Yield Trends Are Insufficient to Double Global Crop Production by 2050

Picture the dinner table in two thousand fifty. There are more people, more income, and more appetite for meat and dairy. Biofuels are still in the mix. The big question is brutally clear: Can we feed that world by lifting yields on the land we already farm, or will we be pushed to plow up more? Many estimates say total agricultural production will have to climb on the order of sixty to one hundred ten percent by midcentury. Historically, yield growth has done the heavy lifting for food security. But the ground truth today is mixed. In roughly a quarter to more than a third of key cropland, yields have stopped improving. That stall shows up in some of the very places where populations are growing fastest or diets are changing the most. So the stakes are not abstract. They’re on the plate. Into that tension, Ray, Mueller, West, and Foley took a refreshingly concrete approach. They didn’t start with a grand model. They started with data—about two point five million statistics, to be precise—spanning roughly thirteen thousand five hundred subnational political units across the planet, from nineteen sixty-one to two thousand eight, with a close look at the most recent two decades. Then they zoomed in on the crops that really move the global calorie needle: maize, rice, wheat, and soybean. These four alone provide about two-thirds of the calories harvested from fields worldwide and, as another way of seeing their weight, around forty-three percent of dietary energy and roughly forty percent of daily protein. If we care about the food system’s trajectory, we care about what happens to these four. The method was intentionally simple so that each assumption was visible. For every local unit, they fit a straight line through twenty years of yield data and read off the slope as an annual change expressed in percent. That’s a non-compounding rate: think of it as adding the same amount each year rather than multiplying by last year’s gain. It’s like saying, "We’re tacking on one more bushel per acre each season," not "We’re growing five percent faster than last year’s five percent." Why do that? Because it matches the way a lot of farming gains actually accumulate—stepwise, with improved seeds or better management—without assuming exponential growth. To isolate yield from land expansion, they then held harvested area fixed at the late two thousands baseline and asked: If those linear trends continued, what happens to production in twenty twenty-five and in twenty fifty? Here’s the heartbeat of the paper. Globally, yield trends for those four staples are too slow to double production by twenty fifty with fixed area. On average, maize is climbing at about one point six percent per year, rice at one point zero, wheat at zero point nine, and soybean at one point three. The benchmark for doubling—again, in this non-compounding setup—is about two point four percent per year. If the world kept cruising at today’s speeds, by twenty fifty we’d see maize production up roughly sixty-seven percent, rice up forty-two, wheat up thirty-eight, and soybean up fifty-five. Those are real gains. They’re just not enough to meet a doubling target without something else giving, like expanded land or slower demand. To make those percentages feel more tangible, the team converted the lines into tonnage. Under current trends, global maize would land near one point three four billion tons by midcentury. But compared with where it would need to be to hit a doubling path that year, maize would fall short by roughly two hundred forty-seven million tons. Rice and wheat face even larger gaps: approximately three hundred ninety-four million tons for rice and three hundred eighty-eight for wheat. And you don’t have to wait for twenty fifty to see the wedge open. By twenty twenty-five, shortfalls are already evident across the board, with maize alone trailing a doubling trajectory by on the order of one hundred million tons in that year. The message isn’t doom. It’s accounting. It says, "This is the size of the hill we’re climbing." Of course, nothing about the future is certain, and the authors treated that uncertainty with the same transparency. They used bootstrapping—resampling the twenty-year yield series in each place—to generate ninety percent confidence ranges for the growth rates and the production paths they imply. In an optimistic bound, maize edges up to two point four percent per year, wheat to one point eight, soybean to two point zero, and rice to one point four. Translate that forward and you get maize effectively doubling by midcentury, wheat up about seventy-six percent, soybean around eighty-four, and rice closer to fifty-nine. In a pessimistic bound, maize slows to zero point eight percent, rice to zero point five, wheat to zero point one, and soybean to zero point three. Under that world, wheat barely moves—around a four percent increase by twenty fifty—while maize, rice, and soybean rise by only a few decades’ worth of modest increments. It’s a wide spread, especially for wheat. But across the spread, one thing doesn’t change: You don’t get rice and wheat doubling under plausible yield trajectories without help from somewhere else. A quick aside on why to trust the line they drew through the data. The team checked the diagnostics you’d want: They compared linear to quadratic fits, looked for autocorrelation in the residuals with Durbin-Watson, and tested distributional assumptions with Lilliefors. Not every political unit had a perfect record, so missing years were filled conservatively using recent averages and constrained by higher-level totals. The point wasn’t to win a beauty contest in modeling; it was to build a clear yardstick. When you count growth this way, what do you see? And if you change the yardstick slightly, do you see the same thing? They also asked a second, very human question: Per person, do these gains change the basket of cereals and