When given money, the stats seem to indicate that these people invest that money in both short term (food !) commodities and long term (land, livestock, housing) assets.
Basically, giving money directly is allowing these people to go out and buy some fish for right now, and a fishing pole for tomorrow.
An added benefit is that this is a very lean way to organize aid, since there is no need for dozens of committees and meetings and other organisational overhead to determine the most efficient way to spend. You just let the receivers decide for themselves.
Sure there will be some abuse and some people that go even further off the rails due to drug abuse or bad investments, but most of these people are just as capable as you and me to plan their live, they just need that little boost to get back ON the rails in the first place.
A somewhat similar model is used by Kiva, except that they give the money as a loan, and it's expected to be paid back. Maybe that approach is a bit more in line with your opinion.
In the "fishing pole" analogy: You may give them a fishing pole in the best "central planning" fashion, but they may have little use for a fishing pole because there's practically no fish in the region.
By giving them money, you trust them to decide for themselves whether a fishing pole would be more useful than say a plow.
I think there is a point where people stop trying to be self-sufficient and start relying on handouts. It would be interesting to know what triggers it.
But the assumption that some external provider of aid always knows which kind of aid is better is probably a fallacy.
Another dark side of such issues is that when the production goes over the quotas, then the food is just dumped instead of being given to poorer countries.
I know this can also affect the local farmers, but it just seems bad to throw food away when there is so much hunger in the world.
And if they did buy the fish, are we sure that really was worse than the rod?