Tuesday, July 01, 2008

Discounting rhetoric

In a recent interview, Nicholas Stern turns a criticism of his team's use of discount rates in their report on the economics of climate change on its head:
Applying a 2 per cent pure time discount...means that you assign half the value to somebody born today to someone who is 35 years old. I would not discriminate against future generations like that.
It probably shows just how out of touch I am, but I hadn't seen the issue framed like this before. I'd be interested to know: would some serious thinkers see it as just a rhetorical trick? If they do, would they think it could still be useful?

P.S. 2 July: The comments to this post by Clive Bates, Ian Christie and Roger Levett are well worth reading. But the second of the three is an uncorrected duplicate of Clive Bates's first comment, and should be ignored. Read comments 1, 3 and 4.

4 comments:

Clive Bates said...

Well at risk of making a complete fool of myself, let me declare this to be, in my humble opinion, absurd rhetoric...

First and most obviously, he is comparing two individuals alive today at the present time – that would not normally require discounting, but some approach to ascribing a value to people at different ages – ie. age preferences not time preferences. The calculation of age preferences might need discounting depending on how value is assessed – eg. value by a person expected future earnings, their accumulated life experience to date, their expected personal future welfare – or they may be just regarded as equally valuable, a view that many would hold instinctively.

Rather than use a calculation and assumptions, it may be possible to empirically determine age preferences by trying to find whether people are indifferent to the death of one baby or x grown men today – where x may be less, equal to or more than 1 – depending on attitudes to age. This is a different question to the one needed to assess time preference, which is whether people are indifferent to the death of one person today or x of the same sort of people in 35 years time. These two questions illustrate why Stern's formulation is a distraction rather than a defence of his chosen discount rate.

One area where this sort of thinking is applied is in health, where economists seek to test the value of medical interventions in terms of the welfare they produce… Firstly, some measure of 'value' is needed – perhaps the condition of being dead for a given year=zero, alive and healthy=1, other conditions somewhere between. This allows for measurement of 'QUALYs', which are estimates of the number of quality-adjusted life years there will be with and without the intervention. This can be used to assess cost-effectiveness of the intervention. The discount rate applied to QUALYs has within it an assumption that we value an extra year of life achieved in 10 years more than the extra year achieved next year. I think most people would probably subscribe to that – especially if they were on an operating table while the economists were trying to decide whether it is worth operating.

In terms of revealed time preferences, the medical system shows extremely high pure time preferences for life – spending a fortune in acute intensive care settings and in the final hours at the end-of life, but relatively little on public health interventions that would produce large paybacks in the far future. Applying this sort of framework, a healthy baby would have a higher value because of its longer life expectation. Also, I think it is common practice to adjust for age in 'value of statistical life' calculations used in cost-benefit analysis – again the baby would come out better.

Great games can be played with discount rates. If we care less about a life in 35 years, how much less do we care about a life in 100 or 1000 years? The table shows how that trade off works for a 2% discount rate – with a willingness to sacrifice everyone in about 1200 years from now for anyone today! Stern's example would have been better if he had said 2% means indifference between one person dead today and two in 35 years….

At 2% discount rate how many lives would be equivalent to a single life today?
Years Lives
10 - 1.2
35 - 2.0
100 - 7.2
200 - 53
500 - 19,957
1000 - 398,264,651
1157 - 9,000,000,000

I think this shows the limitations of over-use of discounting...

I maintain that the main problem with Stern's approach is that he is trying to determine what the 'right' time preference should be from first principles and in doing so imposes his abstract ethical framework of his own design in order to do that.

In fact he is no more free to define time preferences in society than he is to define wine preferences around my dinner table. Time preferences exist as an observable trait and are fairly hard-coded into the way we act individually and collectively through politics (we care less about things that are remote in time, distance and genes for rational and probably evolutionary reasons). Even the most moral people typically display higher pure time preferences than the assumptions he makes. Here in Sudan, rates of pure time preference are observably very high – in politics and amongst people – it's about getting through the day.

I think Stern's approach would have been strengthened if he had used real psychological time preferences and showed what consequences flowed from acting in the way we typically do (ie. 6 degrees in the 21st Century, with much suffering in later decades about which we are not too bothered). He could have even used 'green book' methodology as a proxy for how governments act - though real government time preferences are likely to be much higher/short-term than the Treasury Bible tell us they ought to be.

