The coauthors of this paper include widely acclaimed economists, bi...
![Imgur](https://imgur.com/6Q9wPWG.png). This is a great site th...
Our world in data presents energy use across several dimensions. ...
“This paper, the outgrowth of discussions among a group of ecologis...
What does "maximizing the discounted value" means?
Important: “By what criterion should we judge whether consumption ...
Important: "One can think of intertemporal social welfare Vt as a f...
Discount rates are used widely in economics and finance, for making...
The World Commission on Environment and Development report from 198...
This feels like an insurmountable hurdle...there won't be great "su...
Total factor productivity is a measure of productive efficiency tha...
Extremely important parameter for the model, opens up an entire phi...
"Underpricing of Natural Resources Relative to Social Cost". This c...
General findings: "We have identified several factors influencing con...
Lack of data/poor data quality is a major hurdle, and these factors...
Are We Consuming Too Much?
Kenneth Arrow, Partha Dasgupta,
Lawrence Goulder, Gretchen Daily, Paul Ehrlich,
Geoffrey Heal, Simon Levin, Karl-Go¨ran Ma¨ler,
Stephen Schneider, David Starrett and
Brian Walker
I
s humanity’s use of Earth’s resources endangering the economic possibilities
open to our descendants? There is wide disagreement on the question. Many
people worry about the growth in our use of natural resources over the past
century. Some of this increase reflects the higher resource demands from a growing
world population. But it also reflects the growth of per capita output and consump-
tion. During the twentieth century, world population grew by a factor of four to
more than 6 billion, and industrial output increased by a factor of 40. Per capita
y
Kenneth Arrow is Professor of Economics Emeritus, Stanford University, Stanford, Cali-
fornia. Partha Dasgupta is the Frank Ramsey Professor of Economics at the University of
Cambridge and Fellow of St John’s College, both in Cambridge, United Kingdom. Lawrence
Goulder is Professor and Shuzo Nishihara Chair in Environmental and Resource Economics,
Stanford University, Stanford, California. Gretchen Daily is Associate Professor of Biological
Sciences, Stanford University, Stanford, California. Paul Ehrlich is Bing Professor of Popu-
lation Studies, Stanford University, Stanford, California. Geoffrey Heal is Paul Garrett
Professor of Public Policy and Business Responsibility, Columbia Business School, New York,
New York. Simon Levin is Moffett Professor of Biology in the Department of Ecology and
Evolutionary Biology, Princeton University, Princeton, New Jersey. Karl-Go¨ran Ma¨ler is
Director of the Beijer International Institute of Ecological Economics, Stockholm, Sweden.
Stephen Schneider is Professor of Biological Sciences, Stanford University, Stanford, Califor-
nia. David Starrett is Professor of Economics Emeritus, Stanford University, Stanford Cali-
fornia. Brian Walker is a Research Scientist, CSIRO Sustainable Ecosystems Division,
Canberra, Australia. Please address all communications to Kenneth J. Arrow at
arrow@stanford.edu.
Journal of Economic Perspectives—Volume 18, Number 3—Summer 2004—Pages 147–172
consumption in industrialized nations today is far higher than it was 100 years ago,
and some would argue that this is irresponsible in the light of its implications for
resource demands. In the last 100 years, energy use has increased by a factor of 16,
annual sh harvesting by a multiple of 35 and carbon and sulfur dioxide emissions
by a factor of 10. The application of nitrogen to the terrestrial environment from
human use of fertilizers, fossil fuels and leguminous crops is now at least as great as
that from all natural sources combined (McNeill, 2000). If we look at specic
resources and services, such as fresh water, the atmosphere as a carbon sink, and a
wide variety of ecosystem services, evidence suggests that continuing growth in their
utilization rates is unsustainable (Vitousek, Ehrlich, Ehrlich and Matson, 1986,
1997; Postel, Daily and Ehrlich, 1996).
On the other hand, it may be claimed that, just as earlier generations invested
in capital goods, research and education to bequeath to current generations the
ability to achieve high levels of consumption, current generations are making the
investments that are necessary to assure higher real living standards in the future,
despite stresses on the natural resource base. Indeed, historical trends in the prices
of marketed natural resources and the recorded growth in conventional indices of
economic progress in currently rich countries suggest resource scarcities have not
bitten as yet (Barnett and Morse, 1963; Johnson, 2000). This optimistic viewpoint
emphasizes the potential of capital accumulation in the form of increased manu-
factured capital and human capital, as well as technological change, to compensate
for the diminishment of natural resources.
This paper, the outgrowth of discussions among a group of ecologists and
economists, offers an analysis that we hope will go some way toward reconciling the
conicting intuitions. The binocular vision that can be obtained from using both
ecological and economic insights raises questions that might not occur in either
viewpoint alone.
By what criterion should we judge whether consumption is or is not excessive?
Economic analysis has tended to emphasize the criterion of maximizing the present
discounted value of current and future utility from consumptionwhat we will call
intertemporal social welfare. Current consumption is deemed excessive or decient
depending on its level relative to that called for by this optimization problem.
However, analysts with a different perspective have applied a criterion of sustain-
ability, which emphasizes the ability of the economy to maintain living standards. In
this paper, we will examine these criteria for evaluating whether consumption is
excessive and identify factors in the economic and ecological domains that deter-
mine whether or not it is in fact so.
Alternative Criteria for the Evaluation of Consumption
In the framework presented here, the underlying elements of intertemporal
social welfare are consumption (broadly dened) and utility. Then the intertem-
poral social welfare V
t
at time t can be dened as the present discounted value of
148 Journal of Economic Perspectives
the ow of utility from consumption from the present to innity, discounted using
the constant rate
( 0).
1
In focusing at each moment on aggregate consumption,
our framework abstracts from intratemporal equity issues. The equity issues that
preoccupy us are intertemporal.
One of the determinants of V
t
is the productive base, which consists of
societys capital assets and institutions at t. The capital assets include manufactured
capital, human capital and natural capital. The productive base also includes the
knowledge base and societys institutions. Although institutions are frequently
regarded to be capital assets themselves, we will instead view institutions as guiding
the allocation of resourcesincluding capital assets. Institutions include the legal
structure, formal and informal markets, various agencies of government, interper-
sonal networks and the rules and norms that guide their behavior. In what follows,
we will generally employ the term technologyto refer to both the knowledge base
and the institutions that inuence, among other things, the effectiveness with
which this knowledge is put to use. We now discuss two criteria for judging whether
or not current consumption is excessive.
The Maximize Present Value Criterion
One can think of intertemporal social welfare V
t
as a function of initial
conditionsthe capital assets and level of technology at time tand the choices
from time t forward as to how to allocate output between investment and consump-
tion. According to the maximize present value criterion, actual consumption today is
excessive if it is greater than the level of current consumption prescribed by this
optimal consumption path. To put it another way, current consumption is excessive
if lowering it and increasing investment (or reducing disinvestment) in capital
assets could raise future utility enough to more than compensate (even after
discounting) for the loss in current utility.
The optimal path depends, among other things, on the discount rate,
.A
higher value for
, other things being equal, means that less weight is placed on
future utility. The right value of
has long been a matter of debate. Ramsey
(1928) argued that in a deterministic world the appropriate value of
is zero,
implying that the utility of future people ought to receive the same weight as that
1
Let s and t variously denote time (where s t). Let C(s) denote a societys aggregate consumption
and U(C(s)) the ow of utility at time s. Marginal utility is assumed to be positive. Let V
t
denote
intertemporal social welfare at time t,dened as the present discounted value of the ow of U(C(s))
from t to innity, discounted using the constant rate
( 0). Assuming continuous time, we have
Vt
st
U Cs兲兴e
st
ds.
In calling U utility, we are not necessarily subscribing to classical utilitarianism. We are assuming, more
generally, that consumption has a social worth, which we call utility U(C ), and that V
t
is a numerical
representation of an ethical ordering over innite utility streams beginning at t. Koopmans (1972) has
identied conditions on orderings that permit one to express V
t
in the form we are using here.
Kenneth Arrow et al. 149
of people today. Koopmans (1960), however, showed that applying a zero social
rate of pure time preference can lead to paradoxes. Lind (1982) and Portney and
Weyant (1999) contain diverse arguments concerning the appropriate choice of
.
One interesting aspect of the optimal consumption path according to the
maximize present value criterion is that it can be linked, theoretically, to the
outcome of a decentralized market economy. In a fully competitive economy with
a complete set of futures markets and no externalities, and in which individuals
discount their future utility at the social rate
, the time path of consumption will
correspond to the optimal consumption path. Conversely, in a world where markets
for many future goods and many types of risk bearing do not exist and where
environmental externalities prevail, consumption generally will not be optimal.
Nevertheless, it is possible to conduct social cost-benet analysis to judge whether
a policy reform at t increases intertemporal social welfare (Dasgupta, 2001a; Arrow,
Dasgupta and Ma¨ler, 2003b).
The Sustainability Criterion
An alternative yardstick for evaluating time proles of consumption is the
sustainability criterion. The terms sustainability and sustainable development
became commonplace after the report by the World Commission on Environment
and Development (1987), widely known as the Brundtland Commission (after its
chairperson). Sustainable development was dened as development that meets
the needs of the present without compromising the ability of future generations to
meet their own needs.Several interpretations of sustainability are compatible with
this phrase (Pezzey, 1992; Solow, 1992; Heal, 1998; Asheim, 2003). Here we take
sustainability to mean that intertemporal social welfare V must not decrease over
time. Thus, we will say that the sustainability criterion is satised at time t if
dV
t
/dt 0.
