Sendhil Mullainathan and Eldar Shafir are two leading figures in the hot (if occasionally oversold) field of behavioral economics. Mullainathan teaches economics at Harvard and is a MacArthur Fellow. Shafir teaches psychology and public policy at Princeton. This week, they released an accessible short book titled "Scarcity: Why Having Too Little Means So Much," which summarizes some of the best behavioral economics work.
I caught up with Mullainathan this week. An edited transcript follows.
Harold Pollack: What are you trying to do with this book?
Sendhil Mullainathan: We’re really trying to do two things. As you know, I've been interested in poverty for a long time. This book is an outgrowth of me simply trying to understand U.S. poverty, poverty in the developing world, and another form of poverty, which we don't really call "poverty" but you might want to call just "financial distress," which many Americans are actually feeling because of this recession.
As our introduction discusses, there’s been a left turn in my thinking. That left turn was the realization one day that, maybe, when I’ve been looking at poverty, I've been looking at it too narrowly. Maybe poverty is a special case of something else. That something else is scarcity, and anyone who has the experience of “having very little” experiences the same psychology.
To me, that insight was both stupid -- obviously the poor are different from the busy — but also, if true, profound. Things at the edge of stupid and profound always interest me because that's usually where some good ideas sit. So much evidence started to come in that really made me feel, "Wow, maybe this is more on the profound end than on the stupid." That's what led to the book.
The second thing I'm trying to do here is a little more subversive. It's hard to get people to empathize with the poor. You can get some people to sympathize with the poor, but to empathize is actually very hard, because most people are not poor. I realized that scarcity gives you a thread. You can understand some behaviors of your own that you experience under scarcity, and you can almost follow that thread and say, "I can at least imagine what that scarcity must be like if it were really unrelenting. This allows you to almost project yourself more into people’s shoes, and therefore to gain a richer understanding of a world which many of us don't otherwise have access to.
Harold: Many poverty researchers are ambivalent about behavioral economics. There's this sense that behavioral economics focuses too narrowly on individual decisional pathologies: mistakes by poor people, mistakes by workers who don’t put money in their 401(k) accounts, and that the subject neglects larger factors such as mass unemployment.
Sendhil: I think there's a tension between macro and micro. One can say: "Why are you worrying about these micro‑behaviors? Let's worry about these big institutional things." I think it's not an either/or, Sure, there's stuff to be done on the macro level, but if you peel the onion and say, "Let's just look at tuberculosis, it kills so many people." And the numbers are astonishing, and then you say, "Well, why is tuberculosis killing people? At the heart of this is some behavior. You give people antibiotics. Compliance is pretty good for a while, but then drops off.” That's behavior.
Why does poverty persist intergenerationally? Part of the answer has to do with parenting; we know that. Well, what is parenting, but a behavior that a person must undertake? It's something you must do every second of the day, every minute of the day. I don't think this book has much to say about aggregate growth dynamics, innovation, technological change. Those matter, undoubtedly. On the other hand, there's another half of life. All these individual behaviors really do add up to something very big.
Harold: This work may also help us understand in a different way why good programs and interventions actually help people. Why are tight labor markets good for people? By sparing millions of people situations of prolonged scarcity and insecurity, we're helping them lead less chaotic lives that are less decisionally challenged. In addition to the other benefits of sustained employment, we make it easier for people to make smart long-term decisions.
Sendhil: That's right. To take another example, it's very unclear how to think about economic instability. Unemployment is one issue, but consider the increase in hours-variability. People face a lot of variability in hours. The macroeconomist might think: Well, whatever. Labor income goes up. Labor income goes down. But that’s not all. You make a lot of decisions when your wages are tight and when your hours are low that have big consequences.
To take another example, consider the big financial recession we just had. Here we really saw individuals' behavior and choices meet up with the macroeconomy. Here are people sitting on houses that might or might not be foreclosed. They must take loan modifications. They have to make decisions. They must do all of this under conditions of financial distress.
Some people examine the situation solely in terms of interest rates and bailouts. Yes, these matter, but so does the behavior of these millions of homeowners. It's very hard to reach people when they are in conditions of financial distress. That is a fundamental problem. I think it is something that we’ll face again and again.
Harold: That raises another issue. One person’s decisional challenges create a predatory opportunity for someone else. If you weren’t writing your book, but one about marketing dicey financial products, one major strategy would be to approach people when they're ill-equipped to make good decisions and then take advantage of them.
That happens a lot. Your team did one experiment that isn’t actually in your book. You showed that a very large proportion of financial advisers gave people biased and self-interested advice about retirement savings. In the housing crisis, one can cite many examples where homeowners on the verge of foreclosure were victims of similar or worse frauds and exploitative arrangements.
Sendhil: That's right, and it’s a more common problem. Forget about foreclosure. Imagine that your finances are very tight, and you need to make ends meet. Somebody offers you a solution to that problem: Here's a loan.
Like any human being, in that moment, you're really tunneling that problem. You're going to try and solve it. You're not going to ask questions like, "Okay, but what are all the strings that are attached?" It’s as if you are fighting a fire, and someone hands you a bucket of water. You're going to take the bucket and fight the fire.
When you look at payday loans, when you look at all these predatory lending products, they all have that feature going for them. They really are that bucket of water that helps you with the fire that you're fighting. They come with all these strings attached. But that's not what you're looking at in the moment.
Regarding our audit studies of financial advisers, I was astonished by what advisers did when we sent actors who literally had what textbook finance should say is the optimal portfolio -- low cost, index fund, perfectly diversified. The advisers said: "This is good, but I've got this fund over here that you really ought to look to move into..." That was astonishing to me. It's one thing to take someone from cash to a not‑so‑good fund. It's another to actually take someone from the perfect fund to this other thing.
Harold: Tell me about the scarcity trap, and why it's so insidious.
Sendhil: The scarcity trap captures this notion we see again and again in many domains. When people have very little, they undertake behaviors that maintain or reinforce their future disadvantage. If you have very little, you often behave in such a way so that you'll have little in the future.
In economics, people talk about the poverty trap. We're generalizing that, saying this happens a lot, and we've experienced it. For example, sometimes you get really busy. Then you're stuck, and you just can't seem to climb your way out. Dieters experience this, too. It's very hard
Another tragic example concerns lonely people. The lonely are interesting because it’s so tempting to say: "Oh, lonely people. Yeah, those are just losers, or whatever. Those are people who can't make friends." Actually, the data suggests that the vast majority of lonely people don't lack any social skills at all. It's just they found themselves in lonely situations.
You move to a new town and you don't really know anybody. How do you meet people? It's hard to meet people. The longer that persists, now the longer you've been lonely, and then ‑‑ this is the key part with the lonely and the busy and the money and the poor ‑‑ now that you're in that state, your behavior changes, and the way your behavior changes seems to keep you in that state.
There are, I think, a few ways in which your behavior changes. Scarcity draws a lot of attention to itself. That's the key finding that I think motivates everything. When you're experiencing scarcity, your mind automatically focuses on that thing. That focus brings benefits, which we talk about. But it has some costs, too, which help create the scarcity trap.
One cost, for the lonely: If you want to be interesting, the one thing you shouldn't do is really focus on the fact that “I want this person to like me.” That's going to make you very uninteresting. But the lonely, they just can't help but focus on that.
There's this beautiful study in which subjects speak into a microphone and they either think that someone else listening to them, or they think they're just talking. Among the non-lonely, there's very little difference in how third parties would rate subjects’ responses. A third party rates subjects as equally interesting in both conditions. Yet lonely people become less interesting when they think someone is listening. It's sort of a choking effect. That's one kind of scarcity trap.
Harold: That's a painful example for many academic nerds.
Sendhil: You and I wouldn't know anything about that, but yes…
In a different way, this focus is a bit like firefighting. You become so focused on solving this or that problem, that you do so at all costs. You become single‑minded. That inevitably leads you to borrow. "How am I going to solve this problem? Well, I've got to make this deadline, finish this project. You know what? I'll put off this other stuff." Yet by putting off this other stuff, what are you doing but putting tomorrow's self in a deeper time debt?
Poor people face similar traps. They often find some creative solutions that don't involve borrowing. Yet credit is always this very attractive thing; it looks good when you're tunneling.
One of our studies really convinced me that we were onto something. We took this scarcity trap that you find with borrowing among the poor, and we said: "Well, if this is true, we should be able to recreate that exact scarcity trap with borrowing something else." So we had these undergrads come in and play games, like "Family Feud," round after round. We made some of them poor ‑‑ they had very few seconds for each family feud ‑‑ and some of them rich.
When they could borrow, the rich didn't do any worse than when they couldn't borrow. Yet give the poor the ability to borrow, and they do exactly like the poor in the world. They borrow seconds. They find themselves in the scarcity trap, and they get stuck. Of course the word “poor” here is funny. These are all Princeton undergrads, randomly assigned to having fewer seconds to play a game.
It was astonishing how we could re-create all of the financial behaviors of the poor, but with blueberries, in this game we invented called "Angry Blueberries," or with guesses in "Wheel of Fortune." That's what really made us feel like maybe there was something more to scarcity and the scarcity trap, because it's not about poverty and money. That borrowing, I think, is the key feature of the scarcity.
Harold: Somehow there's this combination of tunneling leading to neglect of long‑term consequences. You give another poignant example of Indian street vendors. Every morning, they borrow 1,000 rupees to sell fruit. They earn maybe 1,100 rupees by the end of the day, and they have to pay back the loan. They pay an interest rate of 5 percent per day, half of their profit. Why is it so hard for them to get out from under that apparently insane situation?
Sendhil: Before we get to the vendor example, which seems extreme, let’s go back to what we talked about earlier, about empathy, and how we can all be stuck in that same inefficient time trap. I know that it's inefficient for me to prepare my class at the last minute. I know that. But yet somehow, I'm stuck just like the vendors. I'm doing it on borrowed time, every time, and I can never get out.
What's ironic about these scarcity traps is that, with money and time, they all have this funny feature. People think they're in the scarcity trap because they have too little money, or too little time. That's sort of true, but it’s not exactly true. It’s not exactly true because you're always accomplishing the same work. You're just accomplishing it later. If you could just somehow get one step ahead, that's it. It's not that you have too little time. You have the same amount of time; you're just one step behind.
The vendors face similar challenges. It's not that they have too little money. It's that they're in this debt cycle. If they did the same activity, but did it out of their own money, then they would be one step ahead, just like we could be one step ahead.
I think it's worth introspecting, why do we constantly get stuck? I think it's similar to the vendors' reason. Imagine someone came to you and said, "Harold, here's this gift of time," and now all of a sudden you're free. You have this miraculous feeling of being one step ahead now. But then, slowly, you'll feel yourself creeping back in and a week later, two weeks later, you're back to where you were.
For the vendors, it's exactly the same thing. We went and gave them the gift of money. We bought them out of their debt, and they're not myopic idiots. They didn't fall back the next day. But just like you, it took them like six months, and then they all fell back in.
Harold: If just giving money doesn’t fully help people over the long run, what should we do?
Sendhil: First, we should recognize that the structure of the choice environment itself is often the problem. What keeps you and me in our time debt trap? We're stuck in time debt traps because we have two weeks to do something. Classes start September 3rd, it's only August 15th. I have two weeks to prepare my lectures. Weirdly, you have too much abundance. So that task is not really crying out for your attention. It only cries out for your attention when you get too close. Intuitively good time managers know how to address this problem. They find ways to make these distant deadlines more immediate.
I think the right response to the money trap is similar. A good policy response finds ways to make the inadvertent negative consequences of debt -- which otherwise fall outside our tunneled thinking -- immediately felt back inside the tunnel.
Let me say it differently. Suppose you have a savings account and you experience a financial crunch. This money feels like it's just sitting there. Dipping into it can really help you solve it. The consequences happen outside your tunneled view. We could do a better job of protecting your savings by making you feel the pain of when you spend it. People talk about commitment accounts, but commitment accounts are almost too strong, because what they do is they really mean for you to never access your money. I think there are softer versions.
For example, imagine that you have a savings account ‑‑ and 401(k)s have a little bit of this feature -- in which you can put money in and you can't withdraw from it. Only you can borrow from it. The catch is that you can borrow from it at a high interest rate, paid back into your own account. It’s as if you say: "Yes, you really want the money now. Great. You can take it out as a loan which you have to repay. And when you repay you're just rebuilding the savings, plus some."
Harold: So my future self commits usury against my current self.
Sendhil: That's perfectly put. Effectively, what the money lender or the predatory lender is doing is to be the middle man between you and your future self, and to taking the rents. But why shouldn't your future self come immediately and say, "Look, you want to borrow? Fine. Borrow at high rates, but let's pay me. Don't pay that dude." That type of product is not so different from predatory lending. We know it's going to do well because predatory lending does well. Only the consequences are fundamentally different. Instead of all of that stuff going outside of your own system, it's sticking in your system, and a savings account is being built.
Harold: Similar behavioral insights exist in other interventions. For example, something called contingency management treatment is pretty effective for substance users. You say to a client: "I’ll pay you $10 if you’re able to produce a clean [monitored] urine sample on Thursday." That turns out to be impressively effective, especially in combination with other help. Steve Higgins and his colleagues have done terrific work helping pregnant women reduce their smoking with such approaches.
Sendhil: That example is terrifically right. Many intervention designers don't understand the logic of tunneling. So they provide incentives, but most of the incentive falls outside the tunnel. So you’ll see limited benefits. But give $10 right now, the entire incentive falls inside the tunnel.
I think such examples help us understand a broader puzzle in the literature: How is it possible that so many incentives fail ‑‑ big incentives -- yet some tiny incentives seem to work insanely well? I think the answer is often: Did you get something immediately in the tunnel, or did you not?
Harold: The same reasoning underlies Mark Kleiman’s work regarding the effectiveness of swift and certain (but relatively mild) sanctions in the criminal justice system.
Sendhil: Kleiman’s work is very interesting. It's terrific, but the tunnel is so narrow. Think about swift and certain; you might be able to do even better if you can reach people within that 10‑second tunnel. That's the moment where there is a heated argument. In that sense, punishment and other long-run consequences are so outside the tunnel. It might even be more of an opportunity to think about how we, in those 10 seconds, might intervene when the dude is really in a hot state. How do you inject incentives right into that moment?
Harold: Wow, that is hard to do.
I have one final thing. Welfare reform imposed a five-year cumulative lifetime limit on receipt of federally funded cash aid. I’m uncomfortable with the current lifetime limits for broader policy reasons. Even if you basically accept the incentive argument in welfare reform, that lifetime limit seems specifically non‑designed to deal with the issues that you're talking about.
How would you rejigger these policies to be more effective? To raise a similar example, how should we design unemployment insurance? It also has a pretty sharp long-term deadline that people often hit
Sendhil: This concept of scarcity allows me to go back and forth fruitfully between examples. When I think of the welfare example, I often think of our PhD students. Imagine how bad our PhD student theses would be if we said, "It's year two. Course work is done. Come back in three years with a thesis, and if you want you could always come and ask me questions, but otherwise that's it. You're off." Of course it would be awful. That’s just too distant a deadline.
Harold: You realize you're describing the experience of 30 percent of all graduate students in the United States.
Sendhil: [laughs] Right, exactly. Even in the worst places, though, think of the institutions you set up, such as a lunch where you must make a presentation. Something as simple as that helps to partition the three-year task into a three-month task. At least a student needs to make progress for this lunch.
If we wanted to implement the five‑year lifetime limit, effective policy would start by making this limit more salient, bringing into the tunnel how much of that lifetime limit is left. For example, you could say, "I'm going to send a letter every month that very saliently says, 'You will run out of benefits by this time, at the rate you're going,' or 'At your rate, by the age of 28 you won't have benefits.' " We can play with those, but literally it's almost like what I would like to do is to create this bright, flashing light about the deadline that's in the distance, that comes in.
With graduate students, the same thing is true. If you went to them and said, "You know, you don't have that much time left to finish a thesis," and you can almost even break it into parts. "Students," you would say, "to get to the thesis, you first have to get a paper, and you need to have a paper by year one, and to get a paper you need a project..."
You could do the same thing with the unemployed. You could say, "Look, to get a job you need to have a good resume, and have that prepared. Are you at the resume stage? Now have you done...?" Take the big thing and break it into little parts. I'm not saying this is the entire solution, but it's a low‑hanging fruit because we don't need legislative change.
Harold: One of the tragedies of unemployment insurance is there's this incredible stigma for being out of the workforce for any length of time. Yet the way the system is structured, you're in the tunnel in the earliest period before the stigma is so apparent to you. For your first couple of months that you are jobless, you may be thinking: "I want a job that's like my old job." Unfortunately, in an economy like ours right now, there isn't a job like your old job. Then four months later, prospective employers start saying, "Hey, he's been out of the workforce awhile. What's wrong with him?" All of a sudden, the employer who might have interviewed you three months ago is no longer as interested.
Sendhil: Larry Katz, Linda Babcock, Bill Congdon and I have a paper on behavioral labor policies. We propose one trick that might overcome all this. Suppose we had a UI system that says to people: "Look, whatever job you find, we're going to provide a benefit that ensures you will get wages to match your last job. " We'll do that for three months, five months, six months. Now, to somebody who's worried about wage loss, it's easier for them to say, "Great. I'll take this job and that will at least give me full replacement, much higher than what I'm getting unemployed." That’s a real carrot within the tunnel.
And people may tell themselves: "I'll start looking for another job while I'm in that job." If they find a better job, great! If they don't, well at least we've now moved them into a job rather than continued unemployment.
Harold: I should note that everything we’re talking about would work so much better with better macro policies and a low-unemployment-rate economy.