What is the value of human life? This seemingly academic question is more important than most people appreciate, especially in the context of transport funding decisions.
But before we discuss how the value of human life influences transport funding decisions, let’s first justify the economic concept itself. I’m aware that for many people the phrase “value of life” just sounds wrong, usually because they think that life has, well, infinite value.
On the other hand, it’s easy to point out that people accept certain “risks” all the time; we are if you like “risk-takers.” That’s not to suggest that people are inherently foolhardy, but it is to suggest that people seem to accept the risk of death when the probabilities are low and/or the rewards are high. Every time you cross a road, for example, you are effectively trading-off a small risk of death versus the benefit of being on the other side of the road.
Perhaps even stronger evidence that the value of life concept has some merit can be found in situations where people pay for things that reduce their likelihood of death. When deciding which car to buy, for example, most people will trade-off safety features versus the additional costs. In this case, people are actually revealing, through how much they pay, how “safe” they prefer to be. To put it bluntly, some people will willingly accept increased risk of death when the price of safety is too high.
At this point I should acknowledge that people are not necessarily fully informed of the relative risks associated with their actions. I may also agree that in situations where people do not have full information it may be worthwhile for governments to “nudge” people towards certain choices that have outcomes that are more desirable. Such arguments lie behind government involvement in, for example, advertising campaigns that encourage children to adopt more healthy eating habits. Ultimately, however, the issue of imperfect information suggests people err when evaluating relative risks; it does not suggest the VOL concept has no merit.
The suggestion that life has infinite value also seems absurd when you consider its full implications: If life were infinitely valuable then we would reasonably expect individuals to use all of their resources to eliminate risks to life, and where risks were unable to be eliminated then we would cease to undertake those activities altogether. My personal view is that while many people feel uncomfortable talking openly about the value of life, it is a conversation that every advanced society that is concerned for the welfare of its citizens (both in terms of their safety and their freedom) needs to have.
Putting philosophical issues to one side, government agencies estimate VOL because it’s actually rather practical. Indeed, VOL helps government agencies determine appropriate levels of funding for health and safety initiatives. And outside of the health system, perhaps no other government agency needs to value life so much as the transport sector. This brings us to the ‘nuts and bolts’ of this post. That is, the methodology that it typically used to estimate VOL.
To estimate VOL you actually just need to follow these three simple steps:
- Get a representative sample of NZers;
- Ask them questions about how much they would be prepared to pay (both in terms of direct and indirect costs) to drive on a “safe” road, compared to a goat track; and
- Collate their responses and use some freaky econometrics (preferably non-parametric logit models) to estimate how much people are prepared to pay for reduced change of death.
From these types of “stated preference” experiments NZTA have estimated the value of life for the average New Zealander to be approximately NZD $4 million.
A normal benefit-cost analysis of a transport safety initiative then proceeds by multiplying this VOL by the reduction in transport related deaths attributed to the project. E.g. Project A saves 5 deaths therefore benefits = $4million x 5 lives saved =$20 million in benefits. Voila, with this methodology NZTA is able to estimate the economic benefits of transport safety improvements in comparison to their economic costs. This in turn can inform relative levels of investment in transport safety initiatives compared to, say, public transport and walking/cycling.
But is this a reasonable methodology? I think not. The primary issue I have with the current process for estimating VOL is that it considers VOL from an individual perspective, rather than in terms of one’s societal value. To put it another way, the current methodology asks people how much they value their own life, but it does not ask about the value of that person to other people. Just ask yourself: How much your own mother, father, and partner would collectively be prepared to pay to avoid your death? Quite a lot one would imagine! That is quite a lot “on top” of your own value of life, i.e. what I call “societal VOL” is additive to “individual VOL” – where current methodologies estimate only the latter.
Am I alone in suggesting that NZTA maybe under-estimating the value of life? It seems not. In May 2011 the State Services Commission commented (in a formal review of NZTA):
NZTA may need to check that the costs in its Economic Evaluation Manual (which are based on a social value of life and willingness to pay surveys) appropriately measure the current economic cost of serious injury. For example the cost to ACC of a paraplegic 20 year old is around $10-15 million life time cost.
In this case, the State Services Commission has picked up the fact that the value of avoiding injuries used by the NZTA is at least four times lower than the actual fiscal costs. Either this suggests that ACC is spending too much on paraplegics, or alternatively that the NZTA is underestimating how much society is prepared to pay to mitigate the costs of serious injury (NB: Injuries are a complex kettle of fish because there are arguments to suggest that being seriously injured is worse than dying).
