Rich countries sent $137bn in aid to poor countries in 2014, plus large amounts given through non-governmental organisations. People give because they feel a moral obligation to help those who live and die in destitutione destitute. Yet we still do not know whether these gifts are helping or hurting, or how to use the money to good effect.

The obvious answer is to give to those who suffer most. Yet many of the people in the greatest need are those living live in countries where aid would be wasted — because of war, because their governments exploit their citizens rather than help them or because the state has little capacity to govern. Giving money directly to citizens, or via NGOs, does little to block a powerful “extractive” state.

Aid given to good governments in poor countries can work, but capable governments have other, often better, alternatives, such as borrowing from international capital markets. Whatever the disposition of the government, large aid flows detach leaders from the needs of their people, making them like the autocrats who sit on mining and oil revenuesells whose revenues require no cooperation from their citizens. In Africa alone, there are more than a dozen governments who obtain more than half their revenue from foreign aid.

Aid administrators are impaled on this fork. Their home citizens want an impossible combination: aid, they say, must go to those in most need and aid must be effective.

What if we fund only those projects that have been thoroughly evaluated and shown to work? But evaluations rarely capture unintended consequences, such as the government reducing its provision when foreigners do its job for it, or when accountability to beneficiaries is lost by replacing internal by with external provision. The “what works” agenda also runs of the risk of replacing what (local) people want by what (often foreign) technocrats think they ought to have. It is these unintended consequences that explain why many projects succeed while the country fails.

One solution to the aid dilemma is to work outside the recipient’s borders and provide global public goods. Examples include basic scientific research, particularly in health; better control of the arms trade (why are some European countries who are so enthusiastic about aid so keen to sell arms?); trade policies that help reduce poverty; and the provision of high-quality, unbiased advice to countries negotiating trade deals with richer, larger and better-prepared neighbours.

For the same reasons, I am sympathetic to using aid funds to help refugees and migrants, not because they are poorer than those who are not moving — this is rarely the case — but because the money is spent in the donor country where outcomes can be monitored by the funders and are thus subject to informed democratic debate. Of course, we need to be careful which programmes are reduced when funds are reallocated.

Does this mean that help should only be provided from a distance? Does effective aid mean excluding those whose lives we know could be saved? I have to come to think not.

One of the greatest recent achievements of aid has been that so many are alive today who would be dead but for aid-funded antiretroviral therapies, as well as the many children whose lives have been saved by the use of insecticide-treated bed nets, and by reducing mother-to-child transmission of HIV. Even when aid-funded provision reduces local responsibility, risking future health, the issue becomes a trade-off between lives (certainly) saved today and lives (possibly) lost tomorrow. We should surely decide in favour of lives today, especially if we are careful to do what we can to minimise —I don’t believe we can eliminate—long-term negative consequences.

My belief that we should follow this route has been strengthened in recent conversations with the Global Fund to Fight Aids, Tuberculosis and Malaria. In most countries, the Global Fund is able to exploit the synergy that comes from the fact that their drugs are supplied in clinics that double up for other purposes, such as maternal and child health, thus improving overall health outcomes beyond those immediately targeted.

The Global Fund agrees that it would be better if countries were to fund their own drugs and health systems, and the organisation has had considerable success in working towards country funding and away from international aid. This is an continuing process and it will take time for the Global Fund to close shop, which is its aim. But stopping today would cost many lives, too high a price to accept. Such life-saving programmes should therefore be excepted from the “not within the country” critique of aid.