Episode 29: Patricio Domínguez

 
PD_IADB.jpg

Patricio Domínguez

Patricio Domínguez is a Research Economist at the Inter-American Development Bank.

Date: May 12, 2020

Bonus segment on Dr. Domínguez’s career path and life as a researcher.

A transcript of this episode is available here.

Episode Details:

In this episode, we discuss Dr. Domínguez's work on the interaction of offenders and victims in determining when crime occurs:

"How Offenders and Victims Interact: A Case-study from a Public Transportation Reform" by Patricio Domínguez.

OTHER RESEARCH WE DISCUSS IN THIS EPISODE:

 

Transcript of this episode:

 

Jennifer [00:00:07] Hello and welcome to Probable Causation, a show about law, economics and crime. I'm your host, Jennifer Doleac of Texas A&M University, where I'm an Economics Professor and the Director of the Justice Tech Lab.

 

Jennifer [00:00:18] My guest this week is Patricio Domínguez. Patricio is a Research Economist at the Inter-American Development Bank. Patricio, welcome to the show.

 

Patricio [00:00:26] Thank you again for having me.

 

Jennifer [00:00:29] Today, we're going to talk about your research on the interactions of victims and offenders focusing on robberies of bus drivers in Chile. But before we dive into all of that, could you tell us about your research expertise and how you became interested in this topic?

 

Patricio [00:00:45] So before doing my Ph.D., I studied civil engineer with a major in transportation engineer. And then I was trained as a labor economist and I - which is where I became interested in the economics of crime literature. And, well, I remember that during my Ph.D., I was curious about the relationship between cash availability and crime. And given my background as a transportation engineer, I think it was kind of natural to start looking at a huge reform that happened in Chile and that I had to study very closely, which in addition, I thought it could offer a good way to show how potential victims and offenders interact. So the idea was like - usually we as applied economists focus on the effect of a particular shock and how it affects crime opportunities or how law enforcement agencies work, and we interpret the - an observed variation in crime as driven by the shock. Right. And that allows you to identify the effect of - for example, how law enforcement agency works.

 

Patricio [00:01:57] So all those paper, I think measured - and you can think about the papers that measured the effect of a new sentencing regime, for example, the effect of prisons, police, among others - in all those cases, we usually interpret the officers change in - as driven by potential offenders reacting to the shock, let's say, the new sentencing regime, the police presence, the new set of opportunities available. But something that is much less emphasized in the literature is a potential endogenous reaction from the victim side that it could also explain part of the variation that we observe in crime victimization. So basically, I thought this reform could offer us an interesting case study to analyze this important feature in crime studies.

 

Jennifer [00:02:47] Your paper is titled "How Potential Offenders and Victims Interact: A Case-study from a Public Transportation Reform." And as you mentioned, you frame this study as being about the role that would-be victims play in determining whether crime takes place. And you highlight in the paper that much of the emphasis in the economics of crime literature is on the offenders themselves and on various criminal justice actors. So as a field, we've spend a lot of time measuring the effects of things like policing and punishment. So before your study, what had we known about the role of victims in this process?

 

Patricio [00:03:21] Exactly. So that idea motivates me a lot in this project. So one person that noticed this issue very early in the economics of crime literature was Phil Cook. But many scholars have emphasized that this has been sort of understudied in the previous literature. So the idea is a kind of fundamental source of endogeneity the crime literature. So the idea is like crime rates are not only the result of offenders' action, but also the interaction between their choices and the choices of potential victims. You may include here legal enforcement agencies. So the more important intuition for my research is that the level of effort exhibit by potential victims to protect - let's say the property - that determines the availability of crime opportunities, which in turns depend on the level of risk they perceive. So - and you can also see this in the criminology literature, for example, some insight from Ron Clarke, the Professor at Rutgers. He has advocated for the need to focus on criminal event rather than offenders' choice when study crime. So this makes more explicit the need to incorporate what he calls factors surrounding crime. So the idea that beyond a willing offender, there is an actual victimization requires a vulnerable target and an appropriate opportunity. And from the empirical literature, there are also a few examples that I would like to highlight. For example, there's a paper between Phil Cook and John MacDonald on the business - the effect of business improvement district in L.A. when they show that the actions adopted by private actors can have a substantial effect on crime and crime control policy.

