TL;DR A company that maximizes its bond price instead of its stock price cares about the long-term future and is incentivized to reduce existential risk.
The world that solves the alignment problem has an institution with the right incentives
There is a world in which we solve the alignment problem. We are not in that world.
The world in which we solve the alignment problem has institutions with incentives to solve alignment-like problems. The same way Walmart has incentives to sell you groceries for a reasonable price. In countries with functioning market economies, nobody thinks about “the grocery problem”.
One such institution is one that has incentives to care about the long-term future. I believe that there is an easy way to create such an institution in our current world.
Eternal companies maximize their bond price instead of their stock price
The first insight is that to bet on the end of the world is to take a loan that you expect not to pay back. Hence to bet against the end of the world is to buy a bond that pays installments far into the future. An example of such is a perpetual bond.
When a company sells a perpetual bond they promise to pay regular installments on the bond until the company fails. (If this concept seems alien to you there is a great video by Tom Scott on the concept)
Normal companies try to maximize their risk-adjusted returns (i.e. their share price), but what if we had a company that maximized the price of its perpetual bonds? We can call this type of company an eternal company.
The price of perpetual bonds are determined by interest rates (which the company’s executive has no control over) and the company’s risk of default (which the company’s executive does have control over).
To maximize its bond price an eternal company will try to minimize its risk of default, by
making a modest profit (if a company is losing money it will eventually default), and
growing sustainably (bigger companies are able to diversify their investments making them less likely to default).
The probability a company will default (D) can be decomposed into three factors.
Idiosyncratic risk (I): The risk that the company defaults due to factors particular to the company.
Systemic Risk (S): The risk that the company defaults due to factors particular to the industry or the market as a whole.
Existential Risk (E): The risk that the company defaults because all of humanity goes extinct.
P(D)=P(I)×P(S)×P(E)
A small eternal company will have a very large idiosyncratic risk and will not have much control over systemic and existential risk. A large eternal company may be able to diversify away its idiosyncratic risk and be able chip away at systemic risk (e.g. through regulatory capture). A gigantic eternal company could hold such sway that it may be able to reduce existential risk.
I’m not an expert in corporate bonds. But I think there are some empirical questions that can be asked to test the above theory.
Do “existential scares” such as the possibility of nuclear warfare and climate change make interest rates go up?
Do companies that default on their bonds soon go into bankruptcy?
Do companies that have expensive bonds last longer than companies with cheap bonds?
Do companies that issue a lot of bonds behave more like eternal companies than profit-maximizing companies?
Solve the principal-agent problem with pensions or futarchy
All companies face the principal-agent problem. The investor (principal) wants the company to maximize its profits (or in the case of an eternal company, minimize its risk of default), however the executives (agent) want to maximize their own compensation.
How do we ensure the executives of an eternal company actually try to maximize the company’s bond price and not just award themselves a cushy salary?
I’ve heard people complain that companies are too short-sighted, just trying to maximize their quarterly earnings instead of their expected earnings. Apparently part of the problem is that the executives bonuses are tied to the results of these quarterly earnings.
To align the incentives of the executive and the bondholders, executives of an eternal company will not get annual bonuses, instead they will get generous pensions. Of course, the pensions will be in the form of the eternal company’s perpetual bonds! If company fails the executives lose their pension! Clearly the executives will have the incentive to minimize the risk of default.
An alternative is to make decisions via prediction markets i.e. Futarchy, but that’s orthogonal this proposal.
Eternal companies have a lot of slack
A company trying hard not to default would want to be in a low-risk industry with regular cash flows. Some examples I can think of are
a bank,
a pension fund,
an insurance company, or
a fund that just holds T-bills (do these exist?).
Eternal companies will likely have these policies:
Lots of slack: A well-rested staff is less likely to make silly mistakes, and have time to evaluate non-obvious risks before making decisions.
Low employee turnover: There is a risk that information is lost when the outgoing employee hands over to the incoming employee. This can lead to mistakes.
Cautious hiring practices: A poor hire making poor decisions can put the company at risk.
A generous pension instead of annual bonuses: Explained in the previous section.
Brand-name products: When a company cuts costs and creates an inferior product, they can risk losing customers. An eternal company will prefer to create high-quality products to attract a loyal customer base.
This is not a great idea
While this idea seems to work in theory, it essentially assumes that existential risk is priced into interest rates, but I don’t know if this is the case. Additionally, eternal companies are likely to grow very slowly. It will probably take, say, 100 years for the company to get big enough to care about existential risk more than its idiosyncratic or systemic risk. If you think AI Doom will happen in the next few decades then it’s probably too late for this idea to work.
But, in 500 million years, when our paperclip maximizer encounters an alien civilization that did solve the alignment problem, instead of it saying “The civilization that gave birth to me had no chance.” It would be more dignified if it could say, “The civilization that gave birth to me had fledgling eternal company that would have tried to solve the alignment problem, but they were a few decades too late.”
This is not my best idea and I don’t have any experience in the financial sector (the most likely sector for an eternal company) so I’m not going to start my own eternal company. I just thought I should put this idea out there.
An Eternal Company
TL;DR A company that maximizes its bond price instead of its stock price cares about the long-term future and is incentivized to reduce existential risk.
The world that solves the alignment problem has an institution with the right incentives
There is a world in which we solve the alignment problem. We are not in that world.
The world in which we solve the alignment problem has institutions with incentives to solve alignment-like problems. The same way Walmart has incentives to sell you groceries for a reasonable price. In countries with functioning market economies, nobody thinks about “the grocery problem”.
One such institution is one that has incentives to care about the long-term future. I believe that there is an easy way to create such an institution in our current world.
Eternal companies maximize their bond price instead of their stock price
The first insight is that to bet on the end of the world is to take a loan that you expect not to pay back. Hence to bet against the end of the world is to buy a bond that pays installments far into the future. An example of such is a perpetual bond.
When a company sells a perpetual bond they promise to pay regular installments on the bond until the company fails. (If this concept seems alien to you there is a great video by Tom Scott on the concept)
Normal companies try to maximize their risk-adjusted returns (i.e. their share price), but what if we had a company that maximized the price of its perpetual bonds? We can call this type of company an eternal company.
The price of perpetual bonds are determined by interest rates (which the company’s executive has no control over) and the company’s risk of default (which the company’s executive does have control over).
To maximize its bond price an eternal company will try to minimize its risk of default, by
making a modest profit (if a company is losing money it will eventually default), and
growing sustainably (bigger companies are able to diversify their investments making them less likely to default).
The probability a company will default (D) can be decomposed into three factors.
Idiosyncratic risk (I): The risk that the company defaults due to factors particular to the company.
Systemic Risk (S): The risk that the company defaults due to factors particular to the industry or the market as a whole.
Existential Risk (E): The risk that the company defaults because all of humanity goes extinct.
P(D)=P(I)×P(S)×P(E)
A small eternal company will have a very large idiosyncratic risk and will not have much control over systemic and existential risk. A large eternal company may be able to diversify away its idiosyncratic risk and be able chip away at systemic risk (e.g. through regulatory capture). A gigantic eternal company could hold such sway that it may be able to reduce existential risk.
Will eternal companies work in practice?
In theory, a loan is a bet that society will collapse in the future, but in practice, markets don’t seem to take existential risk seriously. For example the Metaculus community prediction on the chance that humans will go extinct by 2100 is only 2%.
I’m not an expert in corporate bonds. But I think there are some empirical questions that can be asked to test the above theory.
Do “existential scares” such as the possibility of nuclear warfare and climate change make interest rates go up?
Do companies that default on their bonds soon go into bankruptcy?
Do companies that have expensive bonds last longer than companies with cheap bonds?
Do companies that issue a lot of bonds behave more like eternal companies than profit-maximizing companies?
Solve the principal-agent problem with pensions or futarchy
All companies face the principal-agent problem. The investor (principal) wants the company to maximize its profits (or in the case of an eternal company, minimize its risk of default), however the executives (agent) want to maximize their own compensation.
How do we ensure the executives of an eternal company actually try to maximize the company’s bond price and not just award themselves a cushy salary?
I’ve heard people complain that companies are too short-sighted, just trying to maximize their quarterly earnings instead of their expected earnings. Apparently part of the problem is that the executives bonuses are tied to the results of these quarterly earnings.
To align the incentives of the executive and the bondholders, executives of an eternal company will not get annual bonuses, instead they will get generous pensions. Of course, the pensions will be in the form of the eternal company’s perpetual bonds! If company fails the executives lose their pension! Clearly the executives will have the incentive to minimize the risk of default.
An alternative is to make decisions via prediction markets i.e. Futarchy, but that’s orthogonal this proposal.
Eternal companies have a lot of slack
A company trying hard not to default would want to be in a low-risk industry with regular cash flows. Some examples I can think of are
a bank,
a pension fund,
an insurance company, or
a fund that just holds T-bills (do these exist?).
Eternal companies will likely have these policies:
Lots of slack: A well-rested staff is less likely to make silly mistakes, and have time to evaluate non-obvious risks before making decisions.
Low employee turnover: There is a risk that information is lost when the outgoing employee hands over to the incoming employee. This can lead to mistakes.
Cautious hiring practices: A poor hire making poor decisions can put the company at risk.
A generous pension instead of annual bonuses: Explained in the previous section.
Brand-name products: When a company cuts costs and creates an inferior product, they can risk losing customers. An eternal company will prefer to create high-quality products to attract a loyal customer base.
This is not a great idea
While this idea seems to work in theory, it essentially assumes that existential risk is priced into interest rates, but I don’t know if this is the case. Additionally, eternal companies are likely to grow very slowly. It will probably take, say, 100 years for the company to get big enough to care about existential risk more than its idiosyncratic or systemic risk. If you think AI Doom will happen in the next few decades then it’s probably too late for this idea to work.
But, in 500 million years, when our paperclip maximizer encounters an alien civilization that did solve the alignment problem, instead of it saying “The civilization that gave birth to me had no chance.” It would be more dignified if it could say, “The civilization that gave birth to me had fledgling eternal company that would have tried to solve the alignment problem, but they were a few decades too late.”
This is not my best idea and I don’t have any experience in the financial sector (the most likely sector for an eternal company) so I’m not going to start my own eternal company. I just thought I should put this idea out there.