Solving the problem of externalities

Externalities have the appearance of side-effects as losses or gains for third parties. Without a clear mechanism for dealing with externalities, the causative agents of those remain detached from the side-effect losses or gains that they create. And that’s a big problem, because then the incentives of those causative agents are misaligned with the general good of society. This may be one of the greatest, though relatively subtle, problems of our time. Let’s try to solve the problem. We don’t have much better things to do anyway. :smiley:

Measuring externalities

The first part of the problem consists in operationalizing and measuring externalities. Is there a way to deal with externalities that represents them in an easy and appropriate manner? Unfortunately, this doesn’t seem to be the case, since estimating the value of externalities requires a rather holistic perspective on the economy that’s difficult to achieve. For example, the release of greenhouse gasses has global effects that are very much varied and hard to estimate. On the other hand, the creation and free access to digital goods represents a positive externality whose benefit is perhaps even more uncertain. How do you measure the value of a million freely available songs or books? Surely, you can’t just go ahead and attribute a value equivalent to the market price of similar songs or books to those, since otherwise you’d end up in a situation in which every person with access to the internet would possess a value of hundreds of millions of dollars, or more. Perhaps that might not seem too inappropriate, but considering the real sitation that most people don’t have so much money, they couldn’t possibly pay for all the value they get for free, if free digital goods were said to have a value equivalent to market prices of non-free digital goods or equivalent material goods like books.

Excursion on the time/attention dimension

Then, we should also consider that nobody can even make use of freely available digital goods, even if they had a value that we’d estimate in the range of billions of dollars, because people only have a limited amount of time they can allocate to consuming such goods. The more goods are available on a certain (minimal) budget, the more dominant the scarcity of time or attention becomes relative to the scarcity of access. A digital society is dominated by an attention economy, not by a monetary economy! The relative value of time with respect to money is increasing, as we move towards digitizing our goods and services. Sure, there are ways to exchange money for additional time in many respects, for example by outsourcing daily chores and tasks to people who get paid for doing them, for example cleaning, caretaking, shopping, research, investment, and so on. But there are limits to this process. Nobody has more than 24 hours a day, and nobody can outsource the experience of reading a book, or decrease the time required to absorb its knowledge content below a minimal amount defined by the information throughput and processing capability of that person.

Time as measure of value

This excursion serves to demonstrate that it’s hard to simplify the value of externalitites to a purely monetary dimension, since money plays a non-absolute and diminishing role in our economy. This stems from the fact that we can’t pay people in “time” directly, since time is not fungible. Yet, if we accept this line of reasoning, we have to come to the conclusion that the problem of externalities cannot be solved, because externalities impacting the free access of time for third parties cannot be appropriately compensated. Well that remains true, unless we manage to develop technologies that can accelerate time for specific persons relative to the rest of the world! The only technology that would truly qualify for that would be substrate independent minds who could be accelerated nearly arbitrarily. For the sake of simplicity and convenience let’s run with the counterfactual assumption that we possess such a speculative technology and the cost of time is linear with respect to its amount – such a basic assumption is also used in the book The Age of Em by Robin Hanson. In such a situation, we could see each unit of time as a unit of a universal currency. As we today measure the value of goods and services in dollars or Euros, we could measure their value in seconds or hours. In this sense, gaining an hour of time would mean getting your mind accelerated that you experience one hour of subjective time more than the general norm of society, since your mind is accelerated relative to it. Vice versa, a cost of an hour would mean that your mind runs slower, and you experience an hour less than “standard society”.

The universality of the value of money depends on our level of technology

Does this framework help addressing the problem of externalities? Well, with the “fungible hour” we have a more universal measure of value than money nowadays. But of course, if we can trade hours, we can also trade them with money, so the real difference that the assumption of real fungibility of time makes, is to make money more universal as measure of value. If anything, this shows that the usefulness of money as representation of value is limited by our level of technology, rather than by any intrinsic shortcoming. So, let’s assume that we could, in principle, measure the value of externalities in some form of currency, at least given a sufficiently high level of technology.

How would a direct solution look like?

Now, let’s assume that we could precisely measure the cost or gain of any externality for any involved person. Then, we could imagine an institution that precisely transfers the cost or gain created for other persons to the causing person of that externality. That looks pretty much like a perfect solution to the problem of externalitites. What stops us from implementing such a solution is the sheer difficulty of implementing such a system. We cannot properly measure the costs or gains inflicted on other parties. We also cannot effectively coordinate to reflect those costs or gains back to the causative agents of the externalities, especially when those are global externalities. While we can assume that a ubiquitous sensor network and advanced artificial intelligences might be hugely helpful for solving the first problem, the second problem is a political one, since we don’t have a universally accepted global externality compensation institution. Yet, this line of reasoning shows the necessity such an institution for implementing a straightforward solution to the problem of externalities.