Several features of this criterion for sustainability deserve emphasis. First, the
criterion concentrates on the change in V, not on V s level. Second, even if a
consumption path were to satisfy the criterion today and at all future dates, it would
not guarantee that utility (U ) in each future moment will be as high as it is today
(Asheim, 1994). Third, the criterion does not identify a unique consumption path:
the criterion could in principle be met by many consumption paths. Fourth, if
exhaustible resources are sufciently important in production and consumption,
then it is conceivable that no sustainable development program exists (Dasgupta
and Heal, 1979, chapter 7). Fifth, satisfying the sustainability criterion today does
not guarantee that the criterion will be satised at all future times: a given
consumption path may imply a rising V over the interval from now to the next
period, but a falling V when it is evaluated over some future interval of time. Sixth,
in dening sustainable development, there is no presumption that the consump-
tion path being followed is optimal in the sense of maximizing V. And seventh, a
consumption path that satises the sustainability criterion need not be intertem-
porally efcient; an alternative, Pareto-superior path may well exist.
The notion of societys productive base is closely connected to the issue of
150 Journal of Economic Perspectives
whether the sustainability criterion is satised. As dened earlier, the productive
base is the stock of all societys capital assets at t, inclusive of manufactured capital
assets, human capital and natural capital. It also depends on the level of technol-
ogy. For now, we will regard the level of technology as xed; we will consider
technological change below.
Let genuine investment refer to the change in the productive base. Genuine
investment can be expressed as the sum of the values of investments or disinvest-
ments in each of societys capital assets, where the value of each investment is the
product of the change in the quantity of the asset times the shadow value or
accounting price of that asset.
2
Clearly, intertemporal welfare V(t) is nondecreas-
ing at t if and only if genuine investment is non-negative at t.
3
We can refer to
genuine investment as the change in societys genuine wealth.
This requirement that the productive base be maintained does not necessarily
entail maintaining any particular set of resources at any given time. Even if some
resources such as stocks of minerals are drawn down along a consumption path, the
sustainability criterion could nevertheless be satised if other capital assets were
accumulated sufciently to offset the resource decline.
While the criterion for sustainable development is straightforward to express,
implementing it poses severe empirical challenges. Measuring changes in quanti-
ties of capital stocks is very difcult. It is especially hard in the case of stocks of
natural resources such as minerals, fossil fuels, sh or insects. In evaluating the
social losses from reductions in natural resourcesand thus the alternative invest-
ments necessary to offset such lossesone must consider in principle all of the
contributions of natural resources to present and future utility. Such contributions
2
Let K
t
denote the vector of stocks of all capital assets at t. Plainly, V is a function of K
t
. In the particular
case where V is stationary (that is, where t itself does not directly inuence V ), we can write V
t
V(K
t
).
Let K
it
denote the stock of the i
th
capital good at date t. By the chain rule of differentiation,
dV/dt
i
V/K
it
兲共dK
it
/dt
i
p
it
I
it
,
where p
it
(V/K
it
) is the shadow price or accounting price of K
it
, and I
it
(dK
it
/dt) denotes the
rate of change in K
it
. The right-hand side of the expression above is genuine investment.
It is frequently convenient to divide the expression for dV/dt by the value of some one kind of capital
(valued at its shadow price). Thus, if there are just two kinds of capital, denoted, say, by K
1
and K
2
,
respectively, genuine investment measured relative to the rst kind of capital is
dV/dt/p
K
1
K
1
1/K
1
dK
1
/dt p
K
2
K
2
/p
K
1
K
1
兲兴共1/K
2
兲共dK
2
/dt.
This indicates that our sustainability criterion can be expressed as the growth rate of one kind of capital
plus the product of the growth rate of the other kind of capital and an adjustment factor. Note that, from
the denition of shadow prices above, the adjustment factor is the elasticity of K
1
with respect to K
2
along an isoquant of V(K
1
, K
2
). We apply this expression in a later section, where we extend the
expression for dV/dt to account for technological change.
3
This result was proved for optimally managed economies by Pearce and Atkinson (1995) and for
arbitrary economies by Dasgupta and Ma¨ler (2000).
Are We Consuming Too Much? 151
may be direct as, say, objects of natural beauty; or they may be indirect, as in the
contributions of ecosystem services such as water purication, ood control, cli-
mate stabilization, pollination of crops, control of agricultural pests and the gen-
eration and maintenance of soil fertility (Daily, 1997); or they may be both (a
wetland). Measuring these services is no easy task.
Another great challenge is determining how much more of one type of capital
asset would be needed to compensate for the loss of one unit of another type of
capital asset. Ecologists express concerns that natural resources have only a limited
set of substitutes. They fear that to the extent that economists are overly optimistic
about opportunities to substitute other capital assets for certain natural resources,
reductions in their stocks would then receive too little weight. A major goal of
ecological economics is to increase our understanding of the ways in which differ-
ent kinds of natural capital contribute to human well-being and the extent to which
they are substitutable for one another and for other kinds of capital assets.
4
Extending the Sustainability Criterion to Account for Changing Population
In initial discussion of intertemporal social welfare, V, we have abstracted from
the impact of changes in population size. Dealing with a changing population
presents empirical challenges: forecasting the time prole of population is quite
difcult. But the theoretical issues associated with a changing population may be
even more challenging, as many of the theoretical problems remain unresolved.
Should social welfare seek to maximize the sum of the utilities of all individuals, or
should it seek to maximize some form of average welfare over all peoplenow and
in the future? How can one compare the social value of a large population with
higher total utility versus a smaller population with higher average utility?
One way to proceed is to regard population as another asset, in addition to the
forms of capital we have already considered. This approach implies that for any
given conception of intertemporal social welfare, there is an accounting price of
population. Under this approach, genuine wealth is the accounting value of all
capital assets, including population. The sustainability criterion continues to re-
quire that genuine wealth at constant accounting prices must not decline, but now
invokes a broader notion of genuine wealth.
Of particular interest is the case where the intertemporal social welfare func-
tion is taken to be the discounted present value of the ow of total utility divided
by the discounted present value of population size over time, where the discount
rate used in both the numerator and denominator is the social rate of pure time
4
On substitutability, see, for example, Dasgupta and Heal (1979, chapter 7), Ehrlich and Ehrlich
(1990), Daily (1997), Daily et al. (2001), Levin (2001), Heal et al. (2001), and Heal (2003). An
important complication related to assessing substitution possibilities is that the potential for substitution
can vary by location. A natural resource in one place, like a local woodland, is not the same economic
commodity as the same natural resource in another place. Dasgupta (1993) discusses the implications
of local nonsubstitutability for the worlds poorest people, who often have no substitutes available to
them when their local resource base is degraded. For an interchange on the relative importance of
substitution possibilities at aggregate and local levels, see Johnson (2001) and Dasgupta (2001b).
152 Journal of Economic Perspectives
preference,
. This is a form of dynamic average utilitarianism.
5
Dasgupta (2001a)
has shown that if dynamic average utilitarianism represents intertemporal social
welfare, then under certain conditions a nondeclining V
t
is equivalent to the
intuitively appealing requirement that genuine wealth per capita must not decline.
The conditions in question are i) population changes at a constant rate; ii) per
capita consumption is independent of population size (but presumably dependent
on per capita capital assets); and iii) all transformation possibilities among goods
and services exhibit constant returns to scale. We shall make use of this nding
when deriving certain empirical results later in this paper.
Extending the Sustainability Criterion to Account for Technological Change
In the presence of technological change, output and consumption could rise,
and thus intertemporal social welfare V could increase, even if aggregate invest-
ment in terms of manufactured, human and natural capital were negative. How can
we take such changes into account in assessing changes in intertemporal social
welfare, V ?
Here we enter uncharted territory, since there appears to be no prior literature
that makes such a connection. We explore this relationship by examining how total
factor productivity alters the assessment of V. In a rather special modelone with
strong simplifying assumptionswe derive a formula for incorporating total factor
productivity in V. Let
represent the rate of growth of total factor productivity.
is often called the residual, since it is what remains after subtracting the
inuence of other factors on output growth. Let K (without subscript) represent a
composite of societys natural, physical and human capital assets.
6
Let
be the
elasticity of output with respect to K. Later in the discussion we will observe that
genuine savings rates are close to 0. A straightforward argument shows that with
zero rates of savings, total factor productivity growth at rate
raises the growth rate
of intertemporal social welfare (measured in units of capital) by
/
.
7
5
The formula for V in this case is
Vt
t
NsUcs兲兲e
st
ds
t
Nse
st
ds,
where c(s) and N(s) represent per capita consumption and population size, respectively, at time s.
Notice that, to the extent that the denominator of the above form of V
t
is not affected by policy, the
maximize present value criterion implies the same optimum no matter whether total or dynamic
averageutilitarianism is adopted as the conception of intertemporal social welfare. However, it matters
hugely whether total or dynamic average utilitarianism is adopted when one is applying the
sustainability criterion. For further analysis of sustainability under changing population size, see Arrow,
Dasgupta and Ma¨ler (2003a).
6
K can also be regarded as a measure of genuine wealth prior to the adjustment in such wealth to
account for technological change.
7
Assume that K(t), the productive base, is a constant ( K
0
) and that consumption output A
0
K
0
e
rt
for all t. From the denition of V in footnote 2, V is constant when A
0
K
0
is constant. Thus, the elasticity
of A
0
(total factor productivity) with respect to K
0
along an isoquant of V is
1
. In the second equation
Kenneth Arrow et al. 153
Thus, after calculating genuine investment and the associated change in
genuine wealth, one can account for technological change by adding
/
to the
initially estimated growth rate of genuine wealth. It should be noted that there may
be serious problems with published estimates of total factor productivity and its
growth rate,
, because national accounts do not include the economys use of
nonmarketed natural and environmental resources. Suppose that over a period of
time, an economy extracts from its natural resource base at an increasing rate.
Then the estimate of total factor productivity growth would be biased upward: some
of the estimated growth in total factor productivity would actually be due to
increased resource use rather than an increase in the knowledge base or improve-
ments in institutional performance. On the other hand, if the rate of natural
resource extraction were falling, recorded growth of total factor productivity would
understate the true contribution of changes in the knowledge base. Our empirical
investigation later in this paper relies on estimates of total factor productivity from
models that do not consider natural resource extraction. Ideally, estimates of the
productivity residual should derive from models that incorporate such extraction.
8
Does Satisfying One Criterion Imply that the Other is Satisfied?
The two criteria we have offered for assessing whether consumption is too high
reect different ethical considerations. An economic program that satises the
sustainability criterion need not satisfy the maximize present value criterion. Con-
versely, an economic program that is optimal under the maximize present value
criterion might not satisfy the sustainability criterion.
To see this, suppose that aggregate output is a Cobb-Douglas function of
manufactured capital and the ow of an exhaustible natural resource. (Returns to
scale may be either constant or decreasing.) Solow (1974) showed that sustainable
development is technically feasible if the output elasticity with respect to manufac-
tured capital exceeds the output elasticity with respect to the ow of the natural
resource in production. Dasgupta and Heal (1979) showed, however, that so long
as
0, consumption paths that are optimal in the sense of maximizing intertem-
poral welfare V involve consumption approaching zero in the long run; thus, they
do not yield development that is sustainable over the long term.
The differences between the two criteria have potential implications for public
policy. An investment project that passes the social cost-benet test (namely, that its
acceptance would increase todays V ), could result in a decrease in V at some future
date. Therefore, the standard policy remedies for improving economic efciency
in footnote 2, substitute the productive base K for what was termed K
1
and total factor productivity A
0
for what was termed K
2
. Since (1/A)(dA/dt)
, it follows from this equation that the sustainability
criterion (expressed in terms of K ) is the rate of growth of K plus
/
.
8
In an appendix to this paper available at http://www.stanford.edu/goulder, we derive a formula for
correcting the bias in estimating the rate of growth of total productivity induced by the omission of
resources.
154 Journal of Economic Perspectives
like establishing property rights, addressing externalities and so forthdo not
guarantee sustainability.
Empirical Evidence Relevant to the Maximize Present Value
Criterion
According to the maximize present value criterion, todays consumption is
excessive if it is higher than the level of current consumption along the consump-
tion path that maximizes the current intertemporal welfare V. No one can seriously
claim to pinpoint the optimal level of current consumption for an actual economy.
However, theoretical considerations can identify factors that would cause current
consumption to be different in a predictable direction from the optimal level.
Relevant theoretical issues include the relationship between market rates of return
on investment and optimal social interest rates on consumption, and the relation-
ship between market prices of contemporaneous goods (including those of current
capital goods) and the social costs of those commodities.
The Market Rate of Return on Investment and the Social Rate of Interest on
Consumption
For a consumption path in a market economy to be socially optimal, the
market rate of return on investment, i, must be equal to the social rate of interest
on consumption, denoted by r.Ifi exceeds r, markets are biased toward insufcient
saving and excessive current consumption.
It can be shown that, if intertemporal social welfare is given by the form of V
we have postulated here, the social rate of interest on consumption r is given by the
relation r
g, where
is the social rate of pure time preference,
is the
elasticity of marginal (social) utility, and g is the rate of growth in aggregate
consumption.
9
The parameter
reects impatience, as is well known. The second
term in the equation,
times g, may require explanation. One can view
as
accounting for the fact that, to the extent that future generations have higher
incomes, their consumption will be higher and the marginal utility of their con-
sumption will be lower. However,
need not simply reect diminishing marginal
utility of consumption to an individual: it can be interpreted as a social preference
for equality of consumption among generations. Thus, its function can be similar
to the social rate of pure time preference,
.
Does i exceed r? One might argue that this cannot be answered from empirical
observation alone, on the grounds that the choices of
and
are inherently value
judgmentssubjective matters. However, there exists a different approach that
offers a link to empirical behavior. This approach asserts that a typical individuals
preferences should guide our choices of
and
. In particular, it identies the
9
For one derivation, see Arrow and Kurz (1970). For additional discussion, see Arrow et al. (1996).
Are We Consuming Too Much? 155
social rate of pure time preference,
, with the utility discount rate that a typical
individual would endorse. The assumptions of this approach are clearly subject to
debate, but it seems worthwhile to consider what they imply for the relationship
between i and r and thus for the level of consumption.
Several considerations suggest that the typical individuals preferences regard-
ing the utility discount rate will not be reected in the market.
10
One reason is that,
because of externalities, the utility discount rate that most people would endorse as
socially conscious beings is lower than the rate of time preference emanating from
market transactions. Many years ago, Ramsey, Pigou and Harrod insisted that the
only ethically justiable value for
is zero. Solow (1974, p. 9) put the matter thus:
In solemn conclave assembled, so to speak, we ought to act as if the social rate of
(pure) time preference were zero. A less extreme view would be to say that even
if we dont treat
as zero, individuals do derive a positive externality (outside of the
marketplace) from the welfare of future generations. An argument combining the
market rate of return on investment with the externality of caring about future
generations might call for low values of
, in the range, say, of 00.5 percent per
annum.
What about the other term
gin the expression for r? If we again base our
choices on a typical individuals preferences, then
should reect an average
persons elasticity of marginal utility of consumption. The value of
is linked to
,
the intertemporal elasticity of substitution in consumption:
⫽⫺(
1)/
. Halls
(1988) time series estimates of
suggest that plausible values for
might lie in the
range of 24. If we assume per capita growth rates in consumption ( g) to have been
of the order of 1.5 percent per annum, we arrive at 3 to 6 percent per annum for the
term
g. Together, these considerations would tentatively suggest a value for r in
the range 3.0 6.5 percent per annum.
How does this value compare with the market rate of return on investment i?
We need to choose an appropriate market interest rate with which to make the
comparison, but rates of return on different securities vary considerably. The real
rate of return on private capital in the United States (as directed for use in
government project analysis, United States Ofce of Management and Budget,
1992) is 7.0 percent; on equities from 19702000, 7.4 percent; but on Treasury bills
from 19702000, just 1.6 percent. (The last two gures are derived by computation
from the Wharton-Data Resources website.) We want to compare the social rate of
interest on consumption r with a risk-free rate. If one interprets Treasury bills as the
risk-free asset, then the market rate seems as low or lower than the value for r we
10
Hence, the justication for a lower
may remain, despite the fact that equilibrium outcomes from
individualsmarket behavior at rst blush suggests a higher
. On this, see Marglin (1963), Lind (1964)
and Sen (1967). In discussing this issue, we have just referred to an externality related to individuals
concerns for future generations well-being. Elsewhere in this paper, we refer to externalities associated
with the use of natural resources. One could interpret individuals inefcient (excessively rapid)
exploitation of natural resources as evidence that they care little about future welfare. However, we
prefer the interpretation that such externalities make it impossible for individuals to express their
concerns about the future in their individual production or consumption decisions.
156 Journal of Economic Perspectives
calculated above. Thus, this rough comparison provides little support for the
argument that consumption is excessive.
Two other pieces of evidence that market rates of return on investment may
exceed the social rate of interest on consumption come from the incompleteness
of markets and the existence of capital taxes. The absence of a complete set of
risk-bearing markets, for example, implies that risks cannot be pooled perfectly.
Returns from investment are therefore more uncertain, and so recipients (savers)
would be expected to attach a lower value to investments than they would have if
risks had been pooled more effectively. Consequently, the rate of saving is lower.
The absence of complete pooling tends to promote excessive consumption.
Similarly, the taxation of capital income lowers the private return on capital
below the social return and thus discourages saving and promotes excessive con-
sumption. However, the normative impacts of capital income taxation depend on
other taxes and the prices of various inputs to consumption goods. Labor taxes can
discourage labor supply and reduce wage incomes, thus causing current consump-
tion to fall short of the optimal level despite the inuence of capital taxes.
Moreover, the level of current consumption depends importantly on the prices of
the inputs used to produce consumption goods. The pricing of natural resource
inputs seems especially important, as we discuss in the following section.
Underpricing of Natural Resources Relative to Social Cost
The level of consumption depends not only on market interest rates, but also
on the price of current consumption relative to its social cost or, more specically,
the price of consumption goods relative to capital goods. To the extent that
consumption goods are priced below their social cost, consumption will tend to be
excessive.
Some natural resources are consumption goods, others are direct or indirect
inputs in the production of consumption goods, and many are both. The under-
pricing of natural resources may contribute to the pricing of consumption goods
below their social cost. Such underpricing also alters the relative prices of different
consumption goods, thereby leading to inefciencies in the composition (as well as
overall level) of consumption: too much consumption of resource-intensive goods
and services relative to consumption of other goods. Thus, when natural resource
inputs are priced below social cost, both the overall level and the composition of
consumption can be affected in ways that lead to excessive natural resource use.
