speed cameras: LEGACY


This is the third and probably last installment of my speed camera saga. Part 1 is here. Part 2 is here.

One of the nicest things about writing something that connects with people is that many of them share links and thoughts that make you smarter (and would have improved the piece if you’d had them in the first place). I want to briefly discuss four responses I got: two studies that people pointed me toward, some stats that I wasn’t aware of, and one reactive essay.

Red-Light and Speed Cameras: Analyzing the Equity and Efficacy of Chicago’s Automated Camera Enforcement Program

This study takes a deep look at Chicago ATE data. It found that cameras improve safety:

Over the 3-year period from 2015-2017, we estimate that there were 36 fewer KA type injury crashes, 68 fewer type B crashes, and 100 fewer type C crashes across the 101 locations. In all, there were 204 fewer injury crashes. Reductions of type A and C crashes were estimated at around 15% and that for type B injuries at 9%. Overall, speed cameras led to a 12% reduction in injury crashes.

But the study is most notable because, unlike my analysis of DC data, the study’s authors had access to the zip code of citation recipients. This allowed them to examine the incidence of tickets versus Census demographics in a much more defensible manner than the camera/neighborhood spatial association that I criticized the DCPC study for using.

After controlling for various factors, they found that Black households do receive a higher number of citations than white or hispanic households. They did not find disparate placement of cameras, however.

[T]ickets per household for both speed and red-light cameras are higher in majority Black areas, followed by majority Hispanic/Latinx areas, and finally majority White/Other areas. At the camera level, however, we do not find such relationships. Cameras in majority Hispanic/Latinx areas tend to issue fewer tickets than others for both red-light and speed cameras. There is not a statistical difference in ticketing rates between cameras installed majority Black and majority White/Other areas for red-light cameras, and there is weak evidence that rates of ticketing by cameras in majority White/Other areas are lower than those in majority Black areas for speed cameras.

They also found that cameras on big, fast roads issue a disproportionate number of citations, which is consistent with my own findings from examining DC data.

The second part of the study examines these citations’ economic impact in terms of different groups’ level of wealth. Late fees and payment rates emerge as a significant part of the picture.

I think this study provides real evidence of a disparate impact, but doesn’t provide a clear explanation of why it’s occurring. It’s also important to keep the actual scale of the effect in mind: as you can see on the graph above, the per-household gap is about one speeding ticket every ten years. That deserves attention, but should also be kept in perspective.

Do Speed Cameras Save Lives?

I was also pointed toward this paper, by the Spatial Economics Research Centre, which examines cameras in the UK. I appreciate the friendly manner in which it was shared with me by someone who I take to be an ATE skeptic, but I think it helps his case less than he imagined. This study also found that cameras improve safety:

[S]peed cameras unambiguously reduce both the counts and severity of collisions. After installing a camera, the number of accidents and minor injuries fell by 17%-39% and 17%-38%, which amounts to 0.89-2.36 and 1.19-2.87 per kilometre. As for seriousness of the crashes, the number of fatalities and serious injuries decrease by 0.08-0.19 and 0.25-0.58 per kilometre compared to pre-installation levels, which represents a drop of 58%-68% and 28%-55% respectively. Putting these estimates into perspective, installing another 1,000 speed cameras reduce around 1,130 collisions, mitigate 330 serious injuries, and save 190 lives annually, generating benefits of around £21 million.

Rather than confirming that cameras on highways generate outsize numbers of citations, it found that cameras ought to be placed on highways, because that’s where their safety benefits will be greatest:

[I]t is more effective to install cameras along roads at higher speed limits as much larger reductions in collision outcomes are observed

Finally, the study found mild evidence of “rebound” effect outside of camera locations. I think this is why the study was shared with me: my reply-guy was arguing that cameras just push crashes around. I don’t buy that argument, and it doesn’t seem like the paper’s author does, either. Or at least he thinks the problem could be solved by–you guessed it–more cameras:

Beyond 1.5 kilometres from the camera, there are suggestive evidence of a rebound in collisions, injuries and deaths, indicating drivers could have speed up beyond camera surveillance and cause more accidents. These results, which illustrate the limitations associated with speed cameras, suggest that newer prototypes, such as mobile or variable speed cameras, should be considered.

Demographic Safety Data

Eileen S. made me aware that some safety data exists with a race-ethnicity breakdown, albeit with a notable time lag. Eventually I poked around and realized there are quite a few comparable resources for examining the question.

First, consider NHTSA’s data on fatalities per 100,000 people, which shows that the Black community is suffering from our deadly roads at a rate second only to people of native heritage.

NHTSA also provides stats breaking down the percentage of traffic fatalities that are related to speeding:

All of this is looks even worse when you consider different groups’ urban/rural split:

and by academic estimates of demographic differences in vehicle miles traveled (VMT):

(See also here for a longer NHTSA report on this topic)

I would expect lower VMT, tilted toward urban areas, to mean a lower incidence of speeding-related deaths. But that is not what the data shows.

I don’t know why this disparity exists. My hunch is that it has something to do with the kinds of built environments that disadvantaged groups have to settle for. I am quite wary of bringing it up, because of the risk that a reader will mistake it for an argument about blame. That is not my intention: it’s an argument about victimization and suffering. I think it’s essential context for anyone who wants to discuss the possibility of ATE’s disparate impact.

@DCCarViolence’s essay

I want to thank Joseph Oschrin for taking the time to write this response. I appreciate the work he does on Twitter, too. The crux of his argument is that ATE–and in particular, debating the disparate impact of ATE–is an unfortunate waste of time, and that instead we ought to focus on road diets and other interventions to our infrastructure.

I agree that those kinds of changes are the most desirable way to make streets safer. The problem is that they are wildly expensive–not just in dollars and cents, but in years spent planning, compromising, and–as this week has reminded us–suffering abrupt reversals.

Oschrin maintains that cameras are consuming resources that could be spent on other kinds of interventions. I have a hard time seeing the evidence for this. The bottlenecks to reforming our infrastructure are big problems of budgets and politics. A speed camera program, by contrast, seems to mostly require exchanging emails with a vendor.

Cameras are proven to save lives. They pay for themselves. They don’t consume human enforcement resources. And they don’t engage in racial profiling.

The only problem with cameras is that everyone hates them and indulges in motivated reasoning to justify that distaste. I sometimes imagine how an ATE option might work in SimCity: click here to trade popularity for income and safety!

Maybe not the most fun game mechanic. But I think it’s a great real-world trade and that we should keep taking it.

The STEER Act, and a prediction

I didn’t spend much time discussing the economic analysis included in the Chicago study. That’s because, to a significant degree, that argument has already been digested here in DC. Advocates argued convincingly that citations’ safety benefits are coming at a relative cost to lower-income residents that is unfairly high.

This resulted in a period of legislative churn, during which the Council removed some of the usual consequences for not paying your tickets. I think that was a bad idea.

But more recently they passed the STEER Act (currently under congressional review). And I think it’s an impressive response: it will mean that the consequences of our speed enforcement system will increasingly be felt in non-financial ways, including traffic safety classes (aka white collar prison dispensed on an hourly basis), license suspensions and–amazingly–automotive speed governor devices for the worst offenders. I think it’s a thorough response to arguments about economic inequity raised by ATE critics and I am looking forward to its implementation.

I also think it will do approximately nothing to silence ATE criticism. People do not like being punished for bad driving. This is perfectly normal. Nobody likes being punished. Nobody likes to imagine that they are potentially culpable for the vast amounts of death and injury on our roads. Nobody wants to believe that the sense of freedom they experience behind the wheel ought to be curtailed. People will invent new arguments for why the punishment is unjust, or the deaths are imaginary, or the safety benefit is fake, or the privacy impact is unacceptable, or the carbon footprint is too big, or the streetscape’s natural beauty is being destroyed, or the imposition of driving standards threatens traditional masculinity, or something even stupider that I can’t yet imagine.

That’s fine. Sometimes, when someone is wrong, and after you have listened to them with a polite, blank expression, you must look within yourself, accept that you might get yelled at, and summon the courage to ignore them.

About the author

Tom Lee

1 comment

By Tom Lee