There is another intuitive reason to think that something is fishy is going on with how we currently value life. This fishy smell arises when you look carefully at responses to stated preference experiments. In doing so you will typically find an interesting anomaly: Young people have a lower value on life than old people, even though the former have longer left to live. Now some of this can obviously be put down to differences in risk profiles. But some of it is also, I believe, due to the fact that individual value is not equivalent with social values, especially when age differences come into play. Stated differently, society places a high value on the life of young people, even if young people themselves do not value their own existence so highly.
If you asked people on the street whether (given the horrendous) choice they would choose to save the life of a young person over that of an old person then most would reply “yes.” This suggests that society values life quite differently from how individuals value their own lives. Again, in my mind the anomaly of young people having a lower VOL confirms that the current method of estimating VOL is incomplete.
As someone whose own father suffered a horrendous vehicle accident at the hands of a drunk driver, I can tell you from first-hand experience that serious injuries impact not only on the individual’s quality of life, but also the quality of life of those that are close to them. There is, if you like, collateral damage to society when people are seriously hurt or killed.
To sum up: I believe that current methodologies are systematically under-estimating the value of avoiding transport related deaths and injuries. They do so because they focus exclusively on the individual value of life, rather than the wider social value.
So while the value of life may not be infinite, I think it’s certainly more valuable than we currently think.
Processing...
I can actually see why the ACC would value life higher. If someone dies young as a result of an accident then the economy loses out on a certain level of productivity but if that person is seriously injured then not only might we miss out on that productivity we also incur a cost to look after that person. I’m not arguing that the NZTA figure should not be higher but that serious injury and death need to be treated separately.
Yes I agree, there are good reasons for the fiscal cost of serious injuries being higher. But fiscal cost does not equal economic cost, because the latter draws on “willingness to pay” rather than just resource costs.
So the relevant question is: How much are you willing to pay to avoid being killed versus being seriously and permanently injured? It’s an empirical question whether people would rather live, but be seriously and permanently injured, versus dying .
Asking people the question “what are you willing to pay” is not the same as finding out how much they are actually willing to pay. That latter could probably derived from data such as willingness to pay to carry a fire extinguisher in your car, or have fire alarms in your home vs the incremental rate of fatality or serious injury from not having one.
It was interestin that in the UK there were three different goverment agreed values of life (or rather, value of fatality avoided) roads at ~$1.5m, main line railway at $4m and the underground at ~$8m: presumably because people accept that roads are risky, railways should be safe, and that there is a real aversion to being trapped and dying in a deep tunnel.
If safety improvements were valued much higher there could be an interesting shift in the prioritisation of transport projects. Hopefully more would be spent on pedestrian improvements and also cycle infrastructure.
Yes, absolutely. Changes in the value of life could have a drastic effect on the way that urban environments are designed.
I am, quite honestly, frequently disgusted by the unsafe nature of New Zealand’s cities and towns. They’re unsafe for pedestrians, cyclists, public transport users, and often vehicles too.
The more cost effective approach would be to encourage people not to cycle at all.
The most cost effective approach would be to encourage people not to drive at all. We’d save billions on roadway construction and traffic injuries would drop to zero overnight.
That would halt the economy, limiting its cost effectiveness.
I can see the headlines now “Nick halts economy with one hand”.
That hand would be an iron fist, in a velvet glove 😉
In all seriousness, I’ve been doing a literature review recently regarding the effectiveness of cycling promotion to reduce vehicle kilometres travelled, and as a result reduce population exposure to vehicular accidents.
The general conclusion from the literature is that (in Europe at least) the increase in the incidence of cycle injury and the related healthcare costs related to the new cycle-km is lower than that of the car-km they replaced. In other words shifting some of peoples travel to cycling is safer and cheaper than keeping them driving.
A big part of that is due to the fact that driving in a car is actually the most dangerous thing we routinely do, and the injury outcomes are generally very extreme and often permanently debilitating.
The real dangers of driving is a “swept under the rug” issue actually. Among people under the age of sixty, only cancer and cardiovascular disease kill more New Zealanders.