 

Patricio [00:05:12] There's other papers that, for example, emphasize how victims can alter the set of criminal opportunities by hardening an attractive target. There's this paper in the Economic Journal by Ben Vollaard and Jan Van Ours - I hope I pronounced that well - that they evaluate the - a large scale intervention in the Netherlands. It was a regulatory change in the - that required that all new built homes to have a burglary proof windows and doors, and they find a significant reduction in burglary risk in new homes. And similarly, there are a couple of experiments evaluating the impact of LoJack track devices in vehicles. The LoJack is a radio transmitter that tracks - it's a tracking device, which is highly effective to recover a stolen car. So, therefore, it may reduce the reward from stealing - you can think. And there's a paper by Ian Ayres and Steve Levitt in the QJE, they exploit the differential timing in the arrival across the US cities and they found a large reduction in auto theft. So that's important for this setting that they've observed that the installation of the device was unobservable for offenders, which makes them interpret the results as a general deterrent effect. And it's interesting to notice that a few years later, Gonzalez-Navarro, who is a Professor at Berkeley right now, he studied the effect of LoJack in a slightly different setting in Mexico, where only the fourth car that explicitly incorporated the device. So in this case was observable - the installation of the device that could track the car. So - and in this case, he finds a reduction in LoJack equipped vehicles. But as opposed to Levitt, he interprets the effect of the observed effect as a specific deterrent effect.

 

Patricio [00:07:15] And finally, I think there's a few papers regarding the interaction between offenders and victims and whether they can shape level and the nature of crime. There's a couple of work by Brendan O'Flaherty, a Professor at Columbia, and Rajiv Sethi, for example, where they show that - for example, perception about race can account for racial disparity in violent robbery and murder. And there's some research done by Chandler McClellan and Erdal Tekin and Cheng and Mark Hoekstra in the context of the US Stand Your Ground laws where they raise serious concern about the ability to increase public safety by encouraging potential victims to resist an attack. So there is some combination of a study that have addressed this relationship between victims and offenders. But I think - I believe this is something that should be studied more in the literature.

 

Jennifer [00:08:12] Why don't we know more than we do about the role the victims play here? What do you see as the main challenges in figuring out the causal effects of victims behavior on crime?

 

Patricio [00:08:23] That's a good question. I think there are two main challenge here. One is in terms of the identification and the other in terms of the availability of data. In terms of identification, as in many other problems in economics, you would like to identify shocks in order to isolate the effect of potential confounders. But even in some cases, you can think - when you have a shock, you can have like the possibility that many potential agents can react to the shock in a different way. So it's hard to interpret the result as simply, for example, driven by a supply of offenses, which is what typically do in the economics of crime. As I said, crime rates are not only the result of offenders' actions, but also the interaction between their choices and the choices of potential victims. So we should include this endogenous reaction - for example, from the - I think that is something that is well addressed in the literature, is the endogenous relationship regarding law enforcement agencies. There's a lot of very careful studies in this regard. For example, the papers that address the effect of police and crime, detail on Ernesto Schargrodsky in the ER, others: Draca and Steve Machin, Klick and Tabarrok they all - they have focused on particular shocks to the police presence and how they affect crime.

 

Patricio [00:09:51] There are other papers that focus on the changes in police staffing. They're very well addressed, these sources of indigeneity. But regarding potentially endogenous reaction from victims, it is hard to identify a particular shock. And the main limitation here basically has to do with data because data of a victim is really hard. So victims can adapt in many different ways. They can harden an attractive target. They can alter their travel behavior to avoid a specific area that they consider dangerous or they being out at certain hours of the day. They decide, for example, to purchase specific goods in order to not being victimizes or minimize the cost of - associated with victimization if victimization happens. So there's a lot of potential reactions that makes hard to model and for a researcher to analyze or to incorporate in the evaluation. So I think it's a combination of these two challenges where I believe the - this particular research that we are talking about right now is may contribute. So because we have a shock that affect victims precautionary measures. And I believe we have data at a high level of resolution to identify the effect of the change.

