« Yana makes the LA Times | Main | Thanksgiving dinner »
Nordhaus review of Stern on global warming
Here is Bill Nordhaus's critique of the Stern report. Nordhaus argues that Stern's "new" results boil down to the choice of a lower discount rate.
I agree with Stern that the discount rate should be zero or near-zero for resources which will not be reinvested but rather represent alternative consumption streams across the generations. If we are doing normative analysis, however, and considering alternatives to controlling global warming, a' la Copenhagen Consensus, we are by definition considering other investments. In that case the correct rate of discount is given by opportunity costs, which might be quite high, provided the alternative investments will in fact be undertaken. (On this topic, there are a few really good comments here.)
Having pondered the report a bit more, my main question is what it would cost for China and India to cut back on carbon emissions, all relevant institutional changes included in the calculations. In other words, that figure should count costs of persuasion, enforcement, and implementation, not just the cost of one technology rather than another in the abstract. We do not have a good sense of these costs, and given how many basic tasks these economies fail at, I can imagine the cynic citing the figure of infinity. (To consider one analogy, what is "the cost" of getting avian flu out of China?) Furthermore China in particular has a high rate of savings, and both economies have high rates of return on capital. Current compliance costs should thus be compounded, when considering their future importance, at rates considerably higher than zero.
Posted by Tyler Cowen on November 24, 2006 at 05:31 AM in Economics | Permalink
Comments
very good point..........
Posted by: sa at Nov 24, 2006 8:46:41 AM
Regarding discount rates I again cite the green
golden rule argument of Chichilnisky, Heal, and others
that argues for using a somewhat higher, market-based
rate in the short term for dealing with the opportunity
cost of capital issue, but lower rates in the longer
term for dealing with the normative and intergenerational
problems. Offhand, I do not see Stern's rate as too
unreasonable.
Regarding China and India, if one views the problem
as one of a market allocation, as was essentially proposed
in Kyoto, then the very large and unresolved question
becomes what the initial allocation of property rights is.
In practice in the US, say as with the SO2 emissions
trading program, this was determined by grandfathering
in existing patterns. Needless to say this was not
considered acceptable by either China or India, which is
why they stayed out, and their staying out triggered the
very embarrassing Byrd-Hagel resolution, which was even
more embarrassingly passed unanimously in the US Senate,
just so everybody in the world would know who the bad guys
on this issue are, even before W got in and started going
around insulting other countries openly on this matter.
Posted by: Barkley Rosser at Nov 24, 2006 8:57:22 AM
It's hard not to agree that the discount rate should be zero or non-zero, as much as our politicians might like political forecasting to involve imaginary numbers . . .
But doesn't a zero discount rate imply that even something that imposes trivial costs on each future generation should be avoided at catastrophic cost to us?
Posted by: Jane Galt at Nov 24, 2006 9:25:41 AM
"Non-zero" should read "near-zero", I will correct it.
Posted by: Tyler Cowen at Nov 24, 2006 9:45:55 AM
BTW, there are some cost of global warming numbers in
the Stern Report that seem excessive. One that has
been pointed out to me is that tourism revenues in the
Mediterranean would be down 30%. This seems a bit much.
Posted by: Barkley Rosser at Nov 24, 2006 11:50:31 AM
Here's an easy way to reduce carbon-buildup in the atmosphere: Stop recycling. Paper, plastic, yard clippings, etc., are over 50% carbon. A pound of carbon buried in landfills is one less pound of carbon in the atmosphere.
Posted by: Mike Sproul at Nov 24, 2006 2:35:46 PM
Maybe I'm reading it wrong because I don't understand the economics entirely, but Nordhaus seems to be say that it's not just that Stern chose a near-zero discount rate, but also that Stern chose a utility curvature that is inconsistent with the report's discount rate. When Nordhaus plugged the Stern Report's 0.1% discount rate into his DICE-2006 model with a calibrated utility curvature the model called for a carbon tax of only $19.55, as opposed to $159 in a run of DICE-2006 with the utility curvature from the Stern Report.
Maybe it's because I don't entirely understand what a utility curvature is, or what it means to calibrate it, but it seems to me like Nordhaus has more than just a beef with the discount rate. Can anyone explain this, though?
Posted by: Nathan at Nov 24, 2006 3:34:45 PM
A zero discount rate is not a logical choice. It assumes that we have absolute certainty about cause and effect in our world. We don't. We are not even close. The system we live in is far from being a closed like the logic in the Stern model would suggest. The system is open.
A second objection is offered by those who ask about opportunity costs. We see no real treatment of opportunity costs beyond the binary policy choice. A sad straw man. Acceptable for a publication in a new field but hardly acceptable for a multi-trillion decision. As a matter of fact it is appalling.
A related issue is that we don't know what our social objective function is today or in the future. Thus creating opportunity costs is a very complicated question.
A zero social discount function is not appropriate. Nordhaus offers a more logical approach through the DICE model. His three scenarios illustrate well the problem not only with the zero social discount rate but the underlying approach presented by Stern. In summary, the Stern approach is seriously flawed and inappropriate for policy decisions.
Posted by: bee at Nov 24, 2006 5:31:43 PM
Discount rate for, say, strip mining copper ore, should be future economic growth over (future growth of value of location - future value of location after mining)
This much is obvious. But we don't know what the future economic growth or future values will be. That's all it takes to get economists throwing darts to pick 1%, 2.5%, and so forth. The constructive thing to go is simply to do a best guess of ranges of these values and then find an optimal solution based on that.