Ahhahah ahha ah a hahhha hha. That’s not an option because it infringes on people’s freedom of choice. And it’s not that cost-effective because if you ban cycling we’ll all become obese and die from heart disease.
I didn’t call for a ban on cycling, it’s simply “nudging” people towards certain choices that have outcomes that are more desirable.
Like driving less?
You would nudge people against cycling? Seriously? Or have I interpereted your comment incorrectly?
Start sarcasm “Walking as well. Get rid of walking and cycling. Yeah, that’l do it. No more footpaths = more room for road lanes. Yes. What a glorious world.” End
It’s an indisputable fact that driving is safer than cycling.
Rod; not so: How many people die by and because of driving, including cyclists hit by cars? How many people die by cycling every year? The math does not support your assertion.
What you mean is that the cyclist is more vulnerable than the driver. Well apparently, but because of the harm that motorists put themselves and others in the way of, especially through the speed they are travelling at, the outcome is different.
…Only if you concentrate on the deaths from traffic crashes alone and not the other health-related deaths of driving or otherwise. There’s quite a lot of literature around showing that the health benefits of cycling (in life-years gained) exceed those lost from crashes by an order of magnitude.
The other problem with just comparing the “official” crash rate stats by mode is that you are not comparing apples with apples. E.g. comparing the crash rate of mature 40-year-olds driving on a protected motorway vs inexperienced 10-year-olds riding on a busy urban street (there is a reason we don’t let 10-year-olds drive…). What I’d want to know is “what is the relative risk to ME travelling the SAME route by either mode?” As it happens, I’ve been doing some research in that area and it’s not always black and white. For example, per hour of travelling, your average 15-19 year-old in NZ is typically safer riding their bike than being in a car (driving or passenger).
Care to share any stats to prove the ‘indisputable’ part?
Rod, see my comment above. There are very few firmly established ‘facts’ in the science comparing the safety of driving vs cycling, due to methodological issues around what comprises a valid and reliable comparison. But far from being ‘indisputable’ the current data tends towards cycling being safter, all other things being equal.
“If safety improvements were valued much higher there could be an interesting shift in the prioritisation of transport projects. Hopefully more would be spent on pedestrian improvements and also cycle infrastructure.”
But typically road safety and sustainable transport projects ALREADY have greater benefit-cost ratios than most “travel-time reduction” RONS-style projects. E.g. it is not unheard of to have low-cost road-safety treatments with BCRs >> 20. And a basic cycle lane project that helps 100 cyclists/day easily generates a BCR of 5. So, while I’d love to see a lot more emphasis on improving road safety in this country (and better sustainable transport provision would probably help that too), it’s not the $ value of life that is really holding that up.
A very good question, and one that I’m sure Stu or one of the other regulars will expand on in future postings to make the case for public transport versus private vehicle transport.
One thing that became apparent to me on reading the article was that when considering roading safety improvements, NZTA may not evaluate the savings (costs of death/injuries that would be avoided) if people choose to not make a particular private vehicle journey they otherwise would have, whether for economic reasons (congestion charging/higher fuel prices) or because they took an alternative mode of transport that is safer.
Thanks Ash, yes I think it’s an important question. While transport safety trends in NZ are heading in the right direction, we still have a relatively high rate of accidents per capita.
Your comment on the relative impacts on different modes is interesting. In theory vehicle accident cost savings are already able to be considered when evaluating walking, cycling, and public transport projects. This considers how many people switch from driving because of the proposed project, thereby reducing VKT and the risk of accident. The issue is that the benefit that comes out the other end is often very small, because for any given project we are talking about a small number of people switching, which in turn has a small impact on the relative risks.
But if we were to increase the VOL used in these calculations then yes, you would see increased benefits for transport projects that avoided the need to drive.
“we still have a relatively high rate of accidents per capita”
Very high actually, among the worst in the OECD.
touche! 🙂
How much people are willing to pay for safety improvements to avoid the death of some random individual is interesting. I’d also be interested in a study that sussed out what sum the average kiwi would accept to have a loved one killed. One approach seems to undervalue life, while the other perhaps overvalues it.
Fun questions!
Peter, if I was being more specific:
1. People have an individual value; and
2. People also have a value to some other people.
The value of #2 is not about the value of “some random person”. It’s about someone’s mother, father, son, daughter etc etc. So what I am imagining is that number 2 is defined by how “close” someone is to you, in a social sense.