 

Jennifer [00:11:12] Okay, so your study considers bus robberies in Chile and how a couple of changes to the bus system there affected the robbery rate. So let's start the story with a bit of context. What did the bus system look like before the reforms that you're studying?

 

Patricio [00:11:26] Yeah, yeah. So, yeah, let me spend some minutes explaining the details because I think it matters a lot. So what happens. In 2004, the government in Chile was decided to transform the bus system in Santiago that was ranked as the worst public service at the time. So during the pre-reform system was pretty much that the system was pretty much a consequences of a series of reforms that took place beginning in the 80s. They have like a highly privatized and highly deregulated system. So, for example, the industrial organization of the system was very automized that what makes it very hard to make any kind of coordination. So Santiago, which is the city here, Santiago metropolitan area, has a population between 5 and 6 million people at that time. And the public transportation was served by buses, I think, was a main mode of transportation. 25% of all trips were made in buses. And the system has like 8,000 different buses that they were serving 380 roads. And you have like more than 3,000 operators. So it's just a bit more than one operator per bus in the system. So highly automized system. So perhaps the most notorious feature of the system was a lack of integration in every possible dimension, and that was exemplified by the payment mechanisms.

 

Patricio [00:13:07] So let me highlight this fact. One of the byproduct of the lack of integration was what experts in public transportation called the work for the fare. The work for a fare is a system where drivers, even within the same company, serving the same road, they compete for passengers, basically because their salary depends on the number of riders. So this on the street competition was also responsible for a very inefficient road structure, typically a bus route connected to point of the city's periphery. On average, they were 60 kilometers long. So as a result, there was an oversupply of service in highly congested area. For example, in downtown 80% of buses circulated within notorious consequences in terms of traffic congestion and air pollution. So - and more related to my paper, the - this affects the working conditions of drivers and the payment system.

 

Patricio [00:14:10] So let me explain something a bit about that. So passengers pay the - in this system -passengers pay their ticket with cash inside the bus. And they pay that to the driver who, on top of driving, they had to receive cash from passengers, calculate correct change and finally provide riders with a ticket. It was a very complicated system. And all the money was collected in the so-called pecera which is a Spanish word for fishtank. That name came to be precisely because everything was very visible to everybody. So you sit on the bus looking at the money the driver were collecting there and imagine like riding 60 kilometers with this box along with you. So in addition to, under this cash payment regime, drivers were paid based on the amount of money that they collected that obviously it was related to the number of passengers that ride the bus, which made drivers responsible to protect cash.

 

Patricio [00:15:16] So this open boxes was a very attractive target for cash robberies. But perhaps and unsurprisingly, for many of us that can recall how the system worked, there were not too many incidents. But the one that happened were very violent ones and sometimes we'd cover it in the news. So part of this is because drivers adopted some protection measures. Some of them were not the optimal ones. So personally, I can recall many of them carrying a weapon inside the bus. In Chile a gun possession is still very uncommon. But you can recall that at that time they were carrying a sort of baseball bat or even knives to protect themselves.

 

Patricio [00:16:08] So, and as a footnote, let me recall - this is for the American audience that Ronald Clarke has a couple of paper like showing the salient role of public transportation and crime, and he has documented some studies that were done related to this. So there's one that report, for example, that this kind of robbery of fare-revenue to bus driver became a serious problem across many cities in the US in the 60s and early 70s. And after two consecutive shootings of bus drivers in Washington, D.C. and New York City in 1968, the solution proposed was a common anti-crime tool that you can nowadays see everywhere. So the introduction of this exact change for collection boxes like there were along on-board secure boxes into which the fare were deposit. So in my experience as a user of public transportation, this kind of device are in place across many cities in the US. But the implementation of this device was not feasible in Chile - in the Chilean context - because of the specific industrial organization of the system that was very automized. So I think that is like an enough background for the - how the system worked in the past.

