I was reading a story about peak driving over the weekend. In the course of reading the story, I discerned that we here in California drive far less than the average American. In fact, California ranks 41st among the states in per capita driving:

Date are from the Insurance Institute for Highway Safety.

Given the stereotype about California (as a place where everyone drives, always), this was a surprise to me. But then it dawned on me--when one excludes the District of Columbia (which is kind of like a state, just without representation), California is the most urbanized state in the country. And so I drew a scatter plot of VMT per capita against urbanization by state:

The negative correlation is quite apparent. To anyone who might be interested, here is the bivariate regression:

mpc | Coef. Std. Err. t P>|t| [95% Conf. Interval]

-------------+----------------------------------------------------------------

var4 | -81.73815 14.1223 -5.79 0.000 -110.118 -53.35832

cons | 15994.33 1066.959 14.99 0.000 13850.2 18138.47

------------------------------------------------------------------------------

So a one percentage point increase in urbanization is associated with an 81 mile per year reduction in driving. I think the direction of causality is not too big a problem here (it is hard to tell a story that more driving causes a reduction in urbanization). So Matt Kahn, Ed Glaeser and Richard Florida are all right--cities are environmentally friendly!

[BTW, a little Googling led me to a paper that relates to all this].

Date are from the Insurance Institute for Highway Safety.

Given the stereotype about California (as a place where everyone drives, always), this was a surprise to me. But then it dawned on me--when one excludes the District of Columbia (which is kind of like a state, just without representation), California is the most urbanized state in the country. And so I drew a scatter plot of VMT per capita against urbanization by state:

The negative correlation is quite apparent. To anyone who might be interested, here is the bivariate regression:

mpc | Coef. Std. Err. t P>|t| [95% Conf. Interval]

-------------+----------------------------------------------------------------

var4 | -81.73815 14.1223 -5.79 0.000 -110.118 -53.35832

cons | 15994.33 1066.959 14.99 0.000 13850.2 18138.47

------------------------------------------------------------------------------

[BTW, a little Googling led me to a paper that relates to all this].