Recently, a lack of wind and rain has inflicted unusually high smog levels on Paris. On Monday, March 17, Paris combatted the dangerously high levels of pollution by barring cars with even-numbered license plates from the roads. The intent was to ban even- and odd-numbered license plates on alternate days until the smog returned to safe levels. However, the city removed the restrictions after just one day.

Other cities have imposed license-plate bans to address long-term problems with poor results. The best evidence on the effect of license-plate bans comes from Mexico City’s Hoy No Circula program, which has been in effect since 1989. This program bans most cars one day each week based on the last digit of the car’s license plate number. Hefty fines guarantee that the program enjoys near universal compliance. Unfortunately, Hoy No Circula seems to have increased pollution instead of limiting it. Not only did it not decrease pollution during the regulated weekday, daytime hours, it actually worsened pollution during weekends and nighttime hours. The latter is to be expected. Many people probably shifted some of their driving to hours outside the ban. Other evidence shows that the program failed: it did not increase the use of public transportation, and it did not decrease the sale of gasoline.

In general, policies that closely target the final objective have a better chance of success. Many factors determine the pollution caused by a driver; however, the car’s license plate number is certainly not one of them. License-plate bans seem tailor made to create perverse unintended consequences. Given that the predictable nature of the driving restrictions, there existed a costly but straightforward way to completely circumvent the driving restrictions: buy a second car with a license plate that allowed driving on the day of the week that the first car was banned.

Both new car sales and car registrations increased. However, the majority of new car registrations were for used cars. Economic theory suggests that someone who is forced to buy a second car effectively becomes poorer. These people will on average drive older cars. Likewise, the less you use a given car, the less advantage you receive from owning a newer car. Older cars are generally associated with higher emissions. With higher emission vehicles, a license-plate ban could reduce driving without actually reducing pollution.

Despite their other failings, license-plate bans might at least sound like a fair way of allocating the burden of reduced driving. However, when you consider the impact of buying additional vehicles, it’s distributional properties look pretty poor as well. Many drivers obviously can’t afford to buy a second car, and many of the rich already own two cars.

Other Latin American cities have followed Mexico City’s example. Bogotá’s program, Pico y Placa, bans most cars from driving two days a week. Bogotá’s program is interesting in that once a year, they change the days of the week that a given car is banned. However, they do not change the grouping of license-plate digits. This means that the strategy of buying a second car continues to work. Someone who owns cars whose license plate numbers end in 2 and 8 can drive everyday of the week. Each year when the city changes the restricted days, this pair of cars will continue to provide daily access (each car will just be used on different days). Much like the experience in Mexico City, Pico y Placa seems to have increased the car fleet without increasing overall air quality.