The key word is “stress”

February 12, 2024 1:27 AM

Crimes rates have plummeted in the U.S. since the mid-1990s. Most of the credit for this remarkable trend has been given to an enlarged criminal justice system—largely more police, tougher sentencing and a massive prison complex. But we have found a larger and much more powerful explanation: A drop in interest rates and, in particular, long-term interest rates. When interest rates go up, crime goes up. When interest rates go down, crime goes down. from The ‘Startling’ Link Between Low Interest Rates and Low Crime [The Crime Report]
posted by chavenet (38 comments total)

A useful rule of thumb: large, complex phenomena tend to have multiple causes.
posted by ryanshepard at 3:18 AM on February 12 [36 favorites]

From 2016.

The past few years are a good test of this theory - did crime rates go up in the recent round of interest rate hikes?
posted by madcaptenor at 3:52 AM on February 12 [6 favorites]

I agree with ryanshepard. This strikes me as analogous as writing about the "startling" link between war and famine. Low interest rates are a sign of a secure, prosperous society. "Crime" is, crudely speaking, a sign of the opposite of that. "Stress" in this article seems like a conceptual kludge to obscure the fact that the authors have found a striking correlation and don't have a sufficiently rich explanation of how its pieces fit together and are part of a larger whole.
posted by sy at 4:18 AM on February 12 [8 favorites]

historically, criminals have themselves set interest rates, and their gradual disappearance has led to ineffective rate setting. except around 2022, when some criminals got into the federal reserve and jacked rates up.

this hypothesis is supported by a near perfect correlation of 0.77
posted by logicpunk at 4:30 AM on February 12 [14 favorites]

I would need a lot more graphs from a lot more countries, especially graphs which don't follow the American interest rate trend, to find this really convincing. Like... throw Japan in there, at least, and Greece.

It's not impossible that, as they propose, there's a connection between interest rates and costs of goods, and between costs of goods and crime, but for now they're showing a relationship that's weaker than any random spurious correlation you can find.

For now I'm sticking with my half-baked Scarface explanation for the 1980s crime wave. Have you ever watched a documentary about crack dealers from the '80s? They always talk about how much they were inspired by Scarface to be cold and heartless, to go down killing and dying. Economic despair likely would've led to a drug wave either way, but it could've been a much quieter, gentler drug wave if it weren't for that.

(That's obviously a dumb explanation, but cultural factors surely played a part in the whole mesh of things that were happening. It wasn't just Voelker turning the knob at the Fed that did it.)
posted by clawsoon at 4:39 AM on February 12 [6 favorites]

I think lead is a much better suspect.
posted by Thrakburzug at 4:48 AM on February 12 [15 favorites]

But is lead also responsible for the emergence of simple-minded, fascistic, right-leaning politics as the poisoned cohort ages into their 50s and beyond holding positions of power?
posted by seanmpuckett at 4:56 AM on February 12 [9 favorites]

But is lead also responsible for the emergence of simple-minded, fascistic, right-leaning politics as the poisoned cohort ages into their 50s and beyond holding positions of power?

porque no los dos?
posted by lalochezia at 5:01 AM on February 12 [19 favorites]

The past few years are a good test of this theory - did crime rates go up in the recent round of interest rate hikes?

I feel like there was some major global event that shut down society for a year and profoundly affected the economic situation of millions of people in the intervening years that might be taken into account when analyzing crime rates, but perhaps I’m misremembering.
posted by star gentle uterus at 6:03 AM on February 12 [16 favorites]

That graph reminds me of the Pastafarian theory that the decline in pirates led to global warming.
posted by leotrotsky at 6:15 AM on February 12 [5 favorites]

Everybody's saying / all across the nation / correlation, baby / it is not causation
posted by grumpybear69 at 6:42 AM on February 12 [5 favorites]

As far as I can tell we have high interest rates at the moment, but crime in my city (Baltimore) has dropped significantly in the last couple years. I think the same can be said for most cities that have struggled with high crime rates.
posted by cloeburner at 7:11 AM on February 12 [2 favorites]

I feel like there was some major global event that shut down society for a year and profoundly affected the economic situation of millions of people in the intervening years that might be taken into account when analyzing crime rates, but perhaps I’m misremembering.