The underpricing of natural resources can stem from at least three sources.
First, insecure or poorly dened property rights can lead to excessively rapid
resource exploitation if the exploitation does not require much prior investment
(Bohn and Deacon, 2000). Second, natural resource underpricing can arise from
the failure of the market to incorporate the (negative) externalities associated with
the use of natural resources. Examples of such externalities include the various
damages stemming from the use of fossil fuels (such as acid precipitation or climate
change), and the loss of such ecosystem services as ood control, water-ltration
and habitat provision when wetlands are drained for conversion to farms.
Kenneth Arrow et al. 157
Third, use of natural resources may be underpriced because of government
subsidies. The World Banks 1992 World Development Report (Figure 3.2) examined
fossil fuel, electricity and water prices in 32 developing countries. In all but three
of those countries, subsidies caused prices to fall below cost, even before account-
ing for potential externalities. Similarly, the International Energy Agency (1999)
has estimated that in India, China and the Russian Federation, full-cost pricing
would reduce energy consumption by 7, 9 and 16 percent, respectively. In these
countries, most of the departure from social cost pricing is attributed to energy
subsidies. For estimates of aggregate global subsidies on the use of environmental
and natural resources, see Myers and Kent (2000).
The inuence of OPEC on the international market for oil could function as
a counterbalance to the above arguments, potentially raising world oil prices up to
or beyond social cost. However, there is as yet no clear consensus as to whether
current world prices are above or below social cost. In addition, the pricing of oil
might well be the exception to a more general pattern of underpriced natural
resources.
The underpricing of natural resource inputs can also reduce the prices of
investment goods. If such underpricing is especially pronounced for investment
goods, it could promote a higher ratio of investment to consumption. This does not
undo the problem of excessive use of natural resources, which is a more funda-
mental concern. If the natural resource inputs in the production of investment
goods are underpriced, the rates of depletion of natural resources will be too rapid.
The rate of accumulation of manufactured capital may or may not exceed the
optimal rate, but genuine investmentthat is, overall investment, inclusive of
changes in stocks of natural capitalis likely to be insufcient because of the
underpricing of natural resources and the associated excessively rapid depletion of
these resources.
Interdependence in Consumption
Interdependence in consumption can also lead to prices of consumption
below social cost. Building on Veblen (1899) and Duesenberry (1949), a small but
growing body of empirical work (for example, Frank, 1985a, b; Ng, 1987; Howarth,
1996; Schor, 1998) suggests that a persons sense of well-being is based not only on
a persons own consumption, but also on the persons consumption relative to a
reference group. When others consumption rises in comparison, an individual
could suffer from a loss of well-being because that persons relative consumption
now falls. This interdependence in consumption can be viewed as an externality,
which can compel individuals to work harder and consume more to keep up with
neighbors. It is individually rational behavior, but collectively suboptimal. A formal
growth model incorporating these ideas has been developed by Cooper, Garcia-
Pen˜alosa and Funk (2001).
However, interdependence in consumption does not necessarily imply that
people consume excessively. Suppose that the relative consumption effect applies
not only to current consumption, but also to future consumption (and to leisure,
158 Journal of Economic Perspectives
including sleep). Working harder and consuming more today would then improve
ones relative current consumption, but worsen ones relative current consump-
tion of leisure and also ones relative future consumption! Thus, the bias from
interdependence depends on the strength of the effects along various margins. If
the effect on individual well-being of relative consumption is symmetric, operating
equally on relative consumption and relative leisure now and in the future, it may
have no impact on current behavior relative to what would occur if this effect were
absent. In this case, the relative consumption effect operates like a lump-sum tax,
reducing a persons sense of well-being without changing that persons allocation
of labor resources or income.
General Findings
We have identied several factors inuencing consumption and have shown
how they enable us to judge whether consumption is excessive according to the
maximize present value criterion. Several factorsthe inability to pool risks per-
fectly, the taxation of capital income and the underpricing of natural resources
contribute toward excessive consumption. (Below, we will observe that these same
factors also work toward excessive consumption according to the sustainability
criterion.) Among these imperfections, the underpricing of natural resources
strikes us as the most transparent.
Empirical Evidence Relevant to the Sustainability Criterion
A First Step: Measuring Genuine Investment
As mentioned, genuine investment at date t is the sum of the values of changes
in capital stocks at t, evaluated at their accounting prices. It is the change in
genuine wealth at constant accounting prices. The assets to be included are
manufactured capital, human capital, natural capital and the knowledge base
(Dasgupta and Ma¨ler, 2000; Arrow, Dasgupta and Ma¨ler, 2003b). In this subsection,
we begin evaluating whether various nations are meeting the sustainability require-
ment by estimating and observing the sign of genuine investment. In the next
subsection, we will extend the analysis to account for changes in population and
technological change.
A growing body of research now aims to measure genuine investment in
various countries. The lead has been taken by Kirk Hamilton and his collaborators
at the World Bank (for example, Pearce, Hamilton and Atkinson, 1996; Hamilton
and Clemens, 1999; Hamilton, 2000, 2002). Hamilton and Clemens (1999), in
particular, offered estimates of genuine investment for nearly every nation for the
year 1998. They estimated genuine investment by rst adding net investment in
human capital to existing country-level estimates of investment in manufactured
Are We Consuming Too Much? 159
capital. They then made a further adjustment to account for disinvestment in
natural resources and environmental capital.
11
To estimate the accumulation of manufactured capital, Hamilton and Clemens
(1999) used gures for net national saving. To estimate the accumulation of human
capital, they used expenditure on education. To account for disinvestments in
natural resources and environmental capital, they considered the net changes in
the stocks of commercial forests, oil and minerals and in the quality of the
atmosphere in terms of its carbon dioxide content.
12
Interestingly, the authors found that in 1998, genuine investment was positive
in all the rich nations of the world and in many of the poorer nations as well.
However, for 33 of the worlds poor countries, including many of the nations of
north Africa and sub-Saharan Africa, they estimated that genuine investment was
negative.
Here we adopt the Hamilton-Clemens approach as a starting point for evalu-
ating whether selected nations and regions meet the sustainability criterion. Rather
than calculate the gures for a single year (as in Hamilton-Clemens), we have
calculated annual averages over the past three decades, using annual data from the
World Development Indicators published in the World Banks website at http://
devdata.worldbank.org/dataonline. We present our estimates in Table 1. The
table includes poor nations (China, nations in the Indian subcontinent and nations
in the sub-Saharan Africa region), oil-exporting countries (in the Middle East/
North Africa region) and industrialized nations (the United States and the United
Kingdom).
The differences between genuine investment and the standard measure, net
domestic investment, are particularly striking for the Middle East/North Africa and
sub-Saharan Africa regions. In these regions, the loss of natural resources more
than offset the accumulation of manufactured capital (as reected in domestic net
investment) and human capital (as reected in expenditure on education). For the
United States and the United Kingdom, estimated genuine investment exceeded
domestic net investment, since the increase in human capital exceeded the value of
natural resource depletion.
These gures give an initial glimpse of the change in V. We will shortly
consider how the picture is altered when changes in technology and in population
are taken into account. Before doing so, however, it is useful to consider potential
11
Hamilton and Clemens (1999) employ the term genuine domestic saving, which we treat as
synonymous with genuine investment. The two cannot be distinguished from the data employed in their
study, because the data do not include international capital ows.
12
In calculating the reduction in natural resource stocks, Hamilton and Clemens do not include
discoveries of new reserves as an offsetting element. As an accounting matter this seems correct. It is
consistent with the notion that the overall global stock of mineral or fuel resources is giventhat
devoting resources toward exploration, and the subsequent discovery of previously unknown reserves,
does not constitute an enlargement of the stock. At the same time, expenditure on exploration
equipment or structures should constitute a positive investment in physical capital.
160 Journal of Economic Perspectives
biases in the estimation of genuine investment.
13
First, as Hamilton and Clemens
emphasize, one serious problem is the lack of comprehensive data in national
accounts. For many important (and declining) natural resource stocks, data are not
availableat least not for every country. Among the natural resources not included
in the Hamilton-Clemens study are water resources, forests as agents of carbon
sequestration, sheries, air and water pollutants, soil and biodiversity. So there is an
undercount, possibly a serious one.
Second, these estimates (as well as earlier estimates by Hamilton and Clemens)
employ market prices in indicating potential trade-offs among different forms of
capital. To the extent that natural capital is typically underpriced, the use of market
prices could bias the estimate of genuine investment in an upward direction.
Third, these estimates consider only very broad categories of capital. At a high
level of aggregation, one can miss critical bottlenecks that impose limits on substi-
tution possibilities. The rural poor in the worlds poorest countries, for example,
often cannot nd substitutes when their water holes vanish and the local woodlands
recede (Dasgupta, 1993). Aggregated calculations can easily miss these details and
underestimate the accounting prices of local resource bases.
Fourth, in these calculations, disinvestment in environmental capital is calcu-
13
For a more extensive discussion of ways in which genuine investment could be better estimated, see
Arrow, Dasgupta and Ma¨ler (2003b).
