Episode 31: Randi Hjalmarsson

 
Ep31_Hjalmarsson_photo.jpg

Randi Hjalmarsson

Randi Hjalmarsson is a Professor of Economics at the University of Gothenburg.

Date: June 9, 2020

Bonus segment on Professor Hjalmarsson’s career path and life as a researcher.

A transcript of this episode is available here.

Episode Details:

In this episode, we discuss Prof. Hjalmarsson's work on the how punishment severity affects juries' verdicts:

"How Punishment Severity Affects Jury Verdicts: Evidence from Two Natural Experiments" by Anna Bindler and Randi Hjalmarsson.

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 Randi Hjalmarsson. Randi is a Professor of Economics at the University of Gothenburg. Randi, welcome to the show.

 

Randi [00:00:25] Thank you for having me. It's really nice to be here.

 

Jennifer [00:00:28] Today we're going to talk about your research on how the severity of punishment for a crime affects juries' decisions about whether to convict a defendant of that crime. But before we dive into that, could you tell us about your research expertise and how you became interested in this topic?

 

Randi [00:00:45] Sure. So I'm a labor economist by training. Most people don't actually recognize that because almost all of my research is on the economics of crime and empirical legal studies. I would say my research really has two big strands of research. One is studying the causes and consequences of criminal behavior. And I live in Sweden and I do a lot of this research taking advantage of Swedish registry data to study things like what affects individuals' decisions to commit crime, the role of education, the military prisons. And I guess that's a topic for another show. And my other strand of research is looking at jury decision making. So I started this agenda really in my last year of graduate school looking for my first set of jury data. So that was almost 15 years ago. And since then, I've done a lot of papers, first focusing on how juries' characteristics affect verdicts and especially whether or not they do so differentially for different types of defendants. And then I started to get interested in whether other things that shouldn't affect verdicts do—like factors external to the case, such as the characteristics of previous cases or the expected punishment, which is where we are on this paper.

 

Jennifer [00:02:08] So this paper is titled "How Punishment Severity Affects Jury Verdicts: Evidence from Two Natural Experiments," and it's coauthored with Anna Bindler. So before this paper, what did we know about the effect of punishment severity on various criminal justice outcomes?

 

Randi [00:02:24] So, you know, in the economics of crime literature, punishment severity plays a very big role. Right? I mean, it's one of the key factors in Becker's Model of Crime- that we expect expected punishment to deter crime. And that's what most research focuses on, how punishment severity affects the behavior of criminals. And I think much less is known about how punishment severity affects the behavior of other individuals who these criminals interact with in the criminal justice system. So there was a small body of research that looks at prosecutor charge decisions and judge sentencing or pardon decisions and how these respond to punishment severity. And one of my favorite papers is looking at three strikes laws by David Bjerk that finds that prosecutors are almost twice as likely to reduce a felony arrest charge to a misdemeanor when they are constrained by a three strikes law. But there are very few other papers that I would say could causally identify the effect of punishment on criminal justice agent behaviors. And there's really no studies that do so for juries. That's not to say the question hasn't been asked. There's definitely a law literature that studies what they call the "severity leniency hypothesis." Are juries more lenient when the punishment is harsher? But a lot of this literature really lies on on mock juries. And I think they're very hard to generalize to the real world- they're not representatives, the jurors. The cases are very simple, and it's not the same stakes as what might actually happen in a courtroom. And those few observational data studies that do exist simply can't disentangle causality and correlation.

 

Jennifer [00:04:05] And then, as you mentioned, you also have a bunch of other work on juries, so what had we known about the effects of juries and jury composition more broadly?