Also interest rates didn’t start going up until spring 2022 once the pandemic-induced inflation had been in full swing for a while.
posted by rhymedirective at 7:14 AM on February 12 [2 favorites]

This is one of those "Well duh" things. Correlation at least stemming from the same intertwined causes.

Most people aren't evil Bond villains criming for the sake of criming. The fall into crime because they can't meet their needs in the legitimate process because of job loss, housing problems or food uncertainty.
posted by Mitheral at 7:36 AM on February 12 [1 favorite]

I thought it was the legalization of abortion that was a big factor.
posted by MonsieurPEB at 7:58 AM on February 12 [2 favorites]

MonsieurPB, that is the Donohue–Levitt hypothesis (2001), and it's controversial. Abortion rates go down when countries make it legal: report (NBC News, March 20, 2018) "Countries with the most restrictive abortion laws also have the highest rates of abortion, the study by the Guttmacher Institute found. Easier access to birth control drives down abortion rates, the report also finds."
posted by Iris Gambol at 8:20 AM on February 12 [9 favorites]

Over the past 40 years, after the end of the 1970s-early 80s stagflation and efforts to combat it - interest rates are highly linked to to perceived weakness of the job market and perceived inflation. In other words, rates are low/lowering when policy-makers are most worried about unemployment, and high/rising when policy makers are most worried about inflation.

A poissible inference from the suggested correlation is that job loss / difficulty to get jobs is not an important driver of criminality, which is a provocative finding to say the least.

It's somewhat less counterintuitive to suppose that people might resort to crime when prices are rising faster than incomes.

Something in between might be that crime is opportunistic, and the widespread prosperity that tends to be a precondition for higher rates is more inviting to criminals, because there's more to steal, and/or less risky to steal because prosperous people don't invest as much in cops to deter criminals.
posted by MattD at 8:20 AM on February 12 [1 favorite]

The idea that people can be essentially paid to stop committing domestic abuse also has to be controversial and provocative. People are probably somewhere between Bond villains and Jean Valjean
posted by Selena777 at 8:23 AM on February 12 [2 favorites]

But why did it pause in 1985? There was that drop for a couple years and then back to slavishly following the interest rates, so what caused it? Only one thing happened in 1985: the release of George Thorogood and the Destroyers album "Maverick". Similarly, the 2001-2003 slip was certainly due to the 2001 release of Stephen Hough - Saint-Saëns: Complete Works for Piano and Orchestra (Concerto). When faced with distractions like this criminals have no choice but to sit and contemplate the world around them, but does that fit into 144 characters? of course not. stupid msm. #q#logic
posted by Cris E at 8:28 AM on February 12 [8 favorites]

Iris Gambol Yes, I think the Monsieur is aware.

On a larger note, this is a remarkably silly piece, taking a single correlation and imagining all sorts of things about it. It's always worth reminding yourself of https://www.tylervigen.com/spurious-correlations before you do that.
posted by Galvanic at 8:32 AM on February 12 [2 favorites]

This is stupid as fuck and just a horrible way to do research. This is boomer clickbait level. If you're interested in the actual causes of crime rates why not talk to an expert?
posted by MisantropicPainforest at 8:45 AM on February 12 [3 favorites]

Violent Crime correlates at or better than 0.77 with:
Air Pollution in Blacksburg, Virginia, r=0.79
GMO use in corn grown in Missouri, r=0.79
Disney Movies Released, r=0.8
Nasa's budget appropriation, r=0.8
Frozen yogurt consumption r=0.96

