The Misuse of Externality Arguments


If you own a steel mill and want workers to run it you have to offer them terms at least as attractive as their next best option. If you want coal and ore, you have to offer the miners at least what it costs them to get it out of the ground. The costs of producing steel are transferred from the people who produce your inputs to you by the payments you must offer to get them to do so. Similarly, the value of what you produce is transferred from your customers to you in what they pay you for your steel. If the value of what you produce is more than the cost, it pays you to produce it and you do. If value is less than cost it does not pay you to produce it and you don’t. That is a short version—the long version requires a year of price theory and/or a good textbook—of why a market society produces things if and only if they are worth producing.

There is, however, a problem. In addition to consuming labor, ore, and coal and producing steel, you also produce sulfur dioxide, which makes things less pleasant for those who breath downwind of you. That is a cost of producing steel but, in a society without government regulation, tort law, or anything similar, it is a cost for other people, not you. So it may pay you to produce steel even if total cost, pollution included, is greater than total benefit. Similarly, if you produce external benefits for other people, such as the pleasure passers by will take in looking at your beautiful building, you will ignore those benefits in deciding whether to build it. Even if the total benefit is greater than the cost it may not pay you to.

Economists refer to such effects as externalities—negative (external costs) and positive (external benefits). Their existence is one of the arguments economists offer for government interventions in the market. Air pollution is a negative externality, so regulate or tax it. Knowledge produced by basic scientific research is a positive externality, so subsidize it. The argument is correct in theory. Externalities are one example of what I described in Chapter 52 as market failure, so a government that forces people to take account of them can, in principle, improve on the result of the unregulated market.

There is, however, a practical problem. In order to do the job properly, the government needs to know the sign and size of the externality. Without that information, they might provide the wrong tax or the wrong subsidy or even tax something they ought to subsidize or subsidize something they ought to tax.

Today the problem that all the experts insist must be solved immediately if not sooner is global warming. Forty years ago it was population. It was widely asserted that the more people there were on earth the worse off we would all be. Quite a number of people went further, claiming that we were facing immediate catastrophe. In The Population Bomb, published in 1968, Paul Ehrlich wrote that “The battle to feed all of humanity is over. In the 1970s hundreds of millions of people will starve to death in spite of any crash programs embarked upon now. At this late date nothing can prevent a substantial increase in the world death rate... .” While his position was more extreme than that of many others, his work was taken seriously and less extreme versions were as widely accepted then as concerns about global warming are now.

My first piece of published economics was a pamphlet written for the Population Council at the request of its president. He observed that almost all discussion of the issue was coming from one side of the ideological spectrum and was sufficiently open minded to want to see how the question looked from a different viewpoint, so asked me to write a piece on population issues from a pro-market point of view.

The economic issue was one of externalities: If one more child was born, were other people worse off as a result, and if so by how much? My conclusion was that not only could I not tell how large the externality was, I could not even tell whether it was on net negative or positive. Nor, as best I could tell, could anyone else, although quite a lot of people thought they could.

Increases in population have both positive and negative effects. More people means more people to create pollution and commit crimes, more people to go on welfare, but also more people to make inventions, pay taxes, write books and compose music. The effects in both directions are large and, since they are spread over a long and uncertain future, very hard to estimate.

The simple argument for the conventional view, that more people meant less land and natural resource for each, was simply bad economics, at least in the context of a private property society. A new born baby does not arrive with a deed to his per capita share of the world’s resources clutched in his fist. If I want my child to own land on which to live, either he and I will have to produce something of value that the present owner of that land is willing to accept in exchange, in which case obtaining land for him has not made the rest of the world worse off.

Some other arguments went in both directions. The more children, the greater the cost of putting them through the public schools, a cost not born directly by their parents. Children, however, grow up to be taxpayers. Match up the cost of schooling each child with the taxes he will later pay for schools and the two roughly cancel, making a net externality close to zero.

Some externalities are clearly positive. More people means less of the national debt for each to pay.  Some are clearly negative. Overall, I did not see how one could estimate the externalities, positive and negative, accurately enough to sign the sum.

Many years later, I encountered the same problem in the context of global warming. Most of the public argument was about whether global temperature was going up, whether the reason was carbon dioxide produced by human activity, and how large future warming was likely to be. Practically everyone took it for granted that warming was a bad thing, probably a very bad thing.

