Skip to main content

Stop blaming drivers for Mexico City’s smog

Lucas Davis, Professor, Haas School of Business | June 5, 2017

This spring, Mexico City has been choking under some of its worst smog conditions in years. The problem is ozone. Pollution alerts for ozone have been issued repeatedly, triggering “double” driving restrictions that have pulled hundreds of thousands of cars off the road, twice as many as usual. But what if cars are not the problem? I’ve been combing through recent data from Mexico City and the relationship between cars and ozone is tenuous at best.

Note: Mexico City’s Paseo de la Reforma obscured in smog. (Hector Guerrero/AFP/Getty images) 

Big data

If you want to look at a pollutant that is tightly related to driving, take carbon monoxide. As the figure below shows, carbon monoxide levels in Mexico City tend to peak at 8 a.m. or 9 a.m., when the roads are jammed with commuters trying to get to work. Emissions inventories show that 99 percent of carbon monoxide in Mexico City comes from cars, and you can see this in the daily pattern.

The pattern for carbon monoxide also differs across days of the week. The figure shows only Friday through Monday, but this is enough to be able to compare weekdays to weekends. Friday, Monday and other weekdays have the biggest peaks. Saturday and in particular, Sunday, have lower peaks. Again, this reflects driving. After battling traffic all week, people in Mexico City are happy to drive less on weekends.


Ozone has a very different pattern, peaking in the middle of the day when the sun is highest in the sky. There is no peak during the morning commute like you see with carbon monoxide. But even more revealing, notice that the peak for ozone is similar across all days of the week. Weekend ozone levels are just as high as weekday levels, even though many fewer cars are on the road. When people drive less, carbon monoxide levels go down. Ozone levels? Not so much.


Note: These figures were constructed by Lucas Davis (UC Berkeley) using hourly pollution data from Mexico City’s Automated Environmental Monitoring Network. Each observation is mean pollution averaged over all monitoring stations for a specific hour and day, and the figures use all hours between January 2015 and April 2017.

What’s going on?

I’m an economist, not an atmospheric scientist, but I think the data give a clear picture of what is going on here. Ozone is the classic “secondary” pollutant, meaning that it is not emitted directly but instead is formed in the atmosphere as a product of other pollutants.  The basic recipe for ozone is simple. Take volatile organic compounds (VOCs) and nitrogen oxides (NOx). Add sun. Chemical reactions happen. Voila!! You get ozone, a pollutant that not only causes smog but also is very dangerous for human health.

But here’s the deal: You need both VOCs and NOx. If you are short one of these two ingredients, you can’t substitute more of the other. In econ-speak, we’d describe this as a “Leontief” production function. In areas where there are lots of VOCs, ozone formation is “NOx-limited.” A reduction in VOCs will have little impact, because the process already has more VOCs than it can use. Similarly, in areas where there is lots of NOx, ozone formation is “VOC-limited,” and a reduction in NOx will have little impact.


Note: Figure 1 from Sillman (1999).  The solid lines represent ozone production rates of 1, 2.5, 5, 10, 15, 20, and 30 parts per billion per hour.  Thus there are different combinations of NOx and VOC that yield the same ozone production rates.

Surprisingly, there is no consensus in the scientific literature on whether Mexico City is NOx- or VOC-limited. See here and here.  But the recent data provide pretty clear evidence that Mexico City is, in fact, VOC-limited. NOx levels are much lower on weekends, but ozone levels are not.


In fact, Sunday ozone levels in Mexico City are actually somewhat higher than other days of the week, consistent with the backward bending part of the curves above. For places that are severely VOC-limited, ozone production can actually increase when NOx concentrations drop. In other words, when VOCs are the constraint to ozone production, more cars on the road can actually reduce ozone levels!

Given this evidence, it seems crazy to try to reduce ozone levels by restricting driving. Sunday is, in some sense, an extreme version of what could be achieved through driving restrictions. And while many pollutants are indeed lower on Sundays, ozone is not. Driving is not the problem. 

Policy implications

What are the policy implications?  First, drop the double driving restrictions on high-ozone days. There is no evidence that this has any impact on ozone levels. And, more generally, driving restrictions have been widely shown to be an expensive and ineffective approach to addressing air quality.

If you want to reduce ozone in Mexico City, you have to reduce VOCs. VOCs come from all kinds of things. Paints. Solvents.  Adhesives. Cleaning Products. Cosmetics. Even dog poop. Yes, dog poop.  Carlos Álvarez, a chemical engineer at Mexico’s National Polytechnic Institute, has calculated that 250,000 tons of dog poop are “deposited” annually on Mexico City’s sidewalks, significantly contributing to VOC emissions.


Note: Watch your step. A dog walker in Mexico City’s Hipódromo neighborhood. (Source: Airbnb)

In addition to targeting these sources, it would be worth looking again at transportation. But rather than restricting driving, it’s time to look at gasoline regulations.

California provides a particularly useful point of comparison. Los Angeles is similar to Mexico City in that both suffer from high ozone levels and both are VOC-limited. Since 1996, California Air Resources Board (CARB) gasoline has been required throughout the state. Considerably more stringent than U.S. national fuel standards, CARB gasoline must meet strict content requirements for olefins and other highly reactive VOCs.

CARB gasoline has been shown to be very effective at reducing ozone. When CARB gasoline was introduced, it reduced Los Angeles ozone levels by 16 percent, according to research from the Energy Institute’s Max Auffhammer and Ryan Kellogg. California has also achieved additional ozone reductions by enforcing strict requirements for vapor recovery systems which reduce VOC emissions at gas stations when drivers are filling up their tanks.

Similar requirements could work in Mexico City too.  Whatever approach is taken, let’s then evaluate the policy using data. Too much is at stake to continue rolling out the same tired policies. Let’s use modern data techniques to quickly and credibly figure out what works and what doesn’t work.

Comments to “Stop blaming drivers for Mexico City’s smog

  1. Thanks for the interesting article. The Mexico City government must know that their smog is not created by carbon monoxide from cars, yet they have imposed the driving restrictions. Perhaps it’s easier to impose regulations on the common people then to tackle the real issue of ozone, caused by VOCs and NOx. The government would have to target and regulate big industry corporations manufacturing and selling these harmful products, and that would cause political and financial issues for the government.

  2. Human Race, We Have a Problem! We’re not mitigating global warming fast enough because we don’t know how to communicate and we don’t know how to cooperate. So today we are at 410 ppm and accelerating with no long-term attenuation in sight that will produce an acceptable quality of life for future generations.

    Do we have any social scientists who can fix this in time?:

  3. U.S. refineries export more gasoline to Mexico than the amount of gasoline produced by refineries in Mexico.
    Hence U.S. oil companies can hope to increase their already significant gasoline export profits by invoking this Berkeley professor’s cherry-picking science which then ridiculously concludes “Stop blaming drivers for Mexico City’s smog.”

    The Union of Concerned Scientists goes beyond cherry-picking science facts to enumerate a comprehensive list of major pollutants from motor vehicles:

    Particulate matter (PM). These particles of soot and metals give smog its murky color. Fine particles — less than one-tenth the diameter of a human hair — pose the most serious threat to human health, as they can penetrate deep into lungs. PM is a direct (primary) pollution and a secondary pollution from hydrocarbons, nitrogen oxides, and sulfer dioxides. Diesel exhaust is a major contributor to PM pollution.

    Hydrocarbons (HC). These pollutants react with nitrogen oxides in the presence of sunlight to form ground level ozone, a primary ingredient in smog. Though beneficial in the upper atmosphere, at the ground level this gas irritates the respiratory system, causing coughing, choking, and reduced lung capacity.

    Nitrogen oxides (NOx). These pollutants cause lung irritation and weaken the body’s defenses against respiratory infections such as pneumonia and influenza. In addition, they assist in the formation of ground level ozone and particulate matter.

    Carbon monoxide (CO). This odorless, colorless, and poisonous gas is formed by the combustion of fossil fuels such as gasoline and is emitted primarily from cars and trucks. When inhaled, CO blocks oxygen from the brain, heart, and other vital organs. Fetuses, newborn children, and people with chronic illnesses are especially susceptible to the effects of CO.