 

Jennifer [00:17:31] And then - in beginning in 2005, things started to change. So what were the reforms that were made to the system? How did they come about and when exactly do they take place?

 

Patricio [00:17:43] So the main goal for the government was trying to modernize the system in a broad sense, and they believed that integration was key at that time. So they decided, for example, to have like a new industrial organization and bus routes. So they divided the city in 15 zones and each area was serviced by a single company. The entire industry was franchised through an international call for tenders in 2004. And in order to avoid on the street competition, the companies were required to pay fixed salaries to the drivers.

 

Patricio [00:18:18] They also decided to implement a new bus fleet and integrate the system with metro. In order to integrate the system with the metro, the subway, they needed to create a new payment system that can track whether people transfer from one system to another because they wanted to reduce the cost of transfer from buses to metro. So they implemented this debit card in all metro and buses, which is a sort of the same card that you can see, I don't know, in the BART in San Francisco Bay Area, for example, the Clipper card in D.C. In many different places, you have this debit card, and you can put cash in advance and you can ride the system. So let me tell you something a bit more about the implementation because this is key.

 

Patricio [00:19:05] So the original plan was to start all at once in October 2005. And that date was decided because there was a few months before the presidential election. Obviously, there was a huge plan and they wanted to have some credit for that. So, however, that they realized that the deadline was fast approaching and in view of some technical difficulties, they decided to delay - such a delay of the arrival of new buses, the government decided to postpone the full implementation of the program.

 

Patricio [00:19:40] But - and this is key - the government had already signed contracts with the new companies. So in order to avoid penalties related to delays stipulated in the contract that were signed back in 2004, they created so-called transition period that starts in October 2005 to February 2007. So a bit longer than one year. So that period would allow new company to get use of the city. It's important that some companies were foreign companies. So that was something that they will put some value on and that could make some time to have new bus fleet and the debit card fully in place.

 

Patricio [00:20:19] So during this transition period, the new companies were allocated to the old bus route. So in terms of the operation, this meant no major change for passengers: same bus routes, same payment  mechanisms, still in cash, but an important change in terms of the driver salaries because the new companies were mandated to pay fixed salaries.

 

Patricio [00:20:40] So finally, in February 2007, the most significant changes were adopted all at once. There was a called, the big bang day for the new system where all this new 15 service areas were serving the entire city and the old roads were no longer in place. And also the cash system would fully eradicate it. And the only way to enter the bus would be with this debit card. So to sum up, there are three periods that are important here. So the pre-policy period, which we are going to call before 2005 - October 2005, then the transition period, which is a bit longer than one year from October 2005 to February 2007, and the post-policy period. So three periods to keep in mind for that research.

 

Jennifer [00:21:31] So just to summarize that very briefly for the October 2005 change, the key is that you're switching from a system where the drivers basically get whatever money they're collecting from passengers to a fixed salary. So they get paid even if someone steals the bus fares. And then that February 2007 change is getting rid of cash altogether. So as an economist, when you heard about these reforms, what did you have in mind as the potential mechanisms through which they might affect crime?

 

Patricio [00:21:59] Yeah, yeah. This is important because I thought about - a lot about this. So, well, any economist I believe that cares about, like the extent to which agents adapt to these two new regimes - so where, for example, new incentives are in place. So as an applied economist, we also care about whether some theoretical reaction can be observed in the data. So in this case, when I first thought about this as a potential research project, I was curious about whether this provide a good case study on how offenders and victims interact for the same reason that we believe a $100 bill doesn't lie on the ground for too long. I was curious about whether something similar in this particular market for offenses has any empirical basis or not. So in particular, what I wanted to see is how the changes in propensity to protect the cash, which was induced, as you said, by the new driver salary regime, that affected or not, the robbery incidents inside buses and whether or not the offenders, on the other hand, adopt the level of violence they exhibit to these new resistant regime. So basically what I expect in theory, it was during the transition period, an increase in incident driven by drivers no longer protecting the money, but also a decline in the level of violence exhibited by the offenders who no longer need to use or to threaten the drivers with, for example, a more lethal weapon, let's say.

