Risk Analysis, Vol. 20, No.

6, 2000

Intergenerational and International Discounting

Thomas C. Schelling1

A discount rate for the consumption of future generations is typically composed of two parts.
One is a “pure” time preference for immediate over postponed consumption, the other a de-
clining marginal utility as consumption increases. The costs of greenhouse abatement, how-
ever, for at least the first 50 years, will be borne by the developed countries; the benefits will
accrue to the presently undeveloped. Pure time preference always relates to one’s own con-
sumption; it has no relevance here. Consumption transfers over time will be from richer to
poorer, from lower to higher marginal utility. It is a foreign aid program and it ought to have
to compete with more direct foreign aid, which can benefit the very poor rather than their
much-better-off descendants.

KEY WORDS: Greenhouse; climate; discounting; intergenerational

INTRODUCTION utility increments in the year, say, 2050 with costs in-
Economists who deal with very-long-term policy curred in 2000, but also utility increments in the year
issues, such as greenhouse gas emissions over the 2150 with increments in the year 2050.
next century or two, are nearly unanimous in their Since this article argues that “discounting” is not
opinion that future benefits that take the form of ad- the appropriate concept for dealing with the benefits
ditions to future consumption need to be discounted of reduced greenhouse gas emissions in the distant
to be commensurable with each other and with the future, it should be stated that this author finds tradi-
consumption foregone earlier to produce those ben- tional discounting perfectly appropriate for compar-
efits. There is a near consensus among these econo- ing costs and benefits of, for example, hazardous-
mists that the appropriate discount rate should be waste cleanup, as with the U.S. Superfund program.
conceptualized as consisting of two components.(1–4) In that kind of program, discounting with appropriate
The first is pure time preference and, according to rates of interest is crucial to determining which sites
Fankhauser,(2) deals with the impatience of consumers are worth cleaning up, how much they should be
and reflects their inborn preference of immediate over cleaned up, and when or in what order of priority
postponed consumption. The second component re- cleanup should occur. In that kind of program, those
flects the changing marginal utility of consumption who pay the costs are saving and investing—foregoing
with the passage of time, and is decomposed into a rate some current consumption—in order to reap future
of growth of consumption per capita and an elasticity benefits, along with their children and grandchildren.
of marginal utility with respect to consumption. The It makes sense to “optimize” the “investment port-
two components—pure preference for early over later folio” by reference to appropriate discount rates. The
utility, and declining marginal utility with growing per case is made that global greenhouse gas abatement is
capita consumption—are used to compare not only not like cleaning up one’s own land for one’s own
benefit. Costs incurred for greenhouse gas abatement
1
School of Public Affairs, University of Maryland, College Park, need to be discounted; benefits need an altogether
MD 20742; ts57@umail.umd.edu. different treatment.

833 0272-4332/00/1200-0833$16.00/1 © 2000 Society for Risk Analysis

834 Schelling

PURE TIME PREFERENCE to matter, and of course kinship does. Kinship dis-
tance has both horizontal and vertical dimensions;
Any time preference pertinent to discounting just as children are closer than grandchildren, chil-
the long-term benefits of greenhouse gas abatement dren are closer than nieces and nephews. Time just
cannot have anything to do with impatience. The al- happens to correlate with vertical distance.
leged inborn preference for earlier rather than later The crucial point is that decisions to invest in
consumption is exclusively concerned with the con- greenhouse gas–emissions abatement for the benefit
sumer’s impatience with respect to his or her own of future generations are not “saving” decisions—not
consumption. decisions about postponing one’s own consumption—
I feel no impatience about an increment of con- but are instead decisions about redistributing in-
sumption that will accrue to people whom I shall come, one’s own income. To invest resources now in
never know and who do not now exist in the year reduced greenhouse gas emissions is to transfer con-
2150, compared with an increment that will accrue to sumption from present-day people—whoever those
the people whom I shall never know and who do not people are who are making these sacrifices—for the
now exist in the year 2100, or even in the year 2050. benefit of people in the distant future. It is very much
Reasons can be given for preferring a boost to con- like making sacrifices now for people who are distant
sumption in 2025 to the same boost of consumption in geographically or distant culturally. Deciding whether
2075, but it is hard to see that it has anything to do one cares more about the people who will be alive in
with impatience and the inborn preference of imme- 2150 than the people who will be alive in 2050 is a little
diate over postponed consumption. In 2025 my oldest like deciding whether one cares more about people in
son will be the age I am today and his brothers a little one continent than in another, or about English-
younger; with a little luck they will be alive and speaking people more than people who speak other
healthy and my grandchildren will be the ages that languages, or about those with whom one shares his-
my children are today, and my great-grandchildren tory and culture more than those who do not. People
(whom I do not yet know) will have most of their lives do have preferences about whom to help; the prefer-
ahead of them. Seventy-five years later they will all ences show up in charitable giving, in foreign aid, in
be strangers to me. My genes may be as plentiful in immigration policy, and in military intervention.
the population at that later date but they will be What is under consideration here is very much
spread thinner. I probably would prefer the benefits like a foreign aid program with some of the foreigners
to accrue to my own grandchildren rather than to being descendants who live not on another continent
their grandchildren, but when I remind myself that but in another century.
my grandchildren’s happiness may depend on their
perceived prospects for their own grandchildren, my
MARGINAL UTILITY
“time preference” becomes attenuated.
Actually, time may serve as a measure of “dis- The second component of the proposed discount
tance.” The people who are going to be living in 2150 rate is the rate of change over time of the marginal
may be considered “farther away” than the people utility of consumption. The argument for including
who will be living in 2050. They will also be different that component must be that in transferring, or redis-
in racial composition and geographical distribution tributing income, an important goal is to maximize
from the people alive today. In redistributing income the aggregate utility of consumption over time. The
via transfer payments—providing foreign aid, con- expectation is that, on average, the marginal utility of
tributing to charity, and so forth—people are ex- global consumption will decline over time as a result
pected to differentiate, and do differentiate, among of rising consumption per capita. Resources invested
recipient peoples according to several kinds of dis- now out of present incomes will benefit people in the
tance or proximity. One is geographical: Americans future who are expected to be better off than those
are expected to be more interested in their own cities alive today—an unaccustomed direction for redis-
than in distant ones, their own country rather than tributing income!
distant nations. Another is political: East Coast Both within countries and among countries, civ-
Americans are more interested in the people of Los ilized governments are expected to redistribute to-
Angeles than in the people of Quebec. Yet another is ward the poorer countries and toward the poorer el-
cultural: Some people are closer in language, religion, ements of their own populations. The argument for
and other kinds of heritage. Sheer familiarity seems transferring consumption from the poor to the rich,
Intergenerational Discounting 835

or from the decently well off to the much better off, have been halved—using the popular but arbitrary
would be that the resources transferred grow in the logarithmic utility function—Chinese marginal utility
process, and grow so much that though the marginal will still be many times greater than that of the cur-
utility of the recipient is lower than that of the donor, rent populations most likely to pay for greenhouse
the magnitude that the gift achieves in transit more gas abatement.
than compensates. In neglecting to disaggregate, the optimization
There is not much room for this idea in contem- models assume that those who pay for abatement and
porary transfers. If a poor farmer has some poor soil those who benefit, or whose descendants benefit, are
and a richer farmer has rich soil, it could be argued the same. Because all populations—all nations or
that taking seed from the poor farmer and giving it to regions—are assumed to enjoy increasing consump-
the rich farmer will so enhance the resulting crop that tion per capita, and because investments in abate-
the somewhat utility-satiated rich farmer will gain ment precede the benefits, the benefiting populations
more utility than the poor farmer loses. But that is are assumed to have higher consumption and lower
just an argument for trade: the poor farmer is better marginal utilities than the populations that finance
off selling the seed to the rich farmer, and their joint the abatement.
utility is even higher. The ethical interest arises only if It can be expected, however, that for the first 50
trade is not possible, as when society outfits an adven- years greenhouse abatement will be paid for by the
turer who will emigrate to the New World, become countries that can afford it: the developed countries
rich, and never be heard from again, or as public pol- of western Europe, North America, Japan, and a few
icy contemplates transferring consumption forward others. The beneficiaries of abatement will mainly be
in time to people who have no way to reciprocate. the descendants of those now living in the undevel-
oped countries, for several reasons explained below.
Thus the consumption transfers will be from well-to-
THE NEED TO DISAGGREGATE
do countries that will mainly pay for abatement over
Depreciating the consumption of high-income the coming 50 years to the developing countries that,
future people makes sense; however, a fallacy must though probably progressively better off over the
be avoided here. If average per capita income rises in same time period, will have lower consumption levels
every country for the next hundred years, and if the 50 years from now than the current consumption lev-
poorer populations grow more rapidly than the wealth- els of the developed countries. Benefits from abate-
ier populations, and if most of the economic sacrifices ment during the first 50 years will almost surely be neg-
in the interest of carbon abatement are borne by the ligible compared with benefits during the second 50
countries that can best afford it, the transfers will years. Thus the consumption transfers, despite the
tend to be from the well-to-do people of western Eu- hoped-for uniformly (not uniform) positive growth in
rope, North America, and Japan to the residents of gas domestic product (GDP) per capita everywhere,
what are now called the “developing” countries, who will be generally from rich to poor—that is, from
should be far better off a century from now than they lower marginal utility to higher marginal utility. The
are now. These people though may not be as well-off, implications are startling, but first it should be ex-
during most of the intervening century, as those living plained why the beneficiaries will mainly be the de-
today in western Europe, North America, and Japan. scendants of the populations now poor.
In deciding how to value consumption incre- First, if the benefits of abatement were shared
ments over the coming century or two consumption uniformly over the global population, about 90%
needs to be disaggregated according to the levels of would accrue to the countries now considered unde-
per capita consumption at which they accrue. The op- veloped. The well-to-do are now about a fifth of the
timization models err, on their own terms, in aggre- world’s population; in 2075 the populations in coun-
gating all future consumption and applying a uniform tries now undeveloped are expected to be some-
discount rate for declining marginal utility. Correctly, where between seven eighths and eleven twelfths of
all increments in consumption should be valued at the global population. At that rate those populations
their own marginal utilities. In the optimization will comprise most of the beneficiaries.
models, increments for poor people are discounted Second, material productivity in the developed
equally with increments for the rich; there is no ad- countries currently appears to be substantially immune
justment for the fact that when Chinese per capita in- to weather and climate; production in less-developed
come has doubled and Chinese marginal utility may countries depends much more on outdoor activities,
836 Schelling