Table 1
Genuine Investment and Components as Percentage of GDP
Country
Domestic net
investment
Education
expenditure
Natural Resource Depletion
Genuine
investment
Damage from
CO
2
emissions
Energy
depletion
Mineral
depletion
Net forest
depletion
Bangladesh
19732001
7.89 1.53 0.25 0.61 0.00 1.41 7.14
India
19702001
11.74 3.29 1.17 2.89 0.46 1.05 9.47
Nepal
19702001
14.82 2.65 0.20 0.00 0.30 3.67 13.31
Pakistan
19702001
10.92 2.02 0.75 2.60 0.00 0.84 8.75
China
19822001 (without 1994)
30.06 1.96 2.48 6.11 0.50 0.22 22.72
Sub-Saharan Africa
197482; 19862001
3.49 4.78 0.81 7.31 1.71 0.52 2.09
Middle East & North Africa
197689; 19912001
14.72 4.70 0.80 25.54 0.12 0.06 7.09
United Kingdom
19712001
3.70 5.21 0.32 1.20 0.00 0.00 7.38
United States
19702001
5.73 5.62 0.42 1.95 0.05 0.00 8.94
Source: Authors calculations, using data from World Bank (2003).
Kenneth Arrow et al. 161
lated simply as the estimated damage associated with annual emissions of carbon
dioxide. A ton of carbon dioxide emission is assumed to lead to a loss of $20 (U.S.)
of environmental capital. However, to the extent that newly manufactured capital
is expected to yield increased carbon dioxide emissions and thereby inict
subsequent damage, the value of investment in manufactured capital is reduced.
14
The Hamilton-Clemens approach adopted here ignores this effect. A further
problem is that this approach implicitly links all of the damage from carbon dioxide
emissions to the country responsible for the emissions, whereas in fact a given
countrys emissions of carbon dioxide produce climate impacts worldwide.
Set against these considerations is another issue that is ambiguous in its effect.
The proxy for increases in human capitalexpenditure on educationneglects
depreciation of human capital due to morbidity, mortality and retirement from the
workforce. In this respect, it overstates the increase in human capital. On the other
hand, this proxy ignores skills gained through channels other than formal educa-
tion, as well as improvements in human productivity that are due to expenditures
on health and nutrition. This implies the opposite bias.
While the measure of genuine investment in Table 1 conveys useful informa-
tion, it does not consider changes in population or technology. If population grows
fast enough, the long-run productive base per person could decline even if genuine
investment were positive. On the other hand, accounting for improvements in
productivity would likely brighten the picture. In what follows, we extend our
empirical assessment to consider the impacts of population growth and changes in
total factor productivity.
Population Growth, Technological Change and Sustainability
The rst column in Table 2 reproduces the gures from the last column of
Table 1, which are estimates of average genuine investment as a proportion of GDP
over the interval 19702001. The second column is an initial, unadjusted estimate
of the growth rate of genuine wealth. We arrive at this column by multiplying the
numbers in the rst column by a presumed GDP/wealth ratio for the country or
region in questionthe assumed average GDP/wealth ratio over the three-decade
interval. Published estimates of GDP-wealth (or output/wealth) ratios tradition-
ally have been taken to be something like 0.200.30 per year. However, a wide array
of capital assets like human capital and many types of natural capital are missing
from national accounts. To offset this bias, Table 2 uses the gure 0.15 per year for
poor and oil-rich countries and regions and 0.20 per year for industrialized coun-
tries. Let W refer to per capita genuine wealth. We arrive at the growth rate of W
14
This does not double count. In todays genuine investment calculation, todays emissions of CO
2
constitute a reduction in climate-system capital. This negative investment is the present value of the
damages associated with these current emissions. At the same time, the value of todays investments in
manufactured capital should be the present value of the net services associated with these manufactured
assets. Future emissions and climate damages associated with these assets reduce these net service
streams and imply a reduction in the value of current investment in these assets.
162 Journal of Economic Perspectives
(column 4) by subtracting the population growth rate (column 3) from the growth
rate of genuine wealth.
Column 4s gures for changes in per capita genuine wealth do not account
for technological change. The remaining adjustments in the table are intended to
account for such change as measured through changes in total factor productivity.
Column 5 offers estimates of the growth of the total factor productivity residual, as
reported for the period 1970 2000 in Klenow and Rodrı´guez-Clare (1997).
15
For
the Middle East/North Africa and Sub-Saharan Africa regions, we obtain the
residual by taking a weighted average of the estimates for the countries within each
region, using GDP as weights. In the case of China, we report estimates from Collins
and Bosworth (1996) for a comparable period, since Klenow and Rodrı´guez-Clare
did not offer estimates for this country. The residual was negative only in the
Middle East/North Africa region.
We use the numbers in column 5 to arrive at an estimate of the change in per
capita genuine wealth that accounts for the impact of projected technological
15
Klenow and Rodrı´guez-Clare report the growth rates of a transform of total factor productivity. From
this information we calculate the associated growth rate of total productivity.
Table 2
Growth Rates of Per Capita Genuine Wealth
Country
(1) (2) (3) (4) (5) (6) (7)
Genuine
Investment
as Percent
of GDP
Growth Rate
of Unadjusted
Genuine Wealth
Population
Growth
Rate
Growth Rate of
Per Capita
Genuine
Wealthbefore
TFP Adjustment
TFP
Growth
Rate
Growth Rate of
Per Capita
Genuine
Wealthafter
TFP Adjustment
Growth
Rate of
per capita
GDP
Bangladesh 7.14 1.07 2.16 1.09 0.81 0.30 1.88
India 9.47 1.42 1.99 0.57 0.64 0.54 2.96
Nepal 13.31 2.00 2.24 0.24 0.51 0.63 1.86
Pakistan 8.75 1.31 2.66 1.35 1.13 0.59 2.21
China 22.72 3.41 1.35 2.06 3.64 8.33 7.77
Sub-Saharan
Africa 2.09 0.31 2.74 3.05 0.28 2.58 0.01
Middle East/
North Africa 7.09 1.06 2.37 3.43 0.23 3.82 0.74
United Kingdom 7.38 1.48 0.18 1.30 0.58 2.29 2.19
United States 8.94 1.79 1.07 0.72 0.02 0.75 1.99
Note: These calculations employ the following parameters: output-capital ratio, poor countries/regions
0.15; output-capital ratio, rich countries 0.20;
(share of human and reproducible capital in output)
0.58.
Data for genuine investment, population growth, and GDP growth derive from the World Bank (2003).
The genuine investment percentages of GDP derive from data over the time-intervals indicated in Table
1. The population growth rate is the average rate over the period 19702000.
The estimate for Chinas total factor productivity (TFP) growth is from Collins and Bosworth (1996). For
all other countries or regions, the estimates are from Klenow and Rodriguez-Clare (1997).
Are We Consuming Too Much? 163
change. Here we employ the approximation formula referred to earlier (and
derived in footnote 7). The formula asserts that the contribution of total factor
productivity growth to movements in intertemporal social welfare expressed in
units of overall capitalthat is, the contribution to movements in V/(p
K
K )is
/
*, where
is the growth rate of the total factor productivity residual and
*is
the elasticity of output with respect to all (manufactured, human and natural)
capital.
* can be calculated as the sum of the exponents of each of the types of
capital in the aggregate production function. Ideally, one would want to calculate
* from a production function that includes natural capital. However, we do not
have estimates from such a production function. Instead, we rely on the aggregate
production function estimated by Klenow and Rodrı´guez-Clare (1997), which
includes human and manufactured capital only. The exponents on these two types
of capital are 0.30 and 0.28, respectively. This implies
* 0.58 and suggests that
the adjustment for total factor productivity growth is 1.72 times the value in column
5 of Table 2. This is added to column 4 to yield column 6, our adjusted estimate of
the growth rate of W (per capita genuine wealth), which takes account of total
factor productivity growth.
This adjustment may well overstate the contribution of total factor productivity
growth, for two reasons. First, the omission of natural capital biases downward the
estimate of
* and thus biases upward the total factor productivity adjustment,
/
*. Suppose that in the true aggregate production function, natural capitals
share or exponent were 10 percent, and the shares of human and reproducible
capital were therefore 10 percent lower than indicated above (to preserve constant
returns). In this case,
* would be 0.62 and the adjustment would be 1.61 (rather
than 1.72) times the value in column 5. Moreover, as mentioned earlier, the neglect
of natural capital in the production function implies that
will be misestimated,
with the direction of the error depending on the difference in the growth rates of
natural resource extraction and overall output. We nd that growth rates of energy
extraction (a potential proxy for natural resource extraction) tend to exceed GDP
growth rates over the past three decades, which implies an upward bias in the
estimate of
. Thus, both
and
* may be misestimated in a way that leads to an
exaggeration of the impact of technological change. In future work, we hope to
arrive at improved estimates of the adjustment factor
/
*. However, for the
present paper we employ in Table 2 the values that emerge directly from the
Klenow and Rodrı´guez-Clare (1997) estimates, which appear to be the most satis-
factory estimates currently available.
The Poor World
Our (rather tentative) estimates in Table 2 suggest that parts of the poor world
and some oil exporting regions are not meeting the sustainability criterion. Popu-
lation growth implies signicant differences between the initially estimated growth
rate of genuine wealth (column 2) and the growth rate on a per capita basis
(column 4). And the results in column 6, which account for both population
164 Journal of Economic Perspectives
growth and technological change, differ signicantly from the original genuine
investment calculations.