    Sulfur dioxide (SO2). Power plants and motor vehicles create this pollutant by burning sulfur-containing fuels, especially diesel. Sulfur dioxide can react in the atmosphere to form fine particles and poses the largest health risk to young children and asthmatics.

    Hazardous air pollutants (toxics). These chemical compounds have been linked to birth defects, cancer, and other serious illnesses. The Environmental Protection Agency estimates that the air toxics emitted from cars and trucks — which include Benzene, acetaldehyde, and 1,3-butadiene — account for half of all cancers caused by air pollution.

    Greenhouse gases. Motor vehicles also emit pollutants, such as carbon dioxide, that contribute to global climate change. In fact, cars and trucks account for over one-fifth of the United States’ total global warming pollution; transportation, which includes freight, trains, and airplanes, accounts for around thirty percent of all heat-trapping gas emissions.

    Cal Berkeley somewhere must have an academic affairs panel which can review and rescind staff published material which is bogus.

    • What’s Cal Berkeley? Ive never heard of that school. Must be nice. Anyways he is just commenting on a correlation and not stating that this is cause and effect as evident by the end of the post. “Whatever approach is taken, let’s then evaluate the policy using data. Too much is at stake to continue rolling out the same tired policies. Let’s use modern data techniques to quickly and credibly figure out what works and what doesn’t work.”

  4. “If you want to reduce ozone…”

    This is a frightening instance of not seeing the forest for the trees, one we could be pleasantly spooked by and then continue on our merry way, if it was just a momentary scare and not a specious justification for allowing the truly horrific status quo to continue chugging along.

    Minimizing automobile traffic has so many benefits beyond the short-term atmospheric improvements, from social to economic to, indeed, moral uplift that reducing the arena to a narrow not-an-atmospheric-scientist’s supposed ‘gotcha’ data crunching (more of a selective nibbling, really) is dangerous and deserving of the scientific community’s opprobrium.

    Imagine an economist who demonstrates that rednecks already spend about as much money on test prep services and private tutoring for their kids in high school as they are willing to spend, and therefore there is no point in limiting firearms sales since the savings realized in redneck households on extra guns not purchased will not be spent on more Kaplan courses. Conclusion: don’t limit firearms sales. And don’t blame rednecks for either low college completion rates or high accidental firearms injuries, because this chart right here shows that in some areas people who buy a lot of guns also send their kids to decent colleges. Maybe we want to remove barriers to gun sales, eh?

    The call for more data analytics may be even more insidious — we don’t need a fancy software package to know that car culture sucks the life out of us, killing us quickly in small numbers and a little less quickly in the aggregate. First-world nations have been able to make policy and law that moves toward keeping car culture harnessed and corralled where it should be, while we here in our decline have voices like Prof. Davis continuing to rationalize having given cars free rein for far too long already.

      • Your list of externalities merely scratches the surface of what constitutes the full down side of car culture. How do you price the effects of our children not attending schools where they live, or the debt slavery of workers paying off an automobile loan and insurance perpetually? Of value-added raw materials sitting idle 90% of the day? And taking up real estate space while doing so? Of redistributing our municipal tax revenues to pay for parking lots and road maintenance?

        Meanwhile, your blog post above will be picked up by fossil fuel interests and spread with the headline “Berkeley Business School shows that saving gas is foolish”.

        Regarding restrictions — we used to have a Quaalude crisis in America, and now we don’t. We put restrictions on the producers because we decided, as a nation and a society, that the consequences of not limiting their freedom were worth it. Shouldn’t you be part of the educational apparatus that helps people realize that car culture is similarly hazardous to our society? Or are you part of the ‘enjoy responsibly’ crowd when it comes to abusing driving?

  5. Because you position CARB gasoline regs in L.A. versus nonCARB gasoline in Mexico City,
    you should also provide data on ozone levels in L.A. as you have for Mexico City.
    If ozone levels in L.A. have the same “time delayed” peaking as you say occurs in Mexico City,
    then there is a cause-and-effect and your conclusion/recommendation would have to be modified.

Comments are closed.