Discounting Future Utility & Climate Change

by: Seth Baum

Fri Jan 19, 2007 at 00:57:03 AM UTC

It is standard practice in contemporary mainstream economics to discount future utility (quality of life), much like one discounts future income. This practice is highly controversial within the economics and philosophy communities. My sense so far is to not discount future utility and to handle the consequences of this using different means.

Why should we care? For example, if we discount future utility at the suggested rates, then climate change is at most a small problem prescribing small action; if we do not discount future utility, then climate change becomes among the most serious problems humanity currently faces, prescribing more drastic action. That's a pretty big difference.

While I still don't buy them, there are reasons given for discounting future utility. These include the possibility that future people won't exist (i.e., that something will come up and we'll be extinct), that considering infinite amounts of time otherwise poses problems, and that humans tend to discount the future naturally. More on this can be found at the New School's History of Economic Thought website or in books by Partha Dasgupta & Geoffrey Heal, among others. A more scathing critique of discounting can be found in this article by Frank Ackerman in the interesting "post-autistic economics review". (More on them some other day.)

The difference between discounting and not discounting is quite apparent in the debate between economists Nordhaus (who discounts) and Nicholas Stern (who discounts only negligibly) on the severity of climate change and how strongly we should respond to it. The debate was sparked by the recent Stern Review on climate change, which stated (from its Executive Summary (pdf), p.2):

The evidence shows that ignoring climate change will eventually damage economic growth. Our actions over the coming few decades could create risks of major disruption to economic and social activity, later in this century and in the next, on a scale similar to those associated with the great wars and the economic depression of the first half of the 20th century.

Stern assessed climate change to be a much stronger problem than other economists, such as Nordhaus, apparently due primarily to his much lower discount rate. Nordhaus and his colleagues agree that more should be done about climate change, but he stresses the difference between them and Stern on how much should be done in his response (from Nordhaus's working paper The Stern Review on the Economics of Climate Change (pdf) p.15,21):

While the findings of such mainstream economic assessments may not satisfy the most ardent environmentalists, if followed they would go far beyond current global emissions reductions and would be a good first step on a journey of many miles.
...
The radical revision of the economics of climate change proposed by the Review does not arise from any new economics, science, or modeling. Rather, it depends decisively on the assumption of a near-zero social discount rate. The Review's unambiguous conclusions about the need for extreme immediate action will not survive the substitution of discounting assumptions that are consistent with today's market place.

Which is correct? I still agree with those who argue on ethical grounds that future utility should not be discounted, and I prefer a model reflects this. However, I'm not satisfied by either model, as both appear to base their analysis in monetary units (e.g. dollars) instead of in units of utility (e.g. quality-adjusted life years).

Finally, let's step back to remember:

All models are wrong; some are useful.
- George Box

Mr. Box is right: Neither model is completely correct. However, given that Nordhaus does agree with Stern that we should be doing more to mitigate climate change than we currently are, both models may be useful by encouraging us to reduce our greenhouse gas emissions. From a utilitarian perspective, that certainly does seem like a good idea.

Medical community doesn 't discount future utility?

In my research for the post USA: $50,000 per QALY, I came accross the paper The Cost Effectiveness of Voluntary Prenatal and Routine Newborn HIV Screening in the United States (pdf), Journal of Acquired Immune Deficiency Syndromes 25:403-416. It appears to discount future money but not future utility. Thus, it seems that while today's economists often discount future utility, the medical community does not. So far, I'm with the medical community on this one.

Incidentally, the JAIDS paper discounts future money at 3% per year. This is the same rate that Nordhaus (initially) discounted future utility, and it seems to be a fairly standard rate. While they don't explicitly say so, it really seems to me like Nordhaus & co. are discounting future utility because they reflexively treat it like it was money. Certainly, the medical community would not share this reflex. If anything, it could be accused of sometimes using life years instead of quality-adjusted life years (the JAIDS paper does this), but this appears to be an honest matter of it not always knowing how to perform the quality adjustment, and it does seem to note the distinction.

by: Seth Baum @ Sat Jan 20, 2007 at 00:50:46 AM CST

Weitzman 's take

Martin Weitzman, Harvard Econ, wrote this (pdf) in response to the Stern Review. His "five basic points" are (p.2):

1: The discount rate we choose is all important. 2: We are a lot less sure about core elements of discounting than we commonly admit because critical puzzles and ambiguities are yet unresolved. 3: Structural uncertainty that manifests itself in the thick tails of probability distributions- not risk- is what likely matters most. 4: Gathering information about thick-tailed uncertainties representing rare disasters should be a research priority. 5: The Stern Review may well be right for the wrong reasons.

