Around 24 years ago, crime rates in the US started to mysteriously decline. Since then, crime has fallen by roughly half, with violent crimes plummeting by roughly 51 percent and property crimes decreasing by about 43 percent.
Almost as soon as the crime decline began, criminologists started trying to figure out the reasons. Different theories have come in and out of vogue over the past couple of decades. Some theories, such as mass incarceration, seemed very sound in the 1990s, but have been called into question as more data has come in. Some surprising theories, such as one related to the decreasing lead content in gasoline, have recently gained momentum.
In a massive report released in February 2015, the Brennan Center for Justice looked at more than a dozen of these explanations and ran statistical analyses to show how much of the crime decline can be attributed to each one of them. Their conclusion is that a lot of things had some effect on crime in the 1990s — but there was no smoking gun.
But even the Brennan report, by its own admission, only tells part of the story. It didn’t account for some additional theories touted by criminal justice experts, such as the idea that technology keeps people indoors. So we rounded up some of the Brennan Center’s ideas, plus a few they didn’t mention, to analyze 16 popular theories for the plummeting crime rate.
1) More criminals are getting put in prison
The case for: It seems intuitive. The incarceration rate’s been rising; the crime rate’s been falling. Surely this is because people are being locked up who’d otherwise be committing crimes out on the streets.
Several academic studies have found that increased incarceration had a big impact on reducing crime. In particular, Steven Levitt (of Freakonomics fame) wrote a paper in 2004 that concluded that 58 percent of the drop in violent crime during the 1990s was due to incarceration.
The case against: These studies were based on older data that only included a few years of the crime decline. Levitt acknowledged he couldn’t account for the point of diminishing returns: there are only so many serious criminals out there, and after a certain point the people getting put in prison aren’t people who’d be committing crime after crime on the street. The higher the incarceration rate gets, the less it matters if you increase that rate even more. Studies that examine more recent data, after the point of diminishing returns has been hit, find that incarceration wasn’t nearly as influential.
“Incarcerating violent people has a big effect on violence,” John Roman, senior fellow at the Urban Institute’s Justice Policy Center, said. “But most people we incarcerate aren’t violent.”
The diminishing returns aren’t only about who’s being put in prison, but how long people remain there. The research suggests that people age out of crime, so letting them out of prison 10 or 20 years down the line — instead of the longer sentences applied today — might not pose a threat to public safety. “Crime is a young man’s endeavor,” Brian Elderbroom, senior fellow at the Urban Institute’s Justice Policy Center, said in December. “It’s not surprising that someone who commits a crime at a young age would be a completely different person by the time they’re in their 30s.”
The other problem with this theory is that incarceration rates were increasing for years before crime started going down.
The bottom line: A small effect. Criminologists now tend to believe that incarceration accounts for a fraction of the drop in crime, but no more. The Brennan Center report estimates that incarceration played even less of a role than that: up to 12 percent of the drop in property crime during the 1990s was due to the rise in incarceration, but it was probably more like 6 percent. And it contributed to 1 percent, at most, of the continued property crime decline in the 2000s. Furthermore, the Brennan Center concluded that the rising incarceration rates through the 1980s had already locked up the truly violent criminals, and the point of diminishing returns was hit even before the crime rate started to fall.
2) More police are on the streets
The case for: More police make it easier for law enforcement to respond to crimes and apprehend criminals. If they’re visibly patrolling, they can also help prevent crime by deterring would-be wrongdoers. Research on specific areas, as well as the US as a whole, found that hiring more police helped decrease crime.
The case against: But while the number of police can affect crime rates, crime rates also affect the number of police. When crime rises, cities hire more police in response. It’s extremely tricky to isolate the effect of police on crime from the effect of crime on police.
The bottom line: A small effect. Steven Levitt, in a 1997 paper later revised in 2002, tried to fix this problem by looking at the timing of local and state elections. The logic is that an upcoming election would have an effect on police hiring, but wouldn’t affect crime on its own — so it could show cases where more police did have an impact on crime. Brennan Center researchers evaluated his findings and others to show that up to 10 percent of the drop in crime in the 1990s was caused by hiring more police. As with incarceration, though, it looks like hiring more police has more of an effect on reducing property crime, like robberies, than violent crimes.
3) Broken-windows policing
The case for: Three words: New York City. New York implemented a host of policing reforms in the early 1990s under Mayor Rudy Giuliani, but the most famous was “broken-windows policing” — an emphasis on enforcing minor “quality-of-life” offenses to make neighborhoods feel safer and encourage residents to feel pride (as well as deter people from committing more serious crimes). And sure enough, crime declined precipitously in New York during the Giuliani era.
The link between broken windows policing and the crime decline has taken a beating among criminologists in recent years — so much so that two of its early champions (one of them current NYPD Commissioner Bill Bratton) recently wrote a defense of it. The essay says that other theories use “macro data sets” and “fail to grasp how crime is managed in dense, urban settings.” The implication: you can only tell that broken-windows policing works if you saw it up close.
The case against: There is a reason that this explanation has fallen out of favor. First of all, crime declined in plenty of places that weren’t New York City during the 1990s — meaning New York’s trends probably weren’t totally the result of the NYPD. For another thing, broken-windows policing was just one aspect of the policing reforms implemented in New York or other places during this time; crime-tracking system CompStat, for example, also came online in the mid-1990s. How can you separate how much of the crime drop was due to one thing, or to another introduced at the same time?
The bottom line: Too difficult to tell. Ultimately, different departments define “broken-windows policing” differently and implement it in different ways — and, again, often alongside other changes. It’s true it’s hard to tell why crime declines in cities, but that applies to broken-windows policing as much as it applies to other macro explanations.
4) Police have improved policing in other ways
The case for: CompStat, the crime tracking system that gave rise to “data-driven policing,” was popularized in the 1990s — and many cities that started using it saw serious results. CompStat’s usually associated with high-tech, detailed data collection on crime trends. But most of the time, the data collection is paired with a management system, where information and trends are discussed in meetings, divisions quickly try new tactics to address crime as they come up, and police are held accountable for keeping crime down. CompStat is still used in many, if not most, police departments, and it’s been widely praised by police executives. (It’s also been associated with stats-juking scandals.)
Some research also suggests that certain police tactics do decrease crime. Hot-spot policing — which directs police not to focus on individual people, but instead to be present and visible in the places where crime is most likely to occur — has solid empirical support. And studies suggest that in some contexts, “community policing” — focusing on partnerships with community members first, and enforcing the laws against people second — can have positive results.
George Tita, a criminologist at the University of California in Irvine, argued that police’s ability to better target certain neighborhoods and offenders has built a deterrent to violence. “I think because police are more effective in what they do,” Tita said, “the probability of seeing some action taken after your group is involved in violence is much higher than it was 20 years ago.”
The case against: In the early 1990s, before crime started to decline, many criminologists had given up on the idea that what police did had any impact on crime whatsoever — at least as far as preventing it was concerned. Police tactics have changed between then and now, but that doesn’t mean they caused the crime decline. It’s fully possible that what look like improvements in policing from this perspective could have just been phenomena that happened alongside a decline in crime for other reasons. Add to that the same problems that you have in analyzing broken-windows policing — that police tactics are rarely implemented the same way in two different places — and it becomes a hard case to make.
“I hate to say it, but I don’t think it matters,” Roman of the Urban Institute said, arguing that different cities used very different policing tactics even as all of them saw drops in crime. “There’s a lot of correlation, but not much causation.”
The bottom line: CompStat helped; with other tactics, it’s less clear. The Brennan Center’s report examines crime rates in 50 American cities before and after they started using CompStat. In all, it says CompStat played a small part in the crime decline since the early 1990s. That’s a huge difference from the assumption that police have no impact on crime at all.
It’s harder to analyze other changes in policing tactics, like hot-spot policing and community policing, for the same reasons it’s hard to analyze broken-windows policing. Hot spot policing, in particular, has a relatively large body of research supporting it (for criminological research on police tactics, at least) — but it’s not enough to assess its impact nationally.
5) More permissive gun laws
The case for: Here’s the logic: if a would-be robber thinks there’s a good chance that the person he wants to rob has a gun and will shoot him, he might think twice about the robbery. And many states passed right-to-carry laws in the mid- to late 1990s, just as crime started to fall.
The “more guns, less crime” hypothesis is associated with economist John Lott, who wrote a 1997 study that showed dramatic results. The paper concluded that enacting a right-to-carry law in every state would prevent nearly 1,600 murders a year.