Externalities as basis for basic incomes and merit-based incomes

Such a solution would imply that each person is compensated for the negative externalities caused by any other person. And also that any positive externality that one creates is reflected back to oneself as additional income. While the first could provide the basis for a negative externality based unconditional basic income, the latter would be the basis for additional incomes based on merit. In such a system taxation would become completely obsolete, respectively be replaced by a purely rational compensation system for externalities. The more we transition our real systems of taxation into that direction, the more appropriately can we deal with the problem of externalities.

Contemporarily possible approximations

Since we are still far away from being able to measure the value of externalities properly, or compensating everyone for externalities, we need to look at approximations for that solution that we could implement with our current levels of technology and political integration.

Carbon and land value taxes

The first approximation are taxes that are motivated by the idea of compensation for externalities. Examples for those are in particular carbon and land value taxes. Since everyone is affected by the cost of those externalities (in the second case as loss of the opportunity to make use of a certain piece of land for free), everyone should be compensated in the form of a universal basic income. In the case of carbon taxes, such an income should ideally be global in nature, perhaps most practically implementable with a digital currency, once the technological hurdles of getting everyone real access to those have been overcome. In the case of land value taxes, the income could be local in nature, since people usually prefer to make use of land in their (political) proximity.

Data incomes

Another possibility would be patients who make their medical data available for any interested researcher. This positive externality of available data should be reflected back to the patients who share it. Such an income would be a data-based income that perhaps may become high and universal enough to make universal health insurance obsolete – at least for those who opt in to such a system. Similar data incomes could be considered for personal data that people make available via social networks.

Usage based incomes

Digital goods can only provide value if they are actually consumed. Digital goods that lie around idle are pretty much worthless, with their only value stemming from their potential of being used. So, it seems to be rather rational to measure the use of digital goods and pay their creators in proportion to that use.

Donations and reputation incomes

One problem with the previous approach is that the subjective value of digital goods can vary dramatically. If people are able to express their preceived subjective value of the use of those digital goods directly, then that would be a solution that makes a lot more economic sense, as long as people are rational about their indication of subjective value, and actually care about doing that at all – boundary conditions which one cannot take for granted. Such appreciation of subjective value can form in the form of donations, or of product ratings. While the first approach is gaining in popularity via platforms such as Patreon, the latter could be translated into reputation-generated incomes with a system like Quantified Prestige.

Is inequality an externality?

And if yes, what kind of externatlity is it? One can argue that it’s a positive externality, since it provides motivation for improving or maintaining one’s status in society through intelligent effort (and therefore hopefully creating value for society which occasionally takes the form of positive externatlities). As long as we can’t generate such motivation through technological means, that may remain a valid argument. On the other hand, high levels of inequality seem to come with societal problems with their own attached economic costs. In that case, inequality might be seen as negative externality. There may be a level of inequality that’s “externality neutral”, in the sense that its values as positive and as negative externality are exactly equal. In any case, in an economy that is based on the compensation of externalitites, inequality would be pushed back to this equilibrium point of externality neutrality through its externality compensation mechanisms. As side-effect the idea of externality compensation therefore results in a “natural” way to define “optimal” levels of inequality.


Certain kinds of basic incomes, additional incomes, merit-based and reputation-based incomes can be interpreted as externality compensation incomes. On the other hand, certain kinds of taxes like carbon taxes or land value taxes can be seen as externality compensation taxes. The intention to solve the problem of externalities in a rational way leads to the view of considering such interventions as desirable economic policies. The principle of externatlity compensation should be a pillar of the economics of the future.


thank you for putting in your thoughts about externalities.

Am i right, that your idea is something like connecting all the brains, and the punishment for bad externalities is that your brains computing power is used by someone else, while you gain computing power of the other brains by good externalities?
Also you could use a less dramatic version, positive externalities could grant you access on some sort of bot network or a google cloud to do a computation for you or to give a task to a person with a bad externality.

Still another problem exist, using an externality is sometimes like recycling junk, so basicaly you are forcing people who recycle to pay for the junk. I dont know if that is always a good idea. On the other hand, paying for an externality is a method to stop exploitation of free work, which is obviosuly a good thing. So i havent decided yet if paying for an externality is good or bad :slight_smile:

Instead of going after externalities alltogether, it might be an idea to constrain that to certain aspects, like damage done to the ecology. That of course can be handled by higher taxes on certain products and laws.
On the other hand products that are helping the ecology should get funded.
that is all happening already though.