Imagine a function that has you at the middle, with a value of $4 million, which declines as you move further away (in a social sense). The first social step drags in your parents/children – who might value you $2 million each. Then the next social step drags in your friends/co-workers. Then the third social step brings in general acquaintances.
Of course eventually you will be so far removed from the individual that they are essentially “random”, but in getting to that point you have gone past a lot of people for whom the relationship to the person who was killed is much stronger, and hence their value is much higher.
Bah, it looks like this did post after several failed attempts yesterday. And now it just blew my somewhat lengthier and more thoughtful response out the airlock. Let’s see if I can sum up my response.
These are good points with respect to how VOL changes with respect to perspective / degrees of separation. With valuations you could expect something of a hierarchy: Family > Friends > Strangers > Sociopaths. However, my point was to highlight a different bias in economic valuations that will affect your final VOL numbers regardless of who on your continuum is doing the valuing.
The question is really one of willingness to pay versus willingness to accept. Let me formulate a different example. How much do you think New Zealanders are willing to pay, per little brown kiwi bird, to save them? Using a variety of techniques, such as the survey you suggest, we could come up with a kiwi bird unit-cost.
Now that we have the cost of a kiwi bird, imagine that I desperately want to hunt one of the cute little guys. Certainly, it should not be a problem for me to come for a visit for New Zealand, pay this fee, and kill one. Right?
As it turns out, economic experimentation has borne out that the same little brown kiwi bird will have two distinctly different prices–one price that people would be willing to spend to save the, and another higher price that people would be willing to accept for me to kill one. You mention both concepts in your essay, but they are different numbers–very different.
Economics professor Frank Ackerman has written much about the kinds of systematic biases that exist in valuation approaches such as the one you suggest.
Your objection to NZTA’s benefit-cost analysis approach is that their valuation of life is incorrect, but I could just as easily assert that your valuation method (or any number of a hundred) is equally incorrect. But I don’t think the real question here is to figure out which of the myriad ways of generating an incorrect value for a human life is superior. I think the real question is to answer a policy question.
I don’t think benefit-cost analysis is always inappropriate, but I do think that if your usage relies on valuations that have such a well-known and terrible bias, that you should maybe consider some better tools as the basis for public policy. My former professor, Don Miller, has written much about decision-making in planning, particularly about multi-criteria evaluation methodologies.
Interesting, that subjective “Value of Life” can be determined simply by asking people how much they would be prepared to pay to assure THEIR OWN safety. And interesting that an even higher value might be assigned by those near-and-dear to the life being valued.
But what is the subjective value of a random life to the population at large? In other words, how much would we be prepared to pay to assure the safety of some person totally unknown to us? I suspect the answer is ‘not very much’, and this is surely what underpins much of what is accepted on the roads in terms of the human toll.
Quite frequently we hear comment to the effect that “We need risk in order for life not to become bland and boring”, and often this will be trotted out as an argument against curbs on risk-laden motoring freedoms in the name of safety. Now while it may be true that everyone should have the right to choose a level of risk in their OWN lives which they are comfortable with, and balance that against their OWN safety, what happens all too often on the roads is that the consequences of risk-taking are transferred to OTHERS. It is one thing to engage in, say, an extreme sport where the risk is only to yourself. It is quite another to engage in a dangerous activity which imposes risks on others and particularly if they are non-willing participants or innocent 3rd-parties. And therefore those who argue for risk, are either only thinking of themselves (if they indeed think at all), or else are consciously assigning a lower value to the lives of unknown others, than either they would assign to themselves or their own near-and-dear, or than those unknowns and their near-and-dear might assign to themselves.
So what is the subjective value of an unknown other’s life? And has this demonstrably ‘low’ value whatever it may be, become the de-facto figure that allows road transport to continue business-as-usual, with a level of 3rd-party risk far beyond any other peace-time activity? And with a penalty regime for death and injury caused to 3rd parties far laxer than that in other areas of life or on other transport modes? If each of those (approx) 300 deaths and 3000 serious injuries occurring annually on NZ roads were subjectively valued by all-and-sundry at the “near-and-dear” rate, then serious pressure would come on road transport to clean up its act or else hand over to safer modes.
