupas and Robinson. While looking at banking in rural Kenya they found that “Market women use these (savings) accounts to save up to increase the size of their business” (Dupas et al. 185). This means that savings accounts have a direct effect on business which has a direct effect on the economy as a whole. Savings accounts can also help smooth consumption curves. In other words, savings accounts help make sure that your expenditures do not fluctuate too much after an unforeseen emergency such as a health emergency. Savings accounts have also been linked to a way out of Poverty. Mullainathan and Shafer wrote a paper entitled “Savings Policy and Decision Making in Low income households” where they conclude that access to formal savings accounts may “provide an important pathway out of poverty” (Mullainathan et al 9). With all these papers on the importance of savings, it is easy to see that helping the poor in developing countries. Debit cards, encourage people to save in these intuition’s due to the fact that people can see their account balance. This ability lets them trust the institutions more. It is important that governments utilize this information and this technology in order to help others.


Works Cited:

Bachas, Pierre, Paul Gertler, Sean Higgins, and Enrique Seira. “Banking on Trust: How Debit Cards Enable the Poor to Save More.” NBER Working Paper Series 23252 (2017): n. pag. Web. 24 Apr. 2017.

Bukari, Jeff. “Using Your Debit Card Might Actually Make You Richer.” Fortune. March 22, 2017.

Dupas, Pascaline, and Jonathan Robinson. “Savings Constraints and Microenterprise Development: Evidence from a Field Experiment in Kenya.” American Economic Journal: Applied Economics 5.1 (2013): 163-92. Web.

Messer, Lynne C. “Natural Experiment.” Encyclopædia Britannica. Encyclopædia Britannica, Inc., 07 Oct. 2016. Web. 24 Apr. 2017.

Mullainathan, Sendhil and Eldar Shafr, Savings policy and decision-making in low-income households, in Insu‑cient Funds: Savings, Assets, Credit and Banking Among Low-Income Households, 121145 (New York City: Russell Sage Foundation Press, 2009).

“Projects & Operations.” Projects : Support to Oportunidades Project | The World Bank. World Bank, n.d. Web. 24 Apr. 2017.