oilseeds people encounter in markets? Using the United Nations' medium-variant population projection as the demographic backdrop, the answer is sobering. Faster maize and soybean yield growth might not do much for per-capita harvests in the near term, with a chance of an uptick later on. For rice and wheat, where yield increases hover around one point zero and zero point nine percent per year, per-capita harvests are likely to be flat through twenty fifty if area stays fixed. That doesn’t just matter for global tallies. It matters for households whose food budgets hinge on the price of a bag of rice. Now, global averages are like the view from thirty-five thousand feet. Useful, but they can mislead. Ray and colleagues brought the camera down to the county and province level and mapped where yields are racing ahead and where they’re stalling. In North and Central America, the story is a patchwork. In the United States, you can draw a diagonal of maize optimism: much of North Dakota and Mississippi, slices of northeastern South Dakota and northwestern Minnesota—all showing yield gains at or above that two point four percent doubling pace. Nationally in this region, the United States leads in maize yield improvement, with Canada following, then Cuba and Mexico. Cross the invisible line into parts of Central America, and the texture changes. Honduras and Nicaragua, where maize supplies about twenty-seven and twenty-five percent of daily dietary energy, are clocking around zero point five percent annual gains—not enough to keep up with population growth in per-capita terms. Panama, which gets about seven percent of dietary energy from maize, is in a similar bind. And Guatemala is flashing a red light: Maize yields there have been trending downward at roughly minus zero point seven percent per year. Rice shows a similar chessboard of realities. In the United States, rice yields have been improving at around one point two percent per year, and in Mexico at about one point one. But in Nicaragua and Panama—countries where rice contributes roughly sixteen and twenty-four percent of dietary energy—those gains, about zero point nine and zero point two percent per year respectively, may be outpaced by population, pointing toward declining per-capita rice harvests. Elsewhere in the region, rice yields are actually slipping. The Dominican Republic, Costa Rica, and Haiti are seeing declines between about zero point one and zero point six percent per year. Cuba, interestingly, is a counterexample: If population falls and yields rise modestly—around zero point nine percent per year—per-capita rice availability could expand even without a surge in raw yield growth. Wheat and soybean add their own wrinkles. Doubling-rate wheat gains appear in only a scattering of U.S. counties—think eastern South Dakota, parts of Nebraska, pockets of northeastern Kansas, and down into western Mississippi and Louisiana. Nationally, Canada's wheat yields are rising about one point three percent per year, the United States around zero point eight, and Mexico near one point one. For soybean, much of the U.S. is on an upward track, with some North Dakota counties and scattered sites touching doubling-rate gains. But zoom out and the national picture is steadier: Roughly one point two percent annual gains in the United States, about zero point two in Canada. Taken together, these maps are not just cartography. They’re a to-do list. They tell you where a push on agronomy and access to inputs could pay off and where a plateau is turning into a problem. There’s a crucial design choice running through all of this: holding harvested area fixed. It’s a way of saying, "Let’s see what yields alone can buy." And the implication is sharp. In countries or regions where yields are stagnating—remember that global estimate of twenty-four to thirty-nine percent of key cropland with no recent improvement—demand growth will translate into pressure to expand cropland unless yields accelerate. That pressure isn’t costless. New fields often come from forests or grasslands, with big consequences for biodiversity and carbon. So where does that leave us? With realism and a route forward. The realism is that, as Ray and colleagues showed, current yield trajectories—one point six percent per year for maize, one point zero for rice, zero point nine for wheat, one point three for soybean—do not add up to a doubling by twenty fifty on existing land. Even under rosier scenarios, rice and wheat in particular lag behind the path that would close the gap on their own. The route forward is two-pronged. On the supply side, sustainable intensification—closing the gaps between what farmers harvest now and what agronomy says is attainable—matters. That’s not a slogan; it’s a set of practices: improved varieties, better timing and targeting of fertilizer and water, integrated pest management, and attention to soil health. And the maps in this work are a guidebook for where those investments could yield the most grain per dollar and per kilogram of carbon. On the demand side, there’s flexibility we sometimes ignore. Diets that lean a bit more toward plants can lower the grain demand embedded in animal feed. Cutting food loss and waste—along the supply chain and in kitchens—also shrinks the production needed to deliver the same nutrition. None of those choices erase the need for higher yields. But together, they reduce how steep that climb has to be, and they ease the land-use and emissions trade-offs we’ll otherwise face. It’s tempting, at this point, to daydream about silver bullets: miracle seeds, climate-proof everything, a global turn to hydroponics. The value of this paper is that it doesn’t ask you to believe in miracles. It gives you a clear baseline, grounded in millions of observations, and says: Here’s where we are, here’s how far we have to go, and here’s where effort could make the biggest difference. For a world that eats every day, that kind of grounded map is a gift.