Another problem with the methodology is that people are notoriously bad at estimating a wide range of things, and so the answers they give are inconsistent with reality. The only reason many/most people buy lottery tickets or insurance is that they are appallingly consistently bad at estimating small probabilities – they estimate the likelihoods to be much greater than they truly are. This is basic psychology. Depending on how you present the question, people can be just as bad at estimating very common things. I once saw someone take out his handkerchief, fold it in a diagonal, roll it up and present it as a unit of measure, and then ask someone who had recently seen a horse to estimate how long the horse was in terms of these handkerchief units. The horse was predictably gargantuan because that is simply the way people are at quantifying unfamiliar things, or familiar things in unfamiliar terms.
Surely revealed preference is more realistic than stated preference. I think harder about things when I am putting my life on the line and/or spending money, compared to when I am just answering hypothetical questions.
Yes I agree. But issues with imperfect information possibly become more pertinent for revealed preferences, because you have less control over the situation.
Is there no actuarial basis to NZTA’s calculations? Really? There are solid actuarial models for estimating future value of life based on foregone earnings (with associated taxation of economic activity impacts), and I was sure that there was much more scientific basis than such bollocks “guessing” to how the cost of a death was calculated.
For something so important, empirical, testable modelling should be the norm.
It turns out that lost earnings make up only a small proportion of the social costs of a fatality. NZ used to use this kind of “human capital” approach to valuing life before shifting to the current Willingness-To-Pay approach (which increased the value about ten-fold). I don’t have the original research to hand, so I’m not clear whether the valuation was a personal risk one or valuing a loss to others, which some commentators seem to be concerned about. I’d put some faith too in the WTP method used in the research; typically they don’t just ask people for a direct $value.
BTW, I don’t think that the SSC argument about the “$10-15 million life-time cost of a 20-yr-old paraplegic” is that relevant to the value of life debate; this is not your typical “serious” injury. Which is why the average social cost for serious injuries is currently ~$390k.
If anyone wants a bit more background on this, there are a couple of useful links at MoT:
http://www.transport.govt.nz/ourwork/land/landsafety/thesocialcostofroadcrashesandinjuries/
http://www.transport.govt.nz/research/Documents/UTCC%20_Value_of_statistical_life.pdf
Good points Glen. You’re may be right that the costs of a paraplegic is not particularly relevant, because it’s not a “typical” serious injury. And the ~$390k figure would be some sort of weighted-average.
I was really just trying to highlight that other people have raised “questions” about the way we value life/injuries, but I did not make that very clear in the post.
We all get into terrible moral knots with this issue. Especially if a purely utilitarian valuation is made. For example the killing of a young adult who has just left the education system and has therefore enjoyed a huge investment by society and is about to spend the rest of their working life paying it off [so the theory goes] could be considered much much more ‘expensive’ for us all than an newly retired person about to enjoy a lengthy retirement on national super. Of course this kind of thinking leads directly to supporting mass euthanasia; kill off the costly and all the other lunacies of the user pays mindset. Dark math indeed.
The problem is that society clearly makes the judgement that the usefulness of the car is so great that it will put up with the enormous list of dis-benefits that accrue from the world made in its shape…. including a frankly horrendous level of tragic death and dismemberment…. a level that we wouldn’t put up with for virtually any other benefit. Is this because we all make the calculation that it will happen to someone else?
About our falling crash stats. I’m certain that the biggest cause of the decline of these figures is to do with the increasing level of urbanisation and driving on busier and therefore slower roads. The UK has always had way lower stats than us and way busier roads. Especially narrow city lanes. This also accounts for the higher level of courtesy there than here; you just have to give way or no one would get anywhere. We are such poor drivers because we just aren’t used to sharing our relatively empty roads and streets. Very hard to die crawling along bumper to bumper on a congested motorway, except perhaps by road-rage!
There’s also a very convincing argument that our serious injury/fatality stats are declining because of better vehicle engineering, with improving levels of crash-survivability. That’s particularly apparent if one looks at when things like air bags and side-intrusion beams became commonplace in conjunction with trends to fatality rates. We got the huge win by dividing the Auckland motorways (the Fire Service rescue tender at Manukau used to be the busiest fire appliance in the country, attending all the serious collisions on the Southern Motorway) and now the wins are incremental ones that are, Police statements notwithstanding, mostly down to roads and vehicles rather than significant changes in driving behaviour.
Toss in big leaps in trauma care – serious injury stats are relatively constant even though fatality levels are decreasing – and it’s apparent that we’re not particularly driving better, it’s just that we’re surviving more crashes that would previously have contributed to the road toll.