The case against: Lott’s paper is very influential, but it’s also become increasingly suspect. Two later papers looking at Lott’s data but adding more, or slightly tweaking the model he used in his analysis, have found that it’s really hard to say that guns have had as much of an effect on crime as Lott suggests. And Lott’s done a terrible job of defending his integrity — at one point, when scholars started asking about a survey Lott had supposedly conducted but lost the data for, Lott created a sock-puppet persona named Mary Rosh to attack his critics online.
Meanwhile, other research has suggested that, if anything, having more permissive gun laws — or increasing gun ownership, more generally — might increase crime levels.
The bottom line: No effect. Most researchers studying the link between gun laws and crime have found no reason to think that guns deserve credit for the crime decline.
6) Improvements in the economy
The case for: When people don’t have jobs or aren’t making money, crime might look like a profitable alternative. Research has shown that increases in unemployment, for example, lead to small increases in property crime. And some criminologists are enthusiastic about the correlation between inflation and rising crime levels.
The case against: The big problem with this argument: the recession of the late 2000s. Unemployment was higher in 2009 and 2010 than it was in the crime-wave 1980s and early 1990s. But crime didn’t spike — instead, surprising many, it continued to decline.
The bottom line: Still unclear. It depends on what measure of economic well-being we’re talking about. The Brennan Center, in line with past research on the subject, finds that rising income reduced crime since the 1990s — accounting for up to 10 percent of the crime drop. Unemployment matters a little less (up to 5 percent). Reduced inflation, meanwhile, might have had a slight effect on property crime — but it probably didn’t affect violent crime.
One interesting question: how much does it matter that the economy is actually improving, as opposed to people simply feeling that the economy is improving? There’s some suggestive evidence that rising “consumer confidence” (a statistic based on surveying people about the economy) brought crime down not only in the 1990s, but the 2000s as well.
7) People are using less cash
The case for: If a criminal sees someone with cash, he’s probably more likely to attempt a robbery than if the person is carrying a credit card. This makes sense: real dollars are anonymous and untraceable, while a credit card can be cut off with one phone call.
Surveys show Americans are using less cash than they did before, as people use debit cards, credit cards, and other electronic payments more often. One study found this could have an impact on crime: researchers at the University of Missouri at St. Louis and Georgia State University found that Missouri counties that moved to electronic welfare benefits saw bigger crime drops than those that stuck with cash.
The case against: As with many of the exotic theories about the crime drop, there’s just not enough empirical evidence showing people moving away from cash had a significant role.
The bottom line: Still unclear. Richard Rosenfeld, a University of Missouri in St. Louis criminologist who was involved in the welfare study, said there needs to be more evidence. In particular, he’d like to see the results replicated outside of Missouri.
8) Technology is keeping people inside, not out committing crimes
The case for: With the advent of new technologies like the internet, cellphones, and video games, children and teens are able to better entertain themselves indoors.
“What I think video games, the internet, and what-have-you did is take juveniles off the streets and put them into homes,” Tita of the University of California in Irvine said. “What happens when you don’t have kids hanging out in the corner? You remove the potential for victims and offenders on the street.”
Roman of the Urban Institute argued that violent video games in particular may let people get the “rush” of acting out without actually getting into violent situations. “If you talk to the person on the street, they’ll say of course [violent video games] make people more violent,” Roman said. “If you talk to a sociologist, they’ll say [people] will substitute the game from acting it out in reality.”
Technological advancements like the cellphone and online messaging have also made some illegal activities safer, avoiding encounters that could have easily turned violent in the past. Instead of going to dangerous open-air drug markets, drug users can now call their dealers and meet them somewhere safe, maybe even in their own home.
The case against: While researchers, particularly Tita, claim to have seen this kind of effect in the field, they acknowledge there’s just not good research in this area.
The bottom line: Still unclear. Experts say the theory is plausible, but they can’t be too sure without the proper research.
9) Gentrification is taking over crime-ridden neighborhoods
The case for: For much of the 1980s, many inner-city areas were hubs of violent activity with barely any job opportunities. But in recent years, upwardly mobile immigrants and wealthier Americans have moved back into these neighborhoods, bringing jobs and economic prosperity with them.
The key, Roman of the Urban Institute said, is these immigrants and gentrifiers didn’t push older residents out of inner-city neighborhoods. Instead, they all integrated into a more prosperous environment. This is important not just because it offers potential economic opportunities to longtime residents, but also because it didn’t push the poor into other neighborhoods, where economic hardship and violent crime may have taken root once again. (There’s some debate about whether this is true for all places witnessing the effects of gentrification, such as Washington, DC, and Prince George’s County.)
Roman pointed to research that shows economically and racially segregated areas tend to have a lot more problems with crime. He cited the differences in crime in New York City, which has seen a lot of gentrification in recent years, and Chicago, which remains one of the most economically and racially segregated major cities in America. While New York City has seen homicides drop dramatically over the past few decades, Chicago’s homicide rate has seen a smaller decline.
The case against: While the research shows that gentrification isn’t anywhere near as harmful or displacing as some critics argue, it also indicates that gentrification is very rare. If gentrification is helping push down crime in inner-city neighborhoods, it’s not happening everywhere — or even in very many cities.
As Roman acknowledged, it’s hard to know, from a research perspective, how much of an impact gentrification has had on the crime rate. “It’s very difficult to test this all empirically,” he said.
The bottom line: Still unclear. Roman’s hypothesis certainly makes intuitive sense, and it’s backed by data from segregated and gentrified neighborhoods. But there’s very little research showing the direct impact of gentrification on crime, making it hard to gauge just how much of an effect is really there.
10) Alcohol consumption has declined
The case for: The correlation is there: alcohol consumption per capita has been declining since 1977, and hit particularly low levels during the 1990s and 2000s. (It’s gone up slightly since then.) And alcohol abuse is definitely correlated with crime: according to one estimate, 40 percent of violent criminals in state prison were under the influence of alcohol when they committed their crimes.
The case against: Social scientists typically use beer consumption as a proxy for alcohol consumption generally. But while beer consumption has kept declining through the late 2000s, wine and liquor consumption has increased. So the existing research might be getting less useful at measuring how much alcohol people are consuming, which could skew analyses of its effect on crime.
The bottom line: Some effect. The Brennan Center analysis, as well as past research, shows that there’s a strong association between alcohol consumption and crime. But in the specific case of the crime decline of the 1990s and 2000s, alcohol use simply didn’t decline enough to explain that much of the drop. In all, the Brennan Center estimates, less alcohol consumption explains about 5 to 10 percent of the drop in crime in both the 1990s and the 2000s.
11) Psychiatric pills reduced violent behaviors
The case for: Researchers David Finkelhor and Lisa Jones argued that psychiatric medications, such as antidepressants and anti-ADHD drugs, helped decrease violent acts against and by children by improving people’s moods and behavior.
The case against: Rosenfeld of the University of Missouri in St. Louis calls this theory plausible, but he said there’s no significant research to support it yet. That doesn’t mean these medications didn’t contribute to the drop, but there’s no evidence at the moment — beyond simple correlation — that they did.
The bottom line: Still unclear. Rosenfeld said the issue needs research, although he wouldn’t put it at the top of his priority list.
12) Crack consumption has declined
The case for: Crack consumption and sales definitely fueled crime and disorder in the 1980s. While other “drug epidemics” have succeeded crack, it’s possible that none of them have been as destructive — not least because crack is so cheap to make and sell.
“The popular myth is that it’s all addicts committing crimes to support their habits,” Roman said. “There’s absolutely some of that, … but a much bigger part of the story is the drug-selling network — and the crime that went between the network and the gangs, and the crime that went around the sales.”
The case against: Sure, it’s possible. But cracking down on crack didn’t stop illegal drug consumption or sales. It’s called the balloon effect: cracking down on drug trafficking in one area merely shifts it to other places, since demand among users and sellers for the substances never really declines.
The bottom line: The data is incomplete. Surprisingly, this is one of the harder theories to study — because most of the best data on drug use was collected as a response to the crack epidemic, it doesn’t cover the worst years of it.
13) America’s gangs have gotten less violent
The case for: The thinking is simple: gangs have realized that shootings are bad for them. Knowing that police are much more likely to come down on their neighborhoods or organizations if they take part in a shooting, gang members have tried to avoid violence.
“The violence is simply bad for business,” Tita of the University of California in Irvine said. “If one of their friends is part of a shooting, it’s going to make it tougher for them to put money in their pocket.”
The case against: There isn’t good research on whether gangs are getting less violent overall and how much of an impact that’s having on crime. Tita acknowledges this: “No studies have been carefully constructed to give me 100 percent confidence or even 90 percent confidence that any one of these things I’m telling you is true.”
The bottom line: Still unclear. Intuitively, the idea makes sense. But without adequate research, it’s hard to know just how accurate it is.
14) The population’s just aging out of crime
The case for: There’s certainly a correlation between the decline in crime and the increase in the median age of Americans. The crime wave hit when the baby boom was in its teens, twenties, and thirties — when people are generally more likely to commit crimes.
The case against: The implication of this theory is that crime rates among people in a particular age group didn’t change that much — that there was less difference between the rate at which 20-year-olds committed crimes in 1990 and 2010 than there was in thenumber of 20-year-olds at that point. But that’s not really what the evidence shows. During the crime wave, crime rates for each particular age group rose; after the early 1990s, crime rates for each particular age group fell.
The bottom line: A small effect. The Brennan Center’s analysis suggests that the number of 20- to 30-year-olds in the population does have a small impact on crime (though not the number of teens). But we’re talking about a difference of a few percentage points in the crime drop during the 1990s — and nothing during the 2000s, when there weren’t big demographic changes in age.
15) Legal abortion is preventing would-be criminals from being born
The case for: If you’ve heard of Freakonomics, you know this one. In 1994, when the crime rate started to drop, a child born in 1973 was 21 — and a child who wasn’t born in 1973, because the mother got a newly legal abortion after Roe v. Wade, was one fewer 21-year-old in the population. It’s a very appealing argument to people who hate obvious answers. It’s also, interestingly, one of the few theories on this list with strong empirical support from other countries: studies as far away as Romania have found the same effect.
The case against: There are two big problems with the abortion theory. One of them is that abortions didn’t suddenly go from 0 to 60 when abortion was legalized. Before Roe v. Wade, illegal abortions still happened; after the decision, there were still plenty of people who chose not to get them. There was also a supply problem: “It wasn’t like Roe v. Wade was decided and suddenly there were a million places to get an abortion,” Roman of the Urban Institute said.
The second problem is age. In the 1990s, the crime rate didn’t just go down among people born in 1973 or later. It went down among a bunch of age groups at once. There’s evidence that youth crime rates influence the crime rates of older adults, but that doesn’t mean that the people who turned 21 in 1994 were powerful enough to suppress crime among people a decade older.
This theory also makes assumptions about abortion that aren’t necessarily true. The unspoken assumption of the abortion theory is that abortions necessarily prevent unwanted children, and unwanted children necessarily commit more crime. Those aren’t proven assumptions.
The bottom line: Probably some effect in the 1990s. Like many of these theories, there’s empirical support that abortion reduced crime to some extent in the 1990s. (Because there isn’t detailed-enough data on abortions, the Brennan Center couldn’t quantify exactly how much.) But crime continued to decline in the 2000s, after the Roe v. Wade generation was out of prime criminal age — making it unlikely that abortion explains why crime continued to drop through the first 15 years of the 21st century.
16) Lead in gasoline caused criminality, and now it’s been reduced
The case for: This is another newly popular theory, thanks in part to coverage fromKevin Drum at Mother Jones and others. Just like the timing of Roe v. Wade, the timing of the removal of leaded gasoline from America’s filling stations, which decreased lead exposure among children born around and after 1975, correlates strongly to the cohort of children who hit peak criminal age around the mid-1990s. There’s also a body of psychological research tying lead exposure to more aggressive behavior.
The case against: The lead theory has the same problem as the abortion theory: in the 1990s, even people who had been exposed to lead as children started committing fewer crimes. That indicates that while lead exposure may well have been a factor, it isn’t the “real answer” that it’s often characterized as.
The bottom line: Probably some effect in the 1990s. Again, the Brennan Center didn’t have enough data to quantify lead’s impact, but past research makes a good case that it had some effect. It just probably doesn’t explain the majority of the crime decline in the 1990s. And in the 2000s, when everyone under 30 had grown up exposed to less lead, the lead theory helps even less.