I think this is the best intro to investing for altruists that I've seen published. The investment concepts it covers are the most important ones, and the application to altruists seems right.
(For context: I used to work as a trader, which is somewhat but not very relevant, and have thought about this kind of thing a bit.)
Some tentative thoughts on real estate (I haven't thought about this much):
The real estate market is presumably less efficient than the stock market, so it's easier for careful individuals to make market-beating investments. (It also makes it easier to make investments that underperform the market, but I'd probably expect EAs to be better at this than the average home buyer, though I'm not entirely sure.)
My guess would be that the real estate market is overall less investable than global stock markets (REITs only cover a limited fraction of the overall real estate market), such that more altruists owning houses would lead to a better approximation of a global market portfolio.
If these points were true, that would mean that the introductory example of Carol renting out a home could actually be a good idea (if Carol is an altruist).
Even if you can beat the market by buying a basket of houses (which I'm not sure is true), buying a single house has probably 2-3x the risk of the broad real estate market and 3-4x the risk of the global market portfolio, assuming real estate works similarly to stocks (which is probably a reasonable assumption). So it still seems like a bad idea, for the reasons discussed in the essay.
It might make sense if a bunch of individual EAs buy real estate such that the overall portfolio is well-diversified. I don't expect this to happen in practice, because EAs are geographically concentrated in a small number of cities, so if people own investment properties in the cities where they live, the overall EA real estate portfolio will be too concentrated in those cities.
It might make sense if a bunch of individual EAs buy real estate such that the overall portfolio is well-diversified.
Yeah, this is what I was trying to say. Perhaps I was unclear.
I don't expect this to happen in practice, because EAs are geographically concentrated in a small number of cities, so if people own investment properties in the cities where they live, the overall EA real estate portfolio will be too concentrated in those cities.
EAs can diversify the overall EA real estate portfolio by thinking about where other EAs are likely to buy houses. E.g., they should avoid buying a house if they moved to an EA hub city, but they should buy (or avoid selling) a house in their hometown, especially if they come from a place that doesn't have a lot of EAs.
In addition, buying houses in EA hub cities might be somewhat of a hedge against a change in living costs in those key locations, such that overweighting these could actually be more beneficial than harmful.
Anyway, all of this is a bit of a nitpick, I generally agree with the overall direction.
This is a really excellent piece of work on bringing these concepts to a broader audience. I'm quite interested in long-term investment modelling so I'd like to offer my thoughts. Of course, the below isn't advice, so please don't make investment decisions purely on my comments below.
It's great that you are thinking about how to adjust standard investing concepts based on the notion that it is the total altruistic portfolio that matters, which is formed in a decentralised way. I agree this adds to the rationale for being "overweight" the company that the investor founded, or investing in individual properties. This is not how a typical investor thinks, so there is likely scope to think further along these lines. Either to improve coordination between EA investors, or to better implement a decentralised solution by departing from standard investment concepts.
I think your idea extends to alternative investments. Common wisdom in institutional investment is that it requires greater governance capabilities to invest in the more diversifying assets, such as infrastructure, some hedge funds, unlisted (commercial or residential) property, and private equity. That is, they require greater expertise, more time spent on investment processes, necessitate more careful cashflow management due to illiquidity, and potentially other challenges. And that greater governance capabilities are rewarded - see https://link.springer.com/article/10.1057/jam.2008.1. If an EA investor cares only about the overall altruistic portfolio and is capable of making/managing such investments, then it might make sense to overweight them. Some of them might be accessible through pooled funds.
In the article you rely on the standard deviation of annual returns as a measure of risk. But long term risk isn't well captured by that. Taking a step back, risk should ultimately be defined based on altruists' utility function over spending at different points in time. For example, there might be "hinge" moments when altruistic spending is especially effective. Imagine there is going to be a massive opportunity in 100 years to influence the creation of AGI by altruistic spending. In that case, we don't really care if the annual standard deviation of returns is high. We care only about the probability distribution of the 100 year return.
There is a limit to the ability of leverage to magnify returns. This is partly because of the asymmetry of returns. For example, if you start with $100, then experience -50% return then +50% return, you end up with $75. Assuming you readjust your borrowing amount regularly alongside changes in the asset value, this effect is magnified by leverage and detracts from the overall return. See https://holygrailtradingstrategies.com/images/Leveraged-ETFs.pdf for more.
