California has pledged to be a national leader in the fight against climate change and the housing crisis. But a new report from UC Berkeley’s Terner Center and a companion analysis from California YIMBY show that the state’s own rules may be undermining both goals. At issue is how the state defines “low-VMT neighborhoods” — areas where residents drive fewer miles, helping to reduce emissions. These neighborhoods are supposed to qualify for streamlined housing approvals under state law. But flawed and outdated data models mean that many climate-friendly, high-opportunity urban neighborhoods are being excluded.
The Terner Center report, Aligning Housing with Climate Goals: The Importance of Measuring VMT, authored by Rachel Strangeway and Zack Subin, examines how different tools for estimating vehicle miles traveled (VMT) influence what housing gets built — and where. Their findings are stark: California’s current model misidentifies large swaths of urban areas that should be eligible for fast-track housing development. And it systematically disadvantages the very neighborhoods where more housing would best serve both environmental and social equity goals.
VMT plays a central role in aligning land use with the state’s climate targets. Under Senate Bill 743, California evaluates the environmental impact of housing developments partly through how much driving the new residents are likely to do. Projects in areas with VMT at least 15 percent below the regional average may qualify for CEQA streamlining, enabling faster construction. But if VMT is mismeasured, neighborhoods that are walkable, transit-rich, and already support low emissions could be wrongly excluded from streamlined approval — and the homes needed to make those neighborhoods more accessible simply don’t get built.
The Terner Center study compared three different tools for estimating VMT across California neighborhoods: the State model, which relies heavily on pre-pandemic data from regional agencies; Replica, a simulation platform that uses anonymized mobile device data to model travel behavior in near real-time; and LATCH, a national model based on 2017 household travel survey data. Of the three, the State model is the most commonly used for CEQA streamlining — and it’s also the least accurate, the report finds.
In Los Angeles, for example, the Replica model identified 62 percent of the population as living in low-VMT areas. The State model identified just 40 percent. In San Jose, similarly, dense and transit-accessible neighborhoods were recognized as low-VMT by Replica, but not by the state’s outdated tool. The difference is not academic: it determines where housing gets approved, how fast, and with how much legal risk for developers and local governments.
One of the biggest problems lies in how VMT thresholds are defined. California currently evaluates neighborhoods against their regional VMT averages. In car-dependent regions like the Central Valley or Inland Empire, this makes it relatively easy for a neighborhood to qualify as “low-VMT.” But in already low-driving regions like the Bay Area, the regional baseline is so low that many otherwise climate-friendly areas narrowly miss the cut.
This creates a perverse outcome: the greener a region already is, the harder it is for any neighborhood within it to qualify for VMT-based streamlining. The Terner Center found that using a statewide VMT baseline — rather than regional — would increase eligibility by 40 percent in the Bay Area. In other words, California’s own framework is punishing the regions that have already done the most to reduce driving and support sustainable development.
Beyond the climate implications, this also undermines equity. The YIMBY blog post highlights that the State model often fails to identify neighborhoods designated as “high opportunity” by the California Tax Credit Allocation Committee’s (CTCAC) Opportunity Map. These neighborhoods typically have strong public schools, good transit, and access to jobs — places where more housing could provide life-changing access for lower-income families. But because they often exist in regions with already-low VMT, they’re penalized by the current methodology. Meanwhile, some car-dependent, lower-resource areas are greenlit, simply because their regional VMT baseline is higher.
The Replica model, by contrast, shows much stronger alignment with both actual traffic volume data and with California’s stated equity goals. It not only maps travel behavior more accurately, but also highlights neighborhoods that overlap with high-opportunity areas. These are exactly the places where more housing could advance both environmental and social objectives — yet they are being left out of the fast-track process.
There are clear steps the state can take to fix this problem.
First, California should modernize its VMT modeling framework by adopting tools like Replica that incorporate current, public data — including anonymized smartphone mobility data, which better captures how people actually move through cities today. Pre-pandemic models based on outdated regional data fail to reflect post-COVID behavioral shifts, including increases in remote work and changing transit patterns.
Second, the state should rethink its strict reliance on VMT as the only way to determine environmental eligibility for streamlined housing. Walkability scores, residential density, and transit proximity are also meaningful proxies for lower emissions and could offer more equitable and accurate alternatives for identifying climate-friendly development areas.
Third, and perhaps most urgently, California should replace the regional VMT baseline with a statewide comparison, especially in regions like the Bay Area where low driving is already the norm. This change would allow the state to reward — not punish — places that have long supported sustainable growth and efficient transportation.
Misidentifying low-VMT areas doesn’t just slow housing approvals. It reinforces patterns of exclusion and inequality. By blocking new homes in transit-rich, high-opportunity neighborhoods, California effectively maintains barriers to integration and undercuts its own climate targets. CEQA streamlining, if implemented correctly, should unlock housing where it does the most good — where it reduces emissions and expands access. Instead, under current rules, it’s too often doing the opposite.
Aligning housing policy with climate goals means ensuring that the right data drives the right decisions. As the Terner Center and California YIMBY make clear, the stakes are high. If California continues to use flawed models, it will miss the neighborhoods that matter most — and the homes, families, and futures that depend on them.
The problem isn’t “flawed data”. The problem is that the state purposefully stopped using Level of Service (LOS) in regard to greenhouse gasses. This seems to have been a political decision, rather than a science-based decision.
You don’t need to be an engineer to know that vehicles idling in traffic create much higher amounts of greenhouse gasses per mile than vehicles which are not hindered by traffic.
That’s also related to the reason that motor vehicles get better gas mileage on freeways (vs. in town), despite traveling much, much faster.
Take a look at the traffic backed up on I-80 on some afternoon, and tell me if you think they’re emitting greenhouse gasses despite not moving at all. You can witness those freeway backups from any of Davis’ freeway overpasses, though I wouldn’t recommend hanging out there too long (due to the other poisonous gasses emitted by those vehicles stuck in traffic on I-80).
I also find it amusing that the YIMBYs in the article cite “equity” as a reason to force housing in areas that are already impacted by congestion, since the organizations that sponsor them are the same ones which contributed to “inequity” in the first place (e.g., the technology industry, which also has a history of not hiring very many women – regardless of skin color). But more importantly, note how their argument dismisses entire cities and regions and the people who won’t, or can’t move to their imaginary eco-paradise.
The problem is that you are not a data analyst nor a subject matter expert like the Terner Center.
From their study:
“All of this suggests that how “low-VMT” is defined, as well as the data and models that are used to measure local VMT patterns, are increasingly important to where new housing is incentivized. Methodological changes can alter which neighborhoods are defined as low-VMT, thereby influencing which developments may be provided streamlined CEQA entitlement processes… We find large differences in the low-VMT maps depending on which VMT model is used, as well as whether neighborhood-level VMT is compared with a regional or state VMT baseline.”
That’s a comparison comparing VMTs with other VMTs; not whether or not LOS should also be part of the equation. Before citing experts, you might want to try to understand what they’re comparing.
Again, this isn’t even in question. Vehicles stuck in traffic create more greenhouse gasses per mile, than those which aren’t stuck. No one denies that – experts, or not.
Those in places like the Bay Area are almost certainly creating more greenhouse gasses per mile than someone out in the valley – after accounting for different types of vehicles, etc. Terrain would also be a factor regarding efficiency (e.g., hills vs. flat land, etc.).
This would also be true of businesses serving those respective residences, after accounting for any differences in vehicle types, etc.
While idling in traffic emits more per mile, urban residents still produce fewer total emissions because they drive less overall. That’s why California shifted from LOS to VMT: to prioritize emissions reduction, not just traffic flow.
The Terner Center report shows the state’s current model undercounts walkable, transit-rich neighborhoods like those in LA and the Bay Area, even though they align best with climate goals.
The goal isn’t to move cars faster — it’s to reduce how far and how often people need to drive.
David, while I agree with your conceptual description, there is a serious flaw in California’s approach. The analysis of VMT for Village Farms illustrates the flaw. The model makes assumptions about the geographic distribution of the working residents of the area being analyzed, and that distribution is substantially affected by the availability of jobs in the local community. If there are no available jobs then the residents of the area being analyzed will have to commute to other communities in order to be employed. So the assumption of the percentage of residents with local jobs is a key factor.
Further, it isn’t as simple as looking at the total number of jobs in the local market. The analysis needs to segment those total jobs by income level in order to determine the number of jobs with income levels necessary to cover the housing costs for the residents at a level of no more than 30% of their total household income.
BAE reported that the average price of the “affordable” units in Village Homes is projected to be $740,000. In Davis a $740,000 sale price home generates over $77,000 of housing costs. Divide that by 30% and you get a household income of $231,000. How many jobs in Davis have an annual salary of $231,000?
So the chances of a new resident landing a $231,000 job in Davis are very slim.
As a result every transportation study should show the impact of VMT from a range of local employment percentages, as well as a similar mileage range for the percentage of residents who have to commute to get a job that is sufficient to pay their housing costs.
You’re right that a $740,000 “affordable” unit tied to a $231,000 income is misaligned — but that’s a housing affordability failure, not a failure of VMT policy. What we need are more affordable, infill homes and better data models that account for economic realities — not a return to outdated LOS metrics that reward sprawl and punish urban density.
David, I’ll respond to your comment in reverse. Whatever metrics we use, they should be as accurate as possible and understandable as possible. VMT appears to look at traffic from a collective community, regional and societal perspective. LOS appears to look at traffic from an individual and neighborhood perspective.
With that said, the point of my comment above is that regardless of which method is used, the results it reports should be as accurate as possible, robust in identifying scenarios, and easy for the general public to transparently understand. Where the proposed residents are likely to work is a huge driver of the calculated VMTs. If 10% work locally and 90% commute elsewhere to work, the VMTs are going to be very different than if 25% work locally and 75% commute elsewhere to work. The transportation study should prominently display what percentage of local employment they have used. Having a standard sensitivity analysis of a range of assumptions should be the standard/norm.
With that said, both the collective community, regional and societal perspective AND the neighborhood perspective have value/merit. Ideally a transportation study should contain both.
Regarding your first point “that a $740,000 “affordable” unit tied to a $231,000 income is misaligned — but that’s a housing affordability failure” you and I are in complete agreement. $740,000 is not affordable … unless the household has that $231,000 a year of income. To be arguing for approving such a development while believing in social justice in housing is little more than Virtue Signaling. Richard Rothstein in November 2019 showed the over 300 Davis residents attending his talk the housing history of David with racist Red Lining that exclude persons of color. That history of racism has morphed into a history of classism, and the new housing proposals that the City has received recently are specifically designed to perpetuate our housing affordability failures … and even expand them. And I agree with you 100% that that is NOT a failure of VMT policy.
“The goal isn’t to move cars faster — it’s to reduce how far and how often people need to drive.”
I think the ‘goal’ is to build baby build!
“With that said, both the collective community, regional and societal perspective AND the neighborhood perspective have value/merit. Ideally a transportation study should contain both.”
BINGO, MW!
“You don’t need to be an engineer to know that vehicles idling in traffic create much higher amounts of greenhouse gasses per mile than vehicles which are not hindered by traffic.”
Also important reason not to have abandoned LOS: life sucks when you’re stuck in traffic. Life really really sucks when you are stuck in traffic every day.
That’s not really the issue, but it is important to note that if you reduce VMT, you reduce the amount of travel time.
Seems like it shouldn’t be necessary to point out the obvious, but apparently it is – for science and common sense deniers. Note the first two sentences below, which imply that the methodology used to measure greenhouse gas emissions is influenced by politics:
“Policy makers have placed less attention on reducing CO2 emissions by reducing traffic congestion. As traffic congestion increases, so too do fuel consumption and CO2 emissions. Therefore, congestion mitigation programs should reduce CO2 emissions.”
“The key question is how big of an emissions reduction we can get by reducing congestion. This question is difficult to answer, because CO2 emissions, and the fuel consumption that causes them, are very sensitive to several factors. These factors include individual driving behavior, vehicle and roadway types, and traffic conditions. Because of these factors, a table that estimates CO2 emissions based only on a single variable, such as trip distance, cannot provide an accurate estimate. Rather, a comprehensive methodology that takes advantage of the latest vehicle activity measurements and detailed vehicle emission factors can create a more accurate emissions inventory for different types of vehicles and different levels of traffic congestion. With this methodology, we can accurately estimate how congestion mitigation programs will reduce CO2 emissions.”
https://www.accessmagazine.org/fall-2009/traffic-congestion-greenhouse-gases/
(My third comment.)
Your third comment is using a 2009 article to refute a 2025 article. To be simple, contemporary research increasingly shows that total VMT is more important than congestion alone when it comes to long-term GHG emissions.
And yet, the research you’re citing doesn’t even address what you’re claiming – as already pointed out.
The article I cited shows that congestion was ALREADY being purposefully downplayed as of the date of that study (first two sentences):
“Policy makers have placed LESS attention on reducing CO2 emissions by reducing traffic congestion. As traffic congestion increases, so too do fuel consumption and CO2 emissions.”
If your point is that there are (generally) fewer greenhouse gasses (on average) from driving one mile in congested traffic, vs. 10 miles in free-flowing traffic, then that would make sense.
But it’s obviously not true when comparing MILES driven.
“If your point is that there are (generally) fewer greenhouse gasses (on average) from driving one mile in congested traffic, vs. 10 miles in free-flowing traffic, then that would make sense.”
This is the entire point of prioritizing VMT over LOS. They want to incentivize shorter commutes and the possibility of people using transportation that is not a single occupancy vehicle.