I stll didnt work though the QP paper, i have to shovel time free to do it. soon, i promise :stuck_out_tongue:

going after the externalities is basically close to the socialist program, which wants to go after the capitalist profit which is some sort of externality for owning the working place, if you want to use this perspective.

That’s an interesting thought, but I don’t think it’s feasible to reallocate the computing power of human brains like that. You can’t just use 20% less of your brain, so that someone else can amplify his brain with yours.

What I had in mind, was distributing rights to have your mind transferred to a faster computational substrate. Let’s say we invent some kind of optical chip that can think as much as a human brain does, but does that a million times faster. Then having your mind transferred onto that chip even temporarily, would grant you a huge amount of additional time (though the downside would be increased subjective lag when communicating with the outside world). Perhaps there’s an architecture that’s even a billion times faster, but it’s expensive because it requires very special cooling and equipment. Then switching between the two fast substrates would still make a huge difference (a factor of 1000).

Yes, we could start with that kind of implementation of “computation” or “time” as a currency. Let’s just delegate certain cognitive task to a cloud-based exocortex.

That’s an interesting interpretation. Usually, people give junk away for free, if they can. Sometimes, people have to pay others to get rid of certain kinds of junk. In some cases, only this payment makes recycling economically feasible. If we forced people to pay for junk they use, then few people would be interested in getting/buying junk, so we would have a huge junk accumulation problem. And that problem would drive black market junk disposal prices upwards.

It all depends on whether junk is a positive externality or a negative externality. Common wisdom says that junk is a negative externality, and people should therefore have to pay for producing junk. If however, junk actually turns out to be a positive externality, then the conclusion should be that people are stupid for disliking junk. Perhaps we are all very stupid in that regard, but I’d like to have at least some evidence that points into that direction.

For selfish people, paying is always bad. Why pay for something that you could get for free? :smiley:

Why restrict oneself like that? A more holistic approach can fix more problems at once.

Yes, but the problem with that, is that there’s no clear basis for the amount of such taxes, or the amount of funding in those areas. If we could measure the value of externalities properly, we would have a rational basis for reallocating money in an “obviously fair” way. Granted, that’s a big “if”, but we need to start with some idealizations, if we want to establish new paradigms in economic thinking.

Profit is not an externality of owning a working place, it’s the quintessential reason for owning a working place. Intended outcomes usually don’t count as side effects / externalities.

the point is, that externalities are basically left behinds, for example if i leave information about me while i am buying a certain product, the information might help the company to get a better profile of a typical costumer, while for me this information is totaly worthless, that means if the company wont use the information i wont gain anything.

You could think of externalities like unused value nobody is claiming. Like the heat caused by my processor. If somebody finds use for this energy i wont lose anything when he just takes it.

on the other hand it would be fair to give a share to those people causing the externality, which is propably the point you want to make,

i agree.

its a matter of viewpoint. the profit might be the reason for the capitalist to run the workplace, for the worker the profit is unintresting, unless it is directly connected with his wage. Especially workers with an unlimited contract and no carrier prospects are pretty unintrested how much their company is actually gaining from their work.

I want to expand a bit on the problem of externalities. In particular, I’ll focus on the problems with measuring externalities, and the implications of rigorously enforced externality compensation.

Measuring externalities

Externalities are economic side-effects of actions. This implies that if we want to operationalize, measure, and quantify externalities, we have to track all side-effects of any relevant action. Since most side-effects will unfold in the more or less distant future, this is quite complicated. We could only reliably track side-effects located in the future once we reach that future. If we try to account for externalities unfolding in the future, we would have to predict the future to a sufficiently detailed degree. Of course, that’s a very challenging task and riddled with uncertainties.

There are basically two alternatives:

  1. We track externalities as they unfold, and compensate the causative agents of those externalities just in time, as those effects occur.
  2. We predict the future effects of certain actions and compensate causative agents of externalities based on expected future effects.