especially agriculture, and is potentially much more models assume, over the next 100 years marginal
susceptible to adverse effects of climate change. utilities of the beneficiaries will be much higher
Therefore, besides outnumbering the descendants of during the first 50 years — before abatement bene-
the currently developed countries, they can suffer fits become significant— than in the second 50
greater greenhouse gas damage per capita. (In abso- years. This factor substantially tilts the priority to-
lute terms, the more developed countries could incur ward whatever investments can raise living stan-
more lost GDP per capita, though the damage might dards in the first and second generations. Those
not be noticeable.) investments are likely to be direct investments in
Finally, the currently developed countries enjoy economic development, which should also reduce
GDP per capita ten times or more that of the unde- dependence on climate, rather than investment in
veloped; during the second half of the coming cen- climate stabilization.
tury they will probably still be ahead by a factor of Even more drastic, if marginal utilities will be
four or more. So the marginal utility of consumption higher in the fifth decade than in the sixth, in the third
of the poorer nine tenths of the population will be than in the fourth, and in the first than in the second,
several times that of the richer tenth, and the benefits today’s undeveloped populations have stronger
in utility increments from material consumption will claims, on the basis of marginal utility, than the pop-
therefore be overwhelmingly inherited by the de- ulations two or four generations in the future. Once
scendants of those who are currently poor. For exam- the world’s population is disaggregated by income
ple, if Chinese per capita income increased at 4% per level, it becomes logically absurd to ignore present
year for the next 50 years and 2% for the following 50 needs and concentrate on the later decades of the
years, and U.S. per capita income increased at 1% coming century. An initial interest in climate and its
over the hundred years, Chinese per capita income impact on welfare should not insulate one from alter-
would still be less than half the U.S. level at the end of native means to the same end. That then means that
the century. At those rates of improvement, the Chi- no framework for considering the benefits and costs
nese will be approaching the present U.S. level to- of greenhouse gas abatement should isolate itself
ward the end of the century; before then, interest will from the opportunity cost: direct investment in the
probably be lost in further increments. economic improvement of the undeveloped coun-
Earlier it was stated that the implications are tries. Abatement expenditures should have to com-
startling. One of them has already been mentioned: pete with alternative ways of raising consumption
virtually all the benefits from enhanced consumption utility in the developing world.
will accrue to countries that will not participate much Greenhouse gas abatement, which is largely
in financing the abatement. The transfers will be from identified with energy policy, is insulated in optimiza-
the currently rich to the descendants of the currently tion models from economic development. An answer
poor, who will—when the benefits begin to be felt— to how much (and when) to abate is given in optimi-
be much less poor than they are now but still poorer zation models independently of what else is going on.
than the descendants of the currently rich and proba- When greenhouse gas abatement is identified as a
bly still significantly poorer than the abatement- mechanism for making income transfers to future
financing countries are now. generations—especially to those whose consumption
Another is that the implicit “discount rate” based levels are still comparatively low—it should have to
on marginal utility comparisons will be negative. The compete with transfers for investment in economic
currently popular optimization models can’t show development.
this negative rate, because GDP per capita is assumed Carbon dioxide abatement is probably “target
to rise everywhere. Even if it does, however, disaggre- efficient,” helping mainly those society would prefer
gating shows that the beneficiaries will be both poorer to help. Poorer countries are usually going to be more
and more numerous than those who finance the incre- vulnerable to climate change than wealthier coun-
ments in consumption. tries. But direct investments in public health, birth
control, training and education, research, physical in-
frastructure, water resources, etc., can also be di-
THE DEVELOPMENT CONTEXT
rected to target populations, so abatement does not
A third implication is that if GDP per capita necessarily have the advantage.
continues to increase in most of the developing It is doubtful that developing countries would
world, as can be expected and as the optimization choose to defer consumption increments to later
Intergenerational Discounting 837

generations, whether what is deferred comes out of REFERENCES

their own resources or out of resources made avail- 1. Cline, William R. (1992). The economics of global warming.
able by wealthier countries. If offered a choice of Washington, DC: Institute for International Economics.
immediate development assistance or equivalent in- 2. Fankhauser, Samuel. (1993). The social costs of greenhouse gas
emissions: An expected value approach. University College and
vestments in carbon abatement, it is likely that po- University of East Anglia, Centre for Social and Economic Re-
tential aid recipients would elect the immediate. So search on the Global Environment, Norwich, U.K.
if we, the developed, elect carbon abatement for 3. Manne, Alan S. (1995). The rate of time preference: Implica-
tions for the greenhouse debate. Energy Policy, 23, 391–394.
their benefit, it is we who choose their descendants 4. Nordhaus, William D. (1994). Managing the global commons:
over themselves. The economics of climate change. Cambridge, MA: MIT Press.
© 2000 Society for Risk Analysis