Leverage has a strong role in the Capital Asset Pricing Model theory you're using. The theory does however assume away various challenges to do with leverage, like the one above. In general, it is uncommon for institutional investors (pension funds, university endowments, charitable foundations, etc) to directly borrow to invest. However, they may outsource it to a money manager, e.g. a hedge fund, who can access a decent borrowing rate on their behalf and who has the expertise to manage it. I'm not saying that leverage should never be used by EA investors. Rather, I would be quite careful before deciding to use it.
When actuaries model (commercial) real estate, it's normally assumed that both its risk and expected return are somewhere in between those of shares and bonds. Arguably, real estate has characteristics of each, as it is an asset used for productive enterprise, and since leases typically provide regular fixed rental payments. Nevertheless, I would look to property indices' historical data for guidance.
Certainty equivalence may not be the right concept for measuring the value of moving all EA investments to a global market portfolio. I would instead compare the sharpe ratios. If you want to put an expected dollar figure on it, one way would be to calculate the increase in expected return you could achieve while holding risk constant. This avoids needing to make an assumption about investor risk preferences, which the certainty equivalent concept relies on.
I haven't read all your footnotes so perhaps some of the above is mentioned there. Nevertheless, I hope my comments are helpful and I am glad people in EA is actively thinking about this. Happy to chat more if you are interested.
I have no comment on whether it's a good idea to build the global market portfolio with leveraged ETFs, but since you asked:
You can use the etf.com screener to find ETFs matching your criteria. I just searched on there and based on the 10 minutes I spent looking, I think this is about the closest you can get:
This is still not really the global market portfolio, but it's at least kind of close. Also a couple of these ETFs are really small, so they'll have high trading costs.
If you're aiming for only, say, 1.5x leverage, you can buy some of the above (starting with the ones with the lowest fees) and fill the gaps with correspondingly larger amounts of other asset classes.
What are your thoughts on using max drawdown instead of volatility, or the Sortino ratio instead of Sharpe? Personally I'm more partial to both of them, maybe in part because it makes planning for the future easier, but maybe it's also giving into the endowment effect?
Portfolio Visualizer allows you to minimize max (historical) drawdown for a given target return.
IMO the ulcer index is the best measure of volatility that matches what people intuitively care about. It essentially measures the frequency and severity of drawdowns (the linked page explains it in more detail).
I didn't discuss the ulcer index in this post because in theory, investors with isoelastic utility should care about standard deviation, not drawdowns, and I lean toward the belief that people's focus on drawdowns is somewhat irrational (although probably somewhat justified by the fact that most asset returns are left-skewed). But broadly speaking, if you use the ulcer index as your measure of risk, concentrating in a small number of assets looks even worse than if you use standard deviation, so the case for diversification is even stronger.
My thinking is that donating during drawdowns might be particularly bad, both personally and for your longer term donation strategy, since you're selling low and "locking in" large losses in your portfolio. So minimizing drawdown allows you to better plan your budget and donations, and allows you more flexibility in timing your donations. You might find a particularly good donation opportunity during a drawdown period that will only be available during that period, but it'll be extra costly (personally and to future donations) to donate then, so avoiding such drawdowns seems like an especially good thing to do.
Also, Sharpe penalizes extreme upside compared to Sortino, which seems weird to me.
Is it actually the Sharpe ratio that should be maximized with isoelastic utility (assuming log-normal returns, was it?)?
But broadly speaking, if you use the ulcer index as your measure of risk, concentrating in a small number of assets looks even worse than if you use standard deviation, so the case for diversification is even stronger.
My thinking is that donating during drawdowns might be particularly bad
This is true, and the standard deviation fully captures the extent to which drawdowns are bad (assuming isoelastic utility and log-normal returns). Increasing the standard deviation is bad because doing so increases the probability of both very good and very bad outcomes, and bad outcomes are more bad than good outcomes are good.
Is it actually the Sharpe ratio that should be maximized with isoelastic utility (assuming log-normal returns, was it?)?
Yes, if you also assume that you can freely use leverage. The portfolio with the maximum Sharpe ratio allows for the highest expected return at a given standard deviation, or the lowest standard deviation at a given expected return.
A question I often have about the index-fund approach: if we were in a past historical moment, wouldn't VC-style investment on moonshots perform better than index investing?
e.g. if we were in 1910, wouldn't it be better to invest in a portfolio of moonshots that included Ford and Edison, rather than in a broad market index? (Most of the moonshots would fail but Ford and/or Edison would bring in massively outsized returns)
The answer sort of depends on what you mean by moonshot, but under one reasonable definition, it's actually the opposite: investing in potential moonshots would have resulted in worse performance than an index fund. Or to put it another way, boring companies tend to outperform exciting companies.