A list of things that correlate with Violent Crime.
posted by shenkerism at 11:22 AM on February 12 [7 favorites]

Frozen yogurt consumption r=0.96


Well, duh. The toppings contain potassium benzoate.
posted by Huggiesbear at 11:32 AM on February 12 [6 favorites]

(That’s bad)
posted by Huggiesbear at 11:36 AM on February 12 [9 favorites]

I believe in the lead hypothesis. In DC right after 9-11 they change the water treatment process because of concern that the giant chlorine tanks were a big terrorist target. The change caused lead to leech out of the pipes in the poorer parts of DC. Now 20 years later we've seen a surge in violent crime, mostly committed by people who would have been at highest risk for exposure. The drop in crime in the 1990s correlates to 20 years after we took lead out of gas and paint.
posted by interogative mood at 12:16 PM on February 12 [2 favorites]

The past few years are a good test of this theory - did crime rates go up in the recent round of interest rate hikes?

Relevant:

"At some point in 2022 — at the end of 2022 or through 2023 — there was just a tipping point where violence started to fall and it just continued to fall," said Jeff Asher, a crime analyst and co-founder of AH Datalytics.

In cities big and small, from both coasts, violence has dropped.

"The national picture shows that murder is falling. We have data from over 200 cities showing a 12.2% decline ... in 2023 relative to 2022," Asher said, citing his own analysis of public data. He found instances of rape, robbery and aggravated assault were all down too.

posted by clawsoon at 1:34 PM on February 12 [2 favorites]

I'm not sure why an article from 2016 is being posted here, but James Austin is a well-known and respected researcher in the field. If you want more recent work from him, there's plenty on his website, including updates with more recent data and analysis of COVID and crime rates.

Forecasting US Crime Rates and the Impact of Reductions in Imprisonment: 1960–2025


SJC - Impact of COVID Post Pandemic and Beyond March 2023
posted by gingerbeer at 1:36 PM on February 12 [3 favorites]

Related, I have no idea why the article linked in the post has someone else's name as the author? Here, on Austin's website, is the original article.

At any rate, if you're actually interested in the underlying data and methodology, there's plenty in his reports to dig into. It's neither click bait nor spurious correlations. I don't agree with everything Austin says, but he's a legitimate researcher on the topic.
posted by gingerbeer at 2:07 PM on February 12 [2 favorites]

James Austin is a well-known and respected researcher in the field

Wow. Because that’s the work of a hack who makes Leavitt look respectable. Is criminology really that bankrupt?
posted by Galvanic at 4:05 PM on February 12 [2 favorites]

I don’t care who wrote it, and I’m making no claims about the credibility of the author. It’s shit research.
posted by MisantropicPainforest at 6:07 PM on February 12 [2 favorites]

Would you unpack that some? This isn't my field, but I'm in an adjacent one, and have used his research before. I'd prefer not to do that if it's not sound! I also know little about the methodology for forecasting and would like to know more. When you read through the methodology section in this report, for example, what are the flags that stand out to you as "shit research" or the work of a hack?
posted by gingerbeer at 9:32 AM on February 15

The author of the link in the OP is just plotting two time series and saying one is driving the other. That's it. There's no smoking gun to tell if its hacky or bad, any more than flipping a coin and using the result to tell you if its hot or cold outside. Its 'not even wrong', just nonsensical.
posted by MisantropicPainforest at 10:07 AM on February 15 [1 favorite]

I think I'm asking about the underlying research that he's done, not the very short summary in the 2016 paper. When the methodology is described as below, I don't have enough expertise in this kind of forecasting to say why the prior research connecting the 2 time series isn't valid.