Had he started with real world time-preferences, he would have held up a mirror to our civilisation and showed us the consequences of the short-termism and forced us to start making different collective choices, if we didn't wish to accept the long term consequences of our inbuilt short-termism with respect to the climate. In a way, he needed to make us sit up and realise that our time preferences need to change for moral reasons, rather than conduct his analysis on the basis of moral assumptions that are not grounded in reality.

The way he did it meant that the analysis was overly reassuring – because it was done with the implicit assumption that we are acting morally, we do have low time preferences about care a great deal about the future, when in practice we don't.

Regards

Clive

Clive Bates said...

Well at risk of making a complete fool of myself, let me declare this to be, in my humble opinion, absurd rhetoric...

First and most obviously, he is comparing two individuals alive today at the present time – that would not normally require discounting, but some approach to ascribing a value to people at different ages – ie. age preferences not time preferences. The calculation of age preferences might require discounting – eg. value by their future earnings, their accumulated life experience to date, their total personal future welfare – or they may be just regarded as equally valuable. It may be possible to empirically determine age preferences by trying to find whether people are indifferent to the death of one baby or x grown men today – where x may be less, equal to or more than 1 – depending on attitudes to age. This is a different question to the one needed to assess time preference, which is whether people are indifferent to the death of one person today or x of the same sort of people in 35 years time. These two questions illustrate why his formulation is a distraction rather than a defence of his chosen discount rate.

One area where this sort of thinking is applied is in health, where economists seek to test the value of medical interventions in terms of the welfare they produce… Firstly, some measure of ‘value’ would be needed – perhaps the condition of being dead for a given year=zero, alive and healthy=1, other conditions somewhere between. This allows for measurement of ‘QUALYs’, which are estimates of the number of quality-adjusted life years there will be with and without the intervention – and this can be used to assess cost-effectiveness of the intervention. The discount rate applied to QUALYs has within it an assumption that we value an extra year of life achieved in 10 years more than the extra year achieved next year. I think most people would probably subscribe to that – especially if they were on an operating table while the economists were trying to decide. In terms of revealed time preferences, the medical system shows extremely high pure time preferences for life – spending a fortune in acute intensive care settings and in the final hours at the end-of life, but relatively little on public health interventions that would produce large paybacks in the far future. Applying this sort of framework, a healthy baby would have a higher value because of its longer life expectation. I think it is common practice to adjust for age in value of life calculations used in cost benefit analysis – again the baby would come out better.

Great games can be played with discount rates. You might use it to say pure time preferences imply that we are indifferent to one life lost today and a larger number (x) of lives lost in the future. The table shows how that trade off works for a 2% discount rate – with a willingness to sacrifice everyone in about 1200 years from now for anyone today! Stern’s example would have been better if he had said 2% means indifference between one person dead today and two in 35 years….


Years from now and lives traded for one life today

10: 1.2
35: 2.0
100: 7.2
200: 52.5
500: 19,956.6
1000: 398,264,651.7
1157.45: 9,000,000,000.0



I maintain that the main problem with Stern’s approach is that he is trying to determine what the ‘right’ time preference should be from first principles and imposes his own abstract ethical framework in order to do that. In fact he is no more free to define time preferences in society than he is to define wine preferences around my dinner table. Time preferences exist as an observable trait and are fairly hard-coded into the way we act individually and collectively through politics (we care less about things that are remote in time, distance and genes for rational and probably evolutionary reasons). Even the most moral people typically display higher pure time preferences than the assumptions he makes. Here in Sudan, rates of pure time preference are observably very high – in politics and amongst people – it’s about getting through the day.

I think Stern’s approach would have been strengthened if he had used real psychological time preferences and showed what consequences flowed from them (i.e. 6 degrees in the 21st Century, with much suffering in later decades about which we are not too bothered). In doing so, he would have held up a mirror to our civilisation and showed us the consequences of the short-termism and forced us to start making different collective choices, if we didn’t wish to accept the long term consequences of our inbuilt short-termism with respect to the climate. In a way, he needed to make us sit up and realise that our time preferences need to change for moral reasons, rather than conduct his analysis on the basis of moral assumptions that are not grounded in reality. The way he did it meant that the analysis was overly reassuring – because it was done with the implicit assumption that we are acting morally, we do have low time preferences about care a great deal about the future, when in practice we don’t.