Straightforward enough, yes?
Posted by: bhauth at Nov 24, 2006 5:44:28 PM
bee.
The Stern report does not use a zero discount
rate, it uses one equal to 0.1, "near zero,"
but not equal to zero.
However, an objection that Nordhaus raises
is that this is low enough so that what is
driving results is what is happening more
than 200 years in the future. Given the
enormous uncertainties about our knowledge
of things that far in the future (will there
even be any humans around?), it seems a bit
much to have results that are driven by
such an outcome, and this is due to the
very low discount rate combined with
assumptions about the economy that are
simply not very defensible. We just do
not know what will be going on then.
Posted by: Barkley Rosser at Nov 24, 2006 5:57:13 PM
As far as I can tell, Nordhaus is simply wrong when he says that Stern uses a 0.1 discount rate. It is true that Stern suggests that a 0.1 discount rate is appropriate when considering simply the issue of intertemporal utility comparisons (in other words, from a welfare point of view, we should not value our own welfare more highly than that of future generations). But he is explicit about the fact that we need to adjust for the time value of money, and the fact that future generations are likely to be significantly wealthier than we are. Although he curiously doesn't state explicitly anywhere in the report exactly what the discount rate is, the technical appendix to the discussion of discounting suggests that he's using a discount rate of around 3%, which actually seems quite reasonable.
As far as I can tell, this means that much of Nordhaus' critique is based on a complete misreading of Stern's math.
Posted by: K. Williams at Nov 24, 2006 8:43:44 PM
As far as I can tell, Nordhaus is simply wrong when he says that Stern uses a 0.1 discount rate.
It's a shame the Stern Report doesn't say what discount rate it's using in Chapter 2 where it talks about discounting, but Nordhaus is right that it uses 0.1%. Look at Chapter 6, where the Stern report calculates the cost of Global warming. They use δ = 0.1%.
Posted by: Nathan at Nov 24, 2006 10:25:37 PM
I am concerned about the practicality of the Stern proposals. One of the downsides of implementing a carbon pricing regime is that the benefits are shared by all, but the disbenefits are disproportionally shared by poor people as they are the ones whose lives will be most improved by faster economic growth. This is a perfect recipe for conflict; wars and such like can cost a lot more than 1% of growth. Likely, because of this conflict of interest it will be very hard to persuade poor countries to adapt the proposals, they will be natural sceptics. Will the west be prepared to invade India and China to get them to see it our way? I don't think so. So the only answer can be technology. However government directed technology programmes have a poor record.
Perhaps an effective method to develop the needed new technology would be for the rich countries to club together and offer a prize of, say, $100bn, for a technology answer that solves the problem (either low cost clean energy or some way to remove CO2 from the atmospher). The prize would be payable to the inventors once they had, say, through commercial application reduced CO2 in the atmosphere by a certain percentage. In order to deal with the future discount problem the prize is created now and invested so that it becomes more valuable over time. In further twist, to deal with the sceptics, it would only be payable if anthropegenic global warming thesis could proved to be correct, otherwise it reverts to the donors.
Posted by: ChrisA at Nov 25, 2006 8:11:30 AM
Nathan, look at p. 50 of the report (it's the Appendix to Chapter 2). The discount rate there (which is an average adjusting for various possible growth paths, etc.), clearly starts at around 3.5%. You're misreading the report. The 0.1% is the "utility discount rate," which adjusts for changes in intertemporal utility. That is different from what you could call the consumption discount rate, which adjusts for changing value of money over time, etc. That rate is, again, somewhere around 3%. Look at p. 162 in Chapter 6, where they talk about "the dsicounted consumption path." Again, Nordhaus (and you) misunderstand the math in the report.
Posted by: K. Williams at Nov 25, 2006 9:51:44 AM
Barkley-
I am aware that Stern used 0.1 at the discount rate (I said zero for convenience). The choice for all practical purposes is the same (Yes, I know that actually using zero would not play well with the math).
Again the logical problem is that it creates a world of near certainty across time. This is absurd. Given that the dynamic of a discount rate is exponential yet the thinking he is employing is linear we have the problem that Nordhaus reveals with the DICE model.
Posted by: bee at Nov 25, 2006 2:11:14 PM
I think there are two economic models to seperate.
First a general equilibrium or optimal portfolio model where the interest rate equals the discount rate plus the change in the marginal rate of utility. As Stern uses d=0.1%, log utility with average growth 2%, this would imply an interest rate of 2,1%, which may be be little low but not extremly. However, he would have had to take alternative investment opportunities into account and choose the whole consumption path. This is not what he did. His consumption path is exogeneous.
Second, a net present value analysis. Suppose there is a loss of income in the future. How much money does it need now to compensate for this loss? To do this, the present generatio has to save e^(-rt) income units - independent of utility - where r is the interest rate. If the future loss is uncertain, the untility function would matter, but still the correct discount rate would be given by r.
I would conclude that if one chooses a whole consumption path, one can use a social utility function. If on the other hand one wants to compare returns of different investment opportunities then one has to compute the present value. Maybe for some risk adjusted probability measure.
Posted by: Dirk at Nov 26, 2006 6:45:17 AM
