Impact investing 2

Some analogies between fiscal and monetary policy, humanitarian aid and development finance, and charity and impact investing

1. Fiscal and monetary policy

We use both fiscal and monetary policy to improve domestic welfare:

1. Fiscal policy: We can increase tax on the rich people who earn >$100k/y, and then redistribute it to poor people who earn $10k/y as welfare payment (or earned income tax breaks). The short-term welfare gains are really impressive, because due to logarithmic utility of consumption, we have the very simple rule that a dollar is worth 1/k times as much if you are k times richer (and that doubling someone's income is worth the same amount no matter where they start). This is also very scalable- we can easily send every person a stimulus check and spend a lot of money that way. But, generally, the fiscal multiplier is small, as people can't use capital as effectively as bigger firms.^[1] Transfer payments multipliers are often smaller than government spending multipliers.^[2] People are often very risk averse and just squirrel the money away and not much happens.

The government can spend money on public goods like health or education with higher sROI, but this is more difficult: subsidizing education is also subsidizing childcare, which increases labor force participation rate. And subsidizing health might be really good, because suffering/pain differs fundamentally from other goods in economic models, where severe pain triggers near-complete price insensitivity and willingness to sacrifice enormous portions of wealth for relief- far exceeding what logarithmic utility models would predict. Or the government can subsidize R&D. The drawback is that it's less scalable and there are steeper diminishing returns. But some economists have argued that spending on health, education and R&D in rich countries is not very cost-effective any longer on the current margin, because we already spend so much and are hitting diminishing returns. For instance, only 20% of labor productivity growth in the US since 1988 has been driven by R&D spending

And so, both redistribution via transfer payments or spending on public goods have a large and robust short-term welfare effect but lower long-run growth as the money is taken away from the private sector, which is better at stimulating long-run growth. Trickle down economics has a bad reputation, but it works (e.g. the 2017 US tax cuts increased long-run GDP by 0.6%, which is pretty good). And growth causes a lot of welfare, it’s just that the effects are very delayed and by the time the trickle down effects help the poorest in society, they are no longer as poor as the poorest people now. So it can make sense to give people a lot of short-term relief now.

2. Monetary policy: An example is a central bank printing $1T, then loaning it out to private banks at a below-market interest rate (say 0% though it could be negative). Why is that good?

1. Inflation is a tax on money and encourages investment. While poorer people are affected more by inflation (e.g. as they spend a larger share of their money on gas and food), their wages also go up more and unemployment goes down, so the poorest, say 10%, usually gain more from having more jobs and money than everyone else loses from having a little less.
2. The fiscal multiplier of such a policy is high: banks are incentivized to recoup their investment and will loan to firms that succeed. This is very scalable and easy to do. You just need to lower interest rates and it will have a very large effect on the economy.

Why would the state make concessional loans and lose money? Why not loan at market rates? For this would have no additionality, the state would just be another lender. These concessional loans pay for the service that banks provide to the state. What is the service? Banks find firms that perform well, so that the state is not in the business of 'picking winners'.

This gets at a central tension between left wing and right wing economics: robust short-term welfare effects of redistribution (which the left favors) vs. larger, but slower long-run effects of growth via trickle down economics (but that only benefit richer future people). In the end it boils down to discount rates and the eta parameter in diminishing returns (is a dollar really only 10x more valuable to someone on $10k/y than $100k/y, or is it maybe more?).

2. Aid and development finance

Analogously, we use both aid and development finance to improve foreign welfare:

1. Aid: The prime example is things like humanitarian aid like disaster relief in cash or in kind (e.g. food aid, short-term reconstruction relief). We spend ~$26B/y on that. We also spend on public goods like social infrastructure and services ($65B/y) to develop the human resource potential and improve living conditions. Health aid like malaria nets is $13bn, but $6bn was spent on perhaps higher leverage technical assistance. Half of all $180B/y in aid is spent like this.^[3] But the fiscal multiplier is generally considered small.^[4]
2. Development Finance: An example of development finance is giving out loans through development banks. For instance, the World Bank's International Finance Corporation,^[5] loans to foreign private firms and public private partnerships. Development finance institutions give out ~$50B/y.^[6],^[7] Development banks also give loans to poor countries' governments, which offer concessional loans to poor countries.^[8] 

These concessional, below market rate loans have the benefit of creating more growth, but the drawback is that, unlike cash-transfers, you cannot target the poorest of the poor. These loans also create strategic alliances, and are probably better for global growth because it gets countries trading with richer countries. There is also a trade-off between financial returns and social impact. If you're getting market rate returns over the long-term, there's likely no 'additionality.'^[9] If there's additionality, returns will likely be below market rate. I think some development banks have not really understood this (e.g. from 2012-2018, one UK development bank had average returns 10.3% between 2012-2016, and likely no additionality).^[10]

3. Charity and Impact investing

Analogously, we use both charity and impact investing to improve welfare:

1. Charity: Take randomista interventions in global development. We can give to health charities via Givewell and save a life for ~$5k. Or we can give to GiveDirectly, which is even more scalable and saves a life for $300k^[11] and it has a 2.5x economic multiplier in the general equilibrium, but that is not that much.^[12] [13]
2. Impact investing: Stimulating economic growth in poor countries might be much more effective than funding randomista interventions.^[14] It's very difficult to find good impact investing opportunities. ESG investment on the stock market is unlikely to have big effects and the impact of ESG investing is often overstated and hyped ('you can have above market rate returns and do a lot of good and then you can reinvest the returns to create a perpetual motion like do gooding investment vehicle; in contrast if you make grants you just lose all your money immediately'. This sentiment neglects that the social returns of grants are higher).^[15],^[16] Corporations can sometimes do a lot of good: for instance, GlaxoSmithKline published a paper recently on 'Introducing new vaccines in low- and middle-income countries: challenges and approaches'- giving them concessional loans might be good. IPOs, too, can have counterfactual impact unlike investing in publicly traded firms that are priced correctly. For instance, Opti-Harvest helps farmers use sunlight to increase yield, improve water use, reduce labor, and mitigate environmental impacts, but they withdrew their IPO— perhaps a missed opportunity.

Concessionary investments are on a spectrum from slight financial sacrifice at one end, to a grant to a company at the other. The lower financial returns an investor accepts for the sake of social impact, the higher the opportunity cost of impact investing. The opportunity cost of an investment is what the investor could otherwise have done with the money, which could be donating straight away, or investing to give: socially neutral investing and donating the profits later.

4. A few more observations

• Generally, non-profits can make profit, and for profits do not have to make profits, there's no principled difference between these two structures.
• Subsidizing highly scalable, zero marginal cost (global) public goods, entrepreneurship: it seems like sometimes corporations can do things that have a very high cost-effectiveness because their products have zero marginal cost. For instance, Moskovitz & Collison could subsidize Asana and Stripe Atlas for poor countries—that might be really effective, because software is often ~zero marginal cost and in fact such price discrimination might be profit maximizing.^[17] 
• Along similar lines: 'Old-school philanthropists' like US liberals had that figured out by being 'overexposed' to media companies and using them to spread their message. Some media companies have a relatively low market cap. For instance, WordPress bought Tumblr for just $3M, Iheart media, the biggest podcast company, and the NYT only have market caps ~$0.5B and $10B respectively. A billionaire philanthropist could buy these firms and have outsized influence on the discourse. These days, silicon valley, which has a much bigger impact on the discourse, is less aligned with the US left.
• Low marginal cost of research and high returns to research: If you have a foundation anyway like the Gates foundation with 1k employees then the returns to researching investment opportunities might be very high because the marginal cost of spotting investment opportunities is very low. In other words, say Gates pays a grant officer to find non-profit opportunities in the WASH (Water, Sanitation, and Hygiene space). One day at a conference she bumps into a driven entrepreneur with an idea of how to chlorinate water at scale at a very low price. Given her philanthropically subsidized expertise has almost zero marginal cost, she can ex ante find returns that are in excess of market-rates. And so, the cost-benefit ratio is very high.
• Similarly, if you run a corporation you might be well placed to make grants in their area of expertise with low marginal cost. This also applied to frontier AI labs, where there’s been a lot of con controversy recently around whether people should work for them. But if you’re an altruistically-minded person it might make sense to exploit opportunities in for-profit corporations to do things that would have a low marginal cost for the firm, but an outsized benefit to humanity (e.g. responsible scaling policies).

Tab 2

“3.2. Low-cost but high-benefit changes

Within the bounds of what market forces allow, and what companies and the public see as acceptable, there could be minor design changes that yield large social benefits at negligible cost to competitiveness or user satisfaction.

This is especially true for rare situations. Constitutional crises don’t happen often, so market pressures won’t directly shape how an AI behaves during one. But that AI behaviour could be hugely consequential.

It would also be true in situations where users don’t care all that much about the behaviour. Perhaps they find some AI’s encouragement to reflect on their values mildly annoying, but not nearly enough to switch to a different AI.”

https://x.com/NunoSempere/status/1998366827152650636 

https://www.anthropic.com/news/anthropic-rwanda-mou 

• Donating our open-source alignment tool \ Anthropic https://www.anthropic.com/research/donating-open-source-petri