Pakistan, Bangladesh, India and Nepal show positive values for the growth rate
of W. However, estimated growth rates for per capita genuine wealth are projected
to be negative in the Sub-Saharan Africa and Middle East/North Africa regions. At
an annual rate of decline of 2.6 percent in genuine wealth per annum, the average
person in the sub-Saharan Africa region becomes poorer by a factor of two about
every 25 years. The ills of sub-Saharan Africa are routine reading in todays
newspapers and magazines. But they are not depicted in terms of a decline in
wealth.
In the Middle East/North Africa region, per capita genuine wealth is
estimated to be declining at an annual rate of 3.8 percent, after adjusting for
technological change. For this region, the estimate for the growth rate in total
factor productivity is negative (0.23 percent). Hence, adjusting for techno-
logical change for this region further increases the estimated decline of per
capita genuine wealth. It is worth noting that the negative estimate for total
factor productivity growth depends closely on the value one attaches to ex-
tracted energy resources. Alternative assumptions about future energy prices
can signicantly inuence this valuation and the estimate of total factor pro-
ductivity. Thus, the uncertainties in the estimates are especially great in the case
of the Middle East/North Africa region.
The results for China are strikingly different from those just discussed. The
unadjusted growth rate of per capita genuine wealth is signicantly positiveabout
2.1 percent. The estimated growth rate of total factor productivity (3.6 percent) is
high compared with the other countries examined, helping to augment the growth
rate of W. However, Chinas gure could be biased upward: the estimates of
genuine investment do not include soil erosion or urban pollution, both of which
are thought by experts to be especially problematic in China.
How do changes in wealth per head compare with changes in conventional
measures of economic progress? The far-right column of Table 2 contains estimates
of the rate of change of GDP per head over the interval 1970 2000. In all of the
listed countries except for China, the growth rate of per capita GDP is considerably
higher than the estimated growth rate of per capita genuine wealth. In the Middle
East/North Africa Region, per capita GDP growth is positive, while per capita
genuine wealth is falling. For all of these countries except China, an assessment of
long-term economic development would be signicantly off the mark if it were to
look at growth rates in GDP per head.
One might infer from the table that several poor countries are consum-
ing too much. Such a conclusion would be off the mark. In many poor
nations, the production of both capital goods and consumption goods is
highly inefcient. The countries simultaneously suffer from low levels of
genuine investment and consumption, and in the most important sense of
the term, these nations do not overconsume. One cannot assure a satisfac-
tory quality of life in these nations simply by devoting a larger share of
Kenneth Arrow et al. 165
productive factors toward the production of capital goods. Indeed, devoting a
greater share of output toward investment could cause considerable misery by
reducing what is already a very low level of per capita consumption. For these
nations, the sustainability problem is part of a larger problem of inefcient
production and low productivity, which accounts for both low genuine invest-
ment and low consumption.
The Rich World
Table 2 also displayed estimated growth rates of W for the United States and
the United Kingdom. These are positive for both countries, although the U.S.
estimate is about a third of the UK estimate. The estimates of growth rates of per
capita GDP are also positive. Thus, for these countries, the differences across the
two indices of economic development are less dramatic.
One might be tempted to conclude that the rich countries are avoiding
consuming too much. But it is important to note that the gures for changes in
per capita wealth for different countries are not entirely independent. The
success of rich countries may in part be due to the failure of poorer nations.
As we noted earlier, natural capital is very frequently underpriced in the market
because property rights may be poorly dened or poorly enforced. In extreme
cases, such capital assets are free. Dasgupta (1990) and Chichilnisky (1994) have
used this fact to argue that countries that are exporting resource-based products
(they are often among the poorest) are to an extent subsidizing the consump-
tion of those countries that are importing these products (they are often among
the richest). Such hidden subsidies would help promote positive growth rates in
per capita wealth in rich countries, while working toward reduced growth rates
in the poorer nations that export resource-based products. High levels of
consumption in rich countries may promote excessive resource degradation in
poor countries, which imperils well-being in the poorer countries. This negative
by-product of rich nations consumption is not captured in existing measures of
changes in per capita wealth.
Sensitivity of Results to the Assumed GDP-Wealth Ratio
One signicant parameter used in our estimates of the growth of W is the
GDP-wealth ratio. This parameter is employed to translate genuine investment into
a growth rate of (unadjusted) genuine wealth. Table 3 indicates the sensitivity of
our estimates to this ratio. The middle column reproduces the results in Table 2 on
the growth rate of per capita annual wealth. The left and right columns, respec-
tively, consider a lower and higher value for the GDP-wealth ratio. For countries
with positive genuine investment, a lower value for this ratio implies a higher
beginning-of-period level of wealth; thus, for a given level of genuine investment, it
implies a lower growth of adjusted per capita genuine wealth. For the regions with
negative genuine investment (the Sub-Saharan Africa and Middle East/North
Africa regions), a lower value implies a higher (that is, less negative) growth rate of
166 Journal of Economic Perspectives
genuine wealth. Results seem fairly sensitive to this parameter. When the output-
wealth ratio is 0.10, the estimated growth of W becomes negative in Bangladesh,
India, Nepal and the United States.
In the light of the sensitivity of these results to this parameteras well as
signicant uncertainties about the underlying data and other parameters that
enter these calculations our conclusions must be very tentative. But despite
the uncertainties, it seems clear that measures of changes in per capita genuine
wealth yield a very differentand often much bleakerpicture of the prospects
for poor nations, as compared with the message implied by changes in GDP per
capita.
Further Perspectives and Conclusions
We have evaluated consumption levels according to two criteria: the dis-
counted present value of the utility stream (the maximize present value criterion)
and the maintenance or improvement of intertemporal social welfare (the sustain-
ability criterion). Although the evidence is far from conclusive, we nd some
support for the view that consumptions share of output is likely to be higher than
that which is prescribed by the maximize present value criterion. We also nd
evidence that several nations of the globe are failing to meet a sustainability
criterion: their investments in human and manufactured capital are not sufcient
to offset the depletion of natural capital. This investment problem seems most
acute in some of the poorest countries of the world.
We would emphasize that insufcient investment by poor countries does not
imply excessive consumption in the most important sense. For many of the poorest
nations of the world, where productivity and real incomes are low, both consump-
tion and investment are inadequate: current consumption does not yield a decent
Table 3
Sensitivity Analysis
Country
GDP-wealth
ratio 0.10
Central
case
GDP-wealth
ratio 0.25
Bangladesh 0.15 0.30 0.93
India 0.01 0.54 1.41
Nepal 0.09 0.63 1.91
Pakistan 0.03 0.59 1.34
China 6.79 8.33 10.20
Sub-Saharan Africa 2.50 2.58 2.82
Middle East/North Africa 3.44 3.82 4.51
United Kingdom 1.49 2.29 2.59
United States 0.15 0.75 1.19
Are We Consuming Too Much? 167
living standard for the present generation, and current investment does not assure
a higher (or even the same) standard for future generations.
This study and all previous studies of which we are aware provide only point
estimates of genuine investment or of changes in genuine wealth. Given the vast
uncertainties associated with the estimates, even when point estimates are positive,
there may remain a signicant possibility that genuine investment is negative. The
uncertainties justify added caution.
The presence of nonlinearities compounds the importance of uncertainty.
The biophysical impacts associated with the loss of natural capital can be highly
nonlinear: these impacts may be small over a considerable range, and then
become immense once a critical threshold is reached. Crossing the threshold
leads to a bifurcation, a situation where the characteristics of the natural
system change fundamentally. For example, shallow clear fresh water lakes can
absorb a low level of phosphorus with little ill effect. However, if more phos-
phorus is added to the lake through sewage or runoff from agricultural land,
more algae grows in the water, less sunlight will reach the bottom, and the green
plants on the bottom will disappear. As a consequence, the bottom sediments,
which contain phosphorus from dead algae, will be less stable, and phosphorous
will be released from the bottom. More discharge of phosphorus from outside
will trigger even more phosphorus from the bottom. This positive feedback will
eventually force the lake to ip or bifurcate from clear to turbid water. Recent
work by ecologists and environmental economists has investigated these dynam-
ics and shown that such ips can occur over as short a period as a month
(Scheffer, 1998; Carpenter, Ludwig and Brock, 1999; Brock and Starrett, 2003).
Another possible bifurcation arises in certain climate models that indicate that
a rise in greenhouse gases might reverse the direction of the Atlantic stream
that now warms northern Europe. Paleo-climatic history shows that such rever-
sals have been common.
16
Nonlinearities in ecosystem dynamics imply the presence of serious downside
risks related to the losses of natural capital. Central estimates of the shadow prices
for natural capital are likely to be too low if one only considers central cases rather
than the entire distribution of potential outcomes from losses of natural capital.
Thus, the expected values of genuine investment or of changes in per capita
genuine wealth might well be lower than the values emerging when one employs
only central values for parameters. Accounting for risk-aversion could lower these
estimates even further.
While there may be uncertainty about whether various countries are meeting
the sustainability criterion, the need for vigorous public policies to support more
efcient consumption and investment choices is unambiguous. Through regula-
tion, taxes or the establishment of clearer or more secure property rights, public
policy can help prices of natural and environmental resources better approximate
16
Mastrandrea and Schneider (2001) have employed a climate-economy model to investigate the
possibility of reversals and to assess the implications for climate policy.
168 Journal of Economic Perspectives
their social cost. These policies can help prevent excessive resource depletion and
promote higher genuine investment. Such policies are justied on efciency
grounds whether or not genuine investment currently is positive.
