It Can't Possibly Be That Easy

Over the weekend, I read Paul Krugman's big essay on climate economics, Building a Green Economy.  In it, he makes the following claim:

Just as there is a rough consensus among climate modelers about the likely trajectory of temperatures if we do not act to cut the emissions of greenhouse gases, there is a rough consensus among economic modelers about the costs of action. That general opinion may be summed up as follows: Restricting emissions would slow economic growth — but not by much. The Congressional Budget Office, relying on a survey of models, has concluded that Waxman-Markey “would reduce the projected average annual rate of growth of gross domestic product between 2010 and 2050 by 0.03 to 0.09 percentage points.” That is, it would trim average annual growth to 2.31 percent, at worst, from 2.4 percent. Over all, the Budget Office concludes, strong climate-change policy would leave the American economy between 1.1 percent and 3.4 percent smaller in 2050 than it would be otherwise.

And what about the world economy? In general, modelers tend to find that climate-change policies would lower global output by a somewhat smaller percentage than the comparable figures for the United States. The main reason is that emerging economies like China currently use energy fairly inefficiently, partly as a result of national policies that have kept the prices of fossil fuels very low, and could thus achieve large energy savings at a modest cost. One recent review of the available estimates put the costs of a very strong climate policy — substantially more aggressive than contemplated in current legislative proposals — at between 1 and 3 percent of gross world product.

Such figures typically come from a model that combines all sorts of engineering and marketplace estimates. These will include, for instance, engineers’ best calculations of how much it costs to generate electricity in various ways, from coal, gas and nuclear and solar power at given resource prices. Then estimates will be made, based on historical experience, of how much consumers would cut back their electricity consumption if its price rises. The same process is followed for other kinds of energy, like motor fuel. And the model assumes that everyone makes the best choice given the economic environment — that power generators choose the least expensive means of producing electricity, while consumers conserve energy as long as the money saved by buying less electricity exceeds the cost of using less power in the form either of other spending or loss of convenience. After all this analysis, it’s possible to predict how producers and consumers of energy will react to policies that put a price on emissions and how much those reactions will end up costing the economy as a whole.

There are, of course, a number of ways this kind of modeling could be wrong. Many of the underlying estimates are necessarily somewhat speculative; nobody really knows, for instance, what solar power will cost once it finally becomes a large-scale proposition. There is also reason to doubt the assumption that people actually make the right choices: many studies have found that consumers fail to take measures to conserve energy, like improving insulation, even when they could save money by doing so.

But while it’s unlikely that these models get everything right, it’s a good bet that they overstate rather than understate the economic costs of climate-change action. That is what the experience from the cap-and-trade program for acid rain suggests: costs came in well below initial predictions. And in general, what the models do not and cannot take into account is creativity; surely, faced with an economy in which there are big monetary payoffs for reducing greenhouse-gas emissions, the private sector will come up with ways to limit emissions that are not yet in any model.

Now, it's important to note that the goal of the Waxman Markey bill is to reduce US carbon emissions by 83% by 2050 (from 2005 levels, so even more than that from 2010 levels). So essentially, the CBO is saying, and Krugman is endorsing, that this level of emissions reduction will have so small an effect on economic growth that it's going to be indistinguishable from noise. I don't dispute that environmental economists think this, but I find it to be a completely facially implausible conclusion. I want to lay out two arguments for why these economists cannot possibly be right. The first is a common-sense argument about what actually has to happen at the level of the lives of individual citizens to bring about such a large reduction in carbon emissions. The second argument is based on looking at what was required to cause significant changes in energy efficiency in past episodes.

Let me start by saying, for any new readers that happen to stumble across this, that I believe the general thrust of the scientific consensus on climate change, and I strongly agree there is an excellent case for decisive action.  See here and here for some past relevant posts.  In general, I like to get my information on climate change from reading Science, Nature, and PNAS, rather than from partisan political sources.  However, I also believe in being realistic about what one is proposing to do, and honest about the implications.

Next, let's think briefly about some implications of the quantitative claims above about economic growth and emissions reductions.  US trend economic growth in recent decades is about 3% a year.  So between now and 2050, in a business-as-usual future that is similar to the recent past, we would expect the economy to grow by 1.03⁴⁰-1 = 225%. So the economy will be about three times as large as it currently is. Some of this will come from there being more people in the US, but more of it will come from the people being wealthier, which of course they generally like to express by having bigger houses, bigger and faster cars, and more advanced technology to fill them both with.

Now, if the economy is going to be a bit more than three times larger, but we are only going to emit 17% of the current level of carbon emissions, then the carbon intensity of the economy - that is the ratio of carbon emitted per dollar of goods and services created, is going to have to be only 5% of the current value.  Next you have to figure that there are certain things in an industrial society that are very hard to do without liquid fuel - construction and agricultural machinery come to mind, along with aviation.  Relying heavily on biofuels is a very dubious prospect in a world that also needs to feed 9 billion (assumed wealthier) people from its limited agricultural land.  So you can probably figure that the residual 5% of carbon emission intensity is all going to go on these kind of specialized uses that are hard to substitute.

Therefore, these goals basically imply that the ordinary living and working of most citizens would be essentially carbon free by 2050.  That is in 40 years time.

Now, I can certainly imagine a middle class lifestyle and workstyle that is carbon free.  The technology is almost there.  For example, we could live in super-insulated passive solar houses, we could drive electric cars to work at our super insulated zero-emissions offices and factories.  The electricity to power our cars, provide for residual heating and cooling needs, and drive our industrial production would all (or almost all) have to come from some combination of renewables and nuclear, rather than the coal and natural gas that form the bulk of it today.  I think if everyone did something along those lines, we could get down to 5% of our current carbon intensity.

But it should be clear that this basically requires replacing almost everything in our society.  Since today, our houses are by and large made from R12 2 x4 stud walls, they pretty much would all need to be replaced to avoid the need for lots of heating/cooling energy.  Ditto our commercial and industrial buildings.  And of course most of current electricity generation infrastructure would need to go too.  Finally, of course, all the cars will have to be replaced.

Now, the lifetime of cars is much less than 40 years, so they will all be replaced anyway; that's not a problem (though there certainly are questions about the ultimate scalability of that many electric cars). But the median age of a house is 35 years. Here's the age of housing as of 2003 according to the US Census Bureau, American Housing Survey for the United States:

As you can see, there's a lot of houses that are more than 40 years old.  Note also that this kind of graph tends to understate the life of houses - most of the young houses are built on greenfield sites on the edge of town, and most of the older houses in town are still there.  So we are going to have to do a lot of extra replacement to get to 5% of current carbon intensity.  Instead of just building big houses on the outskirts of town, we also need to go and replace everything in town.

And of course other kinds of infrastructure tends to last even longer than houses - for example, the median age of current coal plants is 44 years.

Now, think of it this way: suppose you have a certain amount of money to spend over the next forty years that is your share of industrial society's surplus.  You could take that money and either a) tear down your house and replace it with a super-insulated carbon-neutral one of about the same size, or b) add an extra floor and a swimming pool to the house you have and continue to power it with cheap fossil fuels (coal and shale gas, let's say).

I would argue that this is, very roughly, what the choice between business as usual and an 80% reduction in carbon emissions means in personal terms.  I think at the personal level, most of us can understand that if we have to completely replace the house, we're not going to end up with the same amount of house as if we just add to the one we have.

My second argument is based on looking at past history.  In order to get to 5% of our current carbon intensity in 40 years, we need to improve carbon efficiency by an average of 7.2% per year (0.05^1/40 = 0.928).  That's a very large rate of change.  In particular, our main experience with society making serious improvements in energy intensity is as a result of the oil shocks of the 1970s.  I have looked extensively in the past at the effect of those shocks.  Here for example, is the year-on--year rate of change of deployed vehicle fuel economy in the US fleet (see here for methodological details):

As you can see, in the late 1970s and 1980s, we reached a level of fuel efficiency improvements of around 2-3% a year, sustained for a little over a decade.  The peak year was an improvement of 6.5%.  But what was required to kick that off?  Two massive oil-shocks, each of which led to a big recession.  The first was the Arab oil embargo in 1973-74, and the second was the effect of the Iranian revolution in 1979, and the immediately following Iran-Iraq war.  Google shows the effect on US GDP as follows:

So to get a puny 2-3% a year for a decade, just in the liquid-fuel sector, took two major recessions (or three, if you include the 1982 one).

So is it really plausible that we can price carbon high enough to improve our fossil fuel intensity across the whole economy by an average of 7% a year for 40 years, and it have no effect on growth?  That appears wildly implausible to me.