My central claim, in writing Economics in Two Lessons, is that most economic policy issues can be understood in terms of opportunity costs and their relationship to prices. I was talking about 21st century (electric and self-driving) cars, and several of the issues that came up illustrated this point very neatly. Among the objections to 21st century cars are the following
- Since 21st century cars don’t use petrol, governments will lose the revenue needed to fund the road network
- Self-driving cars will cruise around cities to avoid paying for parking, thereby increasing congestion
- Because of the limits of AI, self-driving cars will inevitably kill people
The answer to the first two questions is the same. These problems arise because prices don’t reflect opportunity costs. Opportunity costs arise from cars using the road network, reducing access to others, and from the initial construction of the network, consuming land and resources that could be used for other purposes.
Under current conditions, petrol consumption provides a rough proxy for general road use, while parking charges provide a rough proxy for road use in urban areas, shopping precincts and so on. That relationship breaks down with 21st century cars.
But, this is a self-resolving problem. The reason we used petrol taxes and parking charges was because charging for road use was too hard. With 21st century cars, it’s trivially easy. We can set prices exactly equal to opportunity costs, taking account of time-varying congestion and any other factors we want to.
The dangers of 21st century cars can also be understood in terms of opportunity costs. The question isn’t whether they are perfectly safe, but whether they are safer than the next best alternative – the current mix of human drivers, including the large proportion of incompetents, drunk and drugged drivers.
A side issue that has just occurred to me: is it possible to steal a self-driving car with no manual override? It seems a bit like stealing a train.