Let’s consider an example. Imagine we have a mining company that uses aggressive mining techniques that release potentially toxic chemicals into the environment. Of course, this poisoning of the environment is a negative externality. How do we compensate for this poisoning of the environment? Why is that posioning even bad in the first place? Well, it is to be expected that humans, animals, and plants will suffer ill effects over time from that poisoning. So, we could try to track the health of those entities over time, and let the mining company pay for the damages to their health, as the health problems occur. There are of course many difficulties involved with this approach:

  • How can we be certain that specific health problems really stem from the toxins released by the company? Perhaps the toxins were already present in the environment anyway, and the company merely increased their concentrations. What if the presumed toxins aren’t really as harmful as the current state of our knowledge suggests? What if people merely get sick, because of the nocebo effect, caused by the belief that their environment has been poisoned? Should the mining company get compensated for the payments it had to make that were based on faulty data? And what if the effects turn out to be worse than expected? Would it be fair to punish the company for the bad luck or lack of knowledge that the used chemicals were more dangerous than initially estimated by experts?
  • How do we quantify the damage to the health of various organisms? Of course we can try to treat the affected organisms, but those treatments are likely not to be sufficient for restoring their health to levels prior to the intoxication. How do we quantify damages that can’t be appropriately treated, in particular lethal damages? How do we quantify the damage to the ecosystem at large?
  • Wouldn’t it make more sense to take a preventive approach and treat potentially affected organisms with antidotes to the toxins that are released? Health costs from expected intoxication would be minimized that way. Therefore, the costs of those externalities are potentially reduced.

If we try to take the second approach, even more problems arise:

  • Over what period of time in the future are likely impairments of health considered? 10 years? 20 years? 50 years? 100 years? The indefinite future?
  • Should effects in the future be depreciated due to uncertainty or expected improvements in health care?
  • What if the previously predicted strength of the effects turned out to be wrong? Do we then re-compensate according to our updated knowledge, or do we maintain the principle that a one-time compensation based on the currently best knowledge completes the process of externality compensation?

Yet, it seems to be necessary to choose the second approach of incorporating expected future effects, especially when the effects are likely to be delayed. Imagine that the released toxins only show measurable effects 10 years down the line. For 10 years, the company would be spared from the ill effects of its doing. Once the effects get apparent, it would be suddendly confronted with severe consequences. Investors might anticipate this and reap the profits from the first 10 years, just to cut off their ties with that comapny shortly before the negative effects become apparent. This kind of situation causes the perverse incentive to invest in companies whose negative effects are likely only to emerge in the long-term. Taking expected future externalities into account would fix that problem.

This is going to be very expensive

Yet, if we require tracking present and expected future effects of all relevant economic activity, then the cost of the tracking might quickly outweigh any benefit that rigorously enforced externality compensation might provide. We simply can’t afford to use armies of futurists and domain experts to estimate current and future externalities for each company. Of course we might hope that increasingly sophisticated AI will help us tremendously with this task, but its main effect is only to bring down the costs of tracking and estimation of externalities. We might start with the most easily trackable externalities and then move on to more subtle externalities as technology makes that activity economically viable. There’s no other reasonable way to go about this problem anyway.

Basic income as externality compensation flatrate

A basic income could be considered as the most primitive way to go about externality compensation: Those who pay for a basic income compensate the public for the overall expected negative externalities of their actions. On the other hand, the recipients of basic income get compensated for the overall positive externalities created by all of society. The advantage of this approach is that it reduced the effort of measuring actual externalities to a minimum. In that sense, basic income is a surprisngly cheap way of implementing externality compensation – at least when it comes to the costs of externality accounting.

Expanding externality compensation to update our legal system

We might expand the principle of externality compensation to turn it into the basis of our legal system: Everyone is accountable for all the externalities caused by them. We can count harms to other individuals as externalities, since harming them is rarely the primary motivation for those harms. Instead, individuals seek their own profit, and the harm to other individuals is merely a side-effect of certain ruthless profit-directed actions – for example theft or robbery. Expanded externality compensation merely means that those who cause harm to others are supposed to pay up for those damages.

This would imply that we move away from non-monetary punishment and merely let perpetrators pay for the damage that they have caused. I’m certainly not the first person to suggest this approach, but it’s merit lies in its simplicity: No messy laws and arbitrary punishments. Everything is merely reduced to fines. An intuitive reaction to this idea is that mere fines won’t suffice to represent a sufficient deterrance. Most people would claim they don’t want to live in a world in which billionaires could get away with horrible acts by merely paying large fines. A cynical reply to that objection could be that our contemporary elites frequently do get away with all kinds of crimes already, because they are very unlikely to get persecuted in the first place, and can affort very good lawyers, so in practice the different wouldn’t be that big anyway.