17
Although the
implementation of public policies that satisfy a benet-cost test does not in theory
guarantee sustainability, such policies do not necessarily conict with sustainability,
either. Indeed, our sense is that policies of this sortespecially those that deal with
underpricing of natural resources or environmental amenitieswill improve mat-
ters along the sustainability dimension. When one views desertication in China,
water contamination in Senegal and forest depletion in Haiti, it is hard to imagine
that the establishment of property rights or improved pricing of natural resources
could worsen the prospects of future generations.
Beyond policy action, further research is needed to identify the areas where
current consumption poses a threat to sustainability and to quantify the potential
losses. We need to develop better data quantifying the losses of natural capital and
the potential for substitution between various forms of capital.
Further, to complement the rather simple analytical calculations of changes in
genuine wealth, we need to make more use of disaggregated numerical growth
models. Such models can contain considerable detail in the interaction of various
forms of capital and the services they generate. Current estimates of genuine wealth
depend crucially on the values assigned to shadow prices, yet the empirical basis for
these prices is very weak. Numerical growth models can be used to project growth
paths of economies and the sensitivity of these paths to changes in capital stocks. In
this way, they can generate much better assessments of the critical shadow prices.
Additional information of this kind will help reduce uncertainties about changes in
genuine wealth and clarify the extent to which current consumption levels might
imperil the quality of life of future generations.
y
We are grateful to Geir Asheim, Jack Pezzey and the editors of this journal for very helpful
suggestions, and to Oren Ahoobim and Justin Diener for excellent research assistance. We also
thank the William and Flora Hewlett Foundation for financial support and the Beijer
Institute for Ecological Economics, Stockholm, for sponsoring the workshops from which this
paper emerged.
17
The rationale for many public policies remains strong irrespective of whether genuine investment is
positive or negative. From this, one might conclude that quantifying genuine investment is not very
useful. But measuring genuine investment still has signicant value. By providing an overall scorecard
as to whether a nation is investing enough to sustain the welfare of future generations, it can offer an
important summary assessment that can help mobilize the general public and politicians.
Kenneth Arrow et al. 169
References
Arrow, Kenneth J. 2000. Observations on So-
cial Capital, in Social Capital: A Multifaceted Per-
spective. P. Dasgupta and I. Serageldin, eds.
Washington, D.C.: World Bank, pp. 35.
Arrow, Kenneth J. and Mordecai Kurz. 1970.
Public Investment, the Rate of Return and Optimal
Fiscal Policy. Baltimore: Johns Hopkins University
Press.
Arrow, Kenneth J., Partha Dasgupta and Karl-
Go¨ran Ma¨ler. 2003a. The Genuine Savings Cri-
terion and the Value of Population. Economic
Theory. 21:2, pp. 21755.
Arrow, Kenneth J., Partha Dasgupta and Karl-
Go¨ran Ma¨ler. 2003b. Evaluating Projects and
Assessing Sustainable Development in Imperfect
Economies Environmental and Resource Econom-
ics. 46:4, pp. 64785.
Arrow, Kenneth J. et al. 1996. Intertemporal
Equity, Discounting, and Economic Efciency,
in Climate Change 1995: Economic and Social Di-
mensions of Climate Change. J. P. Bruce, H. Lee
and E. R. Haites, eds. Cambridge, UK, New York
and Melbourne: Cambridge University Press,
pp. 12544.
Asheim, Geir. 1994. Net National Product as
an Indicator of Sustainability.Scandinavian Jour-
nal of Economics. 96:2, pp. 25765.
Asheim, Geir. 2003. Green National Ac-
counts for Welfare and Sustainability: A Taxon-
omy of Assumptions and Results.Scottish Journal
of Political Economy. 50:2, pp. 11330.
Barnett, Harold and Chandler Morse. 1963.
Scarcity and Growth. Baltimore: Johns Hopkins
University Press.
Bohn, Henning and Robert T. Deacon. 2000.
Ownership Risk, Investment, and the Use of
Natural Resources. American Economic Review.
9:3, pp. 526 49.
Brock, W. A. and D. Starrett. 2003. Managing
Systems with Non-Convex Positive Feedbacks.
Environmental and Resource Economics. 46:4,
pp. 575602.
Carpenter, Stephen R., D. Ludwig and W.A.
Brock. 1999. Management of Eutrophication
for Lakes Subject to Potentially Irreversible
Change. Ecological Applications. 9:3, pp. 751
71.
Chichilnisky, Graciela. 1994. North-South
Trade and the Global Environment. American
Economic Review. 84:3, pp. 85174.
Collins, Susan and Barry Bosworth. 1996.
Economic Growth in East Asia: Accumulation
versus Assimilation.Brookings Papers on Economic
Activity. 2, pp. 135203.
Cooper, Ben, Cecelia Garcia-Pen˜ alosa and
Peter Funk. 2001. Status Effects and Negative
Utility Growth. Economic Journal. 111:2,
pp. 642 65.
Daily, Gretchen C., ed. 1997. Natures Services:
Societal Dependence on Natural Ecosystems. Washing-
ton, D.C.: Island Press.
Daily, Gretchen C. et al. 2001. The Value of
Nature and the Nature of Value. Science. July 21,
289, pp. 39596.
Dasgupta, Partha. 1990. The Environment as
a Commodity. Oxford Review of Economic Policy.
6:2, pp. 5167.
Dasgupta, Partha. 1993. An Inquiry into Well-
Being and Destitution. Oxford: Clarendon Press.
Dasgupta, Partha. 2001a. Human Well-Being
and the Natural Environment. Oxford and New
York: Oxford University Press.
Dasgupta, Partha. 2001b. On Population and
Resources: Reply.Population and Development Re-
view. 27:4, pp. 74854.
Dasgupta, Partha and G. M. Heal. 1979. Eco-
nomic Theory and Exhaustible Resources. Cam-
bridge: Cambridge University Press.
Dasgupta, Partha and Karl-Go¨ran Ma¨ler. 2000.
Net National Product, Wealth, and Social Well-
Being. Environment and Development Economics.
5:2, pp. 6993.
Duesenberry, James S. 1949. Income, Saving,
and the Theory of Consumer Behavior. Cambridge,
Mass.: Harvard University Press.
Ehrlich, Paul R. and Anne H. Ehrlich. 1990.
The Population Explosion. New York: Simon and
Schuster.
Frank, Robert. 1985a. Choosing the Right Pond:
Human Behavior and the Quest for Status. New
York: Oxford University Press.
Frank, Robert. 1985b. The Demand for Un-
observable and Other Nonpositional Goods.
American Economic Review. 75:1, pp. 10116.
Hall, Robert. 1988. Intertemporal Substitu-
tion in Consumption. Journal of Political Econ-
omy. 96:2, pp. 33957.
Hamilton, Kirk. 2000. Genuine Saving as a
Sustainability Indicator. Working paper, Envi-
ronment Department, World Bank, January.
Hamilton, Kirk. 2002. Savings Rules and Sus-
tainability. Working paper, Environment De-
partment, World Bank.
Hamilton, Kirk and Michael Clemens. 1999.
Genuine Savings Rates in Developing Coun-
tries.World Bank Economic Review. 13:2, pp. 333
56.
170 Journal of Economic Perspectives
Heal, Geoffrey M. 1998. Valuing the Future:
Economic Theory and Sustainability. New York: Co-
lumbia University Press.
Heal, Geoffrey M. 2003. Optimality or Sus-
tainability? in Economics for an Imperfect World.
Richard Arnott, Bruce Greenwald, Ravi Kanbur
and Barry Nalebuff, eds. Cambridge, Mass. and
London: MIT Press, pp. 33148.
Heal, Geoffrey M. et al. 2001. Protecting
Natural Capital: Ecosystem Service Districts.
Stanford Environmental Law Journal. 20:2,
pp. 33364.
Howarth, Richard. 1996. Status Effects and
Environmental Externalities. Ecological Econom-
ics. 16:1, pp. 2534.
Hussain, Athar, Nicholas Stern and Joseph
Stiglitz. 2000. Chinese Reforms from a Compar-
ative Perspective,in Incentives, Organization, and
Public Economics. P. J. Hammond and G. D.
Myles, eds. Oxford: Oxford University Press,
pp. 24377.
International Energy Agency. 1999. World En-
ergy Outlook1999 Insights. Looking at Energy
Subsidies: Getting the Price Right. London: IEA
Publications.
Johnson, D. Gale. 2000. Population, Food,
and Knowledge.American Economic Review. 90:1,
pp. 114.
Johnson, D. Gale. 2001. On Population and
Resources: A Comment. Population and Develop-
ment Review. 27:4, pp. 739 47.
Klenow, Peter J. and Andres Rodrı´guez-Clare.
1997. The Neoclassical Revival in Growth Eco-
nomics: Has it Gone Too Far? in NBER Macro-
economics Annual 1997. B. Bernanke and J.
Rotemberg, eds. Cambridge, Mass. and London:
MIT Press, pp. 73104.
Koopmans, Tjalling C. 1960. Stationary Ordi-
nal Utility and Impatience. Econometrica. 28:2,
pp. 287309.
Koopmans, Tjalling C. 1972. Representation
of Preference Orderings over Time, in Decision
and Organization. C. B. McGuire and R. Radner,
eds. Amsterdam: North Holland, chapter 4.
Levin, Simon, ed. 2001. Encyclopedia of Biodi-
versity. New York: Academic Press.
