Arnold Kling  

The Sixth Warmest Year

John Locke's Wager... Another Story to Watch...

A major news story is that 2006 is likely to be the sixth warmest year on record. Lubos Motl says that it also is likely to be the coldest year since 2002.

I'm not sure what the right unit of time to use to establish a trend. If we look at 2006 relative to 1986, it's obviously warmer. If we look at it compared to 1998, it's colder. This reminds me of the problem of choosing the right unit of time for macro-econometric modeling--that is, before 99 percent of the economics profession gave up on the idea of macro-econometric modeling.

Comments and Sharing

COMMENTS (10 to date)
Mark Thoma writes:

For econometric questions involving the trend, don't the asymptotics depend upon the span of the data rather then the number of observations?

I must be missing your point.

Bob Hawkins writes:

In climatology, you average 30 years to get one datum. So there is no valid period for comparing a one-year result, because one year isn't data. It isn't even a datum.

Anyone who tells you about one year as if it's meaningful, is trying to sell you something.

Luboš Motl writes:

Thanks for the link.

Well, there can be interesting science at all timescales. I personally find the most important timescale for temperatures to be 1 week - 1 month. This is what determines how much heating oil New England burns this year, whether one can go skiing, etc. and other relatively important things.

My guess is that at the 5-year scale the temperatures will continue to be chaotic. At the 20-year scale, I guess that the warming is a bit more likely but I wouldn't bet your life on it.

The fluctuations of the temperatures appear at all timescales, too. Some of the signals are regular and periodic, others are not. Some of them are understood, others are not.

From the economic viewpoint, I think it is largely irrational to be planning for more than 20 years into the future especially if the expected effects that should influence such a planning are small. One can't make any nontrivial and really wise investments 30 years in advance because there will surely be many changes in the world that will have a more dramatic impact on the evaluation of the quality of such an investment than the topics we talk about today. Would we be overly happy if our lives were constrained by decisions that the people did in 1940 or 1820?

The likely changes in the next decades include new energy technology as well as new technology to possible influence the climate. It would be rather foolish to be spending a lot of money for something that, according to the graphs, doesn't seem urgent in any way even if it is clear that the same actions will be cheaper or much cheaper - and more effective - in the future.

I believe this comment of mine is valid well beyond the climate context. It is great to be thinking about the great grandsons in 2100 but it is bad to be trying to make them less free or less wealthy by stealing the resources and growth potential from the civilization today.

These questions about the timescale are probably more interesting in the context of the financial markets. Different types of investors want to make a profit after different time intervals and they are also sensitive to different trends.

It is interesting to read that 99% of the field gave up macroeconomics modelling. Maybe it's only interesting me because my opinions have been influenced by Václav Klaus, now the Czech president, who is a libertarian and a macroeconomist. ;-) So I guess that although he would agree with most of statements by Dr. Kling, it is conceivable that the death of macroeconomics wouldn't necessarily be one of them. :-)

All the bst

Brad Hutchings writes:

Yeah, 1988 was a scorcher. 1986, by contrast, was very chilly. I'm sure 1987 was somewhere in between!

madsocialscientist writes:

On the macro-econometric modeling front, perhaps you can answer a question. After deriving and working with RBC & NK models for a semester, I see why you probably wouldn't want to feed in a large dataset and generate predictions. However, simulation methods with these classes of models appear to be a rather interesting route. The one thing I could never get about the RBC simulations was why everyone used quarterly data; I mean, it's a long-run model, right?

Any insights on this?

eric writes:

If an ending point for a time sample generates the result, it must have a large standard error. That's why the "stock return mean reversion at 3-5 year intervals" is a weak result: include the 1930's, it's there, exclude the 1930's, it's not.

Snark writes:


The inconvenient fact of the matter is that we need to focus more on what’s for dinner rather than units of time in measuring global warming trends.

Under the circumstances, I’m sure you’ll agree that imposing a “flat” (i.e. flatulence) tax on cattle ranchers is warranted.

Ray G writes:

Flatulence taxes should be saved for posts like that one.

More to the point, short term trends change enough that, as an elementary student in the 70s, I was scared silly by my teachers into thinking another ice age was imminent within my own lifetime.

Now we're warm relative to the 80s, but Greenland is currently cooler than the first few decades of the 20th century.

The hype is so obvious, I really am genuinely surprised how seriously this is being taken as a global "emergency."

Snark writes:
Flatulence taxes should be saved for posts like that one.

Semper Fi, Ray G!

Michael Sullivan writes:

The reason it's seen as an emergency is that we can't just stop climate change on a dime. If the forecasts are correct, what we've already done commits us to a certain amount of significant climate change, and what we do in the next 20 years commits us to even more climate change (how much depends on what we do).

It seems like many of the anti-alarmist crowd are modeling this as a decision that can be easily backed out of in 20-30 years. We have no reason to believe that will be the case, and that's one of the central problems. We'll have a lot more information in 30 years, but at that point, we may *already have committed* to a much greater level of climate change than today. And this "may" is what our central-case estimates are saying *will* happen.

On the question of whether we would like to be constrained by decisions made in 1820 or 1940, I don't see that we have much of a choice in the matter. We *have* been constrained by decisions made in earlier times that make very little sense today.

The current typewriter layout is significantly less optimal than it could be, because QWERTY became standard in a time when reduucing people's ability to type quickly made sense given the extant machinery. PCs were held back significantly by design decisions of IBM and microsoft and others which did not take into account the potential for huge changes with a desire for backward compatibility. Programming decisions in the 70s and 80s around date data consigned us to huge remediation costs in the late 90s (while that remediation was probably overdone, if there had been little or none, the failure costs would have become very significant, if not civilization destroying).

Going back well past the modern era, decisions were made in the 1600s about the rights of africans that still shape our culture in negative ways today, and it is not clear when we will be completely free of that evil legacy.

We are embarked. Everything we do has potential import years down the road, we can only make best guesses about the results, and there is no clear no-effect or all-options-open position.

So what we have to decide now is whether to bequeath our descendants a little more economic growth along with a potential climate change bomb, or a little less economic growth. Note that both potential results have uncertainty. The cost of implementing, say the Stern reports recommendations in the near term, is not very uncertain, but the long term effects on economic growth are not so certain. Perhaps it will spur us to develop better technologies sooner that will lead to a bump in growth in the long run. This is the same kind of uncertainty that exists when considering the potential degree of climate change, and the economic damage from it.

The point is that there is no zero option. Anything we do has the potential to be a big mistake seen from 100 years down the road. We make the best decision we can and our great grandchildren get to live with it. But it looks to me as if decreasing greenhouse gas emissions will decrease our uncertainty about the future rather than increase it and not particularly decrease our options (beyond a one time 1% of GDP cost), so it may be worth doing on the insurance principle even if our expected EV is slightly negative.

Comments for this entry have been closed
Return to top