 

Jennifer [00:23:47] And then for the switch to the cashless system, how should we think about that?

 

Patricio [00:23:51] Yeah. So in this paper where I focus, I use that as sort of validation because here the response would be more mechanical. So you might be worried that in this case, I'm not exactly observing the incidents affecting drivers, but the thing that I - that is report in the data - we can go to the details about the data later - but what I observe is incidents reported in buses. But the incident could have affected the drivers or somebody else. So I use that period in order to validate that. But importantly, what I observe is like a huge decline in the cash-related robbery incidents when the cash was - as a payment mechanism - was replaced. This is important because people were still carrying cash in their pockets, in the wallets. But obviously it was very hard to offenders to steal cash when was not available in this open boxes. So obviously, cash related robberies declined during that period. And one of the - I think the new version of the paper - I tried to look at the potential displacement to other areas, and I used that also to validate the effect. But basically, as you said, I observe a huge decline in cash-robbery incidents.

 

Jennifer [00:25:16] So you're going to use three related approaches to measuring the effect of these reforms: an event study, so basically a pre-post design looking at effects immediately after the change, a difference-in-difference analysis, where you have a control group, and a triple-differences analysis, where you have two control groups. So walk us through each of these strategies and what you see is the pros and cons of each.

 

Patricio [00:25:41] Cool. Yeah. So basically, I have two shocks. Right. One related to the transition period and another one related to the post-policy period. So - and in the first one, I'm interested in how the shock in the salary policy where drivers would - the driver policy was changed from being paid to based on the number of passenger to fixed salary whether that affect the level of the nature of incidents reported in buses. And whether the second shock where cash payment were eliminated or not was eliminated, whether that effect or not, the level of crime that we observe. So I combined three strategies that I believe they complement to each other. For example, the first one, the interruptive time series, I look specifically at the evolution of cash related robberies in buses. So this strategy may be valid if the pre-policy period offers a reasonable counterfactual of what would have happened after the reform. Obviously, this is limited, since one can think that many other contemporaneous changes could also have effect devolution of cash related robberies in buses. So, for example, changes that we're experiencing, the metro system that could have affected the bus ridership. So this approach, although is insightful, is kind of limited in terms - or have a strong assumption.

 

Patricio [00:27:09] So in addition - so in order to isolate the set of - a set of potential confounding factors, I also implement a difference-in-difference strategy. The idea here is to control for - controlling for non-cash robberies, let's say robberies of cell phones. This kind of - if this kind of incidents follow a similar pattern that cash related robberies, and if we believe that this pattern was not altered during the period for any other reason than the changing driver salary policy and the payment making - the payment system, we can identify the effect of the reform of each of the shocks. And I show that, for example, that during the pre-policy period, both type of crime, cash related and non-cash related robberies, they follow a similar trajectory, but they gradually diverge during the transition period. And finally, a cash related incident suddenly drops at the beginning of their post-policy period.

 

Patricio [00:28:14] And finally, I evaluate the analysis using a triple-differences approach, considering the proportional split between cash and non-cash incident reported on buses relative to similar variation observed - regarding incidents reported on street and public spaces. I finally run a set of robustness check such as, you mentioned, an event study design that provides a more transparent description of the temporal evolution of the estimates. Also, I reproduce the main analysis using municipality level data. And finally, in terms of statistical inference, I correct for serial correlation, following a recommendation of the classic work of Marianne Bertrand, Duflo, and Mullainathan, collapsing the data at the period level rather than the weekly level as before. So that's idea of the three approach.

 

Jennifer [00:29:08] Great. And just to highlight for folks, one of the real values I see in the difference-in-difference analysis is that it helps control for changes in ridership. So you might be worried that all of these changes to the bus system might affect the number of people riding the buses. And maybe that's what's driving the change in crime rates. And so controlling for non-cash robberies on the buses is basically controlling for the number of people on the buses, which addresses that potential concern. What data do you have to do all of this?

 

Patricio [00:29:38] Yeah, so I have access to administrative data at the incident level. And I believe, and I'm sure you would agree with me, that this is very important to study crime because you can have a lot of detail about the incidents. In the data that I have they're all crimes reported to the police between 2005 and 2010, which is like a long enough period to analyze the effect of the reform and each record for each incidence, I have information about the time, the date, and the location where the crime was perpetrated. Importantly, in terms of the location, I was able to identify 'bus' as a specific category and the type of good that was stolen. And that allows me to identify whether the main thing that was stolen or attempted to be stolen was cash or not.

 

Patricio [00:30:32] And another important feature of the data is - this is collected by the Chilean police, which is different to other places. It's a very centralized agency - it's a national police, and it's the only one responsible to collect data. For example, the Director of the Chilean police is designated by the president. So I think that makes me confident that this data is very consistent over time and across space, even in a large city as Santiago metropolitan area. And just to give you a context, Santiago is probably the city with the lowest homicide rate in Latin America. It's similar to many European cities, but has a very high robbery rate, which also allows me to focus on a very specific type of event because you have enough variation here to observe, such as the robbery that happens inside buses.

 

Jennifer [00:31:32] All right. So let's talk about the results. What do each of your empirical strategies tell you is the effect of the first reform: moving to fixed salaries for the bus drivers?

 

Patricio [00:31:41] Yeah. So in all the three cases, the results are pretty consistent. On the move to a fixed salaries, the effect is pretty large. I found an increase between 130 and 150% relative to the level of observed during the pre-reform level. So we basically change from 12 incidents per week in the city to 28 incidents per week on average. Important here is the effect during this period is very gradual. And for the post-policy period or the effect of switching to a cashless system - again, I compare here to the pre-policy period. I could have compared to the transition period, but I wanted to be very conservative in this case. And what I found is a reduction on the 70% of total incidents reported in buses, which I believe, again, is pretty large.

 

Jennifer [00:32:39] Yeah. And if I remember correctly, the effects for the triple-differences were a little bit smaller, and this got me thinking - so you can imagine some general equilibrium effects here, right? That is, if you no longer have to carry cash to use on the buses, you might just be less likely to carry cash with you everywhere. And if no one carries cash with them anymore, then robbery everywhere is less productive. And so it might be that your second control group in the triple-difference analysis, which is cash robberies outside of buses, they could be treated somewhat by these changes as well. So in other words, you can imagine those cash robberies falling, too, which would bias your triple-difference estimates downward, and that might explain why they look a little smaller. Does that sound right to you?

 

Patricio [00:33:24] Yeah, yeah. Although the difference is it's not large enough to think about the huge spillovers, for example, to other places. But yeah, you are correct. You're correct. Yeah.

 

Jennifer [00:33:38] Yea. And then, as you said, it did seem like a gradual effect. So when you moved to the fixed salaries and bus drivers no longer have an incentive to protect the cash with their lives because they don't care, they're getting paid anyway. Right. You see in the graph a slow, gradual increase in robberies over time during this period, which suggests some learning on the part of the offenders at least. Is that your interpretation too?

 

Patricio [00:34:05] Exactly. So I was curious about like this figure. One of the thing, by the way - so I encourage the audience to look - take a look at the paper because everything also I tried to report like specific figures to very clean about like where the - or very transparent about the identification strategy  and the results as well. So but basically, you can observe there and you can take a look at the event study that I report. The story that I believe is going on here is like the offenders are learning about this new protection measures that the new salary regime induced on drivers. So this is - so as you might expect - the increase during the transition period is not sharp. It was gradual over time. And - but on the other hand, the change to a cashless system that was implemented all at once, that was pretty sharp. So I try to illustrate those graphically as well in the paper.

 

Jennifer [00:35:11] Yeah, I agree. The graphs are really nice here. It's always nice to see a paper where the graph basically tells the whole story. You know, it's like the numbers in the tables are sort of secondary.

 

Patricio [00:35:23] Yeah, I try to do so, yeah.

 

Jennifer [00:35:26] You also considered the level of violence involved in the robberies that do take place and how this is affected by the reforms we're discussing. So tell us more about that analysis and what you find.

 

Patricio [00:35:38] Yeah, yeah. So that's also related to a sort of prediction that I have in a very simple model that I developed in the paper. So during the transition period you have that basically drivers no longer have incentive to protect the cash. So my question here is - well, do offenders also adapt to this specific margin? So basically, drivers no longer need to carry a weapon, for example, to defend themselves. So and they may oppose minimum resistent in the case of an attack. So the empirical question is - here is - like, do offenders also adapt to this potential response? And to simplify the discussion, I show that conditional on offending, the probability of using - let's say, a firearm or think about any other more lethal weapon - the conditional on offending the probability of using a firearm should decline during this period. So the basic intuition is that the benefit from using a firearm declines because victims are less likely to exhibit a high level of resistance. And actually I test that in the data. And what I found is like although crime increases, the probability of being attacked by a firearm declines by 8%.

 

Jennifer [00:37:02] Putting it all together, what is the punch line of all these analyses? What do you see as the main takeaways from this paper?

 

Patricio [00:37:09] Yeah, so first, I would like to emphasize and this is something that I put in the title of the paper, this is a case study. So - but we can learn a lot of things from case studies. I believe it's clear from the paper that sometimes a considerable portion of the level of crime we observe is because of the things that potential victims do to prevent crime to happen in the first place. So to me, the main takeaways of the paper is - first, private behavior is an important, and probably usually omitted variable in understanding victimization. And second is the idea that although victims can do a lot to avoid being victimized, it may come at a high personal cost. For example, more violence exhibited by offenders. This is, I believe, important for welfare considerations. And I notice, for example, it has been recently emphasized. Phil Cook, who carefully read the paper, he sent me this paper about robbery that he recently wrote where he claims that - well, one goal is to reduce crime, but another important goal is harm reduction. So this additional dimension of welfare, I believe, should be taken into account when evaluating anticrime policies.

 

Jennifer [00:38:30] Has any other research come out since you first released this paper that adds to our understanding of the role that would-be victims play in preventing crime?

 

Patricio [00:38:38] Yeah. You know, one that caught my attention in the sense is a recent working paper by Aaron Chalfin and his colleagues at UPenn. They - in this paper, they analyze the effect of public street light outages in Chicago. By the way, I have a similar work, but looking at the threat of blackouts in Chile, which a different scale, but it's similar in terms of the story. And Chalfin and his colleagues, they find no effect on crime in the same street segment where the public street lightout happened, but they found important spillovers on nearby streets segments during this outage. So it is interesting, they claim that this suggests that crime follows a pattern of human activity. But - in other words, you may think about this - which is what I believe and I talk to Aaron about this - that's precisely what makes the case for this unexpected result, it could be driven by endogenous reaction from the victim side. So this is something - this is something that we have talk about a lot. And I believe that relates in that sense. So what is going on here is like maybe victims are reacting as well in this. So that's why - we are only observing the number of crimes, but that's the result of the interaction between them. So - and I believe that's a careful study that should be considered in the future as well because this story may connect to that.

 

Jennifer [00:40:12] What would you say are the policy implications of your results and the other work in this area? What would you tell policymakers who are listening and maybe thinking, should we leave all of this to victims to fend for themselves?

 

Patricio [00:40:24] Yeah, well, that's important because I believe in a sense, what the paper suggests is precisely caution with regard to policy that seeks to reduce crime based on increasing victim's propensity to resist. Such policy may also induce a substantial increase in the level of violence that offenders exhibit. So as I say that welfare consideration here are direct, I believe. So people do not usually think in terms of these second ordered effect when evaluating a policy. The idea that one goal is reducing crime, but another that in some extreme cases can be even more important is to reduce harm or reduce violence. So if you want to extrapolate the results, you might think about a controversial issue in the US, for example, that I believe to some extent it relates to this paper. But if, for example, the rationale for gun availability in the population, I believe that that's why this paper connects with the research done regarding the Stand Your Ground law by McClellan and Tekin or Mark Hoekstra in the Journal of Human Resources. My message would be like don't put the protection measures in people's responsibility. This should be provided by a public entity that should care about everyone here. Otherwise, you you can have unintended consequences.