Perhaps an even more severe problem arises when perpetrators are unable to pay for the whole damage that they have caused, as is likely to be the case when they are poor and commit quite serious crimes. Would the right procedure be to make them indebted to the legal system or to society at large? How do we deal with that debt? Will criminals be required to pay interest on that kind of debt? Should that debt be adjusted to inflation? How and when should we demand the payment for that debt? Should declaring bankruptcy in such a situation be allowed? Can society decide to cancel the debt? And if yes, what’s the right procedure for that? It may be the case that these practical problems are the main reasons that our legal system has the complicated shape it had right now, rather than being based on monetary externality compensation alone.

Maybe the best way to go about this, is to translate debt that cannot be paid back immediatey into an increased tax rate on their income – until all the debt is paid back. At least once persons are expected to live indefinitely due to life extension technologies, increased tax rates may appear to be a sufficiently deterring punishment even for the most destructive actions. Perpetrators can try to opt out of such penal taxes by comitting suicide, which would equate to a voluntarily accepted death penalty.

Justice as degree of externality compensation enforcement

Thinking about externality compensation could be seen as trying to make the world more just. If we continue this line of reasoning, we could even turn the degree to that we are able to enforce appropriate externality compensation as new definition of justice. This would enable us to turn justice into a measurable and quantifiable concept. Injustice would be accounted as failure to enforce appropriate externality compensation. With this line of reasoning it could make sense to say something like “our injustice level lies at 10% of GDP” (if we will still use GDP as meaningful economic measure in a future that accepts the premise of injustice as failure to compensate for externalities appropriately, which may not be terribly likely).

i really like your idea to transform any punishment into monetary debt.

I want to get back to the problem of measuring the value of negative externalities. How to estimate the value of the harm you’ve created for someone else? Well, there is a method for that which should shed some light into the darkness: You simply ask people how much they would be willing to pay to be spared from that harm if they had the (enforced) choice of suffering from that harm themselves, or paying a certain amount. The amount at which the preferences shift between accepting the harm and accepting the payment is the subjective value of that externality. (There are certainly great citiations for that, but I don’t have any on top my mind and don’t want to invest the time to Google them.)

Of course the problem is that this is subjective. One problem is that intensities of pain in response to certain kinds of harm vary quite a bit. Some people are more prone to experience larger intensities of pain than others. The other problem is that people have different amounts of money to protect themselves from harm. If you are broke, you don’t have the option to spare yourself from any amount of additional injury. The only way out may be to accept a certain amount of debt. Excessive amounts of debt however, may quickly resemble slavery. So, that approach is problematic, too.

Let’s take a rather extreme example to test the limits of this approach: If you kill a person without the primary intention of killing her, this could be seen as negative externality of your primary objective. Now, if you applied our approach above to measure the value of this negative externality, this would come down to the question of how much you would be willing to pay in order not to be killed. First of all, this may seem like a simple question, and intuitively most people would be eager to say “everything”. But as we have seen above, people have vastly different amounts of money, so this cannot be a good universal way to measure the value of an externality of that kind. In order to arrive at a more intersubjective valuation of the negative externality value of killing a person, we do need to allow people to accept a punishment in the form of debt. So, you can pay more than all material possessions you currently have. You can accept a debt that might take decades or even centuries to pay off.

Now if we accepted that line of reasoning and applied it to the consumption of meat from factory farming, we might be inclined rather to focus on the suffering of the animals involved, rather than their deaths. How much would people be willing to pay (including the possibility of debt) in order to get spared the complete suffering of a farm factory animal experienced throughout their lifetimes? Certainly a lot. The real externality value of eating a steak would certainly be far above a thousand euros, or alternatively, accepting years of intense suffering. The number of people willing to accept any of those deals would be negligibly small.

Thus, a policy of adding the value of such externalities to the price of such products would be about as effective as outright banning factory farming, if not more so (since those extra costs come with a moral justification).

But that is just a simple example. What about more complex problems like air pollution and its effects of public health and the climate? In this case, you aren’t inflicting a large harm on a single individual, but a very small amount of harm on a very large number of individuals. In the case of air pollution we could try estimating the marginal increase of the probabilities of getting certain diseases from that extra pollution that you are responsible for, multiply those with the externality value of each disease and sum them all up. Fine, at least that provides us with a lower boundary of the negative externality value of pollution, but what about more indirect effects like the ecological impact of climate change? We’d certainly need a thorough understanding of those impacts, including negative as well as positive effects of anthropogenic climate change.

If we find out that the negative externality value of anthropogenic climate change is vast, then the flipside of that would be, that we should pay vast sums as rewards for measures that prevent or reverse anthropogenic climate change. Geoengineering efforts might quickly become extremely lucrative, suggesting themselves as prime investment with incredible payoffs. This change in economic policy could fix problems with climate change faster than anything else.