Brian Kenny: Hunger, as defined by the United Nations, is a condition in which a person for sustained periods is unable to eat enough food to meet basic nutritional needs. The UN says there are roughly 815 million hungry people in the world today, and according to the United States Dairy Association, 42 million live in the United States. That's an estimated one in eight Americans, or 12.7 percent of all US households, and the irony is there's more than enough food to feed everyone. In 2017 US, frozen storage facilities stockpiled 322 million pounds of butter, 1.5 billion pounds of fruit, and 1.3 billion pounds of frozen poultry. And yet, even with a veritable army of charitable organizations who've made it their mission to bring supply together with demand, feeding America's hungry is complicated business.
Today, we'll hear from Professor Scott Duke Kominers about his case entitled, Feeding America. I'm your host, Brian Kenny, and you're listening to Cold Call.
Scott Duke Kominers is an economist and member of the Entrepreneurial Management Unit at Harvard Business School, and he's joined today by a special guest. Canice Prendergast is an economist who's on the faculty at the Chicago Booth School of Business, and he's dialing in from a studio there. Thank you both for joining me today.
Scott Duke Kominers: Thanks for having us.
Canice Prendergast: Thanks for having me.
Kenny: I'm going to start with you, Scott. I wonder if you could start just by setting up the case for us. Who is the protagonist and what's on his mind?
Kominers: Sure. Feeding America is the third largest nonprofit in America, and they manage a network of over 200 food banks nationwide. All of the food banks have their own individual directors, but Feeding America interfaces and helps them organize and coordinate food distribution. The case actually follows a team. Feeding America sought to redesign the system they use to match donations to different food banks, and convened a team that consisted of Feeding America administrators, food bank directors, and four faculty members from, I hate to say it, another business school. The case in particular follows two of the team from Feeding America and two of the University of Chicago faculty.
Melinda Resser is this extraordinary woman from Feeding America who, at the time, was director of network analytics and really is the original data scientist. She was doing data analytics at Feeding America before there was such a thing as a data scientist. She's been there for decades. Then Melanie Nowacki, who is the National Council administrator who organizes around the network.
From the University of Chicago Booth School, Harry Davis, the Roger and Rachel Goetz Distinguished Service Professor of Creative Management, and Canice Prendergast, the W. Allen Wallis Professor of Economics, who's joined us today.
Kenny: That is great. What prompted you to write the case?
Kominers: I study market design, how we can redesign marketplaces to reduce frictions and create better outcomes for all. When you think about marketplace business today, the first things that come to mind are these big internet businesses. You think about Uber and Lyft and Rideshare or Airbnb and room-sharing as a competitor of the hotel space, but not every marketplace is an Internet platform. Lots of them are in the real world and their designs are just as subtle and often more tangible than the sorts of things that are going on the Internet platforms.
Feeding America knew it had a problem. It was trying to figure out how to allocate food to different food banks, and it knew its system was sort of not an efficient market, and the question was how do you design in a way that gets you closer? It's an extraordinary example of building a marketplace system virtually from scratch, plus it's a fantastic application, right? It's feeding lots of people, saving lives. It's fun.
Kenny: This case takes place in the 2003, 2004 timeframe. Canice, Scott mentioned that you were one of the members of the task force. How did you come to be involved?
Prendergast: I can claim no particular expertise before I got involved, to be honest. It was a case of happenstance. I was luckily approached by Harry Davis, and I think the reason was they thought the tools of economics could be useful to help to answer this question. None of us knew exactly where it would lead, but I think Harry decided I might be the right kind of person to end up having a conversation about it.
Kenny: As you got involved, what was the scope of the hunger problem? Is hunger today in the United States better or worse [then] back in 2003, 2004?
Prendergast: It was just about the same. You're talking one in eight with families who have children. It's closer to one in four that the USDA would say are food-insecure, and we kind of forget two things.
One is that the cost of going hungry is more than just having a stomach that's not full. We know now health implications over the long run effect ... We know that kids can't learn when they're don't have food in their stomach. The long-term implications of hunger are enormous, aside from just the immediate implications that we think of.
The other thing which you mentioned, Brian, is the frustration. We know there is so much food out there. The marvelous thing that Feeding America has done is tap into the resources of both distributors and manufacturers of food because it's really, in some sense, the key answer to the question, and trying to harness that in the appropriate way has really been their great success, but also the great challenge over the last two decades.
Kenny: Scott, can you talk a little bit about their history?
Kominers: They were founded in 1979 by the same person who launched one of the first food banks in the United States. They were formed to coordinate and organize among different food banks. So as people were copying the model and expanding it, it became clear that there were economies of scale [to be utilized], that essentially both there were best practices that could be shared and expanded upon, but also that food is available and being donated to different places. There might be gains from exchange and trade and somehow coordinating a bunch of different food banks together rather than having each fend for themselves, especially since lots of the donor partners are large nationals, sort of supermarket chains or Walmart-type organizations.
Kenny: What was their approach to allocating food at the time that the case begins, before they adopted the market method?
Kominers: Most food that food banks receive is , of course, donated locally, but some amount is donated to Feeding America and then distributed. Of that, Feeding America had to figure out which food banks to send it to, and they were very, very concerned about fairness. They wanted to make sure that they weren't unduly privileging one food bank over another and that they were trying to send the food where there was the largest perceived need. They had calculated these perceived need metrics, called goal factors and goal pounds: the target amount of food and number of pounds of food delivered to each food bank. As donations were offered, they would call a food bank and say, we have an offer of this many pounds of chicken. Would you like it?
“They wanted to make sure that they weren't unduly privileging one food bank over another and that they were trying to send the food where there was the largest perceived need”
The food bank would get the offer and they'd have maybe four or six hours to decide whether to take it, and if they said yes, they'd be responsible for paying transport, going and picking it up, and storing and distributing it. But if they said no, Feeding America would … call the next food bank, but they would treat it as if the food bank had said yes, and it seems a little strange. Right?
Kenny: Yes.
Kominers: The reasoning is, for the sake of fairness, you don't want to ever make it the case that some food bank has some incentive to sort of wait, to try and stay at the top of the list until they get some very specific thing they need. Instead, Feeding America was trying to make sure that every food bank had enough shots of each individual food type, but they were rationing offers rather than the actual donation. If you said you didn't need an offer, they inferred that maybe you just don't have as much food demand. The system was very much push-based and they weren't learning anything about the food bank's actual demands other than perhaps a little bit of information about what they didn't need.
Kenny: Canice, the case introduces a term that I found a little bit ironic, food-rich, in the context of food banks. Can you explain the food-rich versus the food-poor and how that played into the allocation system?
Prendergast: This was something I knew nothing about when we started as well, and it largely had to do with the following. The food banks have a wide range of ways of trying to persuade donors to give food, and Feeding America, at the time, gave probably about 20 to 25 percent of all the food that they received. The difficulty in terms of both the efficiency and the fairness of the system was that there were some parts of the country and some particular food banks that were just very successful at finding other types of food and other amounts of food, and food-rich refers to those food banks, those areas of the country which had a very wide supply of other food they could get.
The difficulty Feeding America had was they didn't really know what [food banks] had or how much they had, so they were essentially a little bit blind as to how much food they should be giving to them. Now some of this involves local distributors, local retailers, but sometimes it just involved personal contact skills of the individual food bank directors. But it meant that there was a big unknown inequity in the system, which was the backdrop to our committee meeting.
Kominers: There was also, incidentally, the problem of transitory demand changes. It might be that you have to turn down a big donation of yogurt this week because your refrigerators were filled up with milk yesterday, but next week you'd be very happy to have a lot of yogurt. Even if you're food-poor, you might still have to turn down a donation that you otherwise might want.
Prendergast: One of the biggest problems for food redistribution in the United States is spoilage, and some spoilage is implicit because a lot of donations only become available close to expiration dates. But there's a considerable amount of spoilage that comes from the fact that food banks may not have enough refrigeration; if you give them something that needs refrigeration, a very valuable piece of food will just go to waste.
Kenny: Do [food donors] take umbrage if somebody doesn't want to take something that they're planning to donate?
Prendergast: That's a terrific question, actually. That was one of the real learning experiences for me. I realized that Feeding America has essentially two constituents: one is the obvious food banks, but the second constituent are the donors, and they're very concerned with donor relations. There were two parts to that. One, they would often twist the arms of food banks to take food that maybe they didn't really want, and that was to keep the donors happy. That was part of the reason why they would penalize food banks if they actually turned down food. The second thing, which I think turned out to be much more important than any of us realized, was that Feeding America was so aware of these relationships that often they would turn down donations rather than offend the donor, because the one thing that donors do not want is for a truckload of food to be sitting on their dock for two or three days.
“The system was very much push-based and they weren't learning anything about the food bank's actual demands other than perhaps a little bit of information about what they didn't need”
Kenny: Yes.
Prendergast: One of the difficulties with the old system, as sort of Scott outlined, is that they would offer it to the person at the top of the list. The person at the top of the list would have six hours to decide whether they wanted to take it, whether they had a truck that could get there in time, and what it meant was that Feeding America could only really offer it to a relatively small number of people before the donor would get upset. It meant at times, hey would turn down donations that otherwise could have gone to feed the hungry.
Kenny: So, the task force, let's turn to that for a second. Canice, I'm going to stick with you for a moment here, as you were there at the beginning. What was the intent of the task force, and what was the dynamic like? It sounds like there was a pretty good mix of people around that table.
Prendergast: It was a wonderful mix. It was a combination of food bank director ... It's very carefully selected to include both very large food banks, but also some of the smaller ones. It had members of the Feeding America staff and it had four faculty members from the university. I think at the beginning when we met, there was a sense that what Feeding America wanted to do was tweak, and tweak could involve the following, which is they would say, "Let's never offer potatoes to Idaho because Idaho has plenty of potatoes, or maybe we should not give as much food to people that have a regional distributor somewhere nearby." So, the sense was that we would stick to the push system rather than anything that looked like a pull system, and the reason for that was the typical reason why not-for-profits use push systems, which is they fear that if they let people choose, they'll just take everything. As a result, they would decide what an individual food bank wanted, and go from there.
The dynamic of the group was wonderful. I think one of the amazing things about this committee was essentially the length of time we got to listen to each other. I think if it has turned out to be a success, the reason was largely because of the willingness not of the academics to listen to the practitioners, but actually the practitioners to listen to the academics. This was a long way removed from anything that they imagined they would do when we started.
Kenny: That makes a lot of sense. There were a lot of ideas, I guess, that surfaced in the context of the conversations. You decided to go with this market-based solution. Scott, can you explain the difference between the push versus the pull approach?
Kominers: Sure. My understanding is they came reasonably quickly to this idea that if only there were some way to give the food banks choice, then a lot of the problems of the push system would go away. They didn't know which food banks wanted which food. They didn't know which were food-rich and food-poor overall. They didn't know anything about transitory demand. If only the food banks could just tell you, this would be great. And moreover, if they had a liquid market, if they had some way of learning which food banks needed which donations, they could figure out which donations to solicit more of and maybe even place some of those donations they were otherwise having trouble placing. The key to all of this seemed to be choice. You had to somehow learn the food bank's preferences. But, of course, you can't just say, as Canice said, "Just tell us your preferences. Take what you want," because you still don't have enough to give everyone everything. You need to somehow figure out some way of balancing fairly between the different food banks.
What they realized was that they could switch to a market-like system. If food banks are bidding, this can reveal their demand, but now there's a different problem, of course… You don't want to create inequality by using a cash market. Instead, they created sort of an internal currency called shares.
Every day there'd be a spot market in donations. The food banks would bid shares, and for each donation the food bank that bid the most would be assigned that donation. They would go pick it up and they would pay their shares into the center of the system, and the collected shares at the end of the day would be redistributed back out to the food banks in proportion to those same goal factors, the need estimates from before.
Now, here inequality smooths itself. You save up shares, you bid on the things you want, and then at the end of the day, you get money back from the system in proportion to how much everyone else spent. This had this miraculous property that if you're a food bank who didn't buy anything today, then you're much more able to buy something tomorrow. At the same time, we learn demand. You learn the food bank's different sort of relative demands for different types of food, and you can even hope to learn which food is most valuable for the system and how much.
Kenny: It makes perfect sense, but, Canice, I'm guessing it wasn't immediately embraced by everybody in the room. What was some of the reactions to this when it was proposed?
Prendergast: I would say not even close. We've all been in meetings where while the other side is being decidedly gracious, you could see glances being exchanged between some of the participants that told you quite how far what we were suggesting was from what they thought we were going to do. And I think it came down not to a philosophical disagreement with the notion that choice could be valuable, because part of the reason why this was introduced realistically is because they were very frustrated with the old system. They were being given food that they already had. They thought other people were benefiting relative to them. There was an enormous amount of frustration, but it was still quite a distance from what they thought we would do.
I think it came down to actually a much more general point at the moment about markets, which is I think there's a sentiment among many people that markets benefit the strong at the expense of the weak, and I think in the context of this, the strong meant the big, more heavily staffed food banks, so the Boston Food Bank, Los Angeles, Chicago, New York, and I think there was a sentiment that anything that we put in place would benefit them relative to a very understaffed place, say, for example, West Virginia, that really doesn't have the staff to manage these things. It was only after talking really for a year that we managed to persuade the participants that what they see as potential inequities could actually be resolved in a way that would benefit everybody.
“I think if it has turned out to be a success, the reason was largely because of the willingness ... of the practitioners to listen to the academics”
Kenny: Wow. One of the things that I really love about this example, we talk a lot here at Harvard Business School, and I'm sure they probably do at the Booth School as well, about the relevance of the research to business and to practice, and this is a great example where you're taking economic theory and you're applying it to a real problem in a way that has very tangible and easily identifiable results.
Prendergast: We do, but let me say one thing, which is we're also riding off the coattails of a large literature in market design, of which Scott is an important contributor.
Kominers: Canice is, too.
Prendergast: And we are just one example of a much broader notion of how we can use the tools of economics to think about how pull systems can work in very unusual contexts.
Kenny: Scott, the case goes into some detail about the operational things that they thought about. There were a lot of considerations that they had about how to actually bring this thing to life in advance of the national meetings. They had to answer some questions before they could really propose this. What were some of those things?
Kominers: Well, first of all, at some fundamental level there is how are these shares going to work? Are they going to manage to hold value? You worry about monetary policy and currency crises and things like this. If you're creating your own currency inside of the system, you want to make sure that savings that food banks save up are going to be valuable tomorrow. There, economic theory comes to the rescue a little bit because individual food items arrive with sufficient frequency and you're playing the game every single day, so It's not like there's large gaps between different stages that might cause your shares to be much more valuable today than they are in the future.
On top of that, of course, there were lots of protections in place for small food banks. As Canice said, one of the big concerns was that the better resourced food banks might be better able to use the system somehow, and so they tried to mitigate as many of those as they could think of, ranging from having the market clear only once so you didn't need to employ a staff person to constantly monitor the system.
Also, they had sort of credit so you could go a little bit over your budget and pay back those shares in the future over credit. Small food banks could band together and bid jointly and then sort of share the transportation costs. Then, of course, for bigger stop gaps, there was a committee that would be set up to adjudicate if there were any major disputes, and also there'd be sort of a separate system in place in case of natural disasters or sort of anything that caused a huge influx of demand in a specific location.
Kenny: Yeah. Canice, you were at the national meeting when they presented this? Is that correct?
Prendergast: No. They actually suggested it would be better if I was not there. In fact, I think what happened at the end of our deliberations, and I think Harry Davis deserves an enormous amount of credit for this, is I think they had been persuaded about the merits of the system, I think to the extent that we also believed. And I think there was a sentiment that this was only going to come to pass if the 215 food banks across the United States were willing to vote in favor of it, and I think the food bank directors on our committee decided that they would be the appropriate proponents, and I think they're absolutely right. I know relatively little about what happened in the meeting other than it was overwhelmingly voted in favor of. We were waiting on the other end of the line, essentially, to hear the answer, and the others seem to have done at least as persuasive a job as I could certainly have done.
Kenny: That's good. That's good. I mentioned this takes place over a decade ago, and there's a B Case, Scott, that you wrote on this. I didn't prepare any questions specific to that, but can you give us a glimpse as to how things have gone over the last 15 years?
Kominers: The B Case comes back in about a decade later and starts to look, actually based on some of Canice's subsequent research with Feeding America, at what happened in the system, and we find several really impressive results. First of all, there's immediate stratification in the share prices for different types of food. Some food, pasta, cereal, things that are shelf stable, high-nutrient density turn out to be worth many, many more shares than things like chips and soda, things that are expensive to transport, hard to store, and not very healthy.
Secondarily, you see stratification in what the food banks do. Some of them, buy lots more of the low share cost items, so they get much larger quantities of some of the food that's easier to obtain in the system and others save up and buy only sort of the bigger ticket items. You see a reflection of the fact that some food banks are more food-rich than others, but the ones who are more food-poor are at least getting lots more food than they were getting before under the old system. Some of Canice's work sort of tries to understand what this meant for the whole equilibrium of the system, even relative to the food they were bringing in and shows that, in fact, it looks like there was sort of a global improvement that basically this made everybody better off.
Kenny: Wow.
Kominers: But then the most extraordinary fact, I think, is what this did for donations coming into the system. If you track the donations leading up to the launch of the new system, there's this huge uptick. Canice, what was it, about 40 percent or something?
Prendergast: It went up from about 250 million pounds to about 350 million pounds.
Kominers: Yes, so a huge increase. Of course, some of that, we can't say that was all totally new donations. Some of it might've been donations that shifted into the system because the market was more liquid, but this idea that if only you had a liquid market, you could solicit more donations, you could feed more people overall, really does seem to have been borne out. The back-of-the-envelope calculation, this means they were feeding roughly 50, 55,000 more people per day.
Kenny: That's amazing. Canice, are you still involved with the group?
Prendergast: I am, actually. I'm trying to do a number of things. Food banking has changed a little bit in the following sense, which is Feeding America has been enormously successful at sourcing, particularly local donations, so the amount of food they're sourcing is going up like crazy, but there's an unusual thing that's happened, which is many of the donations which used to happen in the past are actually now being sold on a sort of thriving secondary market. One of the things that Feeding America has done over the last decade has become a purchasing agent. It uses its buying power and then sells on to food banks. That in itself has a bunch of inequities because some food banks are better able to raise funds than others, so I'm trying to do a bit of work on that. I'm also trying to do another thing, which is I think one of the biggest remaining problems with charitable food distribution occurs at a very local level, which is the poor tend to come to your local school, your local church, and those institutions are not the most effectively run. So, trying to improve things at that very local level I think is something that will be really valuable for somebody to work on.
Kenny: That's fabulous. Well, hopefully, if we have this conversation 10 years hence, the numbers that I read in the beginning of the case will be greatly reduced. Scott, Canice, thank you so much for joining us today.
If you enjoyed this episode of Cold Call you should check out our other podcasts. After Hours features Harvard Business School faculty dishing on the happenings at the crossroads of business and culture. Managing the Future of Work features experts discussing how to survive and thrive in the age of artificial intelligence and learning machines. Subscribe to these and others on Apple Podcasts or wherever you listen. I'm your host Brian Kenny, and you've been listening to Cold Call, an official podcast of Harvard Business School.
Interview recorded October 12, 2018. Transcript edited for clarity and length.