Sorry again for late notice, but I’ll be presenting a video seminar 1pm today at ANU on intergenerational equity. For details contact Ralf Steinhauser on ph: 61 2 6125 4667.
Report: This was a bit of a bleeding edge experience, though it worked OK in the end. The big problem was presenting slides at the same time as video of me talking. ANU was expecting a hardware solution (dual video) while UQ was expecting a software solution (NetMeeting or Bridgit). Fortunately, I had sent the presentation ahead of time, so someone at the ANU end was able to run it for me. But I’ll have to develop a standard procedure for this.
I’ve attached the presentation (in PDF format)here
Some comments from the other side:
– I guess it is harder for people in that format to raise questions during the talk, (could have been the topic but doubt it) I would next time make sure in the beginning to encourage questions and tell them that they should raise their hand as you see them and/or say John and you would take that as rude but finish you thought and take the question
– Your voice was fading in and out a bit. When you said a louder word the speakers tuned down (to not have a feedback I suppose) and our ears had to adjust constantly – not sure if that problem could be solved
– Again the end is a bit weird not much one can do I suppose. I guess theoretically a subgroup of people could come close to the camera and continue the Q&A. Sorry about that though.
Overall, it’s pretty clear that both the technical and the social/conventional aspects of the process need refinement.
On the tech side, while the video of me presenting appeared to be OK, the video of the audience I got, with the same bandwidth, was naturally lower-res. I guess this will improve with time.
Also, Joshua Gans has suggested that it’s silly to confine the audience to people who can get to ANU (or wherever). I’m looking into streaming video over the Internet: it appears this is possible, but not trivial.
With all the problems noted above, it took me two hours and no travel to present a seminar that would normally have taken an entire day, four taxi trips and two plane trips, at a minimum, with an overnight stay if I wanted to avoid an absolutely exhausting experience. Videoseminar is definitely the way of the future.
John Quiggin 
You seem to be arguing for close to zero pure time preference across the generations. Forgive me if I have misunderstood the slides
My counter-argument to that is imagine we were all 20 years with an expectation we would all live to 80 years. I would argue that most people would manage their actions/investments so that they would only agree to an action which deferred utility into the future if that action gave them more utility in the future than the amount they were deferring. ie there is positive pure time preference, and empirical studies indicate it is quite significant. If there is positive pure time preference for most of us in the imaginary 20 year old cohort, then it seems to me logical that it would apply to most of us in an all-age population.
Sure, if everyone were 20 years old and there were no other generations then the whole question of intergenerational equity would be moot.
But the attempt in your final sentence to derive some implication from this case for the case when there are overlapping generations doesn’t work.
I thought your presentation was quite good (despite the technical difficulties). However, it did leave me wanting a stronger defense of the key assumption (4) as fundamental to an ethical system.
Under an ethical system based on the “veil of ignorance” approach one must, essentially by definition, treat all generations equally. As I now understand it, your argument is that if we agree to apply this ethic to all overlapping pairs of generations (Assumption 4), then by transitivity the ethic must apply universally. Hence, zero rate of intergenerational time preference.
I was left wondering if Assumption 4 was really so consistent with social norms. Certainly the generations currently extant and overlapping seem to demand equal treatment. But, those same current generations may also be perfectly willing to treat future overlapping generations inequitably. Why should I not desire that my son be treated as well as myself, while also placing more weight on my son’s utility than on my grandson’s utility? Is that obviously unethical (or inconsistent)?
One could of course argue that it’s not reasonable for current generations to have one standard (of equitability) for themselves and another for the future, but isn’t that back to a universal “veil of ignorance” argument. And if one is willing to apply the “veil” argument universally to start with, does the overlapping generations argument have any additional force?
The idea of the 20 year old population cohort in which everyone lives 60 years is that it can be transformed into an all-age population, because the 20 year olds become 21 and 22 etc. If pure time preference applies in the 20 year old population then it also applies in the transformed all-age population.
johng,
In an overlapping generatios model there is (using your numbers) at any one date (or time interval) at least 1 person who is 20, 21,22,…, 60 years old. Using your numbers there would be 40 generations at any one date (ie overlapping generations). The OL models tend to use only 2 generations (‘young’ and ‘old’) to simplify the notation. (To complicate the notation one could work with days, or months, or minutes, or seconds …).
JQ,
Your powerpoint presentation may raise a few questions on what counts as evidence in ‘evidence-based decision making’. (I am not fond of the phrase ‘evidence-based decision making’ because it can cover very different methodologies – from naive empiricism to methods used which involve an interplay between theory and empirical observations.)
The topic provides a very nice opportunity to bring out that the notion of ‘preferences’ is inseparable from individuals (ie individuals are characterised by preferences) but independent of the institutional environment (eg financial system).
Did anybody raise questions along these lines? In any case, perhaps a thread on ‘evidence-based decision making’ might be interesting.
The claim of this paper summarises something that I believe I have often told my resource economics students based on simpler intuitive reasons than your presentation. As parents we all put a lot of weight on the welfare of our children. Particular groups – such as migrants – put enormous weight on the welfare of their children.