 

Randi [00:04:14] So there's there's a, I'm happy to say, a growing literature now that looks at demographic characteristics of juries on verdicts. So I've done a lot of these papers with my coauthors, Shamena Anwar and Patrick Bayer. And the first paper that we did is looking at the effect of jury race and it's- really studying juries themselves, especially in the US, is very hard because the jury is not random. So one really has to find exogenous sources of variation to study the, you know, exogenous variation in the characteristics of the jury and not the characteristics of the chosen jury. So we do that looking at variation in the composition of day to day jury pool from which the jury is chosen. If there are no Blacks in the jury pool, there can't be any Blacks on the jury. So we we find, I think, really striking results that juries from all white jury pools convict Black defendants significantly more than white defendants. And this gap completely disappears when the jury pool has just one Black member. And we've we've used the same data, this is from felony trials in Florida, we find that older jurors are more likely to convict. In a historical context, we've looked at the introduction of of females to juries, and we we find that adding females to the jury pool from which these juries are chosen increases the conviction rate for sex offenses and increases the gap between convictions for cases with male and female victims. In Sweden, we've looked at political affiliation, which is- so I think one of the amazing things is that we find a lot of the same patterns of an unequal application of justice regardless of the justice system we look at. A system of judges that are politically representative as in Sweden, or a system that is, you know, like the US or historical England. So we find political affiliation matters.

 

Randi [00:06:09] And I think there's one other paper that I would mention that's not bias that I also really like that looks not a jury demographic characteristics, but at another kind of external factor, which is media coverage by Aurelie Ouss and Arnaud Philippe, who look at how media coverage of crime affects sentencing behavior in the short run.

 

Jennifer [00:06:29] So back to punishment severity, so the question you're interested in this paper is whether knowing that the punishment that the defendant could be facing is going to be harsher, does that affect the likelihood that the jury convicts him in the first place? So when you and Anna first started thinking about this project, what were the main challenges that that you had to overcome in order to measure the causal effects of punishment severity on jury verdicts? Are the hurdles here mostly getting the right data or is it finding good natural experiments or is it both?

 

Randi [00:07:02] So this is actually one of the unusual cases where I I had the experiment in mind and went to study the question with it. But had I come at this question from from you know, "I just want to study this question and I don't know about the reform I'm going to study," I think it's clearly a case of an identification challenge, especially when you think about the justice system today. For the most part, more severe crimes are associated with harsher punishments. And if you want to study the effect of the harsher punishment on the verdict, then you have to disentangle that from the characteristics of the offense itself that led to that harsher punishment. And I think that is very tricky to do. And this problem arises in all different kinds of research designs that you can imagine. So if you think about I have two different offenses and—kinds of offenses—and they have different sentences associated with them. And you compare the conviction rates. Well, is- do they have different conviction rates because of the different sentence or the different types of characteristics of the offense? We say, "OK, well, let's step back." Well, in the US, we have different criminal justice system in every state. So suppose we look at an identical offense in every state and they have different sentences in different states. Well, why can't we just do that and compare conviction rates across states? But there are so many other unobservables that differ across these states, even in the justice system, or who is actually going to end up sitting on the jury that we can't disentangle these different sentences from other characteristics.

 

Randi [00:08:32] So what you really want- what we really wanted is a shock to a sentence for a given offense. And because of that comparison across states, we really want that to be in a single jurisdiction or similar, very similar jurisdictions at least, to hold all these other things constant. So we need to find a large enough shock that is measurable and it also has enough observations. I don't think the hurdle is actually data, though it can take patience to find the data. I think it's more a need for, you know, the perfect storm or for everything to align that you have a good natural experiment in a jurisdiction that can provide the data and that there's enough data to have the precision to study it.

 

Jennifer [00:09:13] So you found that in historical London, so this is an economic history paper focused on the effects of changing the potential punishment for various offenses, just as you described during the late 18th and 19th centuries in London. This is a really fun paper to read because of all the interesting historical tidbits that you get along the way. I think you say in the paper that this was a particularly colorful period in London's history. So give us some historical context. What was the criminal justice system like during this period?

 

Randi [00:09:43] So it was really fun to learn about the history here. I have to say, I never thought I would say when I left graduate school that I'm going to write papers on economic history. I never even really liked history myself until I discovered the history of criminal justice. So the system at this period was incredibly dynamic and incredibly different from today. I think I'll just talk about the two aspects that are most relevant to the current paper. So how different then is it today in terms of sanctions and punishments and what trials looked like. Because that is the data that we're studying.

