In a late 2013 study (helpfully summarized here), Morgan Stanley claims annual $1.3 trillion in savings, e.g. thanks to reduced accidents because robots would drive better and improved productivity because humans could do real work instead of drive.
Apparently, these quite substantial sums are strictly based on the way we currently use cars, but the automation of driving could have a much deeper impact on car usage, leading to even more radical savings:
1) Reduced car ownership since cars can automatically switch from one user to another;
2) Reduced road space thanks to fewer vehicles;
3) Reduced parking space: Currently, cars are parked 90% of the time, but driverless cars could be used most of the time. Plus, vehicles could park more tightly;
4) Wealth generated by more productive uses of urban space formerly dedicated to road and parking.
