Stanford Researcher Offers a Modest Proposal for Food Relief

A Congolese girl carried a box of food aid last December at a refugee camp in the eastern part of the Democratic Republic of the Congo. (Phil Moore/AFP/Getty Images)

About one in 10 children in sub-Saharan Africa under the age of 5 is starving, and nearly 40 percent are physically stunted because of chronic malnutrition. The situation in South Asia is even more dire. Yet global shipments of food aid have declined steadily since the late 1990s. In short, there’s not enough food to go around.

So if you ran a relief agency, how would you use your limited resources to save the most lives?

A Stanford researcher may have an answer, though it’s one that at first blush might sound cruel: Deliver food only to the children most likely to benefit from it. His conclusion led The Atlantic to run an article with the headline “The Case Against Feeding Every Hungry Child.”

But it’s more complicated than that. Lawrence M. Wein, a professor of management science at Stanford’s Graduate School of Business, applied mathematical modeling and statistical regressions to a humanitarian issue and found that, in short, aid groups may be looking at the wrong factors when determining who benefits most from food aid. He attaches several caveats to his work, which we’ll get to in a minute, but first the conclusions.

Most important, says Mr. Wein, stunting—which measures a child’s height relative to his age—has been underappreciated as a risk factor for death and severe disability. Aid groups tend to give out food based on a child’s weight relative to his height on the premise that starvation, or “wasting” as it’s commonly called, is a more urgent problem than stunting.

But Mr. Wein encourages aid groups to start using both measures when determining who gets food and how much. How did he arrive at that finding? By crunching the numbers on two sets of data. One, owned by one of his co-authors, Jan Van den Broeck of Norway’s University of Bergen, tracked the weight and height of more than 5,600 young children in the Democratic Republic of the Congo, the former Zaire, over a period of 17 months. The other study looked at the effect of food aid given to children in six villages in Niger, and compared that to six similar villages that received no aid.

Through a series of advanced statistical regressions and other methods no doubt unfamiliar to the average aid worker, he found a strong connection between stunting and mortality risks, particularly if the child was also severely underweight.

“The kids who are stunted are really vulnerable to dying,” he says.

‘All or Nothing’ or ‘Blanketing’

His models also concluded that aid groups may be more effective by taking what he calls an “all or nothing” approach to food distribution. The more common practice is to spread food out to as many children as possible—the “blanketing” approach.

His analysis shows that the all-or-nothing approach, combined with measurements of health that include both stunting and wasting, could reduce death and severe disabilities by 9 percent, compared with the approaches aid agencies take now.

Mr. Wein’s study looks specifically at the effects of ready-to-use, high-protein pastes that deliver critical vitamins and other nutrients. The pastes, which are in wide use today, are remarkably effective at bringing children back from the brink of starvation. His study shows that giving full doses of those foods to the sickest children would save more lives than giving, say, a half-dose to a broader group of children.

“Partial dose does a little,” he says. “But when the dust settles, it’s really the most vulnerable kids who are at the highest risk.”

Now, the caveats. For one, usable data for a study like this are extremely limited. The two studies he used are basically the only two he could work with, he says. For obvious reasons, aid groups are not going to run randomized controlled studies in which they give one set of children food and the others nothing.

Moreover, data collection is not something aid workers are trained to do well or consistently. So while he used rigorous analysis that leave him confident of his findings, Mr. Wein says that in no way is he making policy recommendations.

And there are many good reasons aid agencies might want to take a blanketing approach to aid. For one, it’s a great way to get all the children in a village into a clinic to be examined and inoculated. Politically and culturally, a more selective policy might also be tough to pull off. Finally, aid groups consider a whole host of factors when figuring out food distribution, including how much money they have and how else they can help sick children.

“Even if one buys into the math and existing data,” says Mr. Wein, “it doesn’t necessarily imply that you want to switch from blanket to all-or-nothing.”

Stunting and Starvation

So what do other scholars think of his findings? A couple of economists with expertise in food systems in the developing world say the all-or-nothing argument is not actually the paper’s main contribution.

William A. Masters, chairman of the department of food and nutrition policy at Tufts University’s Friedman School of Nutrition Science and Policy, calls the argument “an obvious mechanical consequence” of having a model tell you which child is the most needy. That is, it’s always true that it is most cost-effective to focus on the most at risk. The problem with executing the concept is a practical one: Relief workers “don’t have X-ray eyes to detect the most needy in front of them.”

For example, in countries where recordkeeping is poor, age is less certain and less easily verifiable than weight. And, as Mr. Wein notes in his paper, if a parent knows that age-to-height ratios determine who gets fed, they have an incentive to lie about how old their child is. Relief workers blanket, Mr. Masters says, “because they don’t have that true model” of need.

Mr. Masters believes the finding on the effects of stunting is the paper’s most significant contribution. It reveals that malnourishment and other chronic health deficits can compromise children’s immune systems and leave them vulnerable to life-threatening illnesses. That’s noteworthy, he says, because stunting is now a much greater problem than starvation.

The United Nation Children’s Fund reports that in South Asia, for example, nearly half of all children under 5 are stunted, compared with 19 percent who are in a state of wasting. The decline in starvation has come through improvements in standards of living and therapeutic interventions, such as the high-nutrition pastes. Stunting, by contrast, has deeper and more complex causes, such as disease and chronic poverty.

Mr. Masters praises Mr. Wein for tackling an understudied issue in a sophisticated way. He notes that the Obama administration is trying to improve food-aid policies, making them more efficient and effective. “Getting the intellectual horsepower of someone like Wein is extremely helpful” to this debate, he says. “I hope it attracts interest in these problems by researchers in business schools and other settings that have not focused on agriculture.”

Thin Data

Christopher B. Barrett, a professor of applied economics and management and an international professor of agriculture at Cornell University, applauds Mr. Wein’s rigorous work while acknowledging the limit of the available data for research like this.

“We have an extremely thin and fragile foundation of data on which to base analysis,” he says.

That doesn’t take away from Mr. Wein’s contribution, says Mr. Barrett. Rather, aid agencies and others should be encouraged to collect and disseminate data on the children they are caring for so that further studies can be done. Whether it’s the U.S. Agency for International Development or Catholic Relief Services, “they really do pay attention to the best science” and modify their policies accordingly.

As for Mr. Wein, provocative ideas are his trade. He made a name for himself after the terrorist attacks of September 11, 2001, when he turned his expertise in operations management to issues of national security. In 2003 he was a co-author of a journal article showing how a release of airborne anthrax could kill 100,000 people in a short time, and argued that “there is no publicly available government population-response plan for anthrax.”

In a 2005 journal article he showed how terrorists could kill tens of thousands of people by placing a few grams of botulinum toxin into milk supplies. The U.S. Department of Health and Human Services called the paper “a road map for terrorists” and tried to get publication blocked.

Mr. Wein’s point in both cases was to use mathematical modeling to show how serious a risk such attacks pose to national security, and how a few key changes by federal authorities could cut such risks. In both cases, he says, those changes were made.

These days his interests run to concerns like maximizing the benefits of blood transfusions and biometric data collection.

“I’ve kind of run out of scary stuff,” he says, “so now I’m back into public health.”


Return to Top