Luckily I’m not an academic or paid for my opinions so I don’t mind being mocked – please put me right!

Clive

Ian Christie said...

Thanks to Clive for an excellent and persuasive comment... Definitely far from foolish, and I can't see where the argument goes wrong on first reading...

I agree that the problem for Stern is that he is not going far enough down his ethical path and still wants to anchor the analysis in terms that econometricians will all endorse in neutral, technical and (in effect) amoral ways. But the basis for action will be ethics and self-interest and cultural psychology, not abstract economic rationality as the prime motivator.

A variation on Clive's analysis in the closing paras is simply to produce credible and widely agreed-upon scenarios for climate based on different temperature increases in different regions and 'hold up a mirror' by asking ourselves whether we would regard it as tolerable to be subjected to them ourselves, assuming no magic adaptation technologies are available. Bertrand Russell once said that it was no use wondering whether we would like to live in a projected future when the point is whether the people of the future would like it. But there are enough human universals across time and space to allow some empathy that reaches into the far future. One approach is to see how many temperature rise scenarios are compatible with any reasonable human expectation of quality of life, especially for the worst off. That is, could any society regard the scenarios to which our current short-termism commits future generations as preferable to what is available now?

All that said, I am under no illusions as to the likelihood of our collectively acting in the interests of far-ahead generations. Indeed, Earthscan is publishing a book later this year by an Australian environmentalist that calls for the abandonment of long-term sustainability ethics in favour of achieving an ecologically stable world through meeting the 'deep needs of the present'. I have no idea how he will make this case but look forward to it.

Roger Levett said...

Whatever problems we find in Stern's 'small print' shouldn't distract us from his political value. Here's what I said in November 2006:

Not very far into the Stern report on the economics of climate change, there's a passage, complete with equations, about the implications of possible exogenous extinction shocks for the appropriate discount rate. In normal English, it says that the bigger the chances of a meteorite coming along and obliterating human life, the more we might as well party on, trashing the planet's life support systems rather than exerting ourselves to keep the place fit for habitation in a future we mightn't have.

This is the way economists think. Caressing arguments like this is...very clever, but in a crass sort of way. It systematises with elegant rigour the intuition that the more likely your squat is to be demolished within the week, the less point there is clearing out the mouldy food at the back of the cupboards, or even aiming carefully in the loo. But it's so pleased with its cleverness that it fails to notice that the analogy is actually misleading. Earth is the only squat we've got, so if there's even an outside chance it isn't about to be demolished, the only sensible course is to avoid making it uninhabitable. Indeed it's a betrayal of the interests of our children not to.

The Stern Review includes hundreds of pages of this sort of stuff. But before howling with despair, consider where it all finishes up. Here are some quotes from the report's summary:

* 'The scientific evidence is now overwhelming: climate change presents very serious global risks, and it demands an urgent global response.

* The benefits of strong, early action considerably outweigh the costs.

* The earlier effective action is taken, the less costly it will be.

* Policy to reduce emissions should be based on three essential elements: carbon pricing, technology policy, and removal of barriers to behavioural change.'

Nothing here, of course, that lots of us haven't been saying for years. But having it said so clearly and starkly, by an eminent Government economist, a knight no less, with hundreds of pages of good science and policy analysis as well as economic voodoo behind it to provide 'thud factor', makes it unignorable by orthodox politicians.

Don't let's even gag on what Stern evidently regards as the most important and persuasive point of all: 'Tackling climate change is the pro-growth strategy for the longer term, and it can be done in a way that does not cap the aspirations for growth of rich or poor countries.' We may think that reducing human suffering, insecurity and death, polarisation of rich and poor, and destruction of beautiful intricate nature might be more compelling reasons than keeping the world safe for Disney, Toyota and Philip Morris. But let's be pragmatic. Economic growth is the religion of our time. A high priest has prescribed (mostly) sense and wisdom. Caesar is listening. Let's endorse the message, rather than poking fun at the pile of steaming chicken entrails used to divine it.