 

Jennifer [00:41:59] Yeah. It also strikes me that a takeaway from both your paper and some of the other ones in this literature is the possibility that new technology will make these decisions easier. Right. So being able to switch from a cash based system to one where everyone carries around a debit card is one example. LoJacks in cars are another example. I'm also thinking about security features and cell phones. So if you steal someone's iPhone these days, it's probably password protected and you can't do anything with it. And so that dramatically reduces the incentive to steal a smartphone like that in the first place. And so it's interesting to think about how these kinds of technological innovations might help us avoid some of these problems. They eliminate the trade-offs.

 

Patricio [00:42:39] Exactly. So one thing is technology and another thing, for example, that I thought about a lot here. So, for example, let's talk about why in the old regime, in the public transportation or maybe what is going on right now in many other cities in Chile and Latin America. So why they don't implement this secure boxes. They don't implement that. They still have - they carry cash. The drivers carry cash. And part of the problem here is the lack of regulations because you may think about both the drivers and the owners of the bus have incentive to keep the cash system - cash payment system in place, even at the expense of drivers' risk. From the owners point of view, this system encourage drivers to control fare evasion, which is a common problem in public transportation. And potentially the cash in place could help drivers to increase their salary by allowing riders to ride without a ticket. So a light regulation in the entire system provides little incentive to implement this very simple and all crime prevention measures. So the final product that you have like is a persistent presence of a highly attractive crime opportunity there. And nobody or maybe nobody can do something really to fix that.

 

Jennifer [00:44:07] And what's the research frontier here? What are the next big questions in this area that you and others will be thinking about in the years ahead?

 

Patricio [00:44:14] Well, this - obviously a hard question, but I imagine that with the availability of new sources of data, we might be able very soon to start modeling at a much more precise way people's behavior. So to the extent that we can just, for example, real time geocode a cell phone data, perhaps we might be able to - this is something that I think may be interesting to look at. To decompose the effect of the police presence that has been very well-established in the literature. But you can decompose the effect in terms of induced responses on victims from potential induces responses on the offenders. So I believe this could potentially improve even more the way that we allocate police officers in the cities.

 

Patricio [00:45:08] Another challenge is how to include victims behavior more carefully in cost-benefit calculations. So presumably there is a lot of it heterogenity in terms like the precautionary measures that different people adopt. And these are costly measures. And from the welfare perspective, we should take that into account, especially if we care about inequality and we care about like the cost that different people are adopting. People - some people may feel particularly unsafe in, for example, when - and I believe my background in transportation can can testify for this - but there are some people that are much more - they don't trust, for example, in public transportation because of the level of insecurity that they feel about that. So considering the measures of the victim's behavior in the cost-benefit calculation, I think is important when we care about the welfare consequences of the new policy change or evaluating different reforms.

 

Jennifer [00:46:14] My guest today has been Patricio Domínguez from the Inter-American Development Bank. Patricia, thanks so much for doing this.

 

Patricio [00:46:21] Thank you Jen for having me.

 

Jennifer [00:46:28] You can find links to all the research we discussed today on our website, probablecausation.com. You can also subscribe to the show there or wherever you get your podcasts to make sure you don't miss a single episode. Big thanks to Emergent Ventures for supporting the show. And thanks also to our Patreon subscribers. This show is listener supported. So if you enjoy the podcast, then please consider contributing via Patreon. You can find a link on our website. Our sound engineer is Caroline Hockenbury with production assistance from Elizabeth Pancotti. Our music is by Werner and our logo is designed by Carrie Throckmorton. Thanks for listening and I'll talk to you in two weeks.

Jennifer DoleacDifference-in-Differences