"We experimented with a large number of models
containing a varying mix of demographic, socioeconomic, and criminal justice variables before
settling on a model that contains only the past year’s imprisonment rate and the current year’s
inflation rate, adjusted by median household income. Prior research has shown that each of these
variables is associated with changes in crime rates over time, and the logic for including them in
our model is fairly straightforward. Increases in the imprisonment rate are expected to reduce
crime on the assumption that punishment incapacitates offenders and deters criminal behavior.
The magnitude of the effect of imprisonment on crime varies widely across studies, however, and some
studies indicate that it weakens at high levels of imprisonment (National Research Council 2014).

Prior research indicates that inflation has strong and consistent effects on crime committed for monetary gain; as retail prices increase, so does the demand for cheaper stolen goods (Rosenfeld and Levin
2016). Inflation is also expected to contribute to both violent and property crime by reducing confidence in government and other institutions (LaFree 1999). Crime rates, especially the property rate,
should vary with purchasing power, which is the rationale for adjusting the inflation rate by median
income (inflation/median household income). "
posted by gingerbeer at 3:04 PM on February 15

The short answer is that this is the kind of research published by social scientists in the 90s that were operating in disciplines without rigorous standards for quantitative research.
posted by MisantropicPainforest at 3:30 PM on February 15

"We experimented with a large number of models containing a varying mix of demographic, socioeconomic, and criminal justice variables before settling on a model"

That approach puts me in mind of the infamous blog post that wrecked the careeer of Brian Wansink (previouslies). For reasons of practicality, I guess some fields are forced to have even lower standards for research design than the psychology of eating behaviour.

(For other reasons that I don't understand, those fields always seem to write their papers sounding like they have complete confidence in their necessarily-very-weakly-supported-because-the-human-world-is-complex conclusions.)

"Prior research indicates that inflation has strong and consistent effects on crime committed for monetary gain; as retail prices increase, so does the demand for cheaper stolen goods"

This is where the economics of it gets weird: Interest rates are raised in order to drive inflation down, so if high inflation leads to high crime we should expect high interest rates to lead to low crime. But that's exactly the opposite of the correlation they found.
posted by clawsoon at 8:42 AM on February 16

“Illusions of Safety,” Mariame Kaba, The Baffler, 13 February 2024
posted by ob1quixote at 9:41 AM on February 16

Would you unpack that some? This isn't my field, but I'm in an adjacent one, and have used his research before. I'd prefer not to do that if it's not sound! I also know little about the methodology for forecasting and would like to know more. When you read through the methodology section in this report, for example, what are the flags that stand out to you as "shit research" or the work of a hack?

Having read your link, I'm not sure I even saw a methodology section for deriving their models, just one for evaluating them. The two models on offer are ARIMA and a hand derived model using "twelve major correlates of crime rates." The paper includes the phrase "interest rate" exactly once, with a presumption that it is positively associated.

When the methodology is described as below, I don't have enough expertise in this kind of forecasting to say why the prior research connecting the 2 time series isn't valid.

Well one simple test one could apply that's been proposed upthread: redo the experiment adding in the past 8 years of data since the 2016 study. Unfortunately, while your citation does that, it the ARIMA model doesn't use the interest rate, just inflation and imprisonment. And the authors are careful to point out that imprisonment has reverse causality and is best correlated as a lag; one wonders if it "cheats" a bit then on the eval set.

Beyond that, I figure in this case, both crime and interest rates are part of a network of economic factors including unemployment, inflation, investment in education, air pollution, increased policing and surveillance, and changing demographics. So I would find a multi-level hierarchical model far more compelling, personally. And informative.

Last point: forecasting and causality are not exactly the same. While knowing causal links is typically useful in forecasting, it's not a requirement. Confounding variables can work just fine for forecasting real world outcomes, but may not accurately describe what happens if we intervene. Austin agrees, BTW: "Nobody would suggest that high interest rates directly cause crime.”
posted by pwnguin at 8:56 PM on February 24

Side question: who wrote this article? The byline at the top says "Stephen Handelman" presently but the archive.org version from 2017 lists "James Austin and Gregory D. Squires."
posted by pwnguin at 8:59 PM on February 24