Since the original post, I've continued to read economists' tortured back-and-forth on future discounting, and I still have yet to find a single compelling reason to discount future utility. I've only skimmed Weitzman's paper, but at a glance he seems to fit in with the typical economist treatments I've seen. I should soon have a more comprehensive response to all that I'm reading.

H/T to Dave Iverson's Ecological Economics blog post for pointing me to Weitzman's paper.

by: Seth Baum @ Wed Feb 14, 2007 at 16:48:32 PM CST

Great discounting references

1) Valuing future life and future lives: A framework for understanding discounting (pdf) by Shane Frederick (2006). This paper provides the best introductory overview of future discounting and our follies with it I've yet seen. It's the first paper I'd recommend to anyone new to the topic.

2) Time Discounting and Time Preference: A Critical Review (pdf) by Shane Frederick, George Loewenstein, & Ted O'Donoghue (2002). This is a large, comprehensive review of the wacky world of future discounting. So far, I've especially enjoyed its "history of future utility discounting" and its massive collection of studies observing how people tend to discount the future. Choice quote from the history section (p.2-3 of pdf):

Each of the major figures in the development of the DU model-John Rae, Eugen von Bohm-Bawerk, Irving Fisher, and Paul Samuelson-built upon the theoretical framework of his predecessors, drawing on little more than introspection and personal observation. When the DU model eventually became entrenched as the dominant theoretical framework for modeling intertemporal choice, it was due largely to its simplicity and its resemblance to the familiar compound interest formula, and not as a result of empirical research demonstrating its validity.

From the observations section, we find studies showing people tend to discount the future at rates from -6% to 190,000% on up to infinity percent. Interesting.

...
I'm fairly determined to get to the bottom of this future discounting mess- and clearly, it is a mess. I started this effort baffled why anyone would recommend discounting future utility and eager to find some good reasons why so many (particularly economists) do. For all the reading I've done, I still have no good answers.

The one plausibly reasonable objection to the undiscounted model (c.f. eqns 3-5 of Total Lifetime Utility at Old Felicifia) is that it can involve comparisons of infinite amounts of utility, which is not as straightforward as comparing finite amounts of utility. If I recall correctly, mathematicians generally say that all "infinities" are equal, i.e. infinity-plus-one equals infinity. (The exception is between countable and uncountable infinities, but that's not relevant here.)

Nick Bostrom has the best shake on this- see in particular Infinite Ethics (pdf). I've only skimmed this so far, but he seems to capture the essence of the quandry quite well. Right now, I'm especially interested in his " Avoiding the indifference problem. It must not be the case that all humanly possible acts come out as ethically equivalent." (p.4).

My initial response to this matter of infinite ethics is to say we should move towards avoiding existential events, extending the lifetime of the biosphere, and space colonization. Again, Bostrom appears to be thinking along similar lines, this time in Astronomical Waste: The Opportunity Cost of Delayed Technological Development. (I haven't read this paper closely either.)

Dasgupta and Heal also discuss this infinity quandry; I found their treatments unsatisfying.

by: Seth Baum @ Thu Feb 15, 2007 at 18:38:24 PM CST

BBD on discounting

From "Population Issues in Social Choice Theory, Welfare Economics, and Ethics" by Charles Blackorby, Walter Bossert, & David J. Donaldson ("BBD" for short; CUP 2005; ISBN 0521825512)

Some interesting remarks on discounting, p.258.

• The "universe is known to be finite", so arguments for discounting using infinte time are "based on an impossibility".

• Discounting can make the present generation worse off:

Consider two alternatives x and y in which three people live and are born in periods 1, 2, and 3. In x, utility levels are 28, 4, and 44; in y, utility levels are 24, 24, and 24. If intertemporal critical-level utilitarianism with a critical level of zero is used to evaluate the alternatives, x is better than y and the utility level of person 1, who represents the first generation, is 28. Now suppose geometric birth-date-dependent critical-level utilitarianism with a critical level of zero and a discount factor of 1/2 is used instead. In that case, values for x and y are 28 + 2 + 11 = 41 and 24 + 12 + 6 = 42, so y is better and person 1's utility is 24. Discounting has made the present generation worse off.

• BBD reject discounting

• Should classical utilitarianism demand great sacrifice (here called "supererogation"), BBD advocate for a "floor" on the present generation's utility as

a far more natural way of dealing with the supererogation issue than by transforming an ethically appropriate value function into one that fails to treat generations or contemporaries impartially.