Lind, Robert C. 1964. The Social Rate of
Discount and the Optimal Rate of Investment:
Further Comment. Quarterly Journal of Econom-
ics. 78:2, pp. 336 45.
Lind, Robert C., ed. 1982. Discounting for Time
and Risk in Energy Policy. Baltimore: Johns Hop-
kins University Press.
Marglin, Stephen A. 1963. The Social Rate of
Discount and the Optimal Rate of Investment.
Quarterly Journal of Economics. 77:1, pp. 95111.
Mastrandrea, Michael and Stephen H. Schnei-
der. 2001. Integrated Assessment of Abrupt Cli-
matic Changes. Climate Policy. 1:4, pp. 43349.
McNeill, John Robert. 2000. Something New Un-
der the Sun: An Environmental History of the Twen-
tieth-Century World. New York: W.W. Norton.
Myers, Norman and Jennifer Kent. 2000. Per-
verse Subsidies: How Tax Dollars Can Undercut the
Environment and the Economy. Washington, D.C.:
Island Press.
Ng, Yew-Kwang. 1987. Relative Income Ef-
fects and the Appropriate Level of Public Expen-
diture. Oxford Economic Papers. 39:2, pp. 293
300.
Pearce, David and Giles Atkinson. 1995. Mea-
suring Sustainable Development,in Handbook of
Environmental Economics. D. W. Bromley, ed. Ox-
ford: Basil Blackwell, pp. 166 81.
Pearce, David, K. Hamilton and Giles Atkin-
son. 1996. Measuring Sustainable Develop-
ment: Progress on Indicators. Environment and
Development Economics. 1, pp. 85101.
Pezzey, John. 1992. Sustainable Development
Concepts: An Economic Analysis. World Bank
Environment Paper No. 2, World Bank, Wash-
ington, D.C.
Portney, Paul and John Weyant, eds. 1999.
Discounting and Intergenerational Equity. Washing-
ton, D.C.: Resources for the Future.
Postel, Sandra L., Gretchen C. Daily and Paul
R. Ehrlich. 1996. Human Appropriation of Re-
newable Fresh Water. Science. February 9, 271,
pp. 78588.
Ramsey, Frank P. 1928. A Mathematical
Theory of Saving. Economic Journal. 38:4,
pp. 54359.
Scheffer, Marten. 1998. The Ecology of Shallow
Lakes. London: Chapman and Hall.
Schor, Juliet. 1998. The Overspent American.
New York: Basic Books.
Sen, Amartya K. 1967. Isolation, Assurance,
and the Social Rate of Discount. Quarterly Jour-
nal of Economics. 81:1, pp. 12124.
Solow, Robert M. 1974. The Economics of
Resources or the Resources of Economics.
American Economic Review. 64:2, pp. 114.
Solow, Robert M. 1992. An Almost Practical
Step toward Sustainability. Resources for the
Future 40
th
Anniversary Lecture, Washington,
D.C.: Resources for the Future.
UNDP (United Nations Development Pro-
gramme). 2003. Human Development Report. New
York: Oxford University Press.
United States Ofce of Management and the
Budget. 1992. Circular A-94 Revised. Washington,
D.C.: United States Government Printing Ofce.
Are We Consuming Too Much? 171
Veblen, Thorstein. 1899. The Theory of the Lei-
sure Class. New York: Penguin.
Vitousek, P., P. R. Ehrlich, A. H. Ehrlich and
P. Matson. 1986. Human Appropriation of
the Product of Photosynthesis. BioScience. 36,
pp. 36873.
Vitousek, P. M., H. A. Mooney, J. Lubchenco
and J. M. Melillo. 1997. Human Domination
of Earths Ecosystem. Science. July 26, 277,
pp. 49499.
World Bank. 1992. World Bank Development Re-
port 1992: Development and the Environment. Wash-
ington, D.C.: World Bank.
World Bank. 1997. Expanding the Measure of
Wealth: Indicators of Environmentally Sustainable De-
velopment. Washington, D.C.: World Bank.
World Bank. 2003. World Development Indica-
tors. Available at http://devdata.worldbank.
org/dataonline, September.
World Commission on Environment and De-
velopment. 1987. Our Common Future. New York:
Oxford University Press.
172 Journal of Economic Perspectives

Discussion

"Underpricing of Natural Resources Relative to Social Cost". This certainly is an incredibly massive issue in the past, present and future. Our world in data presents energy use across several dimensions. Here is a breakdown over time: ![Imgur](https://imgur.com/Ay5l8Rf.png) https://ourworldindata.org/energy General findings: "We have identified several factors influencing consumption and have shown how they enable us to judge whether consumption is excessive according to the maximize present value criterion. Several factors—the inability to pool risks perfectly, the taxation of capital income and the underpricing of natural resources—contribute toward excessive consumption. (Below, we will observe that these same factors also work toward excessive consumption according to the sustainability criterion.) Among these imperfections, the underpricing of natural resources strikes us as the most transparent". The World Commission on Environment and Development report from 1987: https://sustainabledevelopment.un.org/content/documents/5987our-common-future.pdf Lack of data/poor data quality is a major hurdle, and these factors are undeniably important. "First, as Hamilton and Clemens emphasize, one serious problem is the lack of comprehensive data in national accounts. For many important (and declining) natural resource stocks, data are not available—at least not for every country. Among the natural resources not included in the Hamilton-Clemens study are water resources, forests as agents of carbon sequestration, fisheries, air and water pollutants, soil and biodiversity. So there is an undercount, possibly a serious one." Total factor productivity is a measure of productive efficiency that measures how much output can be produced from a certain amount of inputs; it is usually attributed as a main driver of GDP growth rates For more: https://en.wikipedia.org/wiki/Total_factor_productivity This feels like an insurmountable hurdle...there won't be great "substitutes" in ecology, and the authors note this: "They fear that to the extent that economists are overly optimistic about opportunities to substitute other capital assets for certain natural resources, reductions in their stocks would then receive too little weight. A major goal of ecological economics is to increase our understanding of the ways in which differ-ent kinds of natural capital contribute to human well-being and the extent to which they are substitutable for one another and for other kinds of capital assets" What does "maximizing the discounted value" means? Important: "One can think of intertemporal social welfare Vt as a function of initial conditions—the capital assets and level of technology at time t—and the choices from time t forward as to how to allocate output between investment and consumption. According to the maximize present value criterion, actual consumption today is excessive if it is greater than the level of current consumption prescribed by this optimal consumption path. To put it another way, current consumption is excessive if lowering it and increasing investment (or reducing disinvestment) in capital assets could raise future utility enough to more than compensate (even after discounting) for the loss in current utility." Discount rates are used widely in economics and finance, for making present value calculations about things ranging from utility to discounted cash flow. Here is some more background: https://www.investopedia.com/terms/d/discountrate.asp ![Imgur](https://imgur.com/6Q9wPWG.png). This is a great site that visualizes the world population over time. https://ourworldindata.org/world-population-growth Important: “By what criterion should we judge whether consumption is or is not excessive? Economic analysis has tended to emphasize the criterion of maximizing the present discounted value of current and future utility from consumption—what we will call intertemporal social welfare. Current consumption is deemed excessive or deficient depending on its level relative to that called for by this optimization problem. However, analysts with a different perspective have applied a criterion of sustain- ability, which emphasizes the ability of the economy to maintain living standards. In this paper, we will examine these criteria for evaluating whether consumption is excessive and identify factors in the economic and ecological domains that deter- mine whether or not it is in fact so.” “This paper, the outgrowth of discussions among a group of ecologists and economists, offers an analysis that we hope will go some way toward reconciling the conflicting intuitions. The binocular vision that can be obtained from using both ecological and economic insights raises questions that might not occur in either viewpoint alone.” It is absolutely important for this research to have these different perspectives. Present discounted value of current and future utility from consumption refers to all of the utility from consumption in the present plus all future utility derived from consumption, discounted to the present. In order to calculate the present value of future consumption or utility, they discount it to the present using a discount rate and compounding (here is this explained in finance https://en.wikipedia.org/wiki/Discounted_cash_flow). The coauthors of this paper include widely acclaimed economists, biologists, and ecologists (their affiliations and backgrounds are listed below). This unusually diverse collaboration is necessary for this paper (which builds on an understanding of economics, finance, biodiversity, sustainability, and other topics). Kenneth Arrow, one of the authors, is a Nobel laureate who won the Nobel Prize for establishing mathematically, the necessary prerequisites for a free market to have optimal outcomes- that is what "welfare economics" in the title means, it doesn't have anything to do with public assistance. Here is another paper of his: https://fermatslibrary.com/s/uncertainty-and-the-welfare-economics-of-medical-care Extremely important parameter for the model, opens up an entire philosophical debate. "One reason is that, because of externalities, the utility discount rate that most people would endorse as socially conscious beings is lower than the rate of time preference emanating from market transactions. Many years ago, Ramsey, Pigou and Harrod insisted that the only ethically justifiable value for␦is zero. Solow (1974, p. 9) put the matter thus: “In solemn conclave assembled, so to speak, we ought to act as if the social rate of (pure) time preference were zero.” A less extreme view would be to say that even if we don’t treat␦as zero, individuals do derive a positive externality (outside of the marketplace) from the welfare of future generations. An argument combining the market rate of return on investment with the externality of caring about future generations might call for “low” values of␦, in the range, say, of 0–0.5 percent per annum."