Sendhil Mullainathan and Eldar Shafir’s Scarcity: Why Having Too Little Means So Muchis full of interesting insight and experimental results. It presents a novel way of looking at scarcity that extends beyond the typical analysis in economics, the original “science of scarcity”, and will certainly change the way I think about it.
But by the time I reached the end of the book, I was not entirely satisfied. I have new buzzwords and some interesting experiments to think about, but I’m not convinced Mullainathan and Shafir have presented a coherent new perspective on how the world works. Scarcity reminded me of a Malcolm Gladwell book – I got presented with a lot of cool results, they were spun into an interesting narrative that has made me think, but I don’t buy the authors’ main message.
First, the buzzwords, which added to the Gladwell-esque feeling. Scarcity is having less time, money, friends or packing space than you feel you need. (Note the subjective element - scarcity could be caused by relative rather than absolute scarcity_ And it can apply across a range of domains.) Scarcity can have good effects, such as the *focus dividend* when scarcity captures the mind. It can make us experts in the area in which we are scarce, with poor people better able to judge the value of a dollar saved despite the context in which it appears. However, scarcity also causes us to *tunnel*, which is to focus single-mindedly on the scarcity at hand, potentially at the neglect of more important or less timely demands. Scarcity also reduces *bandwidth*, a combination of cognitive function and executive control, by imposing a *bandwidth tax*. When people suffer from scarcity and tunnel, they are also forced to *juggle*, as they move from one pressing task to the next. The way to avoid scarcity is to have some *slack*, an untapped budget we can turn to in times of need.
The most salient example of the negative effects of scarcity comes from a study in a New Jersey mall. Mullainathan, Shafir and Jiaying Zhao presented people with a scenario where they need to fund some car repairs. They then gave the participants tests for fluid intelligence and cognitive control. When the cost of the repairs was $150, high and low-income people scored similarly in the tests. But when the shortfall was $1,500, the performance of the poor plunged – the equivalent of losing 13 or so IQ points. The authors point out that this result has been replicated many times, suggesting it is a robust result. [This experiment is on my list of experiments most likely to fail a pre-registered replication.] By causing the poor to focus on their lack of resources through a high and potentially unmanageable repair bill, the poor’s bandwidth for completing the tests was taxed. In another experiment, people were effectively made rich or poor by the flip of a coin, with the poor demonstrating greater present bias. In that case, the bandwidth tax was clearly not due to the inherent traits of the poor.
Mullainathan and Shafir take these results to mean that we don’t need talent or inherent trait-based explanations for differences between the rich and poor. Instead, scarcity makes the poor perform poorly, leaving them in a scarcity trap. They suggest that the effect of scarcity on bandwidth is a good explanation for differences as it can make sense of diverse empirical facts across behaviour, time and place. But it leaves open the question why they imply that scarcity induced and inherent trait-based explanations are mutually exclusive, or why inherently low bandwidth can’t explain an equally diverse set of facts.
Part of the difference between their and my interpretation is that they are observing a short-term dynamic and extrapolating that to the longer term. My view is that over the longer-term, inherent explanations play a larger role.
For example, when we look at twin studies, we find that IQ and other traits inherently differ between people. Adoption studies show long-term outcomes are typically more representative of biological than adoptive parents. Differences in scarcity of financial or other resources due to differences in adoptive parents have almost no effect on IQ, income, obesity or a range of other long-term outcomes.
Greg Clark’s work on social mobility shows the long-term persistence of status across many generations, despite short-term shocks that would affect scarcity. If Mullainathan and Shafir’s thesis were true, a shock one generation would carry through future generations, rather than seeing the next generation reverting to the underlying status. Under Mullainathan and Shafir’s explanation, we would also see immigrants’ IQ increase when they move country and ease their scarcity (although if scarcity is relative, perhaps their relative position may not have improved). We would see large increases in the IQ of previously poor lottery winners, who would experience a sudden surge in mental resources. In The Son Also Rises, Clark pulls together a few examples of windfalls of this nature and demonstrates the lack of long-term effect. In one case, a lottery of land parcels had no effect on the outcomes of the winners' children.
As a result, I am not convinced that their arguments truly capture the differences between the poor and the rich, or the rushed and the relaxed. Perhaps someone might eat more from time to time due to the bandwidth tax. They might take an ill-advised payday loan when they are stretched for money. But despite this short-term effect, we see little trace of it in long-term outcomes. Something is allowing some people but not others to break the cycle of scarcity.
I also doubt that Mullainathan and Shafir’s description of the poor as suffering from scarcity is generally true. When it comes to time, the poor watch more television, invest less time in caring for their children, have plenty of free time to think about what they will eat, and yet are more likely to be obese. Their characterisation of the poor having a lot on their mind whereas the rich are relaxed despite their more complex employment does not seem particularly strong.
The book is least satisfying when Mullainathan and Shafir start drawing their examples of scarcity from other domains – the padding to turn an interesting idea into a book length argument. Many of them are banal and simply involve scarcity in the way economists might think of it, with no evidence that people were suffering anything more than a lack of time. For example, they talk of the Benihana restaurant, where the chef cooking in front of people sets a quick pace for the meal, allowing more customers to come through. It is a story of scarce time, but provides no interesting insight into their thesis. The fact the Benihana restaurant management took the time to solve their problem suggests they were not overly taxed. Mullainathan and Shafir discuss the crash of NASA’s Mars Orbiter, which had to be launched by a certain date, leading to shortcuts. But what is the evidence that there was tunnelling or taxed bandwidth as opposed to a simple lack of time to do all the checks they would have liked to have done? They don’t present the evidence to distinguish.
When we turn to solutions, many of them do not depend on Mullainathan and Shafir’s argument being true. They are solutions that would be equally useful in dealing with inherently untalented or impatient people. Savings reminders could be useful regardless of the cause of the lack of foresight. Presenting pay day loans in terms of dollars rather than interest rates is standard behavioural economics fare and may work for people lacking bandwidth regardless the cause.
The short-term fluctuation of bandwidth does present some interesting possibilities. They suggest that if education fails due to low bandwidth, we should time education when people can best learn. They also give an example of Kenyan farmers failing to use fertiliser and missing out on large gains to their yields. By getting farmers to pre-purchase the fertiliser when they were flush with cash after harvest, more benefited from the yield gains. (Although again, are their decisions because they are inherently shortsighted or taxed?)
Ultimately, the long-term solution to the costs of scarcity is creating bandwidth and a buffer stock of slack. This sounds sensible, but this point drew me back to an example early in their book. They describe an experiment involving Indian vendors who were provided with money to allow them to escape loans with exorbitant interest rates that consumed much of their income. But despite having their debts cleared, bit by bit they fell back into the scarcity trap. Mullainathan and Shafir accredit their return to the trap to shocks. But why did they not save what they would otherwise have been paying in interest to create a buffer from shocks? Why did the new bandwidth not allow at least some of them to escape? Their income had effectively doubled.
One interesting idea Mullainathan and Shafir leave lying around is whether the bandwidth of whole economies can fluctuate through good times and bad. As a random idea of my own, is the Flynn effect due to the easing of scarcity in the modern world? [No]
Having said the above, the ideas in the book are worth considering, especially for contemplating how short-term mental capacity might be affected by the environment. But extrapolating the results to the long-term despite evidence from twin studies, adoption studies and social mobility analysis needs more. You cannot simply throw the effect of inherent talent and traits out of the window.
*As a postscript, after writing most of this review I searched for other reviews of the book and came across this piece by Tim Harford. Many of the same points – Harford notes the buzzwords and provides a Gladwell reference too. I should note that, as it seems for Harford, the Gladwell comparison is at least part praise.