I disagree with this "floor" system: Why should we insist on a minimum level of well-being for today's initially-well-off at the expense of greater well-being for others, mainly the less-well-off, both today and in the future?

by: Seth Baum @ Sat Feb 24, 2007 at 21:56:43 PM CST

Carbon Tax - Gas Tax Conversion

Stern recommends a carbon tax of $311 per ton CO2; Nordhaus, $17. These convert into gas taxes of US$2.74 or US$0.15 per gallon; 0.55 or 0.03 Euro per litre; 0.37 or 0.02 British Pound per litre.

15 cents per gallon is barely a blip on the radar. That won't change behavior significantly. $2.74 sure as heck will. How big a difference is this in the UK? Is Pounds per litre the right units?

Calculations:
1 gallon of gasoline produces 19.4 pounds of CO2. US EPA
19.4 lbs = 0.0088 metric ton onlineconversion.com (site used for other conversion factors)
-> 1 gallon of gasoline produces 0.0088 metric tons of CO2
17 (dollars per ton) * (0.0088 tons per gallon) = 0.15 dollars per gallon
311 (dollars per ton) * (0.0088 tons per gallon) = 2.74 dollars per gallon

0.15 dollars per gallon = 0.11387 Euro per gallon
2.74 dollars per gallon = 2.08009 Euro per gallon
0.15 dollars per gallon = 0.07641 British Pound per gallon
2.74 dollars per gallon = 1.39580 British Pound per gallon
1 US Dollar = 0.75916 Euro (February 27, 2007)
1 US Dollar = 0.50942 British Pound (February 27, 2007)

0.11 (Euro per gallon) * (1 gallon per 3.79 liter) = 0.03 Euro per litre
2.08 (Euro per gallon) * (1 gallon per 3.79 liter) = 0.55 Euro per litre
0.08 (British Pound per gallon) * (1 gallon per 3.79 liter) = 0.02 British Pound per litre
1.40 (British Pound per gallon) * (1 gallon per 3.79 liter) = 0.37 British Pound per litre

by: Seth Baum @ Mon Feb 26, 2007 at 22:56:28 PM CST

OMB Circular A-094

Possibly the most important document in climate change?

This document from the US Office of Management and Budget dictates how cost-benefit analysis is done by the US federal government, including how time discounting is to be done. I'm told that a main reason why Nordhaus et al are so insistent on discounting is that any time-neutral legislation won't make it through US Congress. Also, if discounting is pretty much required by law here, Circular A-094 (pdf) is where it's penciled in. Don't take this as fact from me, as I'm new to this topic and don't have a great understanding yet. I just wanted to put the idea out there and link to the document.

At first glance, the whole thing smells like my undergrad microeconomics course: Quality of life is measured in dollars and distributional issues don't matter (even if lip service is paid to them). The one area where the two differ is regarding non-US citizens: The micro course cared about them; A-094 doesn't. Here, I'm with the micro course. The section I'm looking at is "6. Identifying and Measuring Benefits and Costs.", p.6-7(5-6):

Both intangible and tangible benefits andcosts should be recognized.
...
(3) International Effects. Analyses should focus on benefits and costs accruing to the citizens of the United States in determining net present value. Where programs or projects have effects outside the United States, these effects should be reported separately.

(4) Transfers. There are no economic gains from a pure transfer payment because the benefits to those who receive such a transfer are matched by the costs borne by those who pay for it. Therefore, transfers should be excluded from the calculation of net present value. Transfers that arise as a result of the program or project being analyzed should be identified as such, however, and their distributional effects discussed. It should also be recognized that a transfer program may have benefits that are less than the program's real economic costs due to inefficiencies that can arise in the program's delivery of benefits and financing.

b. Measuring Benefits and Costs. The principle of willingness-to-pay provides an aggregate measure of what individuals are willing to forego to obtain a given benefit. Market prices provide an invaluable starting point for measuring willingness-to-pay, but prices sometimes do not adequately reflect the true value of a good to society. Externalities, monopoly power, and taxes or subsidies can distort market prices.

Taxes, for example, usually create an excess burden that represents a net loss to society. (The appropriate method for recognizing this excess burden in public investment analyses is discussed in Section 11.) In other cases, market prices do not exist for a relevant benefit or cost. When market prices are distorted or unavailable, other methods of valuing benefits may have to be employed. Measures derived from actual market behavior are preferred when they are available.

Don't get me wrong: Overall, A-094 looks good. We should strive for cost-effectiveness in general, and many, probably most, of the principles outlined in A-094 are smart, sound analysis. The considerable attention paid to factoring on non-monetary effects is encouraging. Compared to, say, this criticism of cost-benefit analysis, I'd take A-094. It's not hard to imagine reforming A-094 to correct the flaws I currently perceive in it.

by: Seth Baum @ Sat Mar 17, 2007 at 14:05:24 PM CDT