by Dean Karlan
The point is that what works in agriculture is highly context-specific, and so advocating a single technical prescription—about what to grow or how to grow it—is unlikely to serve everyone well.
But that doesn’t keep people from trying. If the DrumNet farmers of Gichugu Constituency had met with any extension officers from Kenya’s Ministry of Agriculture, they might have learned this firsthand. The Ministry had a plan for every farmer in the country. Depending on location, it recommended one of twenty-four specific regimens of fertilizer usage and seed varieties.
In the maize fields of Busia District, four hundred kilometers west of Gichugu along the Ugandan border, the official regimen called for hybrid seeds and two kinds of fertilizer—one to be applied at the time of planting and another as top-dressing when the plants were knee-high. The Ministry was confident in its recommendation, which was based on evidence from testing farms; unfortunately, real farmers’ fields didn’t always look like the Ministry’s test plots. They varied in soil chemistry, water, and sunlight exposure. Those variations were substantial enough that farmers could not always reproduce the high crop yields from the Ministry’s experiments.
That might explain why so few of them followed the government guidelines. In a 2000 survey of Busia’s maize farmers, Esther Duflo, Michael Kremer, and Jonathan Robinson found that fewer than one in four had used any fertilizer in the previous year; fewer still had used hybrid seeds. Why were so many ignoring the Ministry’s advice? Had they never heard of the guidelines? Were they just thickheaded? The researchers gave the farmers the benefit of the doubt. Maybe they knew something the Ministry didn’t. Maybe, for the farmers of Busia, following the government’s advice would have been a mistake.
Duflo, Kremer, and Robinson set out to design a simple RCT to find out. They chose hundreds of maize farmers at random and worked with them to set up three small adjacent plots, each about seventeen feet square, in each farmer’s field. Over the next six growing seasons they tested different amounts and combinations of seeds and fertilizers—including the Ministry’s recommended recipe—in two of the three plots, always leaving one for comparison. At the end of each season they measured the output for each plot. When all the data was in, they had a pretty clear picture of the relationship between seeds, fertilizer, and crop yields.
As the Ministry had claimed—and as the farmers knew—higher-quality seeds and greater quantities of fertilizer led to greater yields. Then the researchers went a step further, calculating the net profit for each combination by subtracting the cost of the inputs from the sale price of the final product. That’s when farmers’ choices about fertilizer started to make a lot more sense. The Ministry’s recommended package did produce the highest yield of all, but the inputs were so expensive that the fertilizer investment generated a net loss—about negative 50 percent annually of the amount spent on fertilizer—for farmers. Now, fertilizer isn’t the largest expenditure for the farmers, so losing 50 percent of a relatively small number is not necessarily devastating. But it sure doesn’t help.
So the Ministry of Agriculture wasn’t getting it right, but then neither were the farmers. Most were using either too much—in line with the government’s recommendations—or not enough.
The researchers’ test showed that there was a profitable middle way. It found that using just half a teaspoon of top-dressing fertilizer on each plant (the Ministry called for a full teaspoon per plant, plus fertilizer at planting, plus hybrid seeds) increased crop yields by almost half over comparison plots and, most important, made good business sense. A farmer who followed the half-teaspoon regimen would have earned an annual return of 52 to 85 percent on his fertilizer investment. Though it is again important to note that fertilizer represents a small portion of the overall cost of running a farm, this is a very high annual return on an investment. (For reference, it’s slightly better than the best year in the U.S. stock market in the past eight decades.)
Farmers Are People Too
Many farmers, as though oblivious to these potential profits, continued to till their fields as they had for generations—with little or no fertilizer. The researchers figured that so many people were not simply turning down more money without a good explanation. Why hadn’t fertilizer caught on? They probed the standard economic models, testing explanations about risk aversion and variable returns, but the data just didn’t fit. In the final paragraph of their paper on the profitability experiment they concluded, “There may be a role for non-fully-rational behavior in explaining production decisions (e.g., how much fertilizer to use).”
Non-fully-rational, perhaps, but still thoroughly human. The researchers were just saying that the Kenyan farmers, like most of us, do not behave like Econs. Their thought processes, like our own, are susceptible to all kinds of shortcuts, biases, and filters.
The specific lessons of behavioral economics—the things we can leverage to help farmers—are truths about the way these quirks function. Though you are probably not a maize farmer from Busia, hopefully you will agree that the quirks we will discuss are common to rich and poor alike. What makes them bite so hard in Kenya is the simple fact that the poor have much less wiggle room to make mistakes.
Flood of Information: Inertia and the Status Quo
Sit yourself down on a low wooden stool in a mud house and let your eyes adjust. Notice how dim it is, though the light outside from the midday sun is so intense that the cutout window appears as a blinding white square. Feel the still, still heat. Now consider: Planting time is fast approaching. What should you grow this season? Maize, sorghum, finger millet, soybeans, or cassava? How much of each, and where should they go in your field? How much fertilizer should you buy, and what kind, and from which store?
The sheer number of choices can be debilitating. They all call out to you at once, each pleading its own case, so that no single voice can be heard above the din. Meanwhile, information comes in from all sides. You can see what your neighbors are growing, and you can ask how they chose what they did. You know what you’ve grown the past few seasons and you know how it turned out. Maybe you’ve had a visit from an agricultural extension officer too.
When you’re faced with a flood of information, what floats to the top?
Sometimes the abundance of options and directions leads us, ironically, to choose nothing. As we saw in chapter 3, that’s what American grocery store shoppers did when they were confronted with too many exotic flavors of jam. They passed on preserves altogether.
American shoppers can simply go home without jam, but Kenyan farmers pretty much have to decide on something to grow, even if the options are overwhelming. For them, not choosing usually amounts to not changing, to doing the same old thing.
That begins to explain one of the most widely observed phenomena in behavioral economics: inertia, or the inexorable pull of the status quo. Time after time we see people pass up new opportunities in favor of the familiar. It’s one reason why our farmer, poring over planting options in his dim mud hut, is likely to grow exactly what he grew last season—or why, for that matter, he is likely to be farming in the same way his parents and grandparents did.
Our preference for the usual is pervasive and instinctual. It comes from somewhere outside our rational minds. A California electric company surveyed its customers to decide what kind of service to provide. As it stood, customers who lived in areas with good infrastructure had virtually no power outages, and those who lived in bad-infrastructure areas occasionally lost power, but paid about 30 percent less for electricity. The company was considering upgrading the infrastructure in bad areas, and wanted to know whether customers would be willing to pay more for improved service.
The survey they sent out asked customers to list, in order of preference, six different combinations of price and service quality (the actual price/service combinations experienced in both good and bad areas were included in the six). Tallying the results, they found that, while there was no consensus about which combination was best
, most people had a clear preference for the status quo. About 60 percent of both good- and bad-infrastructure customers ranked their own price/service combination highest.
As we saw in the last chapter, not even economists, whom we have to thank for cataloging the phenomenon in the first place, are immune. When Shlomo Benartzi and Richard Thaler, the duo behind the Save More Tomorrow retirement plan, tracked the activity of hundreds of professors’ retirement accounts, they found that the status quo exerted a strong gravitational pull on them as well. Now, if anyone could be expected to adjust his investment portfolio over time to meet changing needs, it’s an economist. But the professors tended to choose an initial allocation and stick with it, sophistication and fine-tuning be damned. In fact, the average professor in their sample made exactly zero changes to his retirement portfolio over his entire lifetime! How’s that for inertia?
What Stands Out: Recency and Availability
With our behavioral tendencies pushing us to do the same old thing, the odds seem heavily stacked against any kind of change. But each of us, at least on occasion, cuts his inertial ties and ventures out, unfettered, into new territory. Even then, we carry behavioral tics with us. Think back to our farmer in his dim mud hut. Suppose he resolves to do something different, something better, this planting season. How will he decide what changes to make?
If he’s like most of us, he will not fire up an adding machine or run out in search of the latest actuarial tables on crop yields. He’ll look over the fence at his neighbor’s field, or think back to his cousin’s experience trying to grow sorghum last year. Despite the insistence of classical economic models that choices should be made systematically, by dispassionately weighing each alternative in turn, we think in anecdotes. We don’t always see the big picture, laid across the vast axes of space, time, and experience—we see specific examples. Local, recent, and extraordinary events stand out in our minds and weigh far more heavily in our decisions than they should.
A classic example of this phenomenon is the spike in sales of earthquake insurance in the days following a major earthquake. In reality, the chance that an earthquake will destroy your home doesn’t increase after any one incident, but it’s easy to see why tremors send people clamoring for their insurance agents. Fresh images on the six o’clock news—or, even more powerfully, seen firsthand—of collapsed overpasses and mangled buildings flood in and quickly obscure the tiny statistical probability of a quake. Suddenly that policy looks like a pretty good deal.
Daniel Kahneman and Amos Tversky, two pioneers of behavioral economics, used a clever and elegant lab experiment to demonstrate how powerfully specific, salient examples can distort our sense of the overall likelihood of an event happening. Subjects were randomly divided into two groups and had to answer a single question. Those in Group A were asked,
“In four pages of a novel (about two thousand words), about how many words would you expect to find that have seven letters and end with -ing?”
Those in Group B were asked,
“In four pages of a novel (about two thousand words), about how many words would you expect to find that have seven letters and n in the sixth place?”
The average of Group A’s answers was 13.4; Group B’s was just 4.7. It’s strange that subjects in Group B guessed so much lower, because simple logic proves that the number they were looking for actually must be the larger of the two: A list of all the seven-letter words with n in the sixth place would include at least all the seven-letter words ending in -ing and many more. Again, we get it wrong because we think in terms of examples. It is easy to build seven-letter -ing words by conjuring up four-letter verbs and tacking on a suffix, but we don’t immediately think of that strategy when we hear “seven letters with n in the sixth place.” Instead we look for words that just happen to have n next-to-last, like harmony and lasagna (that’s what I came up with, anyway). Since these are harder to come by, we underestimate.
Using Our Behavioral Tics for Good
Behavioral economics gets interesting when we move beyond explaining our choices and onto improving them. That’s exactly the motivation behind innovations like SMarT, stickK.com, and SEED, all of which we saw in the last chapter. We’re far from perfect, but if we can identify the places we often slip up, sometimes we can make tools that help us stay a step ahead of ourselves.
Back in Busia, Esther Duflo, Michael Kremer, and Jonathan Robinson were puzzling over the Kenyan farmers, who appeared to be slipping up left and right. The researchers’ experimental test plots had left no doubt: Farmers could have made more money by putting more fertilizer in their fields. Standard economic approaches hadn’t even identified what was wrong, much less found a solution. The farmers knew about fertilizer and knew where it was sold—so it wasn’t an education or information problem. Fertilizer could be bought in small or large quantity, and at any time of year—so it wasn’t a storage problem. Finally, farmers themselves often talked about wanting to use more fertilizer in the future—so it wasn’t a problem of desire or preferences.
But it was, undeniably, a problem. The facts were plain as day. On the whole, farmers just weren’t using enough.
Duflo, Kremer, and Robinson figured that, if farmers already knew about fertilizer and wanted to use it, maybe they just needed a behavioral nudge in the right direction. So they partnered with ICS Africa, an international nonprofit operating in the area, and developed the Savings and Fertilizer Initiative coupon program (some time later, after evaluating the program, IPA took over operations from ICS Africa on this and other projects).
Representatives of the Savings and Fertilizer Initiative visited farmers at their homes immediately after the harvest and gave them a chance to buy a fertilizer coupon. This way they could pay up front for fertilizer that would be delivered (for free) in time for the following season. At harvesttime farmers were flush with money from selling their crop, and they had agricultural productivity on the brain. If there was ever an occasion when they’d be willing to spend on fertilizer, the researchers figured, this was it.
They were right. Fertilizer use surged by over 50 percent for farmers who had the chance to buy the coupons. Farmers, finally making good on their long-standing wishes to buy fertilizer, grew more crops; and vendors sold over 50 percent more fertilizer without lowering their prices by so much as a penny.
This coupon program is a prime example of the kind of everybody-wins solutions we get from behavioral economics. It’s subtle, cheap, and incredibly effective.
There are two catches. First, it is not a one-shot deal. The problems it addresses—farmers’ tendencies toward procrastination and shortsightedness—crop up season after season. The fix must be equally persistent. That’s the thing about our behavioral tics: They’re often easy to treat, but can also be impossible to eradicate. So when the Savings and Fertilizer Initiative pulled out after a one-season initial test period, all the fertilizer gains disappeared. The farmers went back to square one.
Second, it can only do so much good on its own. The Savings and Fertilizers Initiative is a prime example of the way small, carefully designed products that take into account our behavioral tics can make big differences in behavior. But translating those changes in behavior to significant and lasting increases in income or living standards sometimes requires a bit more brute force. The problems farmers face are endemic and intertwined. It doesn’t do much good to grow more of a crop if you don’t have a good place to sell it, roads to get to the market, prices you can rely on, or brokers you can trust. These are precisely the gaps that the DrumNet program we saw earlier in the chapter tried to fill with its multifaceted approach. How about putting programs like DrumNet together with the Savings and Fertilizer Initiative? Now, there’s an idea with growing potential.
Viral Pineapples and Social Learning
Maybe the way to make more enduring improvements to our decisions is to piggyback behavioral solutions on things we already do. For farmers, one of the essential first steps towar
d choosing what, when, and how to grow is walking outside and looking over the fence at the neighbors’ fields. They get information and inspiration from the people around them, and then they act, inspiring others in turn. It’s a natural feedback loop; it’s trendsetting. Think of viral videos on YouTube. Or even viral . . . pineapples?
In the hands of Chris Udry, a Yale colleague and mentor to me, viral pineapples are a conduit to learning about how people learn. In 1996, Udry and Timothy Conley, a former colleague of Udry’s at Northwestern and guru of spatial econometrics, set out to study how farmers learn to adopt new tools and techniques. They set up in the Akwapim South District of Ghana, an hour north of the capital city of Accra.
It was a good place to work because change was afoot in the gently rolling hills of Akwapim South. For generations, farmers there had raised maize and cassava, rotating their fields between the two crops season by season. But in 1990, the first timid spikes of smooth, succulent leaves could be seen peeking out of the soil. A small portion of farmers, less than one in ten, had begun growing pineapples for export to Europe. As those first intrepid growers enjoyed success, others caught on. By the time Conley and Udry surveyed in 1996, nearly half the farmers in Akwapim South were in the pineapple business.
Those who made the switch had a lot to learn. Compared with maize and cassava, pineapple is a labor- and input-intensive crop. It needs to be planted carefully and calls for more fertilizer. Unfortunately, pineapples don’t come with an instruction manual. The specifics of spacing within a plot, and the timing and quantity of fertilizer applications have to be found out by hand. But not necessarily firsthand—one pineapple grower could spare himself a great deal of trial and error by talking with others, learning from their successes and failures.