You can divide stocks into two types: growth stocks and value stocks. Value stocks are cheaply priced relative to their fundamentals (e.g., they have a low price to earnings or price to sales ratio) because the market expects these companies to be "boring" and not show good earnings growth. Growth stocks are priced expensively because the market expects them to grow. This sounds basically like what you're talking about with "moonshot" companies. If you wanted to systematically invest in moonshots, you could maybe buy the 10% most expensive stocks, because these are the ones the market believes have the most upside potential. But if you did that historically, you would've underperformed the market by a lot—something on the order of 5 percentage points per year. The seminal paper on this is Fama & French (1992), The Cross-Section of Expected Stock Returns.
In theory, savvy investors could identify the most promising publicly-traded growth companies and outperform the market by buying them. But studies on fund managers have found that pretty much nobody can do this.
I'm curious to hear Michael's response, but also interested to hear more about why you think this. I have the opposite intuition- presumably 1910 had its fair share of moonshots which seemed crazy at the time and which turned out, in fact, to be basically crazy, which is why we haven't heard about them.
A portfolio which included Ford and Edison would have performed extremely well, but I don't know how many possible 1910 moonshot portfolios would have included them or would have weighted them significantly enough to outperform the many failed other moonshots.
Right, so if you had good judgment and a reasonable network, I think you could find a lot of the promising opportunities.
Someone would be like "I have a device that produces bright light at a low cost, and I need funding to manufacture it on a massive scale."
So you'd be like "Really? Let me see this device."
You'd inspect the device, see that it does in fact do the thing the inventor says it does, and then after further diligence on the business plan & team you'd invest.
I think this is roughly how to go about separating the real opportunities from the snake oil.
Worth remembering that, especially today, there are hundreds of thousands to millions of other highly intelligent and resourceful people trying to do the exact same thing. So you need to have a reason to believe you can do a better job than they can.
Philippe LeBon was a French civil engineer working in the public engineering corps who became interested while at university in distillation as an industrial process for the manufacturing of materials such as tar and oil. He graduated from the engineering school in 1789, and was assigned to Angoulême. There, he investigated distillation, and became aware that the gas produced in the distillation of wood and coal could be useful for lighting, heating, and as an energy source in engines. He took out a patent for distillation processes in 1794, and continued his research, eventually designing a distillation oven known as the thermolamp. He applied for and received a patent for this invention in 1799, with an addition in 1801. He launched a marketing campaign in Paris in 1801 by printing a pamphlet and renting a house where he put on public demonstrations with his apparatus. His goal was to raise sufficient funds from investors to launch a company, but he failed to attract this sort of interest, either from the French state or from private sources. He was forced to abandon the project and return to the civil engineering corps. Although he was given a forest concession by the French government to experiment with the manufacture of tar from wood for naval use, he never succeed with the thermolamp, and died in uncertain circumstances in 1805.
I think this is the best intro to investing for altruists that I've seen published. The investment concepts it covers are the most important ones, and the application to altruists seems right.
(For context: I used to work as a trader, which is somewhat but not very relevant, and have thought about this kind of thing a bit.)
Thank you, I appreciate the positive feedback, especially from someone as knowledgeable as you!
Some tentative thoughts on real estate (I haven't thought about this much):
If these points were true, that would mean that the introductory example of Carol renting out a home could actually be a good idea (if Carol is an altruist).
Even if you can beat the market by buying a basket of houses (which I'm not sure is true), buying a single house has probably 2-3x the risk of the broad real estate market and 3-4x the risk of the global market portfolio, assuming real estate works similarly to stocks (which is probably a reasonable assumption). So it still seems like a bad idea, for the reasons discussed in the essay.
It might make sense if a bunch of individual EAs buy real estate such that the overall portfolio is well-diversified. I don't expect this to happen in practice, because EAs are geographically concentrated in a small number of cities, so if people own investment properties in the cities where they live, the overall EA real estate portfolio will be too concentrated in those cities.
Yeah, this is what I was trying to say. Perhaps I was unclear.
EAs can diversify the overall EA real estate portfolio by thinking about where other EAs are likely to buy houses. E.g., they should avoid buying a house if they moved to an EA hub city, but they should buy (or avoid selling) a house in their hometown, especially if they come from a place that doesn't have a lot of EAs.
In addition, buying houses in EA hub cities might be somewhat of a hedge against a change in living costs in those key locations, such that overweighting these could actually be more beneficial than harmful.
Anyway, all of this is a bit of a nitpick, I generally agree with the overall direction.
This is a really excellent piece of work on bringing these concepts to a broader audience. I'm quite interested in long-term investment modelling so I'd like to offer my thoughts. Of course, the below isn't advice, so please don't make investment decisions purely on my comments below.
It's great that you are thinking about how to adjust standard investing concepts based on the notion that it is the total altruistic portfolio that matters, which is formed in a decentralised way. I agree this adds to the rationale for being "overweight" the company that the investor founded, or investing in individual properties. This is not how a typical investor thinks, so there is likely scope to think further along these lines. Either to improve coordination between EA investors, or to better implement a decentralised solution by departing from standard investment concepts.
I think your idea extends to alternative investments. Common wisdom in institutional investment is that it requires greater governance capabilities to invest in the more diversifying assets, such as infrastructure, some hedge funds, unlisted (commercial or residential) property, and private equity. That is, they require greater expertise, more time spent on investment processes, necessitate more careful cashflow management due to illiquidity, and potentially other challenges. And that greater governance capabilities are rewarded - see https://link.springer.com/article/10.1057/jam.2008.1. If an EA investor cares only about the overall altruistic portfolio and is capable of making/managing such investments, then it might make sense to overweight them. Some of them might be accessible through pooled funds.
In the article you rely on the standard deviation of annual returns as a measure of risk. But long term risk isn't well captured by that. Taking a step back, risk should ultimately be defined based on altruists' utility function over spending at different points in time. For example, there might be "hinge" moments when altruistic spending is especially effective. Imagine there is going to be a massive opportunity in 100 years to influence the creation of AGI by altruistic spending. In that case, we don't really care if the annual standard deviation of returns is high. We care only about the probability distribution of the 100 year return.
There is a limit to the ability of leverage to magnify returns. This is partly because of the asymmetry of returns. For example, if you start with $100, then experience -50% return then +50% return, you end up with $75. Assuming you readjust your borrowing amount regularly alongside changes in the asset value, this effect is magnified by leverage and detracts from the overall return. See https://holygrailtradingstrategies.com/images/Leveraged-ETFs.pdf for more.
Leverage has a strong role in the Capital Asset Pricing Model theory you're using. The theory does however assume away various challenges to do with leverage, like the one above. In general, it is uncommon for institutional investors (pension funds, university endowments, charitable foundations, etc) to directly borrow to invest. However, they may outsource it to a money manager, e.g. a hedge fund, who can access a decent borrowing rate on their behalf and who has the expertise to manage it. I'm not saying that leverage should never be used by EA investors. Rather, I would be quite careful before deciding to use it.
When actuaries model (commercial) real estate, it's normally assumed that both its risk and expected return are somewhere in between those of shares and bonds. Arguably, real estate has characteristics of each, as it is an asset used for productive enterprise, and since leases typically provide regular fixed rental payments. Nevertheless, I would look to property indices' historical data for guidance.
Certainty equivalence may not be the right concept for measuring the value of moving all EA investments to a global market portfolio. I would instead compare the sharpe ratios. If you want to put an expected dollar figure on it, one way would be to calculate the increase in expected return you could achieve while holding risk constant. This avoids needing to make an assumption about investor risk preferences, which the certainty equivalent concept relies on.
I haven't read all your footnotes so perhaps some of the above is mentioned there. Nevertheless, I hope my comments are helpful and I am glad people in EA is actively thinking about this. Happy to chat more if you are interested.
What's the closest we can get to the global market portfolio with leveraged ETFs?
I have no comment on whether it's a good idea to build the global market portfolio with leveraged ETFs, but since you asked:
You can use the etf.com screener to find ETFs matching your criteria. I just searched on there and based on the 10 minutes I spent looking, I think this is about the closest you can get:
This is still not really the global market portfolio, but it's at least kind of close. Also a couple of these ETFs are really small, so they'll have high trading costs.
If you're aiming for only, say, 1.5x leverage, you can buy some of the above (starting with the ones with the lowest fees) and fill the gaps with correspondingly larger amounts of other asset classes.
What are your thoughts on using max drawdown instead of volatility, or the Sortino ratio instead of Sharpe? Personally I'm more partial to both of them, maybe in part because it makes planning for the future easier, but maybe it's also giving into the endowment effect?
Portfolio Visualizer allows you to minimize max (historical) drawdown for a given target return.
IMO the ulcer index is the best measure of volatility that matches what people intuitively care about. It essentially measures the frequency and severity of drawdowns (the linked page explains it in more detail).
I didn't discuss the ulcer index in this post because in theory, investors with isoelastic utility should care about standard deviation, not drawdowns, and I lean toward the belief that people's focus on drawdowns is somewhat irrational (although probably somewhat justified by the fact that most asset returns are left-skewed). But broadly speaking, if you use the ulcer index as your measure of risk, concentrating in a small number of assets looks even worse than if you use standard deviation, so the case for diversification is even stronger.
My thinking is that donating during drawdowns might be particularly bad, both personally and for your longer term donation strategy, since you're selling low and "locking in" large losses in your portfolio. So minimizing drawdown allows you to better plan your budget and donations, and allows you more flexibility in timing your donations. You might find a particularly good donation opportunity during a drawdown period that will only be available during that period, but it'll be extra costly (personally and to future donations) to donate then, so avoiding such drawdowns seems like an especially good thing to do.
Also, Sharpe penalizes extreme upside compared to Sortino, which seems weird to me.
Is it actually the Sharpe ratio that should be maximized with isoelastic utility (assuming log-normal returns, was it?)?
Makes sense.
This is true, and the standard deviation fully captures the extent to which drawdowns are bad (assuming isoelastic utility and log-normal returns). Increasing the standard deviation is bad because doing so increases the probability of both very good and very bad outcomes, and bad outcomes are more bad than good outcomes are good.
Yes, if you also assume that you can freely use leverage. The portfolio with the maximum Sharpe ratio allows for the highest expected return at a given standard deviation, or the lowest standard deviation at a given expected return.
Thanks for this! Super interesting.
A question I often have about the index-fund approach: if we were in a past historical moment, wouldn't VC-style investment on moonshots perform better than index investing?
e.g. if we were in 1910, wouldn't it be better to invest in a portfolio of moonshots that included Ford and Edison, rather than in a broad market index? (Most of the moonshots would fail but Ford and/or Edison would bring in massively outsized returns)
The answer sort of depends on what you mean by moonshot, but under one reasonable definition, it's actually the opposite: investing in potential moonshots would have resulted in worse performance than an index fund. Or to put it another way, boring companies tend to outperform exciting companies.
You can divide stocks into two types: growth stocks and value stocks. Value stocks are cheaply priced relative to their fundamentals (e.g., they have a low price to earnings or price to sales ratio) because the market expects these companies to be "boring" and not show good earnings growth. Growth stocks are priced expensively because the market expects them to grow. This sounds basically like what you're talking about with "moonshot" companies. If you wanted to systematically invest in moonshots, you could maybe buy the 10% most expensive stocks, because these are the ones the market believes have the most upside potential. But if you did that historically, you would've underperformed the market by a lot—something on the order of 5 percentage points per year. The seminal paper on this is Fama & French (1992), The Cross-Section of Expected Stock Returns.
In theory, savvy investors could identify the most promising publicly-traded growth companies and outperform the market by buying them. But studies on fund managers have found that pretty much nobody can do this.
I'm curious to hear Michael's response, but also interested to hear more about why you think this. I have the opposite intuition- presumably 1910 had its fair share of moonshots which seemed crazy at the time and which turned out, in fact, to be basically crazy, which is why we haven't heard about them.
A portfolio which included Ford and Edison would have performed extremely well, but I don't know how many possible 1910 moonshot portfolios would have included them or would have weighted them significantly enough to outperform the many failed other moonshots.
Right, so if you had good judgment and a reasonable network, I think you could find a lot of the promising opportunities.
Someone would be like "I have a device that produces bright light at a low cost, and I need funding to manufacture it on a massive scale."
So you'd be like "Really? Let me see this device."
You'd inspect the device, see that it does in fact do the thing the inventor says it does, and then after further diligence on the business plan & team you'd invest.
I think this is roughly how to go about separating the real opportunities from the snake oil.
Worth remembering that, especially today, there are hundreds of thousands to millions of other highly intelligent and resourceful people trying to do the exact same thing. So you need to have a reason to believe you can do a better job than they can.
Yes, I think the crux here is how good you think your judgment/taste is relative to that of all the others who are trying.
Historical anecdote: