It's a very interesting question and far too complex for simple numbers that governments try to put on things.
Risk is highly subjective. If we put a multi-million dollar cost per life when thinking about medicines for example, things get very costly. Latest figures (2017) gave about 25000 new cancer patients in NZ. Admittedly not all will be at risk of death, but you could suggest that untreated, they may all be. At $4m per life that is $100bn which is four times the health budget - just for cancer patients (this is obviously over-simplistic). You can do similar numbers looking at the NZTA spend against 350 fatalities per year. Individual projects stack up, perhaps, but a higher level view may not.
But if you look at a probabilistic analysis, a widely used 'broadly acceptable' limit for workers is 1 in 100,000 risk of fatality per year. A cost benefit analysis would suggest $4m is too low. In my view, anyway, but as I say, it is very subjective.

A few years ago I looked at each of the Australian, UK and NZ numbers for the "value of a statistical life" (VOSL) - what we should be willing to pay to prevent one statistical death. A statistical life is not you, your child or (for me) any of my grandchildren. After allowing for the exchange rates at that time each of the values was about NZ$ 4.5 million.
Working as a consultant, my advice to clients was to inflate that to $5 million. Sometimes this caused a change in a system of work, sometimes no change because the number was too large. One client doubled that value to $10 million to force through change. So how much should Pike River been willing to spend on the ventilation system to remove methane? Each shift normally had 30 workers underground. Should they have been willing to spend 30 x $5 million = $150 million? I think the cost of the system was about $2 million. A bargain, unless that would bankrupt the business. Are you then entitled to gamble workers' lives?
Now working as a Lecturer in OHS at Victoria University I am revising the content of the Master's paper on "Hazard and risk" in trimester 2. As part of the teaching I will ask a health economist to talk to the students about the concepts of VOSL and alternatives such as Disability adjusted life years and their application to "reasonable" in the reasonably practicable test.
Perhaps the ultimate valuation test is to get your CEO to think how they would feel if being cross-examined by Crown Law in a prosecution following one or more fatalities. Ask them how much they would wish they had spent. Unsurprisingly, economists refer to this as regret. Read Kahneman's book "Thinking, fast and slow" for more on that.
PS, Victoria has a free fees offer in trimester 2 for study. It's subject to conditions but could be worth using for CPD.

So just to flip this around for some perspective, so far we have answers from $4.37M to $24.8M per life... so in the current situation how much should we be accepting as the fallout from the L4/L3 lockdown response?
Taking Sweden as the example of "do nothing / do minimal" approach that are tappering off at just under 400 deaths per million, which would be approximately equivalent to 1920 deaths in NZ - so 1900 more than what we have achieved.
So at the upper end of the value of human life we have is if it ends up costing the economy less than $47 billion to recover we "made the right choice"... (only $8.3 billion if we used the NZ value of life)

I think the bigger issue is why do we need to put a specific monetary value on human life? Especially when it is usually just a stab in the dark until hindsight comes back to bite us, just ask Robert Eidschun and the Ford Motor Company. With the legislation built around reasonably practicable including taking into account if the reduction in risk is grossly disproportionate to the cost of that reduction, maybe there is a need but it is definitely not as simple as slapping a price tag on death.

And that leads to the other issue - how do you take into account the multiple variables when it comes to valueing the societal cost of an activity - do you value the health of a someone in their mid-20s more than someone close to retirement, or how do you compare a agonising slow painful drawn out death from an occupational illness/desease to an immediate workplace fatality?

The root of the issue I see is generally the real question is being assessed from the wrong direction - we are asking ourselves "what is the least that we are required to do?" rather than "what is the most we a capable of doing?"

I'm sure most people would find it a very different conversation to talk about theoretical illness and deaths compared to a discussion about their own or a family member's illness and death. Priorities have a way of changing when it becomes personal. :wink:

Well done OP. Its not an easy number to find.

It is probably the best number to use in a NZ context as it is the statistical value of life, derived by Min of Transport. It is based on the social cost per fatal crash. Alternative values may be more appropriate for non-transport circumstances.

So it is essentially the value of a life lost in an accident.

It has been developed by Min of Transport and then used by NZ Treasury. If it is good enough for them it should be good enough for us. I can't see any good reason for faffing around trying to come up with another value. If anyone did I'd be challenging the basis for an alternative.

Any faffing around is a purely academic exercise. We could begin, say, recognizing that the average life expectancy in NZ is 82 years so the value of life, per year is $4.7m / 82 = $57,317. Which of course it isn't as the very young and very old likely have less economic value relative to a middle aged person.

And in a Covid context, for example why would we protect the elderly because an 85 years old is past their statistical value of any kind of return. It is still a discussion worth having if for example huge amounts of debt are racked up as a consequence of managed actions.

Incidentally the current (Sept 2019) value is $4,918,898. If you are a data nerd check out Treasury's CBAx model.

So reading up about the statistical value of life (such as the $4.37M stated for NZTA) it would seem like we are potentially using the concept out of context - it should not be use to assess how much is reasonable to spend to save a life, but how much you would have to spend to statistically save a life. These seem like the same but there is one key difference, since it is a statistical value you should not use that as the real value of the reasonable cost for any given situation as each unique set of circumstances will have it's own value of life - average out all of those unique circumstances and you will end up around $4M (for roardng related deaths in NZ) but there will still be a wide spread of actual reasonable values to spend to save a life.

As an example of the cold truth of the use of the statistical value of life it is primarily used to eveluated policy options, for example with a finite budget approving projects with a higher cost per life saved (above the statistical value) could be considered unwise as there should be other projects where you could save more lives for the same money.
Or for assessing if a policy will overall benefit society as a whole - i.e. (coldly) if a set of regulations will end up costing the economy less than the value of the lives it will save we should do it right... but should we if it will cost more even though it will save the same amount of lives?

This also explains why there can be so much variability in the stated values for different agencies, countries, etc.