Stu – I was working in the old MoT at the time the first value-of-life studies were done (late eighties), and the way that this was calculated was a huge advance on what had gone before. From what you say it is certainly worth revisiting, but this time constructing the exercise in order to include the other aspects you mention.
My career in transport started off by looking at road safety issues. The only comment I would make now is that I wonder if we’ve now ‘picked all the low-hanging fruit’. In 1973 the road toll was 843 dead. Last year the road toll was 284 – with twice as many cars on the road as in 1973 – and that was down on an annual average of 375 over the previous three years.
As I don’t have access to the specific numbers, what is the current scope of cyclist and predestrian fatalities?
~ Patrick: one part of why the UK stats are lower is because rates of car ownership are a lot lower, and there is less rural or open highway driving compared with New Zealand.
For more comment: http://www.scoop.co.nz/stories/PA1201/S00001/record-low-2011-road-toll-but-more-to-be-done.htm
The UK also has a significantly newer vehicle fleet (in 2009 the average age was 7.1 years, up from 6.6 years in 2003) than does NZ (13-ish years and counting as of 2011). Newer vehicles are safer, reducing the risk of serious injury or fatality. Even with all other things being equal, which they’re not, the safer vehicle fleet would result in the UK having a lower road toll than does NZ.
An interesting point about car ownership. The stats shouldn’t be vehicle-related-deaths/per-capita it should be vehicle-related-deaths/per-cars-registered
So, if we valued lives more there would a larger push for slowing cars down in pedestrian busy areas such as Symonds St.
Symonds street has recently had a relatively major upgrade, from my understanding safety was a major component of this upgrade (particularly in the stretch between Engineering/architecture and science/rec center/the IC).
Some of my mates were involved in the consultation process due to there role in a faculty student society.
The approach to safety in this area is very much to keep pedestrian and vehicles separate. I feel this approach is appropriate given that the large volume of buses this route carries, and given that it is The NZTA approved overwidth/overheight/overweight route out of the port.
1.8m high fences between the footpath and carriage-way were considered to prevent pedestrians from crossing in less-safe locations. Thankfully they went for a nice garden instead. I think the upgrade has been really good (except the overloaded pedestrian crossing at symonds st/grafton road). Slowing cars is largely pointless given the high volume of non-car vehicles on Symonds street.
What AUSA wanted was all traffic taken off Symonds St between Alten Road and Wellesley Street. What’s been given is still traffic-centric. The only thing that’s changed is that trucks are no longer the dominant vehicle form and there’s now a dedicated high-volume bus lane. Pedestrians are still absolutely second-best on that entire road and, in some ways, it’s been made worse by the landscaping that’s been done outside the 300 and 400 sectors; it’s now much, much harder to cross anywhere other than a sanctified crossing.
My friends were part of the engineering student society.
“it’s now much, much harder to cross anywhere other than a sanctified crossing” That was the idea. The previous pedestrian/vehicle mix was very dangerous (particularly between engineering and science). From a safety perspective I think the addition of a crossing outside engineering and planting to gently encourage pedestrians to use it is a good idea. Ive seen the aftermath of some nasty incidents on that road prior to the upgrade. I don’t think its the best place for a pedestrian utopia, Perhaps High or Queen street?
Given the density of occupation around Symonds St, I think AUSA had the right idea. What has been done reinforces the primacy of vehicles over pedestrians by discouraging pedestrians from going anywhere other than a signalled crossing. A properly pedestrian-focussed plan would’ve lowered the speed limit, put in raised zebra crossings, and generally discouraged vehicles from using the route.
I realise that that’s not consistent with Symonds St’s role in the Central Connector, but I’m pointing out that Symonds St remains absolutely a pedestrian-hostile zone.
John Adams (no not the famous American one from 200 years ago) is very good on how very bad any methodology based on placing $-valuations on the invaluable (like your grandmother) really is. His 1974 paper in Environment & Planning has yet to be bettered on the subject:
http://www.john-adams.co.uk/papers-reports/
http://john-adams.co.uk/wp-content/uploads/2010/02/and-how-much-for-your-grandmother-4.pdf
His little book called simply Risk is brilliant.
Most Cost-Benefit analysis is done simply to allow projects that the prevailing wisdom wishes to see carried out to proceed. Witness RoNS.