I could not and can not see why. The current climate was not designed for us, nor we for it, global climate having varied quite a lot over the history of our species. Humans presently live and prosper across a range of climates much larger than the predicted change. The only a priori reason I could see to expect change to be bad was that we are currently optimized to our current environment. That might be a serious problem for rapid change. But the global warming being projected was at a rate of about a third of  degree centigrade per decade. That should be more than slow enough to let farmers adjust their crops, home owners their housing, to the change at little cost.

If we cannot produce an a priori reason to expect warming to produce net negative externalities, perhaps we can work out the consequences and simply add up their cost. Warming can be expected to raise sea levels a foot or two by the end of the century. Looking at the other side of the equation, human habitability at present is limited by cold not by heat; the equator is populated, the poles are not. An increase in global temperature that raised sea levels by a foot or two would push temperature contours in the northern hemisphere hundreds of miles further north, increasing the area of earth warm enough for human habitation by about a thousand times the area lost to sea level rise. On average, the land gained would be less valuable per square mile than the land lost, but would one expect it to be a thousand times less valuable?

Warming will have other effects negative and positive. Since we are talking about effects over a period of a century or more—William Nordhaus, an economist who has specialized in studying the warming issue, runs one of his calculations out for two hundred and fifty years—it is impossible to make any reliable estimate of their size. Readers interested in a longer discussion of my reasons for thinking that the net effects are at least as likely to be positive as negative will find it on my blog.[1] My point here is a more general one—an explanation of why the attempt to base policy on estimates of external costs and benefits led to mistaken conclusions about population fifty years ago and very dubious conclusions about global warming today.

Consider someone adding up externalities in order to decide what policy should be, whether government should encourage or discourage population growth, tax carbon or subsidize it. If he starts out believing that the net effect of warming is negative he is likely to make generous estimates of the negative externalities, conservative estimates of the positive, and miss some of the latter because he is not looking very hard for them. He ends up honestly convinced that the objective evidence supports the position he started with. If he starts with the opposite belief his calculations will have the opposite bias and he will reach the opposite result.

A few years ago, I came across a striking example of this pattern in the work of William Nordhaus. His research on the effects of global warming found negative externalities sufficient to justify imposing a carbon tax to slow it, although both the externalities he found and the measures he proposed were modest compared to the views of some other researchers, still more so compared to the views of activists such as al Gore.

In order to get even that modest result it was necessary to include in his estimate of costs not only predictable effects such as sea level rise but also possible but unlikely consequences of warming that, if they did occur, would impose large costs. In a book coauthored with Joseph Boyer, the authors wrote that “this approach is taken because of the finding of the first-generation studies that the impacts on market sectors are likely to be relatively limited." [2] Or in other words, without including the costs from unlikely but catastrophic risks, global warming did not seem to be a serious problem.

There is a curious asymmetry to their approach. They took account of low probability high cost consequences of permitting global warming. But, so far as I could tell, they made no similar attempt to take account of low probability, high cost consequences of preventing global warming. That might make sense if we could be reasonably confident that, absent the effects of human action, climate would never change. But we have no grounds for such confidence, since climate has been changing, sometimes quite radically, since long before human beings were able to influence it.

We are currently in an interglacial, a relatively warm period within an ice age that began more than two million years ago. We do not know what causes interglacials to start—or end. Long term estimates of global temperature suggest that it has been trending slowly down for a very long time, possibly since the beginning of the current interglacial, a trend reversed by current warming. It is at least possible that global warming is all that is preventing the interglacial from ending. The result, judging by past glaciations, would be a drop in sea level of more than three hundred feet, leaving every port in the world high and dry. Also half a mile or so of ice over the present locations of London and Chicago.

I do not think the prevention of global warming would be likely to set off that particular catastrophe. It is, however, possible, and there may be other unlikely but possible catastrophes that have not occurred to me. Nordhaus, looking for negative effects of global warming, included the unlikely ones. He did not, so far as I can tell from that book, a later one,[3] or correspondence with him—when writing this chapter I wanted to make sure that I had not misread his work—include any estimate of unlikely effects in the other direction.

Which I take as evidence of the problem of using externality arguments to produce policy conclusions. When balancing costs and benefits, it is only too easy to put an unconscious thumb on the scale.

[1] The blog is To search it for my posts on global warming, use the URL:

[2] Roll the DICE again: Economic Models of Global Warming by Willaim D. Nordhaus and Joseph Boyer, at


[3] A Question of Balance by William D. Nordhaus, at: