Here, I highlight three consequential properties of international agreements to spend a fixed percentage of GDP on both global public goods and bads (global risks):

They are large in scale.
Their game theoretical features solve public good problems (e.g. free-riding)
Their coupling to growth causes differential technological development

We have many international agreements (e.g 500+ on the environment, rivaled only by trade agreements).^[1],^[2] Many are important like the Paris agreement to reduce climate change, the UN's Biological Weapons Convention, etc.^[3]

But here I focus on agreements where countries agree to spend a fixed percentage of GDP on global public goods and global public bads (henceforth referred to as 'agreements'). For instance, OECD countries have agreed to spend 0.7% of GNI on official development assistance ('aid').

These agreements have the following interesting and consequential properties.

Consequential properties agreements to spend a fixed percentage of GDP

Large scale

If many countries agree to spend a fixed, non-trivial part of GDP on an issue (e.g. aid), then this adds up to many billions a year. Thus, these agreements score highly on the 'scale' criteria of prioritization frameworks and are very important. Even if countries fall short on their commitments—which they often do—it often still leverages billions annually over decades (e.g. countries have agreed to spend 0.7% on aid and though many states have yet to meet this target; it has still leveraged a massive ~$150B of aid per year).

Game theory

Solve public good dilemmas

What are public goods? These are social or collective goods that you cannot easily exclude people from enjoying (e.g. clean air) and if you consume the good, it does not affect me from consuming it (e.g. broadcast TV). In other words, public goods have zero marginal costs.

If a country has only 5% of the world population (like the US) and provides global public goods (like knowledge from R&D, a stable climate, global peace and stability, global frontier economic growth, global biosecurity, etc.), which benefits everyone in the world equally, then the country's citizens only reap 5% of the total benefits of this spending.^[4]

If countries cannot cooperate, then it is better to free-ride on the efforts of other countries (and defect on international agreements). Countries can just let others do the basic research and then reap the benefits because IP is hard to protect internationally (e.g. if China funds research into fusion energy or a cancer cure that would benefit everyone, then other countries can just freeride on China's efforts). And so, as a rule, global public goods (and preventing global public bads) are undersupplied, underinvested in, and neglected.

But these agreements improve international cooperation to solve this free rider problem. If all countries agree to contribute their 'fair' share of GDP, e.g. to reduce climate change or find a cure for cancer, then everyone benefits maximally.

From a game-theoretic perspective, stable and effective international agreements are difficult.^[5] But these agreements have the following features that increase cooperation in public goods games. A country can use these features unilaterally to achieve Pareto‐improving outcomes (by self-committing to match other countries' public goods).^[6] Philanthropists might leverage their funds to initialize more global cooperation (perhaps through quadratic matching^[7]).

Reward and punishment: Countries can punish each other for defecting ^[8] (e.g. the US has criticized Germany of free-riding by not spending 2%/GDP on defense as agreed amongst NATO countries and threatened trade sanctions; Germany has criticized the US of free-riding by not spending 0.7%/GNI on aid as agreed amongst OECD countries^[9]) reward each other with better trade deals (e.g. if a country joins the EU it gets access to the EU's free trade bloc, but also has to pay ~0.7%/GNI in EU membership fees, that pay for various international public goods like the EU's Center for Disease Control).
Iterated, infinitely-repeated games: Countries 'play many rounds' of these public goods games. Every year countries check if others have reciprocated and contributed their fair share or if they defected and started free-riding. Over time, countries build up a reputation for cooperation. Finitely-repeated games have uncooperative equilibria: the dominant strategy is for players to defect on the last round, and so rationally foreseeing that the other player will do the same via backward induction, players defect on the first round. However, countries live forever and can play 'infinitely-repeated' trust games.^[10]

Setting the date for meeting the spending target in the future—e.g. EU countries agreed to spend 3.5% of GDP on research by 2025—makes these agreements more politically tractable, because countries do not face sudden, high costs. To see if others cooperate and reciprocate in providing global public goods, a country can start with a low stakes amount. So-called, 'super cooperators',^[11] who take on some of the costs of enforcing the payoff equilibrium, can flip a network from a defect-defect equilibrium to a cooperate-cooperate.^[12],^[13] Such agreements can also lock in a consistent flow of funding for an issue—sometimes for decades.

Transparency: Countries verify if other countries have met their agreements, often by paying their share to multilaterals (e.g. WHO), or by setting up organizations like the OECD which sets up rules as to what counts as aid spending.^[14] Think tanks create indices (e.g. Commitment to Development Index^[15] or the Good Country Index^[16]), which rank countries on how much they cooperate, including their spending, to name cooperators and shame defectors. This can create a 'race to the top' and engage the media and public. A think tank could create a global stability index that measures how countries' policy efforts to increase global stability and reduce risks (e.g. macroeconomic, political stability etc).

Coupling to GDP and differential technological development

These international agreements are pegged to GDP. This means that their budgets (e.g. military, aid, research, etc.) are coupled so that they increase in lockstep with GDP. If we grow GDP, which economies optimize for^[17], we automatically increase these budgets.

This creates some interesting dynamics vis-a-vis differential technological development.^[18] This principle posits that society should 'retard the development of dangerous and harmful technologies, especially ones that raise the level of existential risk and accelerate the development of beneficial technologies, especially those that reduce the existential risks posed by nature or by other technologies'.^[19]

While many have argued that faster growth is the most effective way to improve welfare in the short-run,^[20] it also increases existential risks (via externalities like climate change or emerging technologies that have dual-use risks such as nuclear physics and synthetic biology).

However, as has recently been shown, if we get richer we can spend more on safety to counteract existential risks, and so faster growth might actually decrease existential risks. This is because it gets us faster through our current 'times of perils', where we develop dual-use technologies like synthetic biology, which have societal benefits but also carry substantial risks, and we don't spend enough to counteract them. ^[21],^[22],^[23]

However, safety spending can, but need not increase with growth.^[24]

There are two principal reasons for that:

Some spending might be 'anti-safety spending' and increase existential risk (e.g. spending agreements on the military, or on research if it is dual-use).
It might also be that risks are easier to increase through dangerous technologies than guard against through safety spending.^[25]

Thus, if GDP growth increases or decreases existential risk depends on these agreements.