Mission-correlated investing: Examples of mission hedging and 'leveraging'

Suppose that you have a pot of money that you are 'investing to give' at a particular time horizon^[1]. The base case is just to invest in a diversified 'market' portfolio and give to the most cost-effective option at the horizon. 

This post offers concrete examples of how you can increase the expected outcome by making investments that are correlated with cost-effectiveness at the horizon. That is, 'mission-correlated' investing.

Note that Hauke's original post on mission hedging contains a much more extensive collection of examples. The examples in this post are focused on differentiating between different kinds of mission-correlated investing.

Thank you to Michael Plant and Wayne Chang for encouraging this post.

Background on 'mission-correlated' investments

Mission correlation

As described in this post, an investment has positive mission correlation if its financial returns are expected to be relatively high in worlds where your impact per future dollar is relatively high^[2].

To maximize the expected amount of good you will do with your 'investing to give' money you should overweight such investments. This is because the amount of good you will do is the product of the value of your pot of money and your impact per dollar at the horizon. Overweighting mission-correlated investments increases the correlation between these two variables and thus increases the expected value. 

Similarly, you should underweight investments that have a negative mission correlation.

Mission correlation is independent of the expected returns on any investment. So, for the examples we can assume that all investments are fairly priced with 'market rate' expected returns.

Hedging versus leveraging

I'll use the term 'world state' to refer to the total amount of good in a given world independent of your actions. In principle there could be many world states, but in the examples below I simply split the possible worlds into a good state (with a high amount of good) and a bad state (with a relatively low amount of good).

For some cause areas we might expect impact per future dollar and the world state to have a positive correlation. For others it may be negative. The examples below include both cases.

As detailed in the post linked to above, if the correlation between the world state and impact per dollar is negative then mission-correlated investing will generally result in mission 'hedging'. This is because it will make the bad state 'a lot better' and the good state 'a little worse'. This reduces the volatility of the total outcome at the horizon. 

If the correlation between the world state and impact per dollar is positive then mission-correlated investing will generally increase the volatility of the outcome. We might call this mission 'leveraging' because it increases the expected value and increase the volatility.

Good versus evil

We can refer to making investments with returns that are positively correlated with the world state as 'investing in good'. Making investments with returns with a negative correlation with the world state may then be looked at as 'investing in evil'. 

The word 'evil' is just used here for simplicity and because of precedent. It's not intended to be applied pejoratively to any real investment^[3].

For simplicity we'll assume that whatever factor drives impact per dollar is also the dominant driver of the world state. Then the sign (positive or negative) of the correlation of an investment's returns & the world state will be equal to the sign of the product of the correlations of 'impact per dollar & returns' and 'impact per dollar & the world state'. 

For example, a negative correlation between impact per dollar & returns and a negative correlation between impact per dollar & the world state implies a positive correlation between returns & the world state. So this example investment tends to do well as things get better, making it 'good'.

I use this in the table below to label different cases as 'good' or 'evil'.

Summary of the types of mission-correlated investing

The above considerations are summarized in the following table. The sign of the correlation of returns & impact per dollar determines whether or not you should underweight or overweight an investment. The correlation of the world state & impact per dollar determines whether an investment will be hedging or leveraging. And the product of these correlations determines if we label an investment as 'good' or 'evil'.

Examples

To make the above concrete, I've put together examples for each cell in the table. 

I've used animal welfare as a consistent setting for each case. To make it clear that mission correlation can apply to other cause areas, I've also offered one additional example for each case.

Please understand that these examples are:

• Only directional. I intend to write a future post on the magnitude of mission correlation effects. I think it really depends on the details and the cause area.  It is plausible that some altruists should allocate low double digit percentages of their portfolio to mission correlated investments.
• Only illustrative and hypothetical. As I try to illustrate, in many cases it's possible to come up with plausible arguments for a correlation being either positive or negative. What I've presented here are only intuitive ideas for conclusions that would require significantly deeper investigations to validate.

As a general setup for the examples, we need to define the good/bad world states and associated return expectations. 

For the main example of animal welfare, we can make the good world state one where meat consumption has dropped and consumption of meat alternatives is relatively high (relative to current forecasts). The bad world state is where meat consumption is relatively high and alternatives relatively low.

Also, suppose in all examples that you expect relatively high investment returns for traditional meat producers in the bad state (when more meat is being consumed) and relatively low returns in the good state. That is, returns that are negatively correlated with the world state. You expect the opposite for alt-protein companies - returns that are positively correlated with the world state.

Mission hedging

Now, suppose you expect impact per dollar to be negatively correlated with the world state. For example, if alternatives are culturally accepted and cheaply available (in the good state) then you might expect that finding and convincing the remaining meat eaters to shift will be harder and more expensive.

In this case, directionally, you should do mission hedging.

'Invest in evil to do more good'

The above assumptions imply that meat companies returns will be positively correlated with impact per dollar. So you should overweight these companies as a mission-correlated investment. 

If the bad world occurs you will have more money and higher impact per dollar and be able to do a relatively large amount of good. Hence, it makes the 'bad world a lot better'. 

In the good world you will have less money and do a relatively small amount of good. But because impact per dollar is low in this world, even with an average return you would only have done a limited amount of good.  So mission hedging makes the 'good world only a little worse'. 

Additional example 1

Another example in this category would be investing in AI-related companies as shared by Jonas Vollmer here. 

These companies are 'evil' in the sense that their returns may be relatively high under the 'bad' world state of short-AI timelines with increased risks due to transformative AI. If you also expect higher impact per dollar under short timelines then you will want to overweight these companies to 'mission hedge'.

'Divest from good to do more good'

Alt-protein investments are the opposite of the meat companies in this example. So you should underweight these companies.

Whether you underweight alt-protein companies or overweight meat companies, or both, would depend on further analyses of the options. For example, it could be that most meat companies are diversified food companies that have less correlation with the world state than growth stage alt-protein companies. If so, underweighting alt-protein companies would be relatively valuable.

Additional example 2

Consider emerging markets investments for a global health donor. Emerging market economies doing well is plausibly positively correlated with returns in these markets. Impact per dollar seems likely to go down as these economies improve. So it seems reasonable to expect that returns are negatively correlated with impact per dollar^[4]. Thus, such a donor should underweight emerging markets in their portfolio.

Mission leveraging

Now suppose that you instead expect impact per dollar to be positively correlated with the world state. For example, the good state of the world could be seen as evidence that the alt-protein business model works in Western markets. This could increase your estimates of the impact per dollar of efforts to promote alt-proteins in other markets.

In this case, directionally, you should do mission leveraging.

'Divest from evil to do more good'

In this case, you should underweight meat companies. Thanks to the SRI/ESG movement and associated data and products, this should be a relatively easy position to implement. 

Additional example 3

Arguably putting money into a patient philanthropy fund is a bet on higher impact per dollar under long timelines. While in principle patient funds could be disbursed under short timelines, they will really shine if the optimal disbursement time is far in the future so the fund can grow over a long period of time. Given this, such funds may want to do the opposite of 'Extra example 1' and underweight AI-related companies.

'Invest in good to do more good'

In this case, you should overweight alt-protein companies.

The 'leverage' of this case can be seen as follows. In worlds where alt-protein companies do relatively well and meat consumption is relatively low, then you will get to give a lot of money at a relatively high impact per dollar, doing a lot of good. But, on the downside, you are liable to have lost a lot of money in worlds where alt-protein firms are less successful and meat consumption is high. The upside is better and the downside worse than if you didn't do mission-correlated investing.

To make such a leveraged bet requires comfort with trading off a possibly frustrating downside scenario with being able to do a relatively large amount of good in the upside scenario.

Additional example 4

Suppose the good world state is one where app-based mental health interventions are proven to work relatively well (that is, with even stronger and more robust evidence than we currently have). The corresponding bad world would be where mental health apps are viewed as relatively worse than currently hoped. Then we might expect that in the good world i) returns on mental health app investments will be relatively high, and ii) the impact per dollar of distributing such interventions to low-income households will be relatively high.

That is, we're supposing a positive correlation between returns & impact per dollar and a positive correlation between the world state & impact per dollar. So, the mission-correlated strategy would be for a mental health donor to 'invest to give' in mental health app companies.

Pros and cons

The general benefit of mission-correlated investing is to increase the expected amount of good done. The general difficulty with such strategies is being sure about the (forward-looking) correlations.

The above table and examples illustrate the other pros and cons of both kinds of mission-correlated investing.

With mission hedging you get the benefit of reducing the volatility of the total amount of good. But you are always investing in the opposite direction of what may be morally intuitive and comfortable ('invest in evil' and 'divest from good'). In the examples this means investing in meat companies and AI and divesting from alt-protein and emerging markets.

With mission leveraging you increase the volatility of outcomes which may feel uncomfortable. But you get to invest in a feel good way ('invest in good' and 'divest from evil'). In the examples this means divesting from meat companies and AI and investing in alt-protein and mental health. 

Thus, mission-correlated investing is only appropriate for those altruists who value increases in expected value and the other pros more than the discomfort from these cons. 

1. ^{^}

   Of course, most altruists plan to distribute their giving over time. Assuming that you just have a single, fixed time horizon keeps things relatively simple and focused on the key points of this post. 

2. ^{^}

   To be careful I refer to 'impact per dollar' here instead of cost-effectiveness. Cost-effectiveness is technically 'dollars per impact' though in conversation it often seems to be used either way around.

3. ^{^}

   I think this is only useful for talking about mission-correlated investing in the abstract. In practice I just focus on assessing the correlations and not putting emotive labels on things.

^{^}

A counter argument might say that high financial returns in emerging markets may be associated with an increase in inequality within emerging market countries. This could, unfortunately, increase opportunities for highly cost-effective interventions targeting the least well off.

Show all footnotes