 

Randi [00:10:19] So sanctions, so let's start let's start in the early 1700s, right. So until until the American Revolution in 1776, there are basically two punishments used. You could be transported to Americas. And actually about 70 percent of the trials that we see in London Old Bailey result in a sentence of transportation to the Americas, or you could be sentenced to death. So 20 to 25 percent of sentences were to death. There really were zero prison sentences during this time. Prison was not a part of the criminal justice system in the 1700s. A couple of prisons did exist, but they were not used for sentences. They were used for pretrial detention basically to hold offenders until trial or post trial detention, to hold them upon a death sentence or transportation until the execution or the ship left. So this was really a harsh time of punishment, it was so harsh that it was actually called the Bloody Code, so by the early 1800s there were more than 200 capital offenses. And things that we think of as petty crimes today, like pickpocketing, were actually considered capital offenses.

 

Randi [00:11:34] So that takes us through the 1700s and the other sort of shock in the 1700s that we're going to be using today is is the American Revolution, which I think was sort of a really sort of fun experiment in what happened was that there was an unexpected shock to sentencing because, you know, the American Revolution led to the penal colony saying we're not going to hold any more of your prisoners. So there was simply a halt of transportation in 1776 and they had nowhere to send them. So what they did instead was put them in prisons or on the hulks of ships in the River Thames to clean up the river. And they didn't reestablish really transportation until a new penal colony in Australia in 1787. So we have about a 10 year period of very little, no transportation and then very little as it ramps up again. And one of the things sort of lasting effects of this shock is that this sentence of prison actually persisted after the revolution disappeared and and started to increase so that by the 1870s so by the end of our data, 80 percent of the sentences were to prison. The other big reform during this period is the abolition of capital punishment, what's happened in the early to mid 1800s. So this is also what we are studying today.

 

Randi [00:12:59] So the second aspect of the historical context that I think is important is what trials were like. So in a lot of the way, trials were similar to today in that there were actually jury pools or books of potential jurors who were called in advance to show up at the courtroom. And they even got seated on the jury in some sort of random mechanism where there was a box and they pulled out cards of like size parchment. So there was even then some randomization involved. But that's, I think for my opinion, where sort of the similarities stop. So you weren't just called to the courtroom for one trial. You were called to sit there for one to three weeks. So there was a court session that happened in a month, maybe six to eight sessions a year. You were seated on a jury and that jury tried many cases sequentially day after day. They gave the verdict for at least part of this period immediately after the case. There was no separate jury room. You sat there huddling around each other and then announced the verdict. Trials were really short. There's anecdotal evidence that there were about eight minutes long in the early 1800s on average. And there is really very little evidence presented by the defendant. It was more like the sentencing phase of a modern day trial. And this is largely because until the mid 1820s, there was a presumption of guilt. Right? So there was, you know, very little sort of "present your case" and not until we had a presumption of innocence and then the introduced- introduction of attorneys that this started to change. So because of this presumption of guilt, you know, there's very little pleading guilty, which is something that people always ask about in this paper. So that doesn't play a big role. It's only increased slowly after the presumption of guilt shifted to the burden to the prosecution.

 

Jennifer [00:14:57] OK, so you mentioned that you have two different natural experiments in this paper, so let's focus first on the offense specific abolition of capital punishment. So tell us more about how those changes came about and how they were implemented in practice.

 

Randi [00:15:13] So Sir Robert Peel was the Home Secretary at the time, and he led a really big period of criminal justice reform in the 1820s. And one of the big reforms that he did was to push through the abolition of capital punishment. So it was abolished one offense at a time. It wasn't like they say we're going to abolish capital punishment completely. And for the most part, there is to- at least from our anecdotal perspective, there's no evidence that there was any knowledge of which offense would be next or that there was a push to abolish it for different offenses with the exception of forgery. So the earliest offense was for fraud, it was abolished in 1813. And then, you know, it happens in different years. So say animal theft was in 1832. It was abolished for housebreaking in 1833, burglary and robbery in 1837. And the last one that we're studying goes all the way through arson in 1856. As it was abolished for each offense, then the sentence mostly shifted to transportation, especially in the early 1830s in those earlier reforms. The later reforms might have shifted to prison since the other thing happening at this time was actually the abolition of transportation in the 1850s.

 

Randi [00:16:38] You asked how was it implemented? So what we can look at is is the sort of the timing of implementation in the data. So if we look at all of the sentences and say aggregate them by by year and collapse the data to look at the share of of sentences for each type of offense, we can see a really sharp drop from whatever the level is to zero in the year after the reform happened. So we we identify these breaks in the data. And we also see that these correspond to the exact years in which we can find the law on the record. And there's a separate law for each offense that dates the abolition of capital punishment for that offense.

 

Jennifer [00:17:20] So you mentioned the data. I'm sure some people listening who have tried to get court data from present day and various courts maybe in their jurisdiction are wondering how on earth you're getting court data from the 17- and 1800s. So tell us about the data you are able to use in this project and what information you have and how you put it all together.

 

Randi [00:17:42] So the data, I think, is is really fun, and I can point everyone to a I think a super fun website so that you can play with this data yourselves. So the data comes from the Old Bailey and the Old Bailey is the Central Criminal Court of London and the surrounding county of Middlesex. And it trials all felonies. And don't forget, at this time, pickpocketing was a felony. So this is a lot of trials. And the data is is published in a document called The Proceedings of the Old Bailey. So this is a publication that came out after each court session starting in the late 1600s that was written by the court reporter. And they keep track of the details of the case and the outcome. So the verdict and the sentence and there's often actually a transcript as well. How long that transcript is changed a bit over time—so two hundred year periods over a long time—as the purpose of the proceedings changed. So it was initially sort of produced as as a product of public entertainment that we want to entertain the people by the gory details of these criminal cases. And then eventually it turned actually into a quasi official way of keeping records. And it keeps records for more than 200,000 trials over this two hundred year period.

 

Randi [00:19:02] And I honestly, I can't take credit for a lot of the getting the data together. There's a a group of digital historians—and I never even knew that was a term before I started this project, digital historians. And they digitized the Old Bailey proceedings online. So that's that's the website to to look at. It has a great search engine and lots of data that that you could look at, as well as a lot of institutional history. So they've tagged a lot of variables in these- each each proceeding for each case. So you can easily extract information on the date, the defendant's name, age, gender, their main offense, their verdict and the sentencing information. There's definitely some information there that's not tagged as well. And we had a lot of excellent research assistants help manually code this information that we've used for a few projects, including the judge and the jury names from 1750 to 1822. So you can actually see who the jurors were on each case. I can't see the pool that they were chosen from, but you can see the jury. And that is actually data that's hard to come by today.

 

Randi [00:20:14] So we took all of these 200,000 cases, we cleaned the data, and we coded three main outcome variables that we were interested in. Were they convicted at all? Were they convicted of a lesser offense than they were charged with? And were they acquitted? We have more than 20 offense categories that we look at and we look at these other defendant characteristics. And I'll just quickly give some, I think, interesting statistics that also tell you a little bit about the criminal justice system at this time. So, 73 percent of the cases over this two hundred year period were property offenses and 10 percent violent. Prior to the revolution, it was 88 percent were were property offenses. So this really is a system that is created by the wealthy to protect their property. I think another interesting statistic is that more than 20 percent of the data from this two hundred year period is actually female defendants. And this actually changed a lot over time. So before 1750, 36 percent were female defendants and 13 percent in the last 50 years from 1850 to 1900. It's actually this 13 percent that's very similar to what you might find in felony trial data today. So that's that's an interesting comparison. And finally, conviction rates were much lower back then than they are today, actually. 58 percent of the defendants from 1715 to 1900 were convicted of the original charge and 10 percent of a lesser charge. In some of the jury data sets that I've worked with today, I have conviction rates approaching 90 percent. So, you know, perhaps that already says something about punishment and and the propensity of a jury to convict.

 

Jennifer [00:22:02] How do you use this data to measure the causal effect of punishment severity on juries' decision to convict?

 

Randi [00:22:10] So our research design is a difference-in-difference design. And what that involves for us basically is defining each offense—we have 26 offense categories—as treated or control offense or the year of treatment, basically. And we define, in this case, treatment as the abolition of capital punishment. So for each offense, we identify whether it was always capital like murder or never capital or the year the capital punishment was abolished. And then basically what we do is compare the change in the chance of conviction in the year surrounding the abolition of capital punishment for treated offenses to those that didn't change status in that year. And I think this difference-in-difference design is really important in this context because it was such a dynamic period where there's so many other criminal justice system changes—like the introduction of the police—that that we really need this control group to difference out anything else that might have led to changes in conviction rates. And then we're really focused on the window around the abolition of capital punishment. So we're looking from 1803 to 1871.

 

Jennifer [00:23:20] So what do you find? What is the effect of abolishing capital punishment on the probability that a jury finds a person guilty?

 

Randi [00:23:30] Right. So the probability that we find them guilty, we look at that—as I said earlier, three different measures of guilt. We look at are you guilty at all? Are you guilty of the original charge, the most serious offense? Or of a lesser charge? And before sort of stating the results, I think it's really important to think about what these different measures can capture. If you're a juror and you think a punishment is too severe, what can you do? So you have the choice of acquitting completely or convicting of a lesser charge, which may have a lesser sentence associated with it. But your ability to do this really depends on the type of offense. Is there a lesser charge option available?

 

Randi [00:24:12] So what we find is that the abolition of capital punishment increased the chance of conviction of any charge by 7.6 percentage points. That's about 11 percent relative to the mean conviction rate. But this is actually hiding a lot of heterogeneity in that these results, these effects are really large for violent and sex offenses. Abolishing capital punishment increased the chance of conviction by 37 percent. And there's almost no effect, an insignificant two percent for property crimes. That's not to say there's no effect at all for property crimes, but it doesn't increase the chance of conviction of any charge. What it actually does is increase the chance of conviction of the original charge by 21 percent for property crimes and at the same time reduces the chance of conviction of a lesser charge. So what this tells us is that the juries used the acquittal option for violent crimes, mainly because I think there was no lesser offense option available. For murder, you could convict someone of manslaughter, but for most violent crimes, there's not a lesser offense equivalent. Whereas for property offenses, it wasn't so difficult to find a lesser offense because property offenses were often defined in terms of the value and there were thresholds. So depending on the year that we're looking at, the threshold might be five shillings. So I could convict you of more than five shillings or stealing less than five shillings. And they could avoid a capital sentence by doing that.

 

Jennifer [00:25:45] I suspect this result will be surprising to a lot of people because I think we're used to thinking of that decision about whether or not you're guilty as being completely separate from sort of the sentencing phase. Right. You have facts and a particular trial. And and the sentence shouldn't determine whether your, you know, the fact of whether or not you actually committed the crime or not, which is what the jury should be trying to to uncover here. And so so this really does suggest that juries are factoring all of this in at the same time and they're not actually perhaps making the decision that we think that they're tasked with making. Is that is that the right way to think about this?

 

Randi [00:26:25] I think that's how I perceive it. I also should make clear, actually, that it's not the jury who does the sentence in these cases. The judge does the sentence. So it's what they expect, right. They're they're taking into account what might happen given their actions. They don't have the power to do the sentence themselves.

 

Jennifer [00:26:44] So you mentioned that that this varied across different types of crime. Do you also see any differences across different types of offenders?

 

Randi [00:26:52] Yeah. So I think that one of the main differences is the heterogeneity in offense types. And the other heterogeneity finding that we have that is, I think most interesting, is that by defendant gender. For violent crime, the abolition increased the chance of conviction by 30 percentage points for females and just 18 percentage points for males. So the jury's decision to convict is more sensitive to the sentence for females than males. And we see a similar pattern for property crimes. So what this basically tells us is that before the reform, females were treated more favorably by the jury. There's there's a gap in conviction rates that favors females, something that we come back to in another paper.

 

Jennifer [00:27:37] And then, as you said, there's a bunch of other stuff going on during this period, so you do a bunch of robustness checks to make sure that other factors, such as the quality of evidence presented to the trial, for instance, are not driving your results. So talk a bit about what you do there and how you convince yourselves that it's really the change in punishment severity that's affecting juries decisions.

 

Randi [00:28:00] Right. So there's there's so much to get into here. I'm not sure how much detail to get into, you know, so any difference-in-difference design makes some key assumptions. And basically that assumption is that your control group is a good control group for all the other things that are changing in society at this time that could affect convict conviction rates. Do our treatment and control offenses move in a parallel manner? And the other thing that we are assuming basically is that the timing of the treatment, the timing of the abolition of capital punishment for each offense is random. So we do lots of tests to convince ourselves that these two assumptions, traditional assumptions, hold. That we see- visually inspect—we see very parallel pre-reform trends in conviction rates. In event study designs, we don't see any significant effects in the years leading up to the reform, which suggests that there's nothing happening in terms of nonrandom timing. The results are not driven by an individual offense so we can drop any offense and find the same results, even the largest offensive of larceny.

 

Randi [00:29:06] I think the more interesting sort of assumption is the question of whether something else is happening that changes the composition of cases. So is it is it that the higher conviction rates- is it because the juries make that decision or are they facing a different sample of cases as a result of the abolition of capital punishment? And as we said at the beginning, punishments- changing punishment severity can affect many different individuals in the justice system. And I think that's the main concern that did it affect the offender in a way that changed the composition of criminals that they're seeing? Does it affect your decision to plea? Are you more likely to plea if there's a weaker punishment?

 

Randi [00:29:49] So so we try to break apart each one of these channels. I'm not going to talk about all of them, just give some highlights. In terms of the offender, right, so does this decrease in punishment lead to an increase in offenses? We don't we we we look at this for each offense and we don't see any jumps in the number of offenses around the reforms. But even if it did change, what really matters is whether the quality of evidence changed. It needs to change in a way that actually increases conviction. So it would need to result in a set of cases that increased conviction. So quality of evidence needs to be higher, and that's really hard to test. That's something that we struggle a lot with in our jury data today. How do I measure the quality of evidence? So what we do in this case is take advantage of the actual text that we have in the proceedings, that there's detailed writings of the transcript of the cases and conduct keyword searches of these text for words like "evidence," "police," and "witness," things that might be proxies for the quality of evidence. And we did this for each year and defense category. So we basically create a new data set of proxy of the quality of evidence for each type of offense and year and look at whether or not the quality of evidence increased around the abolition of capital punishment. And we see no evidence of this. In fact, if anything, it's actually negative. So, this is what we do to convince ourselves that we can't really tell another story that's consistent with the findings that we have.

 

Jennifer [00:31:23] OK, let's move on to your second natural experiment, the sudden end of transportation to the Americas as a punishment. You mentioned already that transportation was used as a sentence during this period. But talk more about that. And I guess the main question that seems relevant here is how did people perceive transportation as a punishment relative to the alternatives? Was it better than death but worse than prison? What's the context here?

 

Randi [00:31:48] Yeah, so I think we need to keep in mind that transportation changed. So before the revolution, transportation was to the Americas. I should say that most sentences are for a transportation period of seven years returning before the end of that was a crime in itself. This was the main sentence. And anecdotally, I don't hear so much about this being horrible transportation to the Americas. They were already a developing society. But after this halt and 10 years later, we found a penal colony—eventually, took a while to find a new place in Australia. And this was completely undeveloped. So transportation to Australia was perceived as an incredibly harsh punishment that they often called it transportation for a life beyond the seas. And it's definitely perceived as worse than transportation to the Americas. The voyage took four to six months to get there. The conditions on the ships were pretty bad in terms of illness and death. Even upon getting there, they were put to hard labor in gangs, building infrastructure, and there was harsh discipline upon misbehaving.

 

Randi [00:33:06] How does transportation compare? So, you know, transportation to Australia is pretty bad—worse than transportation to the Americas. We assume that transportation anywhere is less severe punishment than death, but that transportation is a worse punishment than prison, which had an- even even holding conditions constant, prison sentences once they were introduced were typically shorter than seven years someplace else. So that's that's what we take to the data, that transportation is less severe punishment than death, but worse than prison.

 

Jennifer [00:33:44] And then you mentioned that there was a sudden change that made this punishment unavailable, the American Revolution. So how do you use that change to measure the effect of punishment severity?

 

Randi [00:33:58] Right. So so during the revolution, transportation went to zero. So sentences were substituted. And this is where it gets a little messy. It was substituted to both death and prison. So for non-capital offenses, you are only substituted to prison. There was no death option. So there is a clear decrease in expected punishment. But for capital offenses, there was some substitution to death and some to prison. And it's hard to say what happened to capital punishment. So we focus on non-capital offenses. And we ask what happened to conviction rates during this temporary decrease in punishment—so during the revolution—and what happens again when it gets reinstated to Australia? And actually there's like a five year period where it's reinstated in name, but there's nowhere to send people. So sort of an uncertain intensity.

 

Randi [00:34:48] But basically what we do is a before and after design. Unfortunately, we don't have a large enough sample of untreated offenses. So offenses that are not transportation eligible to use as a control group. Almost every offense was, you know- 75 percent of offenses, cases ended in transportation. So basically we capture the reduced form effect of this shock. I think it's a really nice experiment since the shock to sentencing is not driven by the criminal justice system. Right. It's not endogenous to policymakers who want to make sentences less harsh. It's driven by something completely unexpected. But it's not the perfect experiment since we cannot eliminate the effect of anything else changing due to the war. So it's the reduced form effect of this experiment. But at the same time, I think it's informative, especially when we put it next to the capital punishment experiment and find a very similar set of results.

 

Jennifer [00:35:43] Yeah, absolutely. And this is this is the kind of natural experiment where the graphs tell a lot of the story. So you look at these graphs and it's just amazing to see the transportation sentences plummet to zero in 1776.

 

Randi [00:35:59] That's why we have a lot of graphs.

 

Jennifer [00:36:00] Yeah, it is just a beautiful graph. So what did you find? Were the results of the the second analysis in line with what you'd found were the effects of ending capital punishment, back offense by offense that you talked about before?

 

Randi [00:36:13] Yeah, so it was very much the same pattern of results. So we focus on those non-capital offenses and we find that defendants were 8 percent less likely to be convicted when transportation was an option. So prior to the war and we find no effect for capital offenses.

 

Jennifer [00:36:30] And then when transportation started back up again, do you see the effects reversed?

 

Randi [00:36:35] They do in sign, so you see the pattern that you would expect, but the effect wasn't significant.

 

Jennifer [00:36:40] OK, so putting it all together, what are the policy implications of your findings?

 

Randi [00:36:46] Yeah, so for me, the first order policy implication is that policy makers are not the same individuals as those who implement policy. And the effectiveness of the criminal justice laws really depends on many other agents, ranging from police to prosecutors and juries and judges. So this success of a policymaker depends on the behaviors and actions of others to the policy. And I think this is very relevant in the criminal justice system, but also any other context where individuals may avoid actions that they perceive perhaps to result in disproportionate punishment, such as reporting cheating students, for instance.

 

Randi [00:37:32] The other implication, I think, of sort of the heterogeneity analysis is very consistent with the implications of my other demographic jury research, that there really is evidence of an unequal application of justice in this case with respect to gender. That raises questions about the fairness of the criminal justice system today, questions that we struggle with. Well, the fairness of the criminal justice system back then, and these are questions that we still struggle with today.

 

Jennifer [00:38:00] When I teach this paper, I- we usually wind up having interesting discussions about what this could mean for criminal justice reform today and the issue you just highlighted, that the people who, you know, the legislators perhaps who are implementing some policy change are not the ones who are actually implementing that change on the ground. And so that could lead to all kinds of unintended consequences that we might not expect. But I think the other really interesting takeaway for me of this paper is thinking about how if reformers are successful in this current era, dramatically reducing incarceration and reducing prison sentences and not locking people up for life or for very long periods of time, we might unexpectedly see that conviction rates increase. Right. If if everyone in the system then perceives that the punishment really isn't that bad or at least is much better than it used to be, we might wind up seeing that sort of dampening the net effect.

 

Randi [00:38:59] Absolutely. I think I think anybody who wants to study a policy reform needs to think about what is the actual- what was the end result of that policy? A change in a a sentence could end up with no effect at all in terms of expected punishment. And that's what we think, you know, if we think about deterrence, we think, how does how does expected punishment- how does the perception of punishment affect behavior? But perception is a function of not just the punishment, but the probability that you're apprehended, the probability that you're convicted, all these things weight, the actual punishment severity. So to really know what what how expected punishment changes, we need to know how that punishment changes the probabilities associated with receiving that punishment.

 

Jennifer [00:39:50] This paper was published in AEJ Policy in 2018. Are there other papers that have come out since you first wrote this paper that add to our understanding of this topic?

 

Randi [00:40:02] So Anna and I have done a lot of work on on historical London after this paper, so we just found this data set to be really fun and so many more questions that we wanted to answer. And I think the other question that we- first question that we pursued at the same time is this really is is another question of what else affects jury decisions that shouldn't. And we look at whether or not there's path dependency in jury decisions. Right, so we take advantage of this other unique feature that I described in the Old Bailey trials, where that the same jury sits for many trials in a row. And we can actually see the order of the trials and look at how the characteristics of the case before, whether or not there was a capital case and whether or not they were convicted, affects their behavior on the next case. Alright. So we see that there is actually a positive path dependency—that a previous guilty verdict increases the chance of a subsequent guilty verdict by 7 to 14 percent.

 

Randi [00:41:08] So that was the first project and the other result I mentioned in the paper was was this interesting gender gap that we found around capital punishment. So we also wrote another paper in- forthcoming in the Journal of Law and Economics that traces out this gender gap in convictions and in sentencing from 1700 to 1900, so over this two hundred year period. So we see a really persistent gender gap actually, favoring females in both conviction and sentencing that's not at all sensitive to case characteristics. And I see it at the conviction level as we studied in this paper and we also see it in the sentencing level, so by the judges. Females are less likely to be sentenced to a particular punishment as long as there's a more lenient sentencing option available.

 

Randi [00:41:57] And I would say the last project, you know, the last really exciting reform that I found that happened during this time period was the creation of the first police force. So in part using the same data, but definitely in the same time period, we look at the effect of introducing the Metropolitan Police in 1829. So lots of projects. And I suspect there are many more that people can do with these data.

 

Jennifer [00:42:19] And so what is the research frontier here, speaking of other things people might do with the data? What are the next big questions in this area that you and others will be thinking about going forward?

 

Randi [00:42:29] So when you asked me the research frontier, I thought about about not so much what more you can do with this data, but what this data- so we answered a question in a very specific context, right? We looked at punishment severity in Victorian England, where there was a huge shock to to punishment. And for me, what I really think is important is touching back on those policy implications—is is answering these questions in a more modern day context as well. How do changes in punishment affect juries today? How big did the changes need to be, you know, in order for us to find these effects? Is it only for something like capital punishment where you have these sort of jury nullification behavior? Or are there, you know, less extreme sentences that we still perceive as disproportionate, say, perhaps for juveniles? So what would be the effect of reducing sentences or mandatory minimums? And especially relevant today is not so much necessarily the gender differential, but the race differential. Is- do these gaps lead to differential treatment for Black and white defendants?

 

Jennifer [00:43:48] My guest today has been Randi Hjalmarsson from the University of Gothenburg. Randi, thanks so much for doing this.

 

Randi [00:43:54] Thank you for having me.

 

Jennifer [00:44:00] 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 was designed by Carrie Throckmorton. Thanks for listening and I'll talk to you in two weeks.

Jennifer DoleacDifference-in-Differences, Courts