Last week the Bush administration announced a modest initiative for Africa. This week the rich world's leaders gather at the Group of Eight summit to announce a further aid expansion. Meanwhile Raghuram Rajan, chief economist of the International Monetary Fund, has co-authored two papers on aid with Arvind Subramanian, an IMF colleague. Their view? There's no strong evidence that aid boosts economic growth and hence no reason to suppose that aid reduces poverty either.

You have to admire the authors' courage. Everyone from Tony Blair to Brad Pitt is campaigning for more aid; doubting its usefulness is not politically correct, to put it mildly. You have to go back more than 10 years to find a comparably provocative statement from a top aid official -- back to the time when Larry Summers used his perch as World Bank chief economist to endorse the export of toxic waste to poor nations. Not only are Rajan and Subramanian politically incorrect. According to some rival experts, they are analytically incorrect also. Some statistical tests do find that aid produces growth, notably ones that measure only the sort of aid that could be expected to do so, leaving out disaster relief and such like. But although the experts will debate the aid-and-growth evidence for some time, the central argument of the IMF authors is undeniably correct. Aid projects may do good, but they have unseen side effects that hurt. The development folk need to absorb this message.

What are these unseen effects? As aid flows in, it pushes up a country's exchange rate and damages its exporters. Aid projects that hire local workers are bidding up skilled wages, again damaging the export firms that hire from the same labor pool. So an AIDS project or water project may deliver wonderful and visible results while also choking off the export growth that represents the surest route to development. Hence the fact that many statistical tests don't find that aid helps reduce poverty.

It follows that you have to care a lot about whether aid is spent well. A failed aid project is not merely neutral for poverty reduction; it exacerbates the problem. Inconsistent donors who finance the construction of six hospitals but then don't follow up with the resources to make any of them function saddle poor countries with the worst of both worlds: bad health and bad growth rates.

Next, you need to think about the kind of aid you favor. Aid that boosts exporters' productivity -- a road connecting an industrial city to a port -- may be a net plus for growth: The productivity gain outweighs the adverse currency appreciation and wage inflation. By contrast, aid that's spent on education takes time to boost industrial productivity. Too much of it spells short-term problems.

But the biggest lesson from Rajan and Subramanian goes beyond aid. It's that our engagement with global poverty needs to get broader. Because aid has inherent risks and because the benefits of trade liberalization tend to be concentrated in strong developing countries such as China, the world needs to get serious about the third leg of development. As Subramanian and others argue in a forthcoming article in Foreign Affairs, we must focus on rules and policies that we adopt at home that have spillover effects on poor countries.

Call this the "home front" in the battle against poverty. Western banks should be faster at disgorging deposits made by ousted dictators and returning them to their countries. Western drug policy should focus more on treating addicts, controlling the demand for contraband that wrecks countries such as Colombia and Afghanistan. Western oil companies should publish what they pay to governments, reducing opportunities to embezzle the money. Western lovers should only buy diamonds that are certified as coming from legitimate sources -- and that won't finance development-destroying rebels.

That's just the beginning. The home front should also include research on the seeds and medicines that poor countries need. Research dollars that aren't spent in the poor world would be free of aid's harmful side effects. The home front should also address global warming, the costs of which will partly fall on farmers in poor countries. Stretching the definition of home front a little, we need to get better at peacekeeping.

All of which demands a profound shift in our thinking. Over the past 15 years, development organizations have courted public opinion by emphasizing the social services they deliver: They stopped mentioning power plants and roads and talked up schools and clinics. But aid, if it's going to be expanded, needs to shift toward growth-promoting infrastructure, and we need to think beyond aid altogether. The World Bank should have a home-front unit, just as it already has a unit that thinks about trade. Other development groups should see home-front issues as a central part of their mission, not as cute add-ons to their core aid efforts.