At the 2009 International Transport Economics Conference Bruno De Borger, Erik Verhoef, and I were having dinner, Erik raised an interesting question about the use of statistical value of life in evaluation studies. Suppose there is a road improvement which will save 1 life per year, reducing the number of fatalities from 2 to 1 per year (out of 1000 people using the road). Assume all travelers are identical. What value of life should be used in the analysis?
Normally, we would do the equivalent of trying to compute for each traveler what is the willingness to pay for a 50% reduction in the chance of death by driving (from 2 in 1000 to 1 in 1000), and multiply that by the 1000 people whose chance of dying is reduced.
An alternative approach is to figure out the willingness to pay for the driver whose life is saved. So how much would you pay to avoid dying (with certainty) (i.e. what is your Willingness to Pay)? The answer to the first question is usually taken to be all of your resources (you would pay you everything so I won’t kill you).
Alternatively how much can I pay you to allow you to let me kill you (Willingness to Accept)? The answer to this second question is: I would have to pay you an infinite amount of money in order for you to let me kill you.
Both of those sums of money (everything or infinity) likely exceed the willingness to pay to reduce the likelihood of dying with some probability, multiplied by the number of people experiencing it.
In economic terms, we are comparing the area under the demand curve (the consumer’s surplus) for life (which has a value asymptotically approaching infinity as the amount of life approaches 0 (death approaches certainty) for a single individual, with the marginal change in the likelihood of survival multiplied by all individuals (i.e. the the quadrilateral between the y-axis of price and the same demand curve, between Pb and Pa) which describes the change in price for a change in survival).
On the one hand, using the marginal change for everyone rather than total change for the one person whose life is saved, we will give a lower value to safety improvements. On the other hand, the value of life to the individual himself is much higher than the value of life of that individual to society at large.
From Tyler Cowan: Don’t apply positive discount rates to human lives:
“Ben Trachtenberg writes:
This Article presents two new arguments against “discounting” future human lives during cost-benefit analysis, arguing that even absent ethical objections to the disparate treatment of present and future humanity, the economic calculations of cost-benefit analysis itself – if properly calculated – counsel against discounting lives at anything close to current rates. In other words, even if society sets aside all concerns with the discounting of future generations in principle, current discounting of future human lives cannot be justified even on the discounters’ own terms. First, because cost-benefit analysis has thus far ignored evidence of rising health care expenditures, it underestimates the “willingness to pay” for health and safety that future citizens will likely exhibit, thereby undervaluing their lives. Second, cost-benefit analysis ignores the trend of improved material conditions in developed countries. As time advances, residents of rich countries tend to live better and spend more, meaning that a strict economic monetization of future persons values the lives of our expected descendents above those of present citizens. These two factors justify “inflation” of future lives that would offset, perhaps completely, the discount rate used for human life. Until regulators correct their method of discounting the benefits of saving human lives in the future, the United States will continue to suffer the fatal costs of underregulation, and agencies will remain in violation of legal requirements to maximize net benefits.”
I think in practice we have to discount future lives, if the discount rate were zero, then we should do nothing for the present as the infinite future would dominate any calculation. I am dubious health inflation will continue unabated. The discussion on the article is interesting and worth reading.
JW sends me this link from Technology Review and writes “If the sensors and technology for collision avoidance systems can be implemented on smart phones it seems a disruptive way to move robotic vehicle technology forward. It bypasses the vehicle manufacturers and potentially the regulators. Clearly the app does not result in a robotic vehicle, but it may further public acceptance and allow the collection of comparative crash record data, two issues which are much more important than the technology and software in my view.”
App Provides Extra Eyes on the Road – Technology Review:
iOnRoad for Android detect and tracks cars on the road ahead using a phone’s camera and machine vision software. It also draws on a phone’s GPS, accelerometer, and orientation sensors to calculate the distance to other cars, and the speed at which they are traveling.
Just place your device in a mount on the dashboard and start up the app. Then your phone will diligently watch the road ahead, and beep a warning if you get too close to the vehicle ahead, alerting you to hastily brake before any damage occurs.
iOnRoad is a clever idea, and it highlights just how powerful and capable smart phones have become. Just few years ago, such an app would struggle on the fastest smart phone.
In practice, however, I found it a bit distracting. During a drive to Cape Cod last week, with the phone mounted beneath the GPS, my windshield felt cluttered. I kept glancing at the phone whenever a car outline changed from green to yellow (depending on how close I was), in addition to checking the GPS. With continued use of the app my eyes would probably stop drifting over to check how far away each vehicle was. Thankfully, I didn’t get into any near-collisions, and the road was pretty traffic-free.
The app can also work in background mode, so it’ll only sound and show a warning if it detects an imminent collision. So iOnRoad could run behind a GPS app while driving.
The Israeli company behind the app, Picitup, has previously created vision recognition software for to automatically cataloging products (which eBay uses). At first, iOnRoad will be free; and it will be available next month.
Some TED Talks about Transportation
Bill Ford: http://www.ted.com/talks/bill_ford_a_future_beyond_traffic_gridlock.html
Shai Agassi http://www.ted.com/talks/shai_agassi_on_electric_cars.html
Robin Chase http://www.ted.com/talks/robin_chase_on_zipcar_and_her_next_big_idea.html
Sebastian Thrun http://blog.ted.com/2011/03/31/googles-driverless-car-sebastian-thrun-on-ted-com/
Dennis Hong http://www.ted.com/talks/lang/eng/dennis_hong_making_a_car_for_blind_drivers.html
Dean Kaman http://www.ted.com/talks/lang/eng/dean_kamen_on_inventing_and_giving.html
Paul Moller http://www.ted.com/talks/lang/eng/paul_moller_on_the_skycar.html
Gary Lauder http://www.ted.com/talks/lang/eng/gary_lauder_s_new_traffic_sign_take_turns.html
Jaime Lerner http://www.ted.com/talks/lang/eng/jaime_lerner_sings_of_the_city.html
Steven Levitt http://www.ted.com/talks/lang/eng/steven_levitt_on_child_carseats.html
“While most automakers try to fix the problems with today’s tech, Volkswagen is working on tomorrow’s. The future of driving, in major cities at least, is looking more and more likely to be done by high-tech computers rather than actual people, at least if the latest breakthroughs in self-driving vehicle technology mean anything. Internet search engine giant Google has logged some 140,000 miles with its self-driving Toyota Prius fleet and Audi has had similar success with its run of autonomous cars.
Now Volkswagen has presented its ‘Temporary Auto Pilot’ technology. Monitored by a driver, the technology can allow a car to drive semi-automatically at speeds of up to 80 mph on highways.
It works using a combination of existing technology such as adaptive cruise control and lane-keeping assist, rolling them all into one comprehensive function. Nonetheless, the driver always retains driving responsibility and is always in control, and must continually monitor it. In this way, Volkswagen only sees it as a stepping stone towards what seems like an eventual future where nobody will be doing any driving.
In the semi-automatic driving mode, the system maintains a safe distance to the vehicle ahead, drives at a speed selected by the driver, reduces this speed as necessary before a bend, and maintains the vehicle’s central position with respect to lane markers. The system also observes overtaking rules and speed limits. Additionally, stop and start driving maneuvers in traffic jams are also automated.
The good news–or bad, depending on how you look at it–is that compared to the more advanced autonomous driving technologies, Volkswagen’s latest Temporary Auto Pilot is based on a relatively production-like sensor platform, consisting of production-level radar-, camera-, and ultrasonic-based sensors supplemented by a laser scanner and an electronic horizon.
This means that we could see a production version within the next couple of years.”
The main economic difficulty with deployment will be getting the sensor costs down. This is inevitable (if we believe Moore’s Law), but how long it will take is still unclear. The political restrictions are beginning to fall. Forbes reports:
“The State of Nevada just passed Assembly Bill No. 511 which, among other things, authorizes the Department of Transportation to develop rules and regulations governing the use of driverless cars, such as Google’s concept car, on its roads.
As Stanford Professor Ryan Calo notes, this is a big step forward in ensuring that safe, driverless cars become a reality.
Specifically, the law provides that the Nevada Department of Transportation “shall adopt regulations authorizing the operation of autonomous vehicles on highways within the State of Nevada.” The law charges the Nevada DOT with setting safety and performance standards and requires it to designate areas where driverless cars may be tested. (Note that this could take some serious time: Japan, for instance, has been promising standards for personal robots for years and has yet to release them.)
You can read the full text of the law here.”
More stimulus: Buffers for the flighty: ON BALANCE, I agree with the “it’s insane” analysis offered by Matthew Yglesias of America’s refusal to borrow money at historically cheap rates and spend it on infrastructure and other job-generating activities that will need to be undertaken eventually anyway…. [T]echnological change and globalisation have absolutely nothing to do with high unemployment in the construction sector. The people who build things in America will always be Americans, and there haven’t been any revolutions in construction technology between 2007 and 2011…. The reason the construction sector is sitting on the couch playing with the Wii instead of out fixing America is that America isn’t spending the money to do the fixing. America has a $2 trillion backlog of infrastructure maintenance, according to the Urban Land Institute. With the government able to borrow money at ridiculously low 10-year rates, it seems pretty convincing that we should be borrowing that money and spending it now, both to improve that infrastructure and to get the economy going. (Insert sub-argument: yes, but infrastructure programmes take a long time to get underway. Response: did you or didn’t you say this was structural unemployment that will take many years to resolve?)…
Again, only an economist (or The Economist) would think house construction and road construction are near substitutes, and the skills are easily transferred, but the general point remains, we should have an infrastructure bank to fund things that have benefits in excess of costs and can pay back the loans. See Fix It First, Expand It Second, Reward It Third: A New Strategy for America’s Highways