These kinds of weightings are inconsistent with anything other than a miniscule rate of time preference. If I want my 10 year old son to enjoy a nice environment when he gets married at age 26 – and I do – I cannot sensible discount at 3% since I would according to your calculations I would only be assigning him half the welfare weight at that age I assign myself now.
That does not summarise my attitudes. As far as I can think clearly about these issues I want someone (who is not an abstract being but someone alive and kicking today) to be about as well off as I am now in 20 years time. This is not just introspection – my household investment behaviour confirms these attitudes/hopes.
People’s market behaviour and their altruism towards children are easily seen as inconsistent with anything other than extremely low rates of time preference.
I like your use of an overlapping generations framework to make these issues clear. Too often these issues are posed in terms of some distant, abstract future generation which I don’t think is how people think about the future.
John, you’ll be intersted in this article: http://www.theage.com.au/news/case-studies–profiles/eye-spy-a-cost-saving/2008/04/28/1209234761708.html
Harry, how much do you care about your grandchildren who do not currently exist, and who may never come to exist?
Spiros, I think that the overlapping generations logic suggests you care for the welfare of your children and they, recursively, have welfare dependent on their own children etc so you end up optimising over all future generations.
That looks like quite an interesting result. I like it how in the model you have individual rates of time preference but still have a zero rate of social time preference. This seems sensible – due to impatience I often make decisions that reflect a nonzero time preference, but this does not imply that I believe in a nonzero time preference.
If I buy a bottle of wine I will probably want to drink it soon but this does imply that I believe the life of someone in the future is worth less than someone in the present!
Being able to put recordings of videoseminars on the internet would be good. Unfortunately the seminar was at the same time as a talk by someone from the Garnaut Review discussing the impact of high developing country growth on global emissions.
hc, I concur with your intuition on introducing a notion of bequest into an OLG model. However, there is a snag. If the world is finite then there is a problem with the last generation of children. Do you have a solution to this theoretical problem (other than assuming a ‘great surprise’ for the last generation – this assumption would not match well with humans persuing natural science inquiries)
Ernestine, My argument (and I think that of JQ’s) is that the notion of long-term bequests and certainly the idea that the earth will crash into the sun in 4 billion years are irrelevant.
Humans can be portrayed as simple-minded creatures who worry about their progeny now and their rather short-term future. In the case I suggested (and John’s paper illustrated with a numerical example) the only requirement is that you worry about the welfare of your progeny 16 years hence.
In the case of my currently 10 year old son I’d want his environmental welfare to about what mine is now in a couple of decades time.
Unless you attach half the weight to these future people that you attach to yourself now you will discount the future at much less than 3%.
hc,
Thanks for the reply.
1. My point about your argument regarding the overlapping generations model did not involve a “long term bequest”. My point is that your argument: “I think that the overlapping generations logic suggests you care for the welfare of your children and they, recursively, have welfare dependent on their own children etc so you end up optimising over all future generations” is flawed for reasons I have stated (the recursive bit doesn’t work in the last period).
It seems to me it would be easier to go for a temporary equilibrium approach which involves testable hypothesis that, under some conditions, feed into ‘evidence-based decision making’. For example the current population’s preferences for the ‘state of the environment’ of the successor population are ‘elicited’ and decisions are based on the outcome. The successor population may be assumed to do what they want (because the current population is dead; although the current population may influence the successor population through education). There is already an empirical method (‘state contingent valuation’) which goes in this direction.
2. Your solution in paragraph 2 does not work unless you force all people to have children which may be biologically infeasible and undesirable on the grounds that it is dictatorial. There are some questions – would people with many children have a greater concern about ‘the future’ or is it the other way around? (But the ‘children’ argument is good enough as a hypothesis – who knows, childless people may have sufficiently altruistic preferences to care for other individual’s children – friends, relatives)
3. —
4. “Unless you attach half the weight to these future people that you attach to yourself now you will discount the future at much less than 3%”
I have not said anywhere that the social (ie intergenerational) discount rate is or should be positive. On the contrary, I am saying the empirical observation of a positive pure discount rate is nothing but the ‘time value of money’ (which depends on the institutional environment). Empirical data based on financial data would also constitute ‘evidence’ but decisions based on that evidence would lead to different outcomes. See #6 above.
John, you’ve probably commented on this before, but what’s your reply to Nordhaus’ hypothetical:
“Suppose that scientists discover a wrinkle in the climate system that will cause damages equal to 0.1 percent of net consumption starting in 2200 and continuing at that rate forever after. How large a one-time investment would be justified today to remove the wrinkle that starts only after two centuries? Using the methodology of the Review, the answer is that we should pay up to 56 percent of one year’s world consumption today to remove the wrinkle. In other words, it is worth a one-time consumption hit of approximately $30,000 billion today to fix a tiny problem that begins in 2200…the Review would justify reducing per capita consumption for one year today from $10,000 to $4400 in order to prevent a reduction of consumption from $130,000 to $129,870 starting two centuries hence and continuing at that rate forever after.”
I think a time preference function that is close to peoples intuition is 0% for the next 20 years, infinite after that. In other words people can only think so far ahead or feel concern for children already born.
A complication is the uncertainty created by mixed signals. For example Queensland’s simultaneous coal boom and water woes or whether carbon capture will work. That’